Content uploaded by Thomas Cuff
Author content
All content in this area was uploaded by Thomas Cuff on Mar 08, 2016
Content may be subject to copyright.
Content uploaded by Thomas Cuff
Author content
All content in this area was uploaded by Thomas Cuff on Feb 13, 2016
Content may be subject to copyright.
Copyright 1993 Rev. 1.0 1 of 71
THE STM (Scanning Tunneling Microscope)
[The forgotten contribution of Robert Francis Earhart to the discovery of
quantum tunneling.]
by
Thomas Mark Cuff, Engineering Department, Temple University
“Nothing, of course, begins at the time you think it did.”
- Lillian Hellman
“By the way, what I have just outlined is what I call a “physicist’s history of physics,”
which is never correct. What I am telling you is a sort of conventionalized myth-story
that the physicists tell to their students, and those students tell to their students, and is not
necessarily related to the actual historical development, which I do not really know!”
- book, QED… by Richard P. Feynman
§1. INTRODUCTION. ______________________________ p. 001
§2. PRELUDE: ELECTRICAL CONDUCTION IN GASES. _______ p. 004
§3. VERIFICATION OF PASCHEN’S LAW AT SMALL DISTANCES AND THE
RESULTING UNEXPECTED PHENOMENA (THE AMERICAN EFFORT).p. 011
§4. WHAT EARHART et al. ACTUALLY SAW: ‘VACUUM SPARKS’. p. 023
§5. FIELD EMISSION UNDERNEATH THE ‘VACUUM SPARKS’ (THE
EUROPEAN CONNECTION). __________________________ p. 033
§6. CURRENT MEASUREMENT CAPABILITIES CIRCA 1900. __ p. 037
§7. SCANNING, THE MISSING ELEMENT? _______________ p. 041
§8. FIELD EMISSION & ELECTRON TUNNELING THEORY, THE J. J.
THOMSON CONNECTION. ____________________________ p. 043
§9. TUNNELING IN SOLIDS: COHERERS AND SUPERCONDUCTIVITY. ___
________
______________________________________ p. 046
§10. THE ORIGINS OF DC SUPERCURRENTS: TUNNELING vs. METALLIC
BRIDGES. ______________________________________ p. 054
§11. FIELD EMISSION X-RAY TUBES. __________________ p. 060
§12. THE UNREMEMBERED. _________________________ p. 063
§13. MODERN TIMES, TUNNELING HITS ITS STRIDE. _______ p. 068
§1. INTRODUCTION. - In its most basic realization, the STM consists of
an atomically sharp metal tip ‘flying’ over the flat surface of an electrically
conducting plane sample at a distance on the order of 5-10 Å; a low DC voltage
on the order of a 1 V is applied between the tip and the conducting sample
causing an electron tunnel current to flow between the available energy levels in
the tip and plane sample; the tip’s standoff distance (z-direction) from the sample
is usually maintained via a servomechanism, whose reference signal is the
desired magnitude of the tunnel current and whose control signal is applied to a
z-direction piezoelectric crystal driver; and, lastly, the tip can also move parallel
Copyright 1993 Rev. 1.0 2 of 71
(x- & y-direction) to the surface of the sample via x-direction and y-direction
piezoelectric crystal drivers.
In their August 1985 Scientific American article on the STM,
1
the
inventors, Gerd Binnig and Heinrich Rohrer, who shared the 1986 Nobel Prize in
Physics shared with Ernst Ruska for his work in the development of electron
microscopy,
2
gave a very abbreviated and rather superficial history of the
scientific developments which led up to the STM in its present form. The lack of
historical rigor in this Scientific American article was most likely due to the fact
that Binnig and Rohrer, like most practicing scientists, were less well acquainted
with the history of their profession than they were probably aware of. Binnig and
Rohrer’s history of the STM started in 1927 with Clinton J. Davisson and Lester
H. Germer’s experimental verification of the wave nature of the electron, shown
by the diffraction of electrons off the face of a single crystal of nickel; followed by
the work, circa 1950, of Erwin W. Müller on the field-ion microscope (FIM), which
allowed, for the first time, images to be made of individual atoms by inverse
mapping of the intersected trajectories of field ionized helium ions scattered from
a positively charged tungsten tip; and lastly, “…the first experimental verification
of tunneling…”, according to Binnig and Rohrer, by Ivar Giaever around 1960 -
this last assertion was clearly erroneous as will become obvious later on in this
appendix. A much more detailed and relevant discussion of the history of the
STM can be found in Binnig and Rohrer’s Nobel Prize lecture, but even here the
history was not as complete as it should have been.
3
In particular, it did not
address itself to the origins of the phenomenon of field emission which was first
reproducibly observed with a device that presaged the modern STM in many
ways.
1
G. Binnig, H. Rohrer; The Scanning Tunneling Microscope; Scientific American; Vol. 253;
No. 2; August 1985; pp. 50-56.
2
Note, a Nobel Prize may only be split a maximum of three (3) ways, and it is not awarded
posthumously, except in the case where the winner(s) had been notified, accepted the award and
then dies before they can travel to Stockholm to actually receive it.
3
Gerd Binnig, Heinrich Rohrer; Scanning Tunneling Microscopy - From Birth to Adolescence;
in W. Odelberg (Ed.); Les Prix Nobel, 1986; Almqvist & Wiksell International; 1987; pp. 91-111.
[According to Rohrer, “… [he] had been involved for a short time with tunneling between very
small metallic grains in bistable resistors…” This intriguing statement suggests coherer-like
behavior. In order to find out more about this work, I have written to Dr. Rohrer requesting any
references he might have on this subject. His address is: Dr. Heinrich Rohrer, IBM Zurich
Research Laboratory, CH 8803 Rüschlikon, Switzerland. By way of reply, Rohrer sent me 16
reprints of articles on STMs! I have still not been able to determine which of these articles relates
to the “…bistable resistors…” he spoke of in the Nobel Lecture. However, I was able to clear up
something that had always confused me about Les Prix Nobel, 1986, namely why only Rohrer’s
biography appeared - Gerd Binnig’s was nowhere to be found. One of the reprints sent me by
Rohrer was of his and Binnig’s contributions to Les Prix Nobel, 1986 and it contains, in addition to
the their Noble Lecture, the separate biographies of both Rohrer and Binnig. The mystery is
solved, the printer apparently simply forgot to include Gerd Binnig’s biography in the finished
volume.]
Copyright 1993 Rev. 1.0 3 of 71
I should like to propose a somewhat different history leading up to the
invention of the STM. In this history, I shall show how the confluence of scientific
insights coupled with technological advances conspired to make the discovery of
the modern realization of the STM highly probable if not inevitable. The STM
depends on three fundamental pieces of scientific or technological knowledge:
1) the existence of electron tunneling, 2) a stable and precise method for
positioning the metal tip close to the plane conducting surface to be examined,
and 3) a method of measuring steady currents in the picoampere (10
-12
A) to
nanoampere (10
-9
A) range.
Our starting point will be the decade beginning with 1880. This decade
produced Paschen’s law of electrical conduction in gases, which then led directly
to the initial discovery and then verification of electron tunneling between closely
spaced metal electrodes within 20 years. In addition, the 1880s witnessed the
invention of what was and still is the most sensitive method for measuring small
distances, interferometry - specifically the Michelson interferometer and all its
subsequent modifications and adaptations. And as if all this were not enough,
the year 1880 also witnessed the discovery of the phenomenon of piezoelectricity
by Jacques and Pierre Curie. These were all the basic ingredients needed to
concoct the modern STM.
Copyright 1993 Rev. 1.0 4 of 71
§2. PRELUDE: ELECTRICAL CONDUCTION IN GASES. - That air
conducts electricity became obvious once one understood the electrical nature of
lightning; the experiments to prove the electrical nature of lightning were first
suggested by Benjamin Franklin in 1750. The actual experiments were first
performed in France by d’Alibard - Franklin’s famous kite experiments came
later.
4
Beside his experimental endeavors, Franklin also made a significant
contribution to the theoretical understanding of electricity in the form of his one-
fluid theory of electrical phenomena. Franklin’s one-fluid theory of electricity was
able to explain and, hence, organize the rather chaotic collection of observations
of electrostatic phenomena then extant. The two-fluid (“vitreous [+]” & “resinous
[-]”) theory of Charles François Cisternay du Fay or Dufay, which was conceived
somewhat before Franklin’s one-fluid theory, was the other competing paradigm.
The one-fluid theory of Franklin was able to account for electrostatic behavior in
4
There seemed to be some confusion about the actual name of this Frenchman. According
to,
Martin A. Uman; All About Lightning; Dover Publications, Inc.; 1986; pp. 1-6.
his name was Thomas-François d’Alibard, but according to,
Albert Parry (Ed.); Peter Kapitsa on Life and Science; Macmillan Co.; 1968; pp. 38-39 & 245.
the man’s name was Jean François Dalibard. What both books did agree on was that he
[d’Alibard] did perform the experiments on the electrical discharge from clouds before Benjamin
Franklin. In an effort to clarify the identity of d’Alibard, I checked the following reference,
Roman d’Amat; Dictionnaire de Biographie Française, Vol. 9; Librairie Letouzey et Ané; 1961;
p. 1521.
which revealed that it was one Thomas-François d’Alibard (1709-1779) who translated into
French Ben Franklin’s book Experiments and Observations on Electricity. This still did not
answer the question of where Peter Kapitsa [or more likely his editor, Albert Parry] obtained the
attribution for the French translation to “Jean François Dalibard, 1703-99” found in endnote #2,
p.245 of Peter Kapitsa on Life and Science . Endnote #2 referenced two books as the source of
the information on the translation,
Leonard W. Labaree et al. (Eds); The Autobiography of Benjamin Franklin; Yale University
Press; 1964; p. 331.
Alfred Owen Aldridge; Franklin and his French Contemporaries; New York University Press;
1957; p. 21 & 239.
Amazingly, these two books disagreed on d’Alibard surnames. Labaree’s book correctly stated
that Thomas-François Dalibard was the translator, while Aldridge’s book touted Jean François
Dalibard as the translator. Examination of the references in Aldridge’s book revealed that he got
his incorrect information on d’Alibard from,
I. Bernard Cohen; Benjamin Franklin’s Experiments; Harvard University Press; 1941; p. 440.
In the Index of Cohen’s book (p. 440), there was listed one “Jean François Dalibard (1703-1799)”.
The explanation for this appalling state of confusion probably originated in the fact that the French
versions of Franklin’s Experiments and Observations on Electricity are translated “par M.
d’Alibard”. This irritating affectation of dropping the surnames of French authors in this time
period was quite common. The first and second editions of d’Alibard’s translations can be found
at The Library Co., 1314 Locust St., Phila. PA 19107, (215)546-2465; their hours are M-F 9:00
AM to 4:45 PM.
Copyright 1993 Rev. 1.0 5 of 71
simpler manner, and it presaged, by more than a century, the discovery of the
electron - the universal electric fluid.
Conduction of electricity through other gases or mixtures thereof followed
immediately, e.g., Alessandro Volta’s eudiometric pistol, with which he
discovered methane, and Sir William Watson’s discovery of electrical conduction
in rarefied air and the fact that the conduction increased as the pressure of the
gas decreased. The quantification of this behavior attained its most succinct
form - Paschen’s law - as the result of the work of Louis Carl Heinrich Friedrich
Paschen.
5
Paschen’s doctoral thesis of 1888, suggested to him by his advisor,
August Kundt, involved observing the sparking behavior of two spherical
electrodes in a gas; with a fixed voltage applied to the electrodes, the sparking
distance was noted for a given gas pressure. Note, the discharge voltages
plotted by Paschen and the researchers who followed him were always the
minimum voltages which produced a discharge under the conditions of the
experiment. The results of these experiments by Paschen revealed that, for a
fixed voltage, the spark length was inversely proportional to the gas pressure,
i.e., gas pressure versus spark length yields a rectangular hyperbola, see Fig.
E1. As was the case with most scientific laws, Paschen’s law was subject to
much testing to scope out the limits of its validity.
One of the most exhaustive attempts at verifying this law came as a result
of the work of J. B. Peace in England.
6
Peace’s work consisted of measuring
the sparking voltage as a function of the gas pressure and plane electrode
separation; the gas used in all his measurements was air. Peace’s results
confirmed Paschen’s law in the portion of the rectangular hyperbola closest to
the origin, i.e., the most curved part of the hyperbola; the agreement was less
good for the two straight portions of the curve, i.e., the sections corresponding to
low pressure or short spark lengths. Because Peace’s gas pressures went below
2 cm. of Hg, his curves of gas pressure versus sparking voltage for fixed
electrode separation exhibited a monotonic increase in the sparking voltage for
pressures less than what has come to be called the ‘critical’ pressure; Paschen
did not witness this behavior since his lowest pressures exceeded the critical
pressure for the electrode separation he employed; in his spark voltage versus
gas pressure curves at fixed electrode separation, the sparking voltage
monotonically decreased with decreasing pressure.
5
[L.C.H.]F. Paschen; Ueber die zum Funkenübergang…erforderliche Potentialdifferenz [On
the spark discharge…required potential difference]; Annalen der Physik und Chemie; Vol. 37;
1889; pp. 69-96.
6
J.B. Peace; On the Potential Difference required to produce a Spark between two Parallel
Plates in Air at Different Pressures.; Proceedings of the Royal Society (London); Vol. 52; 1892;
pp. 99-114.
Copyright 1993 Rev. 1.0 6 of 71
121086420
0
1
2
3
4
400 V
600 V
800 V
1200 V
1600 V
FIGURE E1 - Paschen Plot Using Data From W. R. Carr.
Spark Length (mm)
Air Pressure (mm Hg)
The nonconformance of some of Peace’s data to Paschen’s law
stimulated a graduate student, W. R. Carr, at the University of Toronto to
replicate the same experiments with better designed equipment.
7
From his
reading of Peace’s paper, Carr deduced that Peace’s apparatus did not subject
the gas to a uniform electric field due to the fringing fields present at the edge of
the plates, which were exposed to the gas and, by Peace’s own admission, were
the site of corona discharges. Carr’s apparatus also employed plane electrodes,
but, unlike Peace’s configuration, the edges of these electrodes were buried in
ebonite (a hard rubber). Carr mentioned that when applying the voltage across
the electrodes, starting at a low voltage and progressively increasing it until a
discharge ensued, it was of utmost importance to wait more than fifteen minutes
between each increase due to the time lag between application of an above
7
W.R. Carr; On the Laws governing Electric Discharges in Gases at Low Pressure.;
Proceedings of the Royal Society (London); Vol. 71; February 11, 1903; pp. 374-376. [Note, this
was only an abstract of the paper that appeared in the Philosophical Transactions of the Royal
Society.]
Idem; On the Laws governing Electric Discharges in Gases at Low Pressure.; Philosophical
Transactions of the Royal Society (London); Vol. 201 (Ser. A); 1903; pp. 403-433.
Copyright 1993 Rev. 1.0 7 of 71
threshold voltage and the breakdown of the gas, an effect first alluded to by
Warburg; the time lag is most pronounced near the critical pressure.
8
With this
improved apparatus, Carr was able to show that Paschen’s law held over all
pressures, even those well below the ‘critical’ pressure. In addition, the family of
plots of sparking voltage versus gas pressure for various electrode separations
revealed the presence of a unique minimum spark voltage, this minimum voltage
was, for a given electrode geometry, only dependent on the nature of the gas,
changing the electrode material and/or spacing did not affect it, see Fig. E2. This
result was very significant, since it meant that for a particular electrode geometry
there was a definite ionization threshold for the gas in the interelectrode volume.
The ability of STMs to work in air is explained by this fact as is the resistance of
electrostatically driven micromachines to breakdown of the air gaps supporting
these fields.
9
The validity of Paschen’s law was also extended by Carr to gases other
than air: hydrogen, carbon dioxide, oxygen, hydrogen sulfide, nitrous oxide,
sulfur dioxide and acetylene; all the gases tested by Carr were found to exhibit a
unique minimum sparking voltage.
10
8
E.G. Warburg; Ueber die Verzögerung bei der Funkenentladung [On the Time-Lag at the
Spark Discharge]; Annalen der Physik und Chemie; Vol. 62 (3rd Series); No. 11; 1897; pp. 385-
395.
9
Re STMs, when asked why STMs are able to function in air without breaking down due to a
gas discharge, a well worn explanation is that when the tip and base electrode are only 5-10 Å
apart, no air can get between them to be ionized. The problem with this rationalization is that it
ignores the breakdown path between the base electrode and the shank of the tip electrode, this
distance is 1000s of Ångströms in length. What prevents the air from breaking down in this
region? The answer is the existence of a minimum discharge voltage.
Re micromachines, “In fact, the breakdown electric field [emphasis added] increases in small
gaps by a factor of over 10 times the macroscopic limit of 3 megavolts per meter, resulting in an
even more favorable scaling for electrostatics.” See,
R.T. Howe, R.S. Muller, K.J. Gabriel, W.S.N. Trimmer; Silicon micromechanics: sensors and
actuators on a chip; IEEE Spectrum; Vol. 27; No. 7; July 1990; pp. 29-35.
This quote implicitly pays homage to Paschen’s law, since in air at STP (Standard
Temperature & Pressure: 20°C & 760 mm Hg) smaller gaps require larger electric fields to initiate
a discharge. Consider the following example taken from Fig. 8 of Carr’s 1903 paper. The
discharge electric field for a 250 µ gap @ STP was 1800 V/250X10
-6
m = 7.6X10
6
V/m, while for a
5 µ gap the required field was 380 V/5X10
-6
m = 76.0X10
6
V/m. I wish to thank Jennifer Coyle-
Byrne for bringing this article by Howe et al. to my attention.
10
Carr used R.J. Strutt’s (the 4th Baron Rayleigh) values for the minimum spark voltage of
nitrogen,
Hon. R.J. Strutt; On the Least Potential Difference Required to Produce Discharge through
Various Gases.; Philosophical Transactions of the Royal Society; Vol. 193 (Ser. A); 1900; pp.
377-394.
Copyright 1993 Rev. 1.0 8 of 71
1086420
0
1000
2000
1 mm
2 mm
3 mm
5 mm
10 mm
FIGURE E2 - Critical Pressure Plots Using Data From W. R. Carr.
Air Pressure (mm Hg)
Spark Voltage (Volts)
A
B
I should like to backtrack for a moment and set the record straight. W. R.
Carr is rightly given the credit for having pointed out the deficiencies in J. B.
Peace’s apparatus and determining Paschen curves for a myriad of experimental
conditions: different gases, pressures, and electrode spacings. But it must be
pointed out that three years earlier, R. J. Strutt (son of J. W. Strutt, a.k.a. 3rd
Baron Rayleigh or simply Lord Rayleigh) had taken similar precautions with his
apparatus when he measured the minimum discharge voltage for a number
gases, including the then very hard to come by gas, helium.
11
Strutt’s
11
Hon. R.J. Strutt; On the Least Potential Difference Required to Produce Discharge through
Various Gases.; Philosophical Transactions of the Royal Society; Vol. 193 (Ser. A); 1900; pp.
377-394. [Strutt obtained helium by either chemically digesting in sulfuric acid or roasting in a
furnace a mineral called monazite, then chemically purifying the resulting gas given off by the
treated mineral. The main impurity, nitrogen, was removed via the Cavendish method, i.e., pure
oxygen was added to the gas mixture, an electrical spark was then passed through the mixture to
cause the nitrogen and oxygen to react with one another forming compounds which could then be
removed by chemical absorbents with the remaining unreacted gas being composed mostly of
helium. Henry Cavendish originally used this technique with atmospheric air and showed that it
was partially composed of an inert gas or gases; J.W. Strutt (a.k.a. 3rd Baron Rayleigh) and W.
Ramsay redid Cavendish’s work and were able to show that the inert gas was made up mostly of
a new element, argon. They received the Noble Prize for this and other related work.
Cavendish’s other lesser known but significant contributions to science include his experimental
proof of the inverse square law for electric charges, which he did before Charles Augustin de
Coulomb, and his experimental proof of Ohm’s law, which he did before Georg Simon Ohm.
Cavendish did not receive credit for the aforementioned discoveries for a simple reason, he did
not publish all of his results; many were found among his papers after his death.
Note, in 1906 Lord Kelvin used the fact that helium could be obtained from minerals such as
clevite and monazite to argue against the then new theory of radioactive decay and its attendant
transmutation of elements. Kelvin’s reasoning was that just because a mineral contained helium,
this did not prove transmutation - e.g., by alpha particle decay - since the helium could have
simply been sequestered in the mineral from perhaps the atmosphere. Surprisingly, Kelvin was
both wrong and right at the same time. He was wrong about the source of helium in clevite;
clevite and monazite contained uranium and thorium, respectively, and so this accounted for the
Copyright 1993 Rev. 1.0 9 of 71
safeguards against measurement artifacts due to discharges from the edge or
back of his plane electrodes consisted of setting the brass plates into recesses
machined into the face of a pair of ebonite disks so that the electrode’s front
surface was flush with the surface of the insulator. Unlike Carr, Strutt did not
employ electrochemical batteries, but used instead a motor driven Wimshurst
electrostatic machine
12
connected in parallel with a large Leyden (Leiden) jar
(capacitor) and a fluid resistor, this configuration formed a very stable adjustable
power supply with the Leyden jar smoothing out the voltage ripple and the fluid
resistor providing an adjustable load to the Wimshurst machine thus allowing the
voltage to be varied; the discharge of the gas was sensed by the resulting sound
in a telephone receiver in series with the brass electrodes. Strutt mentioned in
passing that the first spark to pass through a gas could require as much a three
times the voltage as the succeeding sparks. Consequently, his measurement
protocol consisted of slowly increasing the applied voltage until the first spark
occurred, ignoring this voltage reading and then decreasing the voltage;
13
the
presence of helium - through transmutation. On the other hand, he was right that certain
substances have a natural affinity for atmospheric helium, see APPENDIX B of this thesis (The
1993 Thesis titled “Coherers, a review” by Thomas Mark Cuff, which is available on
ResearchGate.) where glass and certain metals are shown to be helium loving. For more details
about Lord Kelvin’s disagreement with the radioactive decay theory see,
Florian Cajori; A History of Physics…; The Macmillan Co.; 1935; pp. 298-302.
The theory of radioactive decay seems to have had a special place on Lord Kelvin’s list of
Ten Things I Hate the Most, since it was also responsible for shattering his estimation of the age
of the Earth. The heat produced by radioactive decay was a significant source of heat generation
inside the Earth, and so violated one of Kelvin’s main assumptions. For more information see,
Joe D. Burchfield; Lord Kelvin and the Age of the Earth; The University of Chicago Press;
1990.]
12
Until the advent of the Van de Graaff generator (circa 1933), the Wimshurst influence
machine (a.k.a. Wimshurst machine) was the most reliable (i.e., no spontaneous polarity
reversals), self-exciting, easy to use, high voltage generator around. James Wimshurst (1832-
1903) produced the first Wimshurst influence machine in the early 1880s. It was of a radically
different design than all the influence machines that had gone before it. In fact, Wimshurst
himself took obvious delight in pointing out that his machine incorporated many of the same
features specifically condemned as antithetical to ‘good’ influence machine design by the two
reigning experts on such contrivances, Lord Kelvin (a.k.a. Sir William Thomson) and W.T.B.
Holtz. For more information, see,
C.N. Brown; James Wimshurst, His Machine and its Antecedents; Papers Presented at the
10th IEE Weekend Meeting on the History of Electrical Engineering; Brighton, U.K.; July 2-4,
1982; 6 p. [Note, this paper was impossible to find in the U.S. I ordered the entire proceedings
of the 1982 Brighton meeting through the London office of the IEE, P.O. Box 96, Stevenage,
Herts. SG1 2SD, United Kingdom; these proceedings cannot be ordered through the IEE offices
located in New Jersey, U.S. The price was £11.00.]
13
Between 1946 and 1953 F. Llewellyn Jones investigated this effect, and determined first,
that it was real and second, that it was most likely due to the layer of oxide on the surface of the
electrodes. This oxide layer was thought to trap positive ions, which would then reduce the work
function for electron emission thus reducing the required discharge voltage,
F. Llewellyn Jones; Electrical Properties of Tungsten Oxide Films; Nature; Vol. 157; March
23, 1946; pp. 371-372.
Idem; Electrode Ionization Processes and Spark Initiation; Proceedings of the Physical
Society (London); Vol. 62 (Series B); 1949; pp. 366-376.
Copyright 1993 Rev. 1.0 10 of 71
subsequent cycles of cautiously increasing the voltage until the sound of the
discharge was detected, were followed by recording of the discharge voltage and
then decreasing the voltage, these cycles were usually repeated ten times.
At the conclusion of his paper, R. J. Strutt attempted to relate the
minimum in the discharge voltage versus pressure curve to the corresponding
value of the cathode fall for the same gas.
14
The cathode fall being the voltage
drop of the glow discharge structures associated with the cathode, i.e., the
various dark spaces (Faraday’s, Crookes’ & Aston’s) and the corresponding
intervening luminous areas, but excluding the positive column and those gas
discharge structures associated with the anode. In fact, the cathode fall
represented most of the voltage drop found in glow discharge. In this endeavor
he was encouragingly successful, thus providing some physical insight into the
physical nature of the discharge mechanism.
Once the validity and scope of Paschen’s law had been confirmed, the
next step was to develop a theory that would explain the observations. The
person, whose theory most completely accounted for the observed facts, was the
Irish physicist, John Sealy Edward Townsend. The crux of Townsend’s theory
was that the atoms or the molecules making up a gas could be ionized by single
collisions with charged particles (electrons and/or positive ions) accelerated by
the impressed electric field.
15
Since the resulting ion pair usually consisted of a
F. Llewellyn Jones, E.-T. De La Perrelle, C.G. Morgan; La rôle des électrodes dans le
mécanisme d’amorçage de la décharge électrique dans les gaz. [The rôle of the electrodes in the
initiation of electrical discharges in gases]; Comptes Rendus Hebdomadaries des Seances de
l’Academie des Sciences (Paris); Vol. 232; 1951; pp. 716-718.
F. Llewellyn Jones, C.G. Morgan; Surface films and field emission of electrons; Proceedings
of the Royal Society (London); Vol. 218 (Series A); 1953; pp. 88-103.
Note, Frank Llewellyn Jones also wrote a book on the subject of electric contacts. This is only
the second book that I know of on this particular subject - the first, and by far the most famous,
book addressing this topic was that penned by Ragnar and Else Holm. The citation for F.
Llewellyn Jones’ book is,
Frank Llewellyn Jones; The Physics of Electrical Contacts; Oxford University Press; 1957.
14
The measurement of the cathode fall reached its most sophisticated form in the method
first proposed by J.J. Thomson and then later realized by F.W. Aston,
F.W. Aston; The Distribution of Electric Force in the Crookes Dark Space; Proceedings on the
Royal Society (London); Vol. 84 (Series A); 1911; pp. 526-535.
15
Of the competing ionization-by-collision theories, the one by J.J. Thomson was the most
interesting since it gave insight into his view of the atom. Thomson’s idea was that an atom broke
apart (ionized) as the result of repeated collisions with electrons or ions. He thought that these
recurring collisions eventually destabilized the atom causing it to literally explode. This model
was probably consistent with his “plum-cake” model of the atom (sometimes called the “plum
pudding” model), wherein the electrons of the atom were distributed in a matrix of equal but
opposite charge; the neutron had not been discovered, yet, and so was not part of the model.
See,
Max Jammer; The Conceptual Development of Quantum Mechanics; McGraw-Hill Book Co.;
1966; p. 69.
According to this model, the electrons could be either stationary or moving. From Maxwell’s
work, it was realized that a single electron moving in a closed orbit would radiate away its energy.
Thomson was, of course, aware of this, but pointed out that if the electrons were situated in a ring
Copyright 1993 Rev. 1.0 11 of 71
electron and a positive ion, and because these charged particles, again under
the influence of the impressed field, could produce further ionizations, the system
could be said to possess a form of positive feedback. This self reinforcing
behavior embedded in Townsend’s model nicely explained the existence of self
sustained discharges, e.g., glow discharges and arcs.
§3. VERIFICATION OF PASCHEN’S LAW AT SMALL DISTANCES AND
THE RESULTING UNEXPECTED PHENOMENA (THE AMERICAN EFFORT). -
The question of the nature of light and the medium which supported its vibration,
the ether, began to assert itself in the 1800s. Even a brief history of this effort
could take ten or twenty pages, and so I will not attempt it, here.
16
What I will
discuss, briefly, was Albert Abraham Michelson’s contributions to the study of the
motion of the Earth through the ether.
In September 1880, Michelson arrived in Berlin to begin study for his PhD
at the Humboldt University of Berlin. For his thesis, Michelson decided to
attempt to detect the motion of the Earth through the ether. His thesis advisor,
Hermann Ludwig Ferdinand von Helmholtz, could find no fault with the intended
approach other than the difficulty in maintaining the temperature constant enough
that it did not produce a measurement artifact. Using funds provided by
Alexander Graham Bell, Michelson had the firm of Schmidt and Haensch
construct his first “interferential refractor” or what has since been come to be
known as a Michelson interferometer, and performed his first ether drift
experiment in Potsdam in 1881.
17
In 1887, Michelson and E. W. Morley redid
the Potsdam experiment at Case University in Cleveland with a much improved
interferometer.
18
Despite the incontestable success of his ether drift
and rotated in such away as to maintain their relative positions to one another, the radiated
energy was enormously reduced - in one simple example by as much as 10
16
. The more
electrons in the atom, the better its stability. For atoms with more and more electrons, one ring or
shell would not do, but Thomson’s model allowed for multiple rings or shells.
With this model it would be very difficult for a single collision to ionize the atom, since its
electrons were effectively shielded, both physically and electrically, from the outside world by
being embedded in the matrix of positive charge. For a description of Thomson’s model of the
atom, see,
J.J. Thomson; On the Structure of the Atom: an investigation of the stability and periods of
oscillation…; Philosophical Magazine [and Journal of Science]; Vol. 7 (6th Series); March 1904;
pp. 237-265.
It should be noted that this model was actually rather the favorite among the more classically
disposed physicists. See,
Lord Kelvin; Aepinus Atomized; Philosophical Magazine [and Journal of Science]; Vol. 3 (6th
Series); March 1902; pp. 257-283.
16
J. Lovering; Michelson’s Recent Researches on Light; Annual Report of the Board of
Regents of the Smithsonian Institution; July 1889; pp. 449-468.
17
A.A. Michelson; The Relative Motion of the Earth and the Luminiferous Ether.; American
Journal of Science; Vol. 22 (3rd Series); 1881; pp. 120-129.
18
A.A. Michelson, E.W. Morley; On the Relative Motion of the Earth and the Luminiferous
Ether.; American Journal of Science; Vol. 34 (3rd Series); 1887; pp. 333-345. [Note, the mere
Copyright 1993 Rev. 1.0 12 of 71
experiments and the part played by his interferometer, Michelson never did
patent it. As a result, by the 1890s, Michelson interferometers were being built
commercially by a number of machine shops and began to appear in physics
labs throughout the country. In 1892, Michelson was offered and accepted a
post at the University of Chicago. It is to this University, specifically the Ryerson
Laboratories, that we turn to next in our search for the origins of the STM.
Paschen’s law had been subject to about 10 years worth of experimental
verifications by many workers in the field, and it had held up well under this
intense scrutiny. But around 1898, a PhD graduate student named Robert
Francis Earhart decided, at the suggestion of Albert Abraham Michelson, to see if
Paschen’s law held at extremely small electrode separations, down to a
wavelength of visible light. At first glance this would seem to be a foregone
conclusion given the size of the law’s demonstrated envelope of applicability.
Earhart’s apparatus consisted of a fixed ball electrode and a plane
counterelectrode mounted on a movable carriage which also had a flat mirror that
formed one arm of a Michelson interferometer. In its normal operation, the two
electrodes were brought into contact, as sensed by the passage of current due to
a very small test voltage, the plane electrode was then backed away from contact
with the spherical one, the number of fringes was noted, where each fringe
corresponded to half a wavelength of the monochromatic light being used with
the interferometer. The whole apparatus was ensconced in an airtight box, which
allowed measurements to be made at different gas pressures and with different
carefully dried and filtered gases.
19
The experimental curves generated by
Earhart were as follows: the x-axis was the electrode separation in wavelengths
of sodium light, the y-axis was the value in volts of the sparking potential
difference, and the resulting curves were measured at different pressures for the
same gas, see Fig. E3. There were two regions of interest with respect to
possession of a Michelson interferometer did not guarantee that one could make the kinds of
incredible measurements that Michelson was able to so regularly obtain. In fact, no one was to
equal Michelson’s finesse with his interferometer, until 1921 when Dayton C. Miller redid
Michelson’s ether drift experiment and found a slight positive result. These results stood for more
than a quarter of century, until 1955 when R.S. Shankland et al. with the help of a digital
computer were able to show that Dayton’s positive result was most probably due to temperature
changes,
R.S. Shankland, S.W. McCuskey, F.C. Leone, C. Kuerti; New Analysis of the Interferometer
Observations of Dayton C. Miller; Reviews of Modern Physics; Vol. 27; 1955; pp. 167-178.
The techniques used by Shankland et al. to detect the systematic variations in the fringe shift
measurements are pretty much de rigueur today in the field of SPC (Statistical Process Control)
and involved the use of the autocorrelation function to check the data for nonrandomness.]
19
R.F. Earhart; The Sparking Distances between Plates for small Distances. ; Philosophical
Magazine [and Journal of Science]; Vol. 1 (6th Series); 1901; pp. 147-159. [Note, because of my
interest in learning more about Earhart’s work, I tried to get a hold of his PhD thesis. To this end I
consulted the Dissertation Abstracts Index (1861-1980) CD ROM at Temple University’s Paley
Library, and found that he had gotten his PhD in 1900 from the University of Chicago. Calling the
Reference Desk at the John Crerar Library, (312)702-7874, I found out that Earhart’s 1901
Philosophical Magazine articles was his PhD thesis. From the acknowledgement at the end of
this paper, it was apparent that his thesis advisor was A.A. Michelson and also perhaps S.W.
Stratton; Michelson was head of the Ryerson Laboratory of the University of Chicago at this time.]
Copyright 1993 Rev. 1.0 13 of 71
electrode separation: 1) ~3-5λ, and 2) <3-5λ, where λ was the wavelength of
sodium light. In Region 1 (~3-5λ), all the curves of sparking potential difference
versus electrode separation started out with a high sparking voltage at large
separations, with the sparking voltage decreasing with decreasing electrode
separation in a roughly linear manner. The sparking voltage versus electrode
separation curve maintained the same gentle downward slope at smaller and
smaller separations until ~3-5λ of sodium light at which point the minimum
sparking voltage began a precipitous but still linear decrease with decreasing
separation, i.e., the slope increased dramatically, thus putting a ‘knee’ or ‘elbow’
in the curve. This rather unexpected behavior at small separations was ascribed,
by Earhart, to the presence of an exceedingly thin air film tenaciously clinging to
the surface of the electrodes. According to Earhart’s view of the matter, in
Region 1 (~3-5λ) the gas, which was being broken down, consisted mostly of the
normal bulk gas. However, in Region 2 (<3-5λ), due to the closeness of the
electrodes, the bulk gas was squeezed out of the interelectrode space, and only
the condensed air film on the surface of the electrodes filled the space between
the electrodes and was then subject to electrical breakdown. Because the gas
making up this film was in a different state from the same species in the bulk, this
was supposed to account for the unexpected electrical behavior that Earhart saw
in Region 2. As we shall see later on in this appendix and, in more depth, in
APPENDIX F of the 1993 Thesis titled “Coherers, a review” by Thomas Mark
Cuff, which is available on ResearchGate, where we will discuss the work of
Lester H. Germer, Earhart was almost on the right track about the existence and
importance of the gas and/or vapor films on the electrodes. In the 1950s and
60s, Germer would show experimentally that adsorbed gas films on the
electrodes did indeed make it easier for field emission to initiate the so-called
‘vacuum spark’.
806040200
0
100
200
300
400
500
76 cm Hg
40 cm Hg
25 cm Hg
15 cm Hg
1 cm Hg
FIGURE E3 - An Earhart Plot in Air Using Data From G. M. Hobbs.
Electrode Separation in Wavelengths
Discharge Voltage (Volts)
Region 1
Region 2
Copyright 1993 Rev. 1.0 14 of 71
The first person to confirm Earhart’s findings was a British scientist, Philip
E. Shaw. In 1904, Shaw used a Rube Goldbergish appearing arrangement of six
mechanical levers in tandem to allow him to achieve the same range of small
electrode separations as Earhart; Shaw called his apparatus an electric
micrometer.
20
For reasons which were unclear to me, Shaw’s work was not
mentioned explicitly in Earhart’s followup paper of 1908 or, for that matter, in any
of the other papers written by other American researchers with the exceptions of
Carl Kinsley, John E. Almy and, much later, James W. Broxon.
Before continuing, I should like to mention that while Earhart appeared to
be the first person to try to verify Paschen’s law at distances comparable to the
wavelength of light, his work was in some ways anticipated about 60 years
earlier. After the discovery of the Voltaic pile at the end of the eighteenth
century, natural philosophers noticed that a tiny spark was formed on completing
or breaking a Voltaic circuit. This result held true even in the case where the pile
consisted of a single cell. Michael Faraday suggested that the spark was
actually produced just before contact was made in closing the circuit. He was
quickly challenged by one professional scientist, Dr. Jacobi, and also by a well-
to-do amateur scientific sleuth named John Peter Gassiot. Jacobi in 1838 and
Gassiot in 1840 had ordered constructed, according to their respective
specifications, micrometer driven contrivances which allowed them to bring two
electrodes to within extremely close approaches of one another. With these
devices, they proceeded to lay waste to Faraday’s claim of sparks being formed
before a Voltaic circuit was completed. Their results were quite convincing:
Jacobi showed a null result at a distance of closest approach of 1/20,000 inch,
while Gassiot duplicated this null effect at 1/5,000 inch. Both researchers were
able to demonstrate this negative result even with Voltaic piles consisting of
hundreds of cells.
21
Today, of course, we know that Faraday was right but for
20
P.E. Shaw; The Sparking Distance between Electrically Charged Surfaces. - Preliminary
Note.; Proceedings of the Royal Society (London); Vol. 73; 1904; pp. 337-342. [Besides
replicating Earhart’s work, Shaw had earlier used his electric micrometer to investigate the
behavior of coherers,
P.E. Shaw; An Investigation of the Simple Coherer; Philosophical Magazine [and Journal of
Science]; Vol. 1 (6th Series); No. 3; March 1901; pp. 265-296.
Around 1906, Shaw would again be linked to Earhart, this time because of simultaneous but
independent work they were both doing with regards to the electrical breakdown of oils at small
electrode separations - using again their respective contrivances, which served them so well
during the course of their investigations into gas breakdown at small interelectrode distances,
P.E. Shaw; The Disruptive Voltage of Thin Liquid Films between Iridio-Platinum Electrodes;
Philosophical Magazine [and Journal of Science]; Vol. 12 (6th Series); July-December 1906; pp.
317-329.]
21
Jacobi; On the Galvanic Spark.; Philosophical Magazine and Journal of Science; Vol. 13
(3rd Series): No. 84; December 1838; pp. 401-405.
J.P. Gassiot; An account of Experiments made with the view of ascertaining the possibility of
obtaining a Spark before the Circuit of the Voltaic Battery is completed.; Philosophical
Transactions of the Royal Society; Vol. 130 (Parts 1 & 2); 1840; pp. 183-192. [According to
Gassiot’s biography,
Sir Leslie Stephen, Sir Sidney Lee (Eds.); Dictionary of National Biography, Vol. 7; Oxford
University Press; 1937-38; pp. 935-936.
Copyright 1993 Rev. 1.0 15 of 71
the wrong reasons, while Jacobi and Gassiot were wrong for reasons which
would not unequivocally show themselves for almost another hundred years.
Earhart’s observations at small separations, in particular below the ‘knee’
(i.e., in Region 2, <3-5λ), were verified and extended to even smaller separations
in 1904 by Professor Carl Kinsley,
22
who together with Albert A. Michelson,
Robert A. Millikan, Henry G. Gale and Charles R. Mann formed the University of
Chicago Physics Department and ran the Ryerson Laboratories.
23
Kinsley was
able to restrict his measurements of the electrode separation to Region 2 (<3-5λ)
by employing a modification of the Michelson’s interferometer first suggested by
Professor Clark W. Chamberlain of Dennison University, who was working at the
Ryerson Laboratories on an unrelated project during the summer of 1903.
24
Kinsley began his paper with a review of the laws of electrical discharges in
gases, Paschen’ law, and then went on to talk about the work of Earhart and the
still later work of Glenn M. Hobbs (who had not yet published his findings). In
trying to explain the abrupt linear decrease in the sparking voltage characteristic
of Region 2, Kinsley suggested that it was due to a change in the charge carrier.
Specifically, he believeed that at extremely small electrode separations the
charge carriers were no longer ionized gas molecules, but were instead metal
ions from the electrodes, themselves. He based this hypothesis on the then well
known behavior of coherers of forming metal bridges between the closely spaced
metal particles or electrodes when cohered, and the fact that he had observed
this same behavior in his modification of Earhart’s basic apparatus if the
discharge currents were large. Kinsley was himself quite familiar with the
behavior of coherers, having done some work with them earlier in his career.
25
he was also ‘famous’ for demonstrating in 1844 “…by experimenting with delicate micrometer
apparatus (Philosophical Magazine for October) that [Sir William Robert] Grove’s arguments
against the contact theory of electricity were correct.” In doing this, Gassiot again convincingly
‘proved’ the nonexistence of a phenomenon whose existence we take today for granted. A
complete list of Gassiot’s papers up to and including 1863 can be found in the following,
George E. Eyre, William Spottiswoode; Catalogue of Scientific Papers (1800-1863),
Compiled and Published by the Royal Society of London, Vol. II; Her Majesty’s Stationary Office;
1868; pp. 779-780.]
22
C. Kinsley; Short Spark-Discharges; Philosophical Magazine [and Journal of Science]; Vol.
9 (6th Series); January-June 1905; pp. 692-708.
23
J.L. Michel; The Chicago Connection: Michelson and Millikan, 1994-1921; in Stanley
Goldberg, Roger H. Stuewer (Eds.); The Michelson Era in American Science 1870-1930;
American Institute of Physics; 1988; pp. 152-176.
24
C.W. Chamberlain; Note on the Compound Interferometer; Physical Review; Vol. 23; July-
December 1906; pp. 187-188.
Idem; The Radius of Molecular Attraction; Physical Review; Vol. 31; July-December 1910; pp.
170-182. [Caution, the unit of distance measurement used in this and other papers of this era
was the µµ. Do not - as I did - assume that µµ = 10
-12
m; a µµ was a millimicron, i.e., µµ = 10
-9
m
= nm (nanometer), see, for example p. 693 of Carl Kinsley’s 1905, Philosophical Magazine
article. Why physicists of this time period did not deign to designate a millimicron as the logically
consistent symbol ‘mµ’, I do not know.]
25
C. Kinsley; Coherers Suitable for Wireless Telegraphy; Physical Review; Vol. 12; January-
June 1901; pp. 177- 183.
Copyright 1993 Rev. 1.0 16 of 71
Another interesting fact uncovered by Kinsley was the change in the slope of the
precipitous linear roll off as a function of the treatment of the electrode surfaces,
e.g., due to buffing or polishing. Since we now know that the discharge in this
region was due to electron tunneling, and that the ability of surface treatment to
affect the work function of the metal was shown conclusively by J. Erskine-
Murray in 1898 (see APPENDIX D of the 1993 Thesis titled “Coherers, a review”
by Thomas Mark Cuff, which is available on ResearchGate),
26
it should be
obvious why the slope changed.
Kinsley appended an appendix to his paper where he discussed the
results of his experiments on the properties of the presumably metallic bridge(s)
that formed when the electrodes in his contrivance cohered. Specifically, he
examined how the resistance of the metallic bridges, formed by cohering the
electrodes, changed as they were stretched by slowly pulling apart the
electrodes. From this data he attempted to estimate the diameter of the bridge.
The answer he got was 4.4X10
-5
cm, which was significantly less than the
wavelength of the mercury light he was using, meaning that the bridge would not
be visible with a light microscope. Kinsley admited that he did not understand
the process behind the formation of the bridge.
In order to transfer a significant amount of material across the interelectrode
space much higher currents than one could reasonably expect from field
emission would be needed. The mechanism I wish to postulate as providing the
needed current is something called a ‘vacuum spark, arc or discharge’, see
APPENDIX F of the 1993 Thesis titled “Coherers, a review” by Thomas Mark
Cuff, which is available on ResearchGate. The vacuum spark, besides furnishing
essentially unlimited current, satisfies Occam’s razor by providing us with a
simple way of achieving the required mass transport of electrode material, i.e.,
thermal vaporization - electron tunneling and/or field emission, in contrast, do not
involve any significant thermal effects.
27
But before we digress to the subject of
26
J. Erskine-Murray; On Contact Electricity of Metals; Proceedings of the Royal Society
(London); Vol. 63; 1898; pp.113-146.
27
The only postulated thermal effect related to electron tunneling that I am aware of was the
so-called Nottingham effect,
W.B. Nottingham; Remarks on Energy Losses Attending Thermionic Emission of Electrons
from Metals.; Physical Review; Vol. 59 (2nd Series); 1941; pp. 906-907.
Fortunately for my hypothesis, this effect has not been unequivocally observed experimentally
even though people have assiduously searched for it. See,
G.M. Fleming, J.E. Henderson; The Energy Losses Attending Field Current and Thermionic
Emission of Electrons from Metals.; Physical Review; Vol. 58 (2nd Series); November 15, 1940;
pp. 887-894.
Idem; On the Energy Losses Attending Thermionic and Field Emission; Physical Review; Vol.
59 (2nd Series); 1941; pp. 907-908.
R.H. Good Jr., E.W. Müller; Field Emission.; in S. Flügge (Ed.); Handbuch Der Physik;
Springer-Verlag; 1956; pp. 196-231, in particular, see p. 199.
Additional evidence against thermal effects in electron tunneling and field emission was provided
in 1965 by Bert Halpern and Robert Gomer when they showed that field emission persisted even
at liquid helium temperatures (~4°K),
Copyright 1993 Rev. 1.0 17 of 71
‘vacuum sparks’, I should like to continue for awhile with the work inspired by
Earhart’s original experiment.
During the period 1902 to 1905, Glenn Moody Hobbs also investigated
gas discharges between closely spaced electrodes at Ryerson Labs.
28
Hobbs’
apparatus employed a standard Michelson interferometer, but with a more finely
cut main lead screw to allow control down to a tenth of a fringe and other
mechanical enhancements to reduce backlash. Able to capitalize on the
experiences of Earhart and Kinsley, who worked in the same laboratory as
himself, Hobbs essentially redid all the work of the two previous investigators,
and was able to obtain data which was less influenced by the various
measurement artifacts that made Earhart doubt his own data in the neighborhood
of the knee. The main problem with Earhart’s earlier curves was the uncertainty
about whether or not there was a plateau (asymptote) in Region 1 just before the
knee separating the two regions. Hobbs, by performing Earhart’s experiments at
a number of different gas pressures, was able to prove conclusively that there
was indeed a plateau in Region 1 just before the knee; the plateau was most
easily discerned at air pressures less than normal atmospheric pressure, and the
length of the plateau decreased as the air pressure increased to atmospheric,
see Fig. E3. Hobbs’ verification of the existence of a plateau was very important,
since Paschen’s law predicted such behavior in the case of non-planar (e.g., ball
& plane or point & plane) electrodes.
W. R. Carr was actually the first to point out that a plateau was required of
Earhart’s data. His argument to this effect can most easily be understood by
consideration of Fig. E2 and goes as follows. For large electrode separations,
one is above the critical pressure, e.g., the dotted vertical line labeled ‘A’ in Fig.
E2. In this case, as the electrode separation is decreased at constant pressure it
is obvious that the voltage required to cause a discharge also decreases (see
Fig. E3). But Fig. E2 indicates that the critical pressure increases as the
electrode separation decreases and eventually when the separation approaches
~10λ the critical pressure becomes greater than normal atmospheric pressure.
At this point, we would use the dotted line labeled ‘B’ in Fig. E2 to predict the
behavior of the discharge voltage with decreasing electrode separation. Here, as
B. Halpern, R. Gomer; Field Emission in Liquids; Journal of Chemical Physics; Vol. 43; No. 3;
August 1965; pp. 1069-1070.
The only positive results were obtained in the mid 1960s by a group at the Field Emission
Corporation, McMinnville, Oregon. However, these results were highly questionable due to the
complexity of the experimental arrangement and the large number of assumptions embedded in
their methodology. See,
F.M. Charbonnier, R.W. Strayer, L.W. Swanson, E.F. Martin; Nottingham Effect in Field and
T-F Emission: Heating and Cooling Domains, and Inversion Temperature; Physical Review
Letters; Vol. 13; No. 13; September 28, 1964; pp. 397-401.
L.W. Swanson, L.C. Crouser, F.M. Charbonnier; Energy Exchanges Attending Field Electron
Emission; Physical Review; Vol. 151; No. 1; November 4, 1966; pp. 327-340.
28
G.M. Hobbs; The Relation between P.D. and Spark-length for Small Values of the later.;
Philosophical Magazine [and Journal of Science]; Vol. 10 (6th Series); December 1905; pp. 617-
631+Plate XIII.
Copyright 1993 Rev. 1.0 18 of 71
the electrode separation is decreased at constant pressure, the discharge
voltage should increase, and not plateau out as it was actually shown to do by
both Earhart and Hobbs. The reason why the discharge voltage plateaus,
according to Carr, is that both Earhart and Hobbs used an electrode
configuration consisting of a plane and spherical electrode. As the plane and
sphere approach one another, the discharge instead of taking place at the point
of closest approach, which requires larger and larger voltages as this distance
shrinks, actually starts to occur at points on the sphere removed from the point of
closest approach, and so the discharge voltage does not need to increase and it
remains constant, i.e., plateaus. The same behavior would be found in the case
of a plane electrode and a needle electrode as, for example in the case of an
STM. In fact, this is the real reason behind why the STM can function in air: the
plateau voltage represents the minimum voltage necessary to cause a discharge
to occur through the air and this minimum is quite high, ~300 V, which is larger
than the voltages usually used with STMs. Note, the gas discharges I am
speaking about are classed as self-sustaining discharges, which include both
glow discharges and arcs; the glow discharges, in particular, are the ones which
have the minimum voltage requirement. It is true that gas discharges can occur
at voltages below the minimum voltage, but these Townsend type discharges
usually result from the passage of cosmic rays or other ionizing radiations.
These nonselfsustaining discharges are used by nuclear physicists in various
particle detectors, such as ion chambers, proportional counters and Geiger-
Müller tubes.
Hobbs showed that for a given type of electrode material, the slope of
Region 2 was independent of both the type of gas and its pressure. On the other
hand, the slope in Region 2 did depend on which metal or metals were employed
in the electrodes. Another way of looking at this was that the threshold
voltage at a fixed distance within Region 2 was a function of only the
electrode material and any surface treatments it might have undergone.
These two previous results supported Kinsley’s idea that the reason for the
change in slope in going from Region 1 to Region 2 was due to a change in the
nature of the charge carrier taking part in the discharge. Since we now know that
electron field emission was actually the mechanism which initiated conduction in
Region 2, another result of Hobbs’ makes sense: in Region 1 the onset of the
discharge took a sensible amount of time,
29
while in Region 2 there was not a
noticeable time lag. Hobbs also verified Kinsley’s observation that coherence
took place in Region 2 with sufficiently large discharge currents.
29
This time lag was apparently first discovered by Warburg,
E.G. Warburg; Ueber die Verzögerung bei der Funkenentladung [On the Time-Lag at the
Spark Discharge]; Annalen der Physik und Chemie; Vol. 62 (3rd Series); No. 11; 1897; pp. 385-
395.
Copyright 1993 Rev. 1.0 19 of 71
Around 1906, Earhart revisited his original work and extended it.
30
This
time around, he utilized a pointed electrode and a plane counterelectrode; the
pointed electrode consisted of No. 10 Sharp needles winnowed for their close
adherence to the figure of a ‘master’ needle. With this electrode configuration,
Earhart’s apparatus had many of the characteristics of an STM save the ability to
scan in the x-y plane and the ability to directly measure the currents, this last
shortcoming would be rectified ten years later by Franz Rother. Using this
electrode arrangement, Earhart was able to verify the well known fact that gas
discharges took place more readily between a point and a plane electrode when
the point was negatively charged with respect to the plane electrode. This
behavior was mentioned in the main part of this thesis in the section entitled THE
RESURRECTIONS, when we were discussing Arthur Schuster’s discovery of
nonohmic conduction and Lord Kelvin’s comments on the use of his quadrant
electrometer. Physically this was so because when the point electrode was
negative, the intense electric field gradient caused positive ions to intensely
bombard the surface of the point, releasing copious amounts of secondary
electrons and thus contributing to ionization of the gas. Earhart’s data indicated
that this polarity effect existed only in Region 1, once Region 2 was entered the
discharge voltage versus electrode separation curves coincided.
Thus, by the time Earhart publishes, in 1908, what will be his last paper on
the subject of spark discharges between closely spaced electrodes, his
discoveries will have been verified and/or extended by Shaw, Kinsley and Hobbs.
But 1908 was also to usher in the first papers to present empirical evidence
contradicting some of his findings. While many who were to subsequently redo
his experiment, from 1908 onward, would confirm the results he got for Region 1
(~3-5λ), a few people would present opposing data for Region 2 (<3-5λ).
Near the end of 1908, John E. Almy of the Brace Laboratory, Lincoln,
Nebraska redid Earhart’s experiment, substituting a Fabry-Perot interferometer in
place of the Michelson interferometer, using two tiny platinum balls for the
electrodes [later still, two needles], and observing the discharge directly with a
500X microscope instead of intuiting its existence electrically by measuring the
resulting voltage drop across the electrodes.
31
Almy’s experimental protocol
was to set the electrodes at a known distance apart and then slowly increase the
applied voltage until he saw a steady glow discharge appear about the
electrodes. As was the case with Earhart, Shaw, Kinsley and Hobbs, Almy
30
R.F. Earhart; Discharge from an Electrified Point and the Nature of the Discharge
occurring through very small Distances.; Philosophical Magazine [and Journal of Science]; Vol.
16 (6th Series); 1908; pp. 48-59 + Plate I.
31
J.E. Almy; Minimum Spark Potentials; Philosophical Magazine [and Journal of Science];
Vol. 16 (6th Series); July-December 1908; pp. 456-462. [Almy produced his spherical platinum
electrodes by melting the ends of 0.057 mm diameter platinum wires in an oxygen-hydrogen
flame. Years later, in 1978, Edgar Clayton Teague would use a similar strategy to study electron
tunneling, except he would employ spherical gold electrodes made by melting the ends of 1.25
mm gold wires under UHV (Ultra High Vacuum) using an electron beam instead of the oxygen-
hydrogen torch employed by Almy.]
Copyright 1993 Rev. 1.0 20 of 71
recorded that the minimum voltage necessary to produce a glow discharge was
about 360 V. However, unlike Earhart and company, Almy claimed that this
minimum voltage held even in Region 2 (<3-5λ). This was a very significant
point. Earhart’s results indicated that the minimum discharge voltage in Region
2, rather than remaining constant, decreased linearly. Almy, on the other hand,
said that in Region 2, the minimum discharge voltage was the same as in the flat
part of Region 1. In Region 2, if Almy decreased the applied voltage from say
360 V to 350 V, he observed that the glow discharge went out. From this result,
he stated that Paschen’s Law held even at the extremely small separations
characteristic of Region 2, and further that no discharges were possible below
this minimum of ~360 V. In contrast, Earhart never said that you could not get a
glow discharge in Region 2 at 360 V, what he did assert was that there was
some sort of discharge occurring in Region 2 which had a linearly decreasing
threshold with decreasing electrode separation.
When John E. Almy examined his closely spaced electrodes with his
microscope, his criterion for a ‘discharge’ was the presence of a steady light
indicating a glow discharge. Almy showed that even in Region 2, one could get a
glow discharge provided that the voltage was slightly above the minimum
discharge voltage of Carr. But what about the sub-minimal voltage discharges,
measured by Earhart et al. , in Region 2? Why did Almy not observe any light
from any transient discharges in Region 2? The simple answer, that even though
these transient discharges can and do produce light but that their transitory
nature makes them difficult to see even in the case of Almy with his microscope,
is hard to believe. The reason I doubt this scenario is that these discharges,
although very short lived, are nevertheless extremely bright and should have
been easy to spot, especially by someone who expects them a priori. My
explanation of why Almy failed to observe a light flashes in Region 2 is that, as
would be demonstrated in the 1950s by Lester H. Germer (see APPENDIX F of
the 1993 Thesis titled “Coherers, a review” by Thomas Mark Cuff, which is
available on ResearchGate), smooth clean noble metal contacts do not easily
form ‘vacuum sparks’.
Almy also revealed that when he employed needles for the electrodes, the
glow discharge, for the flat portion of Region 1 and all of Region 2, extended well
back along one of the needles, just as one would expect from W. R. Carr’s
explanation of the plateauing of the minimum glow discharge voltage. Carr
stated that below the critical pressure the minimum glow discharge voltage
should increase with decreasing electrode separation in the case of plane
electrodes. If one or both of the electrodes was not planar, the glow discharge
would tend to creep back along the nonplanar electrode so that the discharge
would not take place across the distance of closest approach, and hence the
minimum discharge voltage would remain constant rather than increasing with
decreasing interelectrode distance.
Copyright 1993 Rev. 1.0 21 of 71
The next person to cast doubts on Earhart’s results was a researcher at
the University of Illinois, Elmer H. Williams. In 1910, Williams revisited Earhart’s
work with an apparatus very similar in design to that utilized by Earhart in 1901: a
ball and plane electrode configuration, the electrode separation measured with a
Michelson interferometer, lead screw actuated motion, and the occurrence of a
discharge sensed by the drop in voltage across the electrodes.
32
Williams’
results match those found by Almy in that his data indicated that the flat portion
of the voltage versus distance curve continued unchanged into Region 2. There
was no ‘elbow’ or ‘knee’ at 3-5λ followed by a linear decrease in minimum
discharge voltage with decreasing distance. The minimum discharge voltage
measured by Williams was ~360 V, and this minimum did not appear to depend
on the electrode material. Williams tried the following combinations: aluminum-
aluminum, brass-aluminum, platinum-platinum, aluminum-silver, platinum-silver
and brass-platinum. Additionally, photomicrographs of the plane cathodes
revealed that the pitting of its surface - due to bombardment by positive gas ions
with the attendant lose of material, i.e., cathodic sputtering - was arranged in a
circle with its diameter increasing as the interelectrode separation decreased.
This behavior was, of course, another confirmation of W. R. Carr’s theory that
below the critical pressure the discharge in the case of nonplanar electrodes did
not take place at the distance of closest approach.
Besides replicating Earhart’s earlier work, Williams also examined the
effect of ultraviolet (UV) light on this arrangement. What he found was that, at
small electrode separations (1.5-5λ), irradiation by UV light caused the minimum
voltage for discharge to decrease with decreasing distance. The effect of UV
light, according to Williams, was to enhance ionization of the gas between the
electrodes, and by doing so to affect a reduction in the value of the minimum
discharge voltage. Because Williams’ results were so different in Region 2 from
those uncovered by the workers at Ryerson Labs, it was not surprising that a
year later (1911) he authored another paper on the same basic topic, in an
obvious effort to buttress the conclusions stemming from his previous paper.
33
In this 1911 paper, Williams investigated the influence of different gases
(hydrogen and carbon dioxide), the influence of pressure on the discharge of air,
and he returned to the effects of UV light on the system. Of the three things just
mentioned, it was what he found with respect to the illumination of the system
with UV which was the most interesting. In the second paper, Williams stated
that he had concerns about the fact that his original UV source, two aluminum
electrodes in a quartz tube, required such a high operating voltage that
significant amounts of radio frequency (RF) energy were also being generated,
and so might be confounding the results. Because of his concerns, Williams
decided to redo the experiment employing a low voltage UV source which
worked by drawing an arc between an iron and a mercury electrode (i.e., a
32
E.H. Williams; The Nature of Spark Discharge at very small Distances; Physical Review;
Vol. 31 (1st Series); No. 3; July-December 1910; pp. 585-590.
33
Idem; Spark Discharge at very small Distances; Physical Review; Vol. 32 (1st Series); No.
6; January-June 1911; pp. 585-590.
Copyright 1993 Rev. 1.0 22 of 71
mercury vapor lamp), andwhich operated at only 14 V. With this new source, the
data indicated that there was no effect on the breakdown voltage, i.e., the
minimum breakdown voltage in Region 2 (<3-5λ) was still ~360 V for air. This
negative result was very interesting, since it unequivocally showed that RF
energy could cause a substantial decrease in the minimum discharge voltage in
the range 1.5-5λ - remembering, of course, that Williams maintained that his
data, in the absence of any illumination, exhibits no ‘elbow’ or ‘knee’ at 3-5λ. It
seems obvious to me that the decrease in the minimum discharge voltage
experienced by Williams, due to the RF energy coming from his high voltage UV
source, was a manifestation of self-restoring coherer behavior. While he did
agree that the effect was due to the presence of RF energy, Williams did not
mention coherers at all, which was very strange given their commercial success
and relative prominence just a few years earlier. That coherence could happen
even when the distance between metal electrodes was on the order of 10,000-
25,000 Å was probably due, in this case, to the large DC bias voltage 100-350 V
and the proximity of an extremely strong RF source. This result suggests that
RF energy can induce a substantive increase in the field emission current
at a constant bias voltage. Why RF energy appears to be able to do this is not
clear.
The last American, that I know of, to reinvestigate Earhart’s work was
Edna Carter in 1914.
34
Carter, like Kinsley, measured her distances with a
Chamberlain compound interferometer, but, unlike the others, her the
experiments were performed at the highest vacuum at her disposal, ~10
-4
torr (1
torr = 1 mm Hg). Due to time constraints, I was unable to pursue this further.
As was mentioned in APPENDIX B of the 1993 Thesis titled “Coherers, a
review” by Thomas Mark Cuff, which is available on ResearchGate, Robert
Williams Wood published his theory of ‘electron atmospheres’ in 1912. The
details of his theory were shown to be wrong. In particular, his idea that these
‘electron atmospheres’ extended for tens of thousands of Ångströms from the
metal’s surface, which he believed were responsible for the low voltage
conduction he observed between metallic edges of scribe marks on mirrors, was
shown not to be in accordance with the facts. Between approximately 1912 and
1922, there arose a cottage industry among physicists who were deluging the
physics community with experimental result refuting Wood’s theory. It was
among some these experiments that I found instances of people redoing or doing
variants of Earhart’s basic experiment. An example of this would be the work of
Englund. In 1914, Carl R. Englund published an account of his repetition of
34
E. Carter; Discharge Potentials across very short Distances; Physical Review; Vol. 3 (2nd
Series); No. 6; January-June 1914; pp. 453-456. [Carter apparently did this work in Europe at the
University of Würzburg under the guidance and with the technical help of one Professor Wien.
The apparatus she employed to allow her to bring the iridium electrodes to subwavelength
distances in a vacuum closely resembles what Franz Rother would ultimately use in his 1926
paper on tunneling and field emission.]
Copyright 1993 Rev. 1.0 23 of 71
Earhart’s experiment.
35
According to Englund this was done at the suggestion
of Robert A. Millikan, while he [Englund] was at the Ryerson Laboratory of the
University of Chicago during the spring of 1911. Sensitive to the criticism that
Earhart’s observed conduction in Region 2 was probably due to the electrodes
being pulled into contact by the large electrostatic forces present at these minute
distances, Englund provided both the movable and fixed electrode with a mirror
thus producing a double fringe system. The end result of all his work was that he
verified Hobbs’ work and by extension Earhart’s work, while at the same time
refuting Wood’s theory.
§4. WHAT EARHART et al. ACTUALLY SAW: ‘VACUUM SPARKS’. - I
have characterized the electrical discharges seen by Earhart in Region 2 as
being due to ‘vacuum sparks’, but I have not clarified what I mean by this term. It
is time to do so. The first point that needs clarification is the term ‘discharge’.
Earhart, Kinsley and Hobbs slowly increased the voltage across the closely
spaced electrodes - thus avoiding any artifacts from the time lag effect - until the
needle on their respective Weston voltmeters started to dance about, indicating a
transient current flow; Shaw employed a telephone receiver shunted by a low
value resistor as his detector.
In Region 1 (~3-5λ), this transient current flow represented the onset of a
glow discharge, where - its name not withstanding - a glow discharge was a
steady ionization of the gas between the electrodes accompanied by a steady
muted light. The onset of the glow discharge marked the border between a
Townsend discharge and a glow discharge; in a Townsend discharge, the
applied electric field was less than what was needed to produce a self-sustained
discharge in the gas, but was enough to support transient discharges caused by
external sources of ionization, such as cosmic ray, natural radioactivity, etc.
In Region 2 (<3-5λ), the discharges Earhart et al. observed were
recognized to be of a different character than the gas discharges observed in
Region 1. In fact, what they had actually observed were ‘vacuum sparks’ in
which field emission, a low current phenomenon, transitioned into a transient
high current discharge similar to an electric arc. As we shall see later on, it was
not until the work of Franz Rother that this became clear.
At this point I need to digress and provide some substantive discussion of
‘vacuum sparks’ and their characteristics. The linearly decreasing discharge
voltage observed by Earhart et al. in Region 2 appears to be a threshold
phenomenon, i.e., a transient burst of current flows only when the applied voltage
attains a certain value, which depends on the electrode separation in a linear
manner. This burst of current must be both large in intensity and reasonably long
lived in duration to register on a Weston moving coil voltmeter or a telephone
35
C.R. Englund; Note on the Electron Atmosphere (?) of Metals.; Philosophical Magazine
[and Journal of Science]; Vol. 27 (6th Series); January-June 1914; pp. 457-458.
Copyright 1993 Rev. 1.0 24 of 71
receiver. Also, these discharges, as we shall see shortly, do generate a blue-
white flash of light rich in ultraviolet (UV) radiation. The first recorded
observations of the ‘vacuum spark’ phenomenon were associated with various
investigations into the production of x-rays.
The discovery of x-rays in 1895 by Wilhelm Conrad Röntgen
understandably electrified the scientific and medical communities, and caused
scientists all over the world to drop what they were doing and try their hand at
this new phenomenon.
36
It should be noted that not everyone partook of the
religion of Röntgen rays - Lord Kelvin, for example declared that, “X-rays will
prove to be a hoax”.
37
In America, Henry Augustus Rowland was
experimenting with various Crookes tubes in an effort to gain firsthand
knowledge of these new rays. In particular, he was interested in localizing the
source of these rays in terms of the electrodes and envelope of the tube. As a
result of this work, he and some other researchers published a series of short
papers on their findings. The most significant paper, from our present
perspective, was a one page note in the journal, Electrical World (New York). In
this April 25, 1896 note, Rowland et al. mention that when the metal electrodes of
36
Note, the phenomena of tunneling and field emission did not show themselves in the
operation of the standard cold cathode x-ray tubes (a.k.a. Hittorf-Crookes tubes). The operation
of these types of tubes were neatly accounted for in terms of gas discharges, only. At
atmospheric pressure, the air in the tube acted like an insulator. As the air pressure in the tube
was lowered, the gas began to conduct and exhibited the luminous manifestation of this
conduction known as the glow discharge; the glow discharge being composed of many luminous
and dark areas. Conduction was predominantly due to cathode rays (electrons) due to their high
mobility compared to that of the heavier, and hence less mobile canal rays (positive ions). The
discovery that cathode rays were actually particles was usually credited to Cromwell Fleetwood
Varley in 1870, even though he appeared to have reached his conclusion with far less than
unequivocal proof. In 1886, E. Goldstein discovered the existence of canal rays
[“…Kanalstrahlen…”] when he put a hole in the cathode, and noticed a luminous glow on the far
side of the cathode, i.e., the side facing away from the anode, radiating through the hole.
It was the collision of the cathode rays with the anode or, in fact, any other material body,
including the glass envelope and the residual gas, in the Hittorf-Crookes tube which produced
the x-rays. As the air pressure was reduced even more, a particular region of the glow discharge
adjacent to the cathode and known as the Crookes dark space appeared and with still lower
pressures, this dark space occupied more and more of the volume of the tube. When the
Crookes dark space filled the whole tube, one had the optimum conditions for the production of x-
rays.
That these types of tubes required the presence of some gas to operate properly was
supported by the following operational observations. The quality, i.e., penetration power or
hardness, of the x-rays increased the longer the tube was operated. This undesirable side effect
was due to the cleanup of the residue gas by the electrodes and glass envelope. As more and
more of the residual gas disappeared, higher and higher voltages were necessary to achieve
significant conduction. At around 4X10
-7
atmospheres, it was no longer possible to pass
discharges through the residual gas with normally available voltages, and so x-rays ceased to be
produced. These early so-called ‘gas’ x-ray tubes - to distinguish them from the Coolidge or
electron x-ray tubes with their ‘hard’ vacuums - usually had provisions for increasing the gas
pressure via slow leaks or arc decomposition of volatile compounds, in order to maintain the gas
pressure in the optimum range over the life of the tube.
37
M. Nicholls; The Perils of Prediction; New Scientist; December 21-28, 1991; pp. 63-64.
Copyright 1993 Rev. 1.0 25 of 71
their x-ray tube were situated about 1 mm apart, the Ruhmkorff discharge was
conducted across the vacuum separating the two electrodes in the form of “…a
faint spark or arc…”; they also mentioned that it was their impression that
copious amounts of x-rays were produced in concert with these ‘vacuum arcs or
sparks’.
38
In what was another remarkable coincidence of scientific discovery,
one Dr. Sydney D. Rowland of 38 Wimpole St., W. London, England, was
reported to have done the following,
Professor Rowland obtained excellent [Röntgen rays] with a
perfect vacuum tube, in which the electrons were placed within
one millimetre from each other. The starting point of the rays is
then less than the one-thousandth part of an inch in diameter,
and gave a shadow of remarkable sharpness.
39
Although, S. D. Rowland does not explicitly mention anything about vacuum
sparks, it was nevertheless clear that his apparatus must have generated them
because: 1) in a tube of “high vacuum”, no gas discharge would be able to occur,
and 2) the extremely small diameter (“less than the one thousandth part of an
inch”) of the source of x-rays - as measured by the apparently insignificant
penumbra cast by the x-ray shadow of an opaque object - was consistent with x-
ray generation via vacuum sparks.
40
Slightly more than a year later in July
1897, Robert Williams Wood, in an article in the the Physical Review, described
the same effect in almost the same words as the two Rowlands.
41
Wood’s
epiphany came about as a direct result of his trying to build a point source x-ray
tube. The doyens of modern field emission, Erwin W. Müller and R.H. Good Jr.,
indicated in their treatise on the subject, that R. W. Woods was the first to publish
38
H.A. Rowland, N.R. Carmichael, L.J. Briggs; Notes on Röntgen Rays; Electrical World
(New York); Vol. 27; No. 17; April 25, 1896; p. 452.
All the above papers can also be found in,
Henry Augustus Rowland; The Physical Papers; The Johns Hopkins Press; 1902; pp. 571-
575. [Rowland pointed out the sometimes unappreciated fact that when energizing an x-ray tube
with a Ruhmkorff coil, one was using a source of AC voltage, specifically a damped sinusoid. For
x-ray tubes of low resistance, the AC current resulted in weak x-rays due to the anode and
cathode flipping back-and-forth between the two electrodes. In the case of an x-ray tube of high
resistance, the Ruhmkorff coil output was so heavily damped that the current was predominantly
only in one direction.]
39
Anon.; Untitled; The British Journal of Photography; Vol. 43; No. 1883; June 5, 1896; p.
356.
40
The book in which I found the reference to S.D. Rowland’s work,
Otto Glasser, Margret Boveri; Wilhelm Conrad Röntgen, and the Early History of the
Roentgen Rays; Charles C. Thomas; 1934; p. 322.
concurred in the assessment that S.D. Rowland’s tube was a forerunner of later ‘auto-electronic’
(i.e., field emission) tubes made by, for example J.E. Lilienfeld.
41
R.W. Wood; A New Form of Cathode Discharge and the Production of X-rays, together
with some Notes on Diffraction; The Physical Review; Vol. 5 (1st Series); No. 1; July 1897; pp. 1-
10.
Copyright 1993 Rev. 1.0 26 of 71
his observations of ‘vacuum sparks’, but from what has been stated earlier this
assertion is clearly incorrect.
42
Note, a complete accounting of the exact mechanisms involved in the
operation of the standard Hittorf-Crookes tube as a source of x-rays has not ever
been made to my knowledge. While it was understood early on that cathode
rays (electrons) produced in the gas discharge generated x-rays when they
collided with the glass walls of the tube or, better still, with the metal anode,
questions still remained about x-rays created by the electrons colliding with the
gas itself and the exact nature of the residual gas clean up during the operation
of the tube. The main reason for these unanswered questions, and others, was
that the Coolidge x-ray tube replaced the Hittorf-Crookes version in very short
order. X-rays were discovered, using the Hittorf-Crookes tube in 1895, but by
1913, William D. Coolidge’s hot filament cathode x-ray tube with its ‘hard’
vacuum had appeared and almost immediately replaced the ‘older’ Hittorf-
Crookes version of x-ray tube.
43
In this respect, the Hittorf-Crookes x-ray tube
42
R.H. Good, Jr., E.W. Müller; Field Emission; in S. Flügge (Ed.); Handbuch Der Physik, Vol.
21; Springer-Verlag; 1956; pp. 176-231, in particular, see p. 176. [Note, so great was, and still is,
the influence of Erwin W. Müller, that subsequent articles, concerned with various facets of field
emission, have also cited R.W. Wood as the discoverer of field emission and/or the ‘vacuum
spark’. See, for example,
W.P. Dyke, J.K. Trolan; Field Emission: Large Current Densities, Space Charge, and the
Vacuum Arc; Physical review; Vol. 89; No. 4; February 15, 1953; pp. 799-808.
W.P. Dyke; Progress in Electron Emission at High Fields; Proceedings of the I.R.E.; Vol. 43;
1955; pp. 162-167.
W.P. Dyke; Advances in Field Emission; Scientific America; Vol. 210; No. 1; January 1964;
pp. 108-116, 118. [Warning, p. 108 of this article referred to, “The German physicist Werner
Schottky…”, when it apparently meant Walter Schottky.]
Note, after having determined that R.W. Wood was not the first person to observe ‘vacuum
arcs’ I almost reversed myself. The popular biography of Wood,
William [B.] Seabrook; Doctor Wood, Modern Wizard of the Laboratory; Harcourt, Brace and
Co.; 1941.
contained an exhaustive bibliography of Wood’s scientific papers and books - 263 of them all
together. The first page of the bibliography contained the following citation,
“16. The X-ray Arc. Electrician, 38 (1896), 289, 371”
which led me to jump to the conclusion that Wood had discovered ‘vacuum sparks’ the same year
as Henry A. Rowland and Sydney D. Rowland. However, upon examining the article called out
in the aforementioned citation, two things became clear: first, the citation was incorrect and
should have read,
R.W. Wood; Experimental Determination of the Temperature inside Vacuum Tubes [Parts I &
II]; The Electrician(London); Vol. 38; pp. 289-290, 322-324.
and second, the article was concerned with the measurement of temperatures inside the various
regions of the glow discharge, not the ‘vacuum spark’.
Warning, Seabrook’s rather breathless biography of the adventures of Robert Williams Wood
made for easy reading, but sometimes at the expense of the facts. Take, for example the famous
incident where Wood exposed N rays as a fake (see the chapter entitled Debunker of Frauds).
This episode, as related in the biography, took a number of liberties with the facts. See,
R.T. Lagemann; New light on old rays: N rays; American Journal of Physics; Vol. 45; No. 3;
March 1977; pp. 281-284.]
43
W.D. Coolidge; A Powerful Röntgen Ray Tube with a Pure Electron Discharge; Physical
Review; Vol. 2 (2nd Series); No. 6; December 1913; pp. 409-430. [The technology, which made
Copyright 1993 Rev. 1.0 27 of 71
shared the same fate as befell the coherer: too short a commercial lifetime to
evince enough scientific interest in the details of its inner workings.
44
While it is not clear to me whether or not Rowland, Rowland and Wood’s
point source x-ray tube ever found commercial success, ‘vacuum sparks’
became an essential tool in the science of spectroscopy. In 1905, R. E. Loving
published his findings on the nature of the ‘vacuum spark’.
45
Besides confirming
the observations of Rowland, Rowland and Wood on the location and
appearance of the ‘vacuum spark and the erosion of the anode with the
concomitant buildup of material on the cathode - the opposite of sputtering,
46
the Coolidge x-ray tube possible, was the ability to fabricate tungsten into filaments. This process
was developed initially for use in electric light bulbs by Coolidge, himself. A fascinating
discussion of the battle waged by Coolidge to domesticate tungsten can be found in,
John W. Howell, Henry Schroeder; History of the Incandescent Lamp; The Maqua Co.; 1927;
pp. 75-122.]
44
The ability of the Hittorf-Crookes tube to clean up the residual gas during operation had a
parallel in the operation of the tungsten filament light bulb. It was known for some time that
modern light bulbs in operation reduced the amount of residual gas present in the glass envelope,
but no one knew where the gases went. This problem came under the scrutiny of Irving
Langmuir, who elucidated much about this process including the location of the occluded gases.
Perhaps his explanations can also account for the clean up that was observed in the Hittorf-
Crookes x-ray tubes. See,
Irving Langmuir; Phenomena, Atoms and Molecules; Philosophical Library; 1950.
Besides the question of precisely what happened to the residual gas upon extended
operation of the Hittorf-Crookes tube in the x-ray mode, there was also the uncertainty about the
source of the x-rays. It was discovered early on that the glass walls produced x-rays when
bombarded by the cathode rays (electrons), and that more efficient generation of x-ray could be
achieved by having the cathode rays impinge on the metal anticathode (anode). But as late as
the mid-1920s, researchers were still collecting experimental data on the radiations given off by
the gas in the tube,
J.J. Thomson; On the Electric Discharge through Gases at very low Pressures; Philosophical
Magazine [and Journal of Science]; Vol. 48 (6th Series); July-December 1924; pp. 1-33.
Idem; Radiation given out by Gases through which Electric Discharges are passing;
Philosophical Magazine [and Journal of Science]; Vol. 49 (6th Series); January-June 1925; pp.
761-786.
A. Dauvillier; Researches on the Electric Discharge in Gases and the accompanying
Radiations; Philosophical Magazine [and Journal of Science]; Vol. 2 (7th Series); July-December
1926; pp. 1046-1052.
J.J. Thomson’s work originated in some observations made by E. Wiedemann in 1895;
Dauvillier’s investigation was inspired in turn by Thomson’s papers. What Thomson and
Dauvillier found was that the gas in the Hittorf-Crookes tube also acted as an anticathode
(anode), and so produced x-rays and Schumann-Lyman radiation (ultraviolet and/or soft x-rays).
Most of radiations from the gas originated in the negative glow and the positive column.
45
R.E. Loving; The Arc in High Vacua; Astrophysical Journal; Vol. 22; No. 5; December
1905; pp. 285-304. [Both R.E. Loving and, in a later paper, Edna Carter, gave Henry A. Rowland
priority for the discovery of the ‘vacuum spark’. See,
E. Carter; The Vacuum-Spark Spectra of the Metals; Astrophysical Journal; Vol. 55; January-
June 1922; pp. 162-164 + 2 plates.]
46
In sputtering, the material of the cathode is eroded by the bombardment of positive gas
ions; the positive gas ions arise due to the ionization of the gas between the electrodes, i.e., the
voltage difference across the electrodes has got to be greater than the minimum discharge
Copyright 1993 Rev. 1.0 28 of 71
Loving was able to empirically prove that the light emitted by the ‘vacuum spark’
was characteristic of the material making up the anode. To accomplish this in an
unequivocal manner, Loving built a vacuum spark gap with the anode and
cathode composed of different metals. The vacuum chamber had a quartz
window to allow light from the violet end of the spectrum to leave, and the
resulting light passed into a grating spectrometer - all Loving’s work was done
with a high voltage DC power supply, and so the electrodes, which were his
anode and cathode, were fixed and well-defined. In one experiment, when the
anode was made of magnesium and the cathode of platinum, no platinum lines
appeared on the photographed spectrum; reversing the electrodes so that the
anode was platinum and the cathode magnesium, produced a spectrum devoid
of magnesium lines. It should be noted that Loving was not the first person to
determine that the ‘vacuum spark’ spectrum was characteristic of the anode
material. In a paper published in 1862, George G. Stokes made the same
observation, and indicated further that William Allen Miller had also done some
work along these same lines.
47
Stokes’ interest in the spectra of ‘vacuum
sparks’ was an offshoot of his work on fluorescence, a phenomenon which was
well known even before his time, for which he provided the first correct
interpretation. Since the fluorescent substances employed by Stokes were best
excited by ultraviolet light, he later went on to study artificial sources of this light
and, hence, found that metal vapor arcs were copious producers of these highly
refrangible rays. While Loving employed the light from the ‘vacuum spark’ only to
investigate the source of the light, Robert A. Millikan and his student Ralph A.
Sawyer coopted Loving’s apparatus for the purpose of measuring the spectra of
metals and other refractory materials.
While the emission spectrum of hydrogen had been completely
characterized up to its highest possible frequency in the UV by Theodore Lyman,
48
it was expected that heavier elements, all of which were metals, would have
emission spectra extending into the far UV and even into the x-ray frequencies.
The problem with verifying these expectations was the lack of a suitable source
voltage for that particular gas. The difference in the way in which the electrodes are eroded
allows one to distinguish between a ‘vacuum spark’ and a gas discharge.
According to J.J. Thomson’s book, Conduction of Electricity through Gases, sputtering was
first observed by Julius Plücker around 1858. In parallel with the sputtering, Plücker also
witnessed the gradual clean up of the gas. However, as the reader has already come to expect,
the history of who discovered what when is a slippery subject. A recent treatise on the subject of
sputtering claims with good justification that William Robert Grove was the first researcher to
report the phenomenon of sputtering around 1853, followed by John Peter Gassiot in 1858. See,
Peter Sigmund; Sputtering by Ion Bombardment: Theoretical Concepts; in R. Behrisch (Ed.);
Sputtering by Particle Bombardment I; Springer-Verlag; 1981; pp. 9-71.
47
Sir George Gabriel Stokes; Mathematical and Physical Papers, Vol. IV; Johnson Reprint
Corporation; 1966; pp. 203-233. [pp. 203-233 contains the abstract, first published in the
Proceedings of the Royal Society (London), of the full paper contained in the 1862 volume of the
Philosophical Transactions of the Royal Society.]
48
The light spectrum of hydrogen consists of lines extending from the infrared (IR) to the
ultraviolet (UV). The line are group in four series named after their discoverers: Brackett (IR),
Ritz-Paschen, Balmer, and Lyman (UV) series.
Copyright 1993 Rev. 1.0 29 of 71
to thermally excite the elements into producing light.
49
That was until, Millikan
and Sawyer resurrected Loving’s ‘vacuum spark’ light source and ensconced it
inside a vacuum chamber together with a Henry A. Rowland concave grating,
thus forming a far UV vacuum spectrometer.
50
Although the ‘vacuum spark’ was
normally tiny, it could be significantly enhanced in intensity by paralleling the
spark gap with large value capacitors, which increased the amount of current
carried by the spark. The fact that this ploy does yield a more robust spark,
substantiates my contention that this was the phenomenon detected by Earhart
49
The use of spectroscopy for the identification of elements - initially just the alkali metals -
was the result of a collaboration between Robert Wilhelm Bunsen and Gustave-Robert Kirchhoff.
An immediate result of this work was the discovery of two new elements, cesium and rubidium.
The so-called ‘Bunsen burner’ was pivotal to Bunsen and Kirchhoff’s spectroscopic research
since it provided them with a convenient source for thermally exciting the various alkali metals
they were investigating. Described in 1857 by Bunsen and Sir Henry Roscoe, the Bunsen burner
utilized the premixing of ~3 parts air with 1 part coal gas (made by carbonization - coking - of
coal, i.e., heating up coal in the absence of air to drive off the volatile fractions) to produce a
flame that was both hot and nonluminous. The nonluminousity of the flame guaranteed that the
only light it produced would come from the externally supplied salts that were used to ‘seed’ the
flame. This light was then passed through a spectrograph, and the resulting emission lines
photographed. Today this type of spectroscopy is still in use for the detection of alkalis and it is
called flame photometry. Cesium was discovered by flame photometry of mineral water by the
process of elimination: the spectral line were assigned to all the known alkalis, and any lines
remaining were judged to be due to a hitherto unknown alkali. Bunsen was then able to isolate
about 17 grams of cesium chloride from ~40 tons of mineral water. See,
Sir. H. Roscoe; Bunsen Memorial Lecture; Journal of the Chemical Society, Transactions
(London); Vol. 77; 1900; pp. 513-554.
It should, of course, comes as no surprise to the more discriminating reader to learn that
Bunsen did not actually invent his burner. The credit for this device is usually given to either
Peter Desdga and/or Michael Faraday.
Note, while the Bunsen burner was a good beginning for a thermal source it had a very real
shortcoming: it was only hot enough (~2000ºK) to excite alkali metals, i.e., elements with small
ionization potentials. Since the emission spectra arose from the transition of excited electrons
falling back to lower energy states, e.g., the ground state, higher temperatures were necessary to
excite the bound electrons of most metals, other than the alkalis. For elements of higher
ionization potential, the electric arc and the high energy spark were used because of their higher
temperatures (“…4000 to 8000ºK or more…”), see,
I.M. Kolthoff, E.B. Sandell, E.J. Meehan, Stanley Bruckenstein; Quantitative Chemical
Analysis, 4th Ed.; The Macmillan Co.; 1969; pp. 997-999.
By the term “high energy spark”, Kolthoff et al. were actually referring to the phenomenon
known to physicists and electrical engineers as ‘vacuum arcs, sparks or discharges’.
50
R.A. Millikan, R.A. Sawyer; Extreme Ultra-Violet Spectra of Hot Sparks in High Vacua; The
Physical Review; Vol. 12 (2nd Series); No. 2; September 1918; pp. 167-170.
Idem; Three Fourths of an Octave Farther in the Ultra-Violet; Science; Vol. 50; No. 1284;
August 8, 1919; pp. 138-139.
R.A. Sawyer; A New One Meter Vacuum Spectrograph Design; Journal of the Optical Society
of America & Review of Scientific Instruments; Vol. 15; July-December 1927; pp. 305-308. [Note,
the idea of using the light from the ‘vacuum arc’ in far UV spectroscopy was rediscovered in 1930
by F. Rother and W.M. Cohn. These two researchers, during the course of working with a field
emission x-ray tube of J.E. Lilienfeld’s design, decided to investigate the nature of the
“Brennfleckstralung [incandescent spots]” produced on the anticathode (anode). Preliminary
results revealed that the spectrum of these spots extended into the far UV, and this led Rother
and Cohn to suggest utilizing this radiation in spectroscopy.]
Copyright 1993 Rev. 1.0 30 of 71
et al. in Region 2, since it means that the currents resulting from these types of
sparks carry enough current that the resulting voltage drop across the gap can
be easily detected by the kind of moving coil meters used by, for example
Earhart. With this device firmly in hand, Sawyer proceeded to obtain the far UV
spectra of every metal he seemed to be able to lay his hands on, and in the
process made a name for himself in the field.
51
Any doubt that the ‘vacuum sparks’ are hot - thermally speaking - should
be dispelled by the following fact, mentioned by Sawyer in his book. In 1936,
Bengt Edlén, using a ‘vacuum spark’, captured the spectrum of Cu (XIX).
52
To
appreciate this feat one must remember that neutral, i.e., un-ionized, copper (Cu)
has 29 electrons, and each time an electron is lost due to the atom being ionized
(or oxidized, in chemistry lingo), the positive charge of the resulting ion increases
by one - Cu (XIX) has lost approximately 1/3 of its normal compliment of
electrons. Note, the orbital structure of copper is assumed to be the following in
order of energy: 1s (2e-), 2s (2e-), 2p (6e-), 3s (2e-), 3p (6e-), 4s (1e-), 3d (10e-),
with orbitals 3s (2e-), 3p (6e-), 4s (1e-), 3d (10e-) comprising the valence band
and containing nineteen electrons (19e-). Thus, it appeared that Edlén was able
to strip all the valence electrons from the copper atom. The shortest wavelength
of UV light produced by a electron reattaching itself to the Cu (XIX) ion was ~47
Å. Calling electromagnetic radiation of this short a wavelength far UV or even far
far UV is definitely stretching the bounds of reasonableness, wavelengths of this
order are best characterized as soft x-rays or - stretching, again - soft soft x-rays.
In fact, the use of the ‘vacuum arc’ by Millikan et al. effectively allowed them to
bridge the gap between the part of the spectrum we call light and the portion
known as x-rays.
53
In this it mirrored the efforts of E. F. Nichols and J. D. Tear
in 1923 to bridge the gap between radio waves and light waves by utilizing a tiny
Righi oscillator.
Returning to the observations of Earhart et al., one could reasonably ask
the following question. Can ‘vacuum sparks’ be formed at atmospheric or near
atmospheric pressures? The case for ‘vacuum sparks’ in the presence of
significant amount of gas is buttressed by the fact that Rowland, Rowland and
Wood observed ‘vacuum sparks’ in tubes evacuated by mechanical pumps,
which could produce at best a vacuum of only 10
-6
atmospheres (~1 mtorr, 1 torr
= 1 mm Hg). The advent of so-called ‘hard’ vacuums (<10
-3
mtorr, <10
-6
torr) had
to await the invention of the mercury diffusion pump, circa 1915, by Wolfgang
Gaede and its independent reinvention and improvement by Irving Langmuir a
51
Ralph A. Sawyer; Experimental Spectroscopy; Prentice-Hall, Inc.;1944.
52
B. Edlén; Na I-ähnliche Spektren der Elemente Kalium bis Kuper, K IX - Cu XIX. [Na I like
Spectra of the Elements Calcium through Copper, Ca IX - Cu XIX.]; Zeitschrift für Physik; Vol.
100; 1936; pp. 621-635.
53
This idea, that Millikan et al. bridged the gap between light and x-rays with the ‘vacuum
arc’, was brought to my attention by the following book,
Herman Goodman; Story of Electricity, and a Chronology of Electricity and
Electrotherapeutics; Medical Life Press; 1928; p. 62.
Copyright 1993 Rev. 1.0 31 of 71
few years later. However, this is still begging the question because I have still
not presented experimental evidence showing that this phenomenon was
observed in air. The best evidence for the occurrence of ‘vacuum sparks’ in air
at atmospheric pressure comes from the work of Lester H. Germer, circa 1950,
which is treated in some detail in APPENDIX F of the 1993 Thesis titled
“Coherers, a review” by Thomas Mark Cuff, which is available on ResearchGate.
Having introduced ourselves to the phenomenon of ‘vacuum sparks’, it is
time for a reality check. To wit, what are ‘vacuum sparks’ and what causes
them? Unfortunately, these are not questions which can be answered
completely, confidently and truthfully, even today. Due to the transient nature of
these types of discharges they were and are difficult to study. However, one
thing that was eventually determined was that ‘vacuum sparks’ were initiated by
field emission. Of course, almost thirty years had to pass between the time of
Earhart’s first paper in 1901 and general acceptance by the physics community
of the existence of field emission. Starting around the late 1920s, researchers
began to find empirical data showing that ‘vacuum sparks’ were preceded by field
emission. This fact was ascertained by the work of at least three groups: Hull &
Burger, Snoddy, and Beams.
54
All three groups observed the behavior of high
potentials across metal electrodes in a vacuum. What they saw was that the
breakdown started as a high voltage, pure electron discharge (field emission),
which then evolved into a low voltage, high current discharge (‘vacuum spark’).
One way to follow this metamorphosis in real time was to use an old but still
reliable technique, the rotating mirror or streak camera. Using this technique,
one could observe the anode lighting up as a result of its bombardment by field
emission electrons, followed by the gradually movement of the luminosity to the
cathode during the formation of the ‘vacuum spark’; the time duration between
the field emission stage and the ‘vacuum spark’ stage of the discharge was on
the order of 1-2X10
-7
sec.
Note, J. W. Beams had earlier shown that a discharge in air followed a
similar course, except that it started with a Paschen (glow) discharge and then
changed into metal vapor arc (‘vacuum spark’).
55
In this case, Beams employed
two different methods to follow the various stages of the discharge: 1) the Kerr
cell method, whose arrangement mimics the methodology utilized by Erich Marx
to measure the speed of x-rays; and 2) the rotating mirror method, which was
discussed in the previous paragraph. With either method, the light from the
discharge was passed through a spectrometer. During the Paschen stage of the
54
A.W. Hull, E.E. Burger; Some characteristics of the discharge between cold electrodes in
vacuum.; Physical Review; Vol. 21 (2nd Series); 1928; p. 1121. [This was only an abstract of
talk.]
L.B. Snoddy; Vacuum spark discharge.; Physical Review; Vol. 37 (2nd Series); 1931; p. 1678.
[This was only an abstract of talk.]
J.W. Beams; Field Electron Emission from Liquid Mercury; Physical Review; Vol. 44 (2nd
Series); November 15, 1933; pp. 803-807.
55
J.W. Beams; Spectral Phenomena in Spark Discharges; Physical Review; Vol. 35 (2nd
Series); January 1, 1930; pp. 24-33.
Copyright 1993 Rev. 1.0 32 of 71
discharge, the spectrograph revealed the presence of the so-called air lines (N
2
emission lines), while the metal vapor arc (‘vacuum spark or arc’) stage was
characterized by the presence of metal lines (metal emission lines).
In 1935, H. W. Anderson did what can best be described as a
macroscopic version of Earhart’s landmark 1901 & 1906 experiment.
56
The ball
and plane electrodes employed by Anderson were placed inside a vacuum
chamber; the separation between the electrodes was on the order of millimeters
and the voltages impressed across them went up as high as 500,000 VDC. For
a fixed separation of the electrodes, Anderson determined the threshold voltage
for breakdown. In Earhart’s experiment, the electrode separations for
measurements in Region 2 were on the order of ~1000 Å and the voltages were
usually <100 VDC. Because the applied voltages, in Earhart’s case, were below
~300 VDC his electrodes did not need to be in a vacuum for reasons we have
already discussed.
One of Anderson’s first results was that the breakdown voltage depended
on the electrode material and/or surface finish, the same result Earhart, Kinsley
and Hobbs had also found for Region 2. Anderson went on, however, to
demonstrate an even more important result, namely that the breakdown voltage
saturated at increased electrode separations. This result was counterintuitive
since it implied that the electric field at the cathode - the site of electron field
emission - decreased with increased electrode separation instead of remaining
constant. Anderson was aware that that it was the electric field which was
responsible for electron field emission. This indicated that while the breakdown
was initiated by electron field emission, it was brought to its conclusion by
another mechanism. Anderson determined that the breakdown was actually the
result of the emission of positive metal ions - perhaps by field desorption - from
the anode in response to the initial burst of field emission electrons from the
cathode. These positive metal ions, which moved from the anode to the cathode
as a result of the applied voltage, increased the field emission by three means: 1)
by reducing the electron space charge between the electrodes; 2) by lowering
work function of the cathode, because of their positive charge, as they
approached the cathode surface; and 3) by producing secondary electrons when
they impacted the cathode surface. Due to their low mobility, compared to
electrons, a single positive metal ion could provide for increased electron field
emission for thousands of electrons, at the least; by measuring the temperature
rise in a thermally insulated electrode, when it was alternately the cathode and
then anode, Anderson showed that the temperature rise was greatest when the
electrode was the anode, implying that most of the current was, in fact, carried by
electrons. The newly released electrons would, in turn, release even more
56
H.W. Anderson; Effect of Total Voltage on Breakdown in Vacuum; Electrical Engineering
(New York); Vol. 54; December 1935; pp. 1315-1320. [For a complete description of the
apparatus, see,
Idem; Apparatus for the Measurement of Breakdown Voltage Between Metal Electrodes in
Vacuum; Review of Scientific Instruments; Vol. 6; October 1935; pp. 309-314.]
Copyright 1993 Rev. 1.0 33 of 71
positive metal ions, which would, in turn, generate more field emitted electrons,
and so forth and so on.
The culmination of all this positive feedback was the production of a
‘vacuum arc or spark’, which resulted in the electrical breakdown of the gap. The
saturation of the breakdown voltage could then be explained as follows. The
electron field emission necessary to initiate the breakdown was always present
since, even though the electric field decreased with increased electrode
separation, the field was always above the minimum needed to cause field
emission. The energy with which the positive metals ions approached and then
impacted the cathode increased with increased electrode separation, since the
total voltage always increased, though, only slowly; the energy picked up by the
positive metal ions in crossing the interelectrode space was, of course,
proportional to the total applied voltage and not the electric field. This paper
provided a beautiful and elegant experimental confirmation of what was
happening, albeit on a smaller scale, in the apparatus utilized by Earhart et al.
§5. FIELD EMISSION UNDERNEATH THE ‘VACUUM SPARKS’ (THE
EUROPEAN CONNECTION). - Although Earhart can rightly be said to have
provided the first quantitative data on the phenomenon of ‘vacuum sparks’ via his
1901 and 1908 papers and the papers that were inspired by these papers, his
later research seemed to concern itself mainly with experiments whose results
were clearly explainable by the normal laws of discharges in gases, i.e.,
Paschen’s law and the like. This was not surprising given that the explanation for
the effects he uncovered at very short distances was not convincingly accounted
for theoretically until the late 1920s, only a few years before he retired. In fact, a
complete explanation of ‘vacuum sparks and arcs’ is wanting, even today. Glenn
Moody Hobbs and Carl Kinsley seemed to have also lost interest in these
electrical discharge experiments at small separations even faster than Earhart
did, because neither of them published any papers on any subjects after 1908.
To be exact, the Science Abstract, Series A showed no citations for either of
these gentlemen between 1909 (Vol. 12A) and 1925 (Vol. 28A). Experimental
interest in electrical conduction at small distances between electrodes appeared
to have pull up stakes and adjourned to Europe, starting about 1910.
On the Continent, Gerhard Hoffmann was the first to verify Earhart’s work,
which he did in 1910.
57
Hoffmann’s work was important in three respects: 1)
because he used a hydraulic actuator using mercury as the working fluid and
later on a magnetic field actuator, see Fig. E4, to control the spacing between the
57
G. Hoffmann; Elektrizitätsüberung durch sehr kurze Trennungastrecken. [Passage of
Electricity across very Minute Air-gaps]; Physikalische Zeitschrift; Vol. 11; 1910; pp. 961-967. Or
see Science Abstracts, Series A; Vol. 14A; 1911; Abstract No. 122. [Hoffmann’s dedication and
love for his instrumentation were no better illustrated than on p. 965 of this article, where he
referred to the electrometer used to measure the discharge voltages as “…einen guten
Kameraden […a good friend…]”.]
Copyright 1993 Rev. 1.0 34 of 71
two electrodes instead of the more traditional lead screw;
58
2) because like all
the previous investigators, he measured the nullpoint (a.k.a. zeropoint, the datum
representing the two electrodes in mechanical contact) before the discharge, but
unlike them he redetermined the nullpoint after the discharge; and 3) because in
the discussion that followed his first paper on this subject, he was asked by Erich
Marx why he had not measured the current through the gap directly? Let us
examine each of these points in some detail.
Hoffmann’s use of a hydraulic and later a magnetic actuator showed that
extremely small movements, free of backlash, could be effected by mechanisms
other than the lead screw. Hoffmann also credited these actuators with allowing
him to be able to revisit the nullpoint. By measuring the change in the nullpoint,
Hoffmann was able to show unequivocally that the discharge eroded the surface
of the electrodes, the amount of erosion of his plane-point electrode pair
depended on the polarity of the applied voltage. This was just what one would
expect of a ‘vacuum spark’ process, where the positive electrode experienced
the greater loss of material. In the case of a plane-point electrode pair, one
would expect that the polarity dependent erosion would be greatest when the
pointed electrode was the anode. Hoffmann’s data did, in fact, show a distinct
asymmetry in electrode wear. Erich Marx’s prescient question about measuring
the discharge voltage and current was answered by Hoffmann in the following
way. Hoffmann indicated that the currents, besides being transient, also
appeared to vary with each discharge even under ostensibly the same conditions
which made them not only difficult to measure, but also of questionable value.
58
In what was an interesting example of what biologists call ‘parallel evolution’, researchers
at the University of Pennsylvania investigating point contact diodes during the Second World War
also built separate hydraulic actuators (again employing mercury) and magnetic actuators to
study the effect of mechanical tapping on these diodes. It had been observed during the course
of manufacturing point contact diodes that their characteristics could be improved by tapping the
diode case with a mallet before filling the structure with a nonhardening wax. The actuators were
used to control the pressure with which the sharpened metal whisker was pressed on the
semiconductor slab in a test fixture in the hope of simulating, in the laboratory, the effects of
tapping by substituting for it mechanical pressure. See,
A.W. Lawson, P.H. Miller, L.I. Schiff, W.E. Stephens; Effect of Tapping on Barrier Capacity;
NDRC 14-???; University of Pennsylvania; September 1, 1943. [Note, this report was originally
classified SECRET, but was declassified by authority of the Secretary of Defense in a memo
written August 2, 1960. The particular copy I have of this report was found in The University of
Pennsylvania Archives, North Arcade, Franklin Field, Philadelphia, PA 19104-6320, (215)898-
7024. I wish to thank Gail Pietrzyk for her help in locating all the NDRC (National Defense
Research Council) reports held in the U. of P. archives.]
Copyright 1993 Rev. 1.0 35 of 71
Perhaps Hoffmann’s most lasting legacy was his verification of Earhart’s
results - results which were very controversial, even though they had been
duplicated by a number of independent investigators. Hoffmann enumerated at
the start of his 1910 paper all the pertinent objections voiced against Earhart’s
methodology, and then proceeded to construct his apparatus so as to obviate
each one of them. The list of objections presented by Hoffmann was as follows:
i) Vibration.
ii) Destruction of the electrode surfaces at the point of
contact used to establish the nullpoint by the
discharge.
Copyright 1993 Rev. 1.0 36 of 71
iii) Impossibility of re-examining the nullpoint after the
discharge due to the severity of the destruction.
iv) Inability of the Earhart configuration to allow for
the measurement of displacement of the fixed
electrode due to electrostatic attraction between the
electrodes.
Hoffmann’s contributions did not stop with his 1910 paper. In 1921, after a
hiatus of more than a decade during which he designed and built a more
sensitive vacuum electrometer, he returned one last time to verify, again,
Earhart’s measurements. This obsessive checking and rechecking was a
hallmark of the really good workers in this particular field: Earhart obtained his
first results in 1901 and then verified and extended them in 1908; Hoffmann, as I
just indicated, verified Earhart’s results in 1910 and then again with some
extensions in 1921; and, as we shall see later on, Franz Rother, who would be
the first to observe pure field emission - as opposed to ‘vacuum arcing’ - using
the Earhart apparatus, obtained his first results in 1911, verified them in 1914
and then rereverified them under even more stringent conditions in 1926.
The next person on the Continent to reexamine Earhart’s work was a
German graduate student named Franz Rother. Using basically the same
apparatus to control the electrode spacing as Earhart, Rother in 1911 did what
no one had yet been able to do, he measured the current flowing between the
two closely spaced electrodes. Rother understood that the transient nature of
the ‘vacuum sparks’ made it difficult to accurately divine their currents. In
addition, he also realized that the ‘vacuum sparks’ changed the electrode
surfaces, and in so doing probably resulted in the arc current varying at each
occurrence. His stroke of genius was twofold: 1) he lowered the DC voltage
applied across the electrodes to the point where the arcs did not occur, and 2) he
apparently assumed that a current would flow even in the absence of any
‘arcing’. Rother’s assumption, that there would be current flowing across the gap
without ‘arcing [“…Lichtbogenbildung…”]’, appeared to be quite daring given that
the concepts of electron tunneling and field emission, in their modern
incarnations, had not yet even been stated. However, as we shall see later on,
the first clear notions about what we today call field emission were only first
enunciated in 1903 by J. J. Thomson, but in strictly classical terms. Note, Rother
averred, at the start of his paper, that the idea for his research originated with his
thesis advisor Dr. Otto Wiener in 1907, but a perusal of the Science Abstracts,
Series A did not reveal any papers by Wiener addressing this phenomenon, at
least directly. Rother’s first paper, which contained all the basic experimental
methodologies and techniques albeit in a very terse form, 3 pages, was
published in 1911 while he was still a graduate student.
59
Three years later,
when he finally earned his doctorate, he published a fairly detailed account of the
59
F. Rother; Der Elektrizitätsübergang bei sehr kleinen Kontaktabständen [Electrical
Conduction at very small Electrode Separations].; Physikalische Zeitschrift; Vol. 12; 1911; pp.
671-674.
Copyright 1993 Rev. 1.0 37 of 71
experimental work, 35 pages compared to the paltry 3 pages in the 1911 article.
60
Although I will not be able to delve into exactly how Rother made the
measurements of these exceedingly small currents due to time constraints, it is
important to, at least, touch on the current measurement abilities of that era.
§6. CURRENT MEASUREMENT CAPABILITIES CIRCA 1900. - The
currents measured by Earhart, Shaw, Kinsley, Hobbs, etc. were relatively
speaking very large since they were due to either glow discharge or a ‘vacuum
spark’ discharge. Rother made use of two methods for measuring very small
currents: 1) the platform galvanometer, which he employed in his later papers;
and 2) the quadrant electrometer method, which he used in his first paper in
1911.
The platform galvanometer (also known as the d’Arsonval movement) was
first presented in a paper in 1882 by Marcel Deprez and Jacques-Arséne
d’Arsonval.
61
This instrument was devised, according to its creators, for the
60
F. Rother; Der Elektrizitätsübergang bei sehr kleinen Kontaktabständen und die
Elektronenatmosphären der Metalle [Electrical Conduction at very small Electrode Separations
and the Electron Atmosphere of Metals].; Annalen der Physik und Chemie; Vol. 44 (4th Series);
1914; pp. 1238-1272.
61
[M.] Deprez, [J.] d’Arsonval; Galvanomètre apériodique de MM. Deprez et d’Arsonval
[Transient galvanometer of Deprez and d’Arsonval]; Comptes Rendus Hebdomadaries des
Seances de l’Academie des Sciences (Paris); Vol. 94; 1882; pp. 1347-1350. [As the reader has,
no doubt, come to expect, Deprez and d’Arsonval were not the first persons to invent the
galvanometer. The progenitor of all galvanometers was discovered around 1819 by Hans
Christian Oersted when he accidentally brought a wire carrying DC current near a magnetic
compass and observed the resulting deflection of the compass needle. All sources agree on the
significance and priority of Oersted’s work, it is on the question of who took the next step that
disagreement appears. According to one source, two months after Oersted announced his
findings, the instrument we now call the galvanometer was invented by Johann Salomo Christoph
Schweigger. See,
Herman Goodman; Story of Electricity and a Chronology of Electricity and
Electrotherapeutics; Medical Life Press; 1928; pp. 27 & 60.
Schweigger’s galvanometer consisted of a magnetized needle suspended somehow and placed
inside a fixed coil of wire. According to a different source,
Forest Klaire Harris; Electrical Measurements; John Wiley & Sons, Inc.; 1952; pp. 44-45.
the galvanometer, consisting of the pivoted magnetized needle arranged inside a fixed coil, was
first described by Claude Servais Matthias Pouillet in 1837; this type of galvanometer was usually
called a tangent galvanometer. An improved version of the tangent galvanometer, called the
moving-magnet galvanometer, was invented by William [?] Thomson in 1858. This version
suspended the magnet from a delicate torsion fiber and surrounded the galvanometer with, for
example an iron cylinder to shield it from the perturbing effects of the Earth’s magnetic field.
According to Forest K. Harris, the moving coil galvanometer was first described by William
Sturgeon in 1836, who apparently began his professional life as a shoemaker; it was improved in
1867 by Thomson when he added a fixed iron core (armature) to the coil; and finally achieved its
modern day form with the addition of pole pieces for the fixed magnet by Deprez and d’Arsonval
in 1882.
Copyright 1993 Rev. 1.0 38 of 71
purpose of detecting small transient currents. Its amazing sensitivity owed much
to its use of a light beam as its pointer or indicator, the torsion fiber suspension
system, and the small gap between the rotating coil and the stationary magnet
due to the shaped pole pieces of the latter. Specifically, this moving coil
galvanometer used a mirror a fixed to the coil to reflect a beam of light from a
stationary light source to a distant, movable scale. As originally conceived by
Deprez and d’Arsonval, the torsion fiber supporting the rotating coil was made of
very fine silver wire [“…à deux fils d’argent…”]. Silver was used undoubtedly
because of its enormous ductility, which allowed it to be drawn down into
extremely fine wires; the silver torsion wires, supporting the rotating coil from
above and below, also served as the electrical connections to the coil.
Unfortunately, silver torsion wires were far from ideal. Modern versions of the
Deprez-d’Arsonval platform galvanometer employed torsion fibers made by
rolling gold (14K or 24K), copper or phosphor-bronze into fine ribbons. Note,
while the use of metal torsion fibers was the rule for engineering instruments, the
highest sensitivity could only be obtained using drawn quartz fibers whose
superior mechanical properties were second to none.
62
The drawn quartz
torsion fibers were usually silver plated to allow the electrical connections to be
made to the rotating coil.
When Rother wrote his magnum opus on field emission in 1926, he
employed a commercially built, high sensitivity, platform galvanometer
manufactured by Hartmann & Braun
63
with a 1 mm deflection equaling 2.6X10
-
11
A (26 pA) on a scale located 2 m away. It was obvious that these commercial
instruments were certainly sensitivity enough to be able to detect field emission
currents which were nominally around 10
-8
- 10
-7
A (10-100 nA). In fact, this
same type of galvanometer would have been able to detect the smaller, by an
order of magnitude, tunneling currents, if it had been so desired. The proof of
this last assertion can be found in a 1904 paper by F. Harms, in which he stated
that he was using a Deprez-d’Arsonval type platform galvanometer manufactured
62
C.V. Boys; Quartz Fibers; Nature; Vol. 42; October 16, 1890; pp. 604-608. [Boys drew his
quartz fibers by the crossbow method, i.e., he heated two piece of quartz until they were soft, one
piece fixed to a lab bench the other on the end of an arrow, and then he touched them together
and fired the arrow which drew out the fiber.]
63
The Braun of Hartmann & Braun - today called Hartmann & Braun AG - was Wunibald
Braun, the eldest brother of Karl Ferdinand Braun. Hartmann & Braun AG started out as a small
instrument making company owned by Eugen Hartmann in the town of Nürtingen. Ferdinand
Braun, who besides being a world class scientist was also a good businessman and recognized a
‘comer’ of a business when he saw it, put his brother in touch Hartmann, and Wunibald became
initially a silent partner in the renamed firm, E. Hartmann & Co. Ferdinand Braun would himself
become involved with the Hartmann & Braun as an idea man. Being a small company, Hartmann
& Braun did not have the capital to invest in research and development of new produces, and so
Ferdinand Braun struck a deal with them to take over some of the R&D work on a consulting
basis. More details can be found in,
Friedrich Kurylo, Charles Susskind; Ferdinand Braun; The MIT Press; 1981; pp. 46-47, 57-
58, 191.
Copyright 1993 Rev. 1.0 39 of 71
by Hartmann & Braun with a sensitivity of 9.04X10
-10
A.
64
Platform
galvanometers continued to be used to measure small currents up until about the
early 1960s.
The current sensitivity of a galvanometer, regardless of its type (pointer
type meters, portable reflecting meters, wall galvanometers, platform
galvanometers, or high sensitivity galvanometers), was given by a number of
different figures-of-merit. For example, Forest K. Harris defined the current
sensitivity as the response/stimulus, where the response was the number of
millimeters (mm) of deflection for a given current expressed in microamperes
(µA). He also used a figure-of-merit called a scale factor, which was the inverse
of the current sensitivity, i.e., it had units of µA/mm and was the number one
multiplied the deflection by to obtain the actual current reading. Note, the
movable scale to galvanometer distance was assumed to have been set to 1 m.
Most authors discussing galvanometers used one of these two figures-of-merit.
65
Harris’ book, which was published in 1952, listed the then current sensitivity
and scale factor for a number of commercially available galvanometers. The
surprising thing about this tabulation was that the best scale factor in 1952 was
0.00001 µA/mm, i.e., 10
-11
A per mm @ 1 m scale distance, which was only
slightly better than that of the galvanometer used by Rother in his 1926 paper.
Thus, it appeared that the ultimate sensitivity of galvanometers was reached
early on, probably before 1926. The limit to the ultimate sensitivity was shown by
Gustav Ising in 1926 to be due to Brownian motion of the rotating coil about its
equilibrium position.
66
Although I will not have the time or space to delve into the subject, I
should like to just mention that the sensitivity of the platform galvanometer could
also be increased by the use of an ultra-high stability, vacuum tube (FP-54
Pliotron), DC amplifier. Using such an arrangement, in 1939, F. R. Abbott and
Joseph E. Henderson were able to make continuous measurements of field
emission currents over the current range, 10
-17
- 10
-5
A.
67
64
F. Harms; Über eine Vorrichtung zur exakten Eichung von Elektrometern für
Elektrizitätsmengen und ihre Anwendung auf die absolute Messung äusserst geringer
Stromstärken [On an apparatus for the exact calibration of electrometers for electrical quantities
and their application to the absolute measurement of extremely small current strength].;
Physikalische Zeitschrift; Vol. 5; No. 2; 1904; pp. 47-50.
65
Walter C. Michels; Electrical Measurements and Their Applications, 7th Ed.; D. Van
Nostrand Co., Inc.; May 1968; pp.20-23 & 32-36. [Caution, even though this is the 7th edition of
this book, it nevertheless contains a significant number of errors and inconsistencies.]
66
G. Ising; Natural limit for the Sensibility of Galvanometers; Philosophical Magazine [and
Journal of Science]; Vol. 1 (7th Series); 1926; pp. 827-834. [Also see,
R.B. Barnes, S. Silverman; Brownian Movement as a Natural Limit to All Measuring
Processes; Reviews of Modern Physics; Vol. 6; No. 3; July 1934; pp. 162-192.]
67
F.R. Abbott, J.E. Henderson; The Range and Validity of the Field Current Equation;
Physical Review; Vol. 56; July 1, 1939; pp. 113-118. [Consult the following articles for detailed
descriptions of ultra-high stability, vacuum tube, DC amplifiers:
Copyright 1993 Rev. 1.0 40 of 71
While the moving coil galvanometer eventually became the instrument of
choice for Rother, it was not the way in which he measured his first field
emission currents in 1911. Rother’s first technique for measuring field emission
currents involved using a quadrant electrometer. The quadrant electrometer,
also known as the Thomson quadrant electrometer, had a long and distinguished
lineage going all the way back to the gold leaf electrometer (a.k.a. electroscope).
The gold leaf electroscope was invented around 1786 by Abraham
Bennet. As the materials technology of insulators evolved during the 19th
century, the electroscope became a more and more sensitive indicator of
electrical charge. The goal of instrument makers and scientists - many times the
same person - was to insulate the metal leaves so well that one could detect a
single electric charge. However, it was soon discovered that no matter how well
one constructed the electroscope, its charge leakage rate could not be brought to
zero; even with the best insulators (amber) and filling the electroscope with dry
filtered air, the charged up leaves invariably lost charge. The first hint as to what
was causing this leakage was provided by a 1879 paper by William Crookes.
68
Crookes built a blown glass electroscope where the metal leaves were
completely isolated from the outside, i.e., they hung on a glass hook which, in
turn, was attached to the inside of the glass envelope. He then evacuated and
sealed off the glass bulb, and finally charged up the leaves by a very clever
process - a process so clever I am still at a loss to explain it. What he found
was that if he immersed the charged electroscope in a water bath that was
grounded, so as to keep excess charge from inadvertently accumulating on the
outside of the bulb, the leaves held their initial charge for more than a year.
Taking a hint from Crookes’ work, other researchers soon discovered that
something was ionizing the air inside their electroscopes and that was why the
charge was leaking off. The ‘something’ turned out to be a combination of
natural radioactivity and cosmic rays. The fascinating story behind this discovery
is, however, beyond the scope of this thesis.
69
As was the case with the galvanometer, the electrometer underwent a
dizzying array of modifications on its way to becoming a truly reliable, useful and
commonplace laboratory instrument. I’d like to explain how Rother adapted a
L.A. DuBridge; The Amplification of Small Direct Currents; Physical Review; Vol. 37; February
15, 1931; pp. 392-400.
A.W. Hull; Electronic Devices as Aids to Research; Physics; Vol. 2; No. 6; June 1932; pp.
409-431.
L.R. Hafstad; The Application of the FP-54 Pliotron to Atomic Disintegration-Studies; Physical
Review; Vol. 44; August 1, 1933; pp. 201-213.
D.B. Penick; Direct-Current Amplifier Circuits for Use with the Electrometer Tube; Review of
Scientific Instruments; Vol. 6; April 1935; pp. 115-120.]
68
W. Crookes; On Electrical Insulation in High Vacua; Proceedings of the Royal Society
(London); Vol. 28; 1878-79; pp. 347-352.
69
C.T.R. Wilson; On the Ionization of Atmospheric Air; Proceedings of the Royal Society
(London); Vol. 68; 1901; pp. 151-161.
Copyright 1993 Rev. 1.0 41 of 71
quadrant electrometer to the task of measuring currents on the order of 10
-14
A,
but I have run out of time.
§7. SCANNING, THE MISSING ELEMENT? - In all of the discussions so
far about the work of Earhart and Rother there has been no mention of the idea
of scanning. This was the one element that would have made their apparatuses
functionally identical with modern versions of the STM. It turns out, though, that
the idea of using piezoelectric actuators to create a raster scanned image of a
microscopically sized area was thought of as early as 1932, if not earlier, in a
slightly different context. This need for scanning arose as a logical consequence
of the near-field microscope, which was proposed as a way to ‘beat’ the Abbe
limit.
The first analytical foray into the question of the resolution of a microscope
was the work of one Ernst Abbe.
70
In a paper published in 1873, Abbe first
expounded his optical resolution model.
71
I have not yet obtained and
translated this paper, but a clear and succinct description of Abbe’s ideas can be
found in an English language paper published in 1906 by Albert B. Porter.
72
Porter’s description was so lucid that I could not improve upon it, so it shall be
presented here verbatim,
70
Ernst Abbe (pronounced “ah-buh”) was a professor of mathematical physics at the
University of Jena, Germany. An alternative criterion for calculating optical resolution was the
Rayleigh criteria, which though derived from different assumptions, yielded the same result.
Rayleigh was, of course, John William Strutt, a.k.a. the 3rd Earl of Rayleigh, a.k.a. Lord Rayleigh.
It must be mentioned that Abbe displayed a sense of social responsibility far ahead of his
time and perhaps of ours. Together with a university mechanic named Carl Zeiss, Abbe in 1866
helped the firm, Carl Zeiss, become world famous for optics. In 1884, Abbe joined Otto Schott
and made the firm, Schott Glass, equally famous. All the while not losing sight of his
responsibilities to the workers in both companies. Besides having secure positions, the workers
shared in the profits, there were pension plans which also covered their families, and in 1900
Schott Glass became one of the first companies to institute an 8 hour work day. Abbe’s position
was best summed up in his own words, “I do not intend to die a millionaire.” And he didn’t, he
willed all his money to the Carl Zeiss Foundation - a charitable foundation. See,
Heinz G. Pfaender, Hubert Schroeder; Schott Guide to Glass; 1983; pp. 11-13.
Friedrich K. Möllring; Microscopy from the very beginning; Carl Zeiss; 1979; p. 63.
71
E. Abbe; Ueber einen neun Beleuchtungsapparat am Mikroscop [On a new illumination
apparatus for the microscope]; Archiv für mikroskopische Anatomie; Vol. 9; 1873; pp. 469-480.
72
A.B. Porter; On the Diffraction Theory of Microscopic Vision.; Philosophical Magazine [and
Journal of Science]; Vol. 11 (6th Series); January-June 1906; pp. 154-166. [Warning, Porter’s
citation of Abbe’s paper in the journal, Archiv für mikroskopische Anatomie, incorrectly gives the
paper’s date as 1837 instead of 1873 - probably a typo.
Porter pointed out that even 32 years after Abbe first proposed his ideas on resolution, they
were still considered controversial by microscopists and were not well known even to physicists.
He also revealed that diffraction can affect even macroscopic seeing depending on the
illumination and the size of the object which was viewed. To this end he demonstrated a very
simple experiment where the human eye replaced the microscope and in which the diffraction
effects could be quite easily shown using a spatial low pass filter.]
Copyright 1993 Rev. 1.0 42 of 71
If a lens is to produce a truthful image of an illuminated object, it
must have an aperture sufficient to transmit the whole of the
diffraction pattern produced by the object; if but part of this
diffraction pattern is transmitted, the image will not truthfully
represent the object, but will correspond to another (virtual)
object whose whole diffraction pattern is identical with that
portion which passes through the lens; if the structure of the
object is so fine, or if the aperture of the lens is so narrow, that
no part of the diffraction pattern due to the structure is
transmitted by the lens, then the structure will be invisible no
matter what magnification is used.
73
As Porter went on to explain, in order to see an object, the object must diffract
the light and then the microscope must have a large enough aperture to collect
the higher order diffraction spectra. If the object was significantly smaller than
the wavelength of the light (~1/2 λ) used to illuminate it, it would not produce a
diffraction pattern and so would be invisible no matter what size the microscope
aperture and magnification. This last point appeared to throw up an
insurmountable barrier to viewing very small objects, at least using light.
However, the insurmountability of this barrier turned out to be illusionary. By
using a clever trick one could significantly extend the resolution of a visible light
microscope.
In an article published in 1928, Edward Hutchinson Synge showed that it
was theoretically possible to image below the Abbe limit by using the following
ploy. He suggested stationing a 10
-6
cm [100 Å] diameter pinhole made in an
opaque screen beneath a thin perfectly parallel embedded and sectioned
specimen at a distance that was a fraction of the hole’s diameter. The pinhole
was to be illuminated from underneath by an intense source of light; any light
which came through the specimen was to be passed through a microscope and
detected by a phototube. The image of the complete specimen was to be formed
by moving the specimen back and forth in a raster scan manner with each
increment of movement being 10
-6
cm.
74
73
Ibid.; p. 154.
74
E.H. Synge; A Suggested Method for extending Microscopic Resolution into the Ultra-
Microscopic Region; Philosophical Magazine [and Journal of Science]; Vol. 6 (7th. Series); 1928;
pp. 356-362
Due to technological limitations Synge’s ideas were not realized during his lifetime and his
original work faded into obscurity. Many years later, his idea was independently rediscovered by
a number of different researchers and the realization of these ideas created the discipline of near
field optics. Synge’s earlier contribution was made known to the microscopy community via a
1990 article by Dennis McMullan,
D. McMullan; The Prehistory of Scanned Image Microscopy, Part 1: Scanned Optical
Microscopes; Proceedings of the Royal Microscopical Society; Vol. 25; No. 2; March 1990; pp.
127-131.
Idem; The Prehistory of Scanned Image Microscopy, Part 2: The Scanning Electron
Microscope; Proceedings of the Royal Microscopical Society; Vol. 25; No. 3; May 1990; pp. 189-
194.
Copyright 1993 Rev. 1.0 43 of 71
This configuration is called a near-field microscope, and it works as
follows. The pinhole should not allow any light to pass through due to its
diameter being only a small fraction of a wavelength of the incident visible light.
But due to strictly classical considerations, an evanescent wave does emerge
from the far side of the pinhole. The specimen is located so close to the far side
of the pinhole, that the diameter of the evanescent wave going through the
specimen is approximately that of the pinhole; the specimen is in the near-field
of the pinhole. Upon exiting the specimen, the evanescent wave, attenuated by
the optical density of the specimen, begins to spread out. The microscope
objective, which is located at a macroscopic distance from the specimen (far-
field), captures the zeroth order (central) fringe of the resulting diffraction pattern.
This central fringe contains the information about the average brightness of the
illuminated spot which is then transmitted to the photoelectric cell located at the
eyepiece. Thus by scanning the sample, a gray scale picture of the scanned
area can be built up.
Initially, Synge proposed using differential screws to move the sample in its own
plane and thus generate the scanned image. But in 1932, he suggested that a
far better method would be to use piezoelectric actuators to do the scanning.
75
Synge said that he got the idea while reading a book by P. Vigoureux. Whatever
his inspiration, the idea was ahead of its time. As I said earlier, I do not know if
Synge was the first person to suggest the use of piezoelectric actuators for doing
raster scanning of microscopic areas. However, it would be amazing if he was
the first given that piezoelectricity had been around since its discovery in 1880 of
the brothers Curie.
§8. FIELD EMISSION & ELECTRON TUNNELING THEORY, THE J. J.
THOMSON CONNECTION. - Surprisingly, the nascent idea of emission of
electrons from cold metals was already out and about in a theory proposed by
Joseph John Thomson. I first became aware of this in a 1910 paper by Elmer H.
Williams.
76
On pages 218 & 219 of this paper, Williams stated that J. J.
75
E.H. Synge; An Application of Piezo-electricity to Microscopy; Philosophical Magazine
[and Journal of Science]; Vol. 13 (7th Series); No. 83; February 1932; pp. 297-300.
76
E.H. Williams; The Nature of Spark Discharge at very small Distances; Physical Review;
Vol. 31 (1st Series); July-December 1910; pp. 216-240. [Williams indicated that he had read
these ideas in the following book by Thomson, Conduction of Electricity through Gases; p. 456.
Unfortunately, Williams did not indicate which edition of the book he was referring to. This was
and still is a very famous book and went through three editions - four if you include the prequel,
J.J. Thomson; The Discharge of Electricity Through Gases; Charles Scribner’s Sons; 1898.
It turned out, that the passage quoted by Elmer H. Williams came from the 2nd edition (1906)
of J.J. Thomson’s book, and was found in the section detailing the work of Earhart, Kinsley &
Hobbs. This same passage could also be found in the 1st edition (1903) on p. 386 and in Vol. II
of the 3rd edition (1933) on pp. 493-494. In the 1st edition of Conduction of Electricity through
Gases, the only work on discharges between closely spaced electrodes discussed by J.J.
Thomson was the first paper by Robert Francis Earhart. At the end of his description of Earhart’s
Copyright 1993 Rev. 1.0 44 of 71
Thomson had proposed a “…possible explanation…” for conduction between
very closely spaced electrodes. The model proposed by J. J. Thomson included
many of the ideas that we associate with the modern theory of electron tunneling
and field emission: 1) some electrons inside the metal were free to move about in
the same way as gas molecules inside a container, 2) these free electrons could
not normally escape from the metal due to the image force,
77
and 3) the image
force could be counteracted by a strong electric field impressed across the
electrodes. The discharge, arc, spark etc. resulting from this model would be,
according to Thomson, “…entirely carried by the corpuscles [electrons] and no
part of it by the positive ions.” In other words, this type of discharge would be
very different from, for example a glow discharge in which the current was carried
by positive gas ions, electrons and sometimes negative gas ions. J. J. Thomson
was well aware of the minimum potential difference needed to ionize a gas, and
at the same time he was also cognizant of the experiments of Earhart et al.
which, in Region 2, could not be explained away as a gas discharge. It is very
possible that Rother had read or heard of these ideas of J. J. Thomson’s, and
had subscribed to them. Note, these same essentially classical ideas, by which
J. J. Thomson attempted to explain the emission of electrons from cold metals,
were resurrected in 1923 by Walter Schottky in his bid to rationalize the
conduction behavior of closely spaced conductors.
In 1923, Walter H. Schottky suggested that the various phenomena
associated with conduction across closely spaced electrodes might be due to the
emission of electrons from cold metals.
78
Schottky’s scenario for emission of
electrons from cold metals was in its essence a semi-classical theory with the
purely classical portion having already been espoused some twenty years earlier
by J. J. Thomson in his book, Conduction of Electricity Through Gases. Note,
the second footnote at the bottom of p. 67 of Schottky’s 1923 article referenced
work, Thomson presciently stated that this work of Earhart was very important and he hoped it
would be pursued.
After finding Williams’ paper, I also happened upon an earlier article by K.E. Guthe which
made mention of J.J. Thomson’s theory of electron emission from cold metals. Guthe was trying
to enlist Thomson’s ideas on cold emission in support of coherer action, which Guthe was actively
interested in. See,
K.E. Guthe; Coherer Action; The Electrician (London); Vol. 54; November 4, 1904; pp. 92-94.]
77
The method of images as used in the study of electricity and magnetism originated,
according to Maxwell,
James Clerk Maxwell; A Treatise on Electricity and Magnetism, Vol. 1; Oxford at the
Clarendon Press; 1892 (3rd Ed.); p. 245.
with Sir William Thomson (a.k.a. 1st Baron Kelvin of Largs, Lord Kelvin) and first appeared in the
Cambridge and Dublin Mathematical Journal in 1848. Note, William Thomson was knighted in
1866 for, among other things, his work on the Atlantic submarine telegraph cable; he became a
peer of the realm - a Lord - in 1892.
78
W. Schottky; Über kalte und warme Elektronenentladungen [On cold and warm Electron
Emission]; Zeitschrift für Physik; Vol. 14; 1923; pp. 63-106. [An English language translation of
this rather lengthy article can be obtained from the National Translation Center, Library of
Congress, Washington DC 20541. The cost is $35.00, and the ID# of the translation is TT-123-
82. Warning, this was not a very good translation!]
Copyright 1993 Rev. 1.0 45 of 71
one of his earlier articles entitled: Über den Einfluß von Strukturwirkungen,
besonders der Thomsonschen Bildkraft, auf die Elektronenemission der Metalle.
The term Thomson Image Force [Thomsonschen Bildkraft] in this title referred to
Sir William Thomson’s (a.k.a. Lord Kelvin’s) concept of image forces and not, at
least directly, to the explanation for emission of electrons from cold metals
postulated by J. J. Thomson in Conduction of Electricity through Gases. The
semi-classical part of Schottky’s theory came from his use of what we call today
‘quantum levels’ to describe the energy levels of the conduction electrons in the
metal contacts, a concept he borrowed from Niels Bohr. The actual full blown
quantum mechanical description of the processes considered by Schottky would
have to wait until the 1928 paper by R. H. Fowler and L. Nordheim; the Fowler-
Nordheim paper was itself only possible after the 1926 début of Erwin
Schrödinger’s series of papers on wave mechanics published in Annalen der
Physik. 1928 was, in fact, a bellwether year for tunneling theory, since no less
than four papers appeared postulating tunneling as an explanation for such
diverse phenomena as the ionization of hydrogen by an externally applied
electric field (J. Robert Oppenheimer), electron emission from cold metals (R. H.
Fowler and L. Nordheim), and radioactive decay by alpha particle emission
(Ronald W. Gurney and Edward U. Condon, and George [Georgii] Gamow).
79
Note, it is not generally appreciated that Gamow’s paper had certain fundamental
79
J.R. Oppenheimer; Three Notes on the Quantum Theory of Aperiodic Effects; Physical
Review; Vol. 31; 1928; pp. 66-81.
R.H. Fowler, L. Nordheim; Electron Emission in Intense Electric Fields; Proceedings of the
Royal Society (London); Vol. 119 (Series A); 1928; pp. 173-181.
G. Gamow; Zur Quantentheorie des Atomkernes [On the Quantum Theory of Atomic
Nucleus]; Zeitschrift für Physik; Vol. 51; 1928; pp. 204-212. [Gamow acknowledged that his ideas
did not come from a vacuum by referencing both the Oppenheimer and Fowler-Nordheim papers.
Note, a few day before Gamow submitted his paper, two American scientists, Ronald W. Gurney
and Edward U. Condon, delivered a preliminary paper to the journal, Nature, which covered the
same topic (α decay) and postulated the same conclusion (that it was due to tunneling) as
Gamow. See,
R.W. Gurney, E.U. Condon; Wave Mechanics and Radioactive Decay; Nature; Vol. 122;
September 22, 1928; p. 439. [This preliminary paper was followed shortly by a more detailed and
quantitative paper,
R.W. Gurney, E.U. Condon; Quantum Mechanics and Radioactive Disintegration; Physical
Review; Vol. 33 (2nd Series); No. 2; February 1929; pp. 127-140.]
In a talk given in 1969, Condon indicated that Gurney was the ‘idea man’ of the Gurney-
Condon team. See,
E.U. Condon; Tunneling - how it all started; American Journal of Physics; Vol. 46; No. 4; April
1978; pp. 319-323.
Many books and papers failed to mention that the tunneling theory of alpha decay should
rightfully be referred to as the Gurney, Condon and Gamow theory, and not simply the Gamow
theory. To add insult to injury, some citations, which did reference all three people, nevertheless
managed to somehow introduce an error or two into their citing of the Gurney & Condon paper.
For example, Max Jammer referred to them as “…R.W. Gurner [sic] and E.U. Condon…”, see,
Max Jammer; The Conceptual Development of Quantum Mechanics; McGraw-Hill Book Co.;
1966; p. 247, footnote 191.]
Copyright 1993 Rev. 1.0 46 of 71
errors in it, which were pointed out and corrected in a later paper by Max von
Laue.
80
§9. TUNNELING IN SOLIDS: COHERERS AND SUPERCONDUCTIVITY.
- Franz Rother extended the work of Earhart et al. to very high vacuums in 1926
using essentially the same apparatus.
81
Starting slightly after the mid 1920s
and following an entirely different line of reasoning having to do with the the
elucidation of the mechanism behind the demodulation (detection) of AM radio
signals by galena point contact diodes, H. Pélabon in France, and later Bruno
Benedetto Rossi together with G. Todesco in Italy, built lead screw driven
contrivances for bringing metal electrodes - usually a ball and a plane - to close
approaches on the order of the distances attained by Earhart, but without the
benefit of a Michelson interferometer to double check the distance (for more
details see the main section entitled WHAT IS KNOWN (PRE-1970)). Coincident
with all this other work, the 1920s also witnessed the first realization of electron
tunneling by squeezing a MOM (Metal-Oxide-Metal) junction, which grew out of
an observation during the course of work at low temperatures.
Heike Kamerlingh Onnes’ pursuit of ever lower temperatures produced the
unexpected phenomenon of superconductivity in 1911, when he discovered that
mercury lost all measurable electrical resistance at ~4.2ºK.
82
Superconductivity
was unexpected in that some of then current theories actually predicted a
parabolic relationship between resistance and temperature. For example,
according to the theory of Lord Kelvin, the resistance would initially drop as the
temperature decreased due to the decreased scattering of the free electrons as a
result of the attenuated thermal motions of the ion cores; eventually, though, the
falling temperatures would cause the free electrons to reattach themselves to
their respective ion core, i.e., freeze out, thus causing the resistance to begin to
increase. The superconductivity of mercury, of course, demolished this theory.
80
M. von Laue; Notiz zur Quantentheorie des Atomkerns [Note on the quantum theory of the
atomic nucleus]; Zeitschrift für Physik; Vol. 52; December 31, 1928; pp. 726-734.
81
F. Rother; Über den Austritt von Elektronen aus kalten Metallen [On the Efflux of Electrons
from cold Metals]; Annalen der Physik; Vol. 81; No. 20; 1926; pp. 317-372, + Plates 2-4.
82
Heike Kamerlingh Onnes won the 1913 Nobel Prize in Physics for his pioneering working
in low temperature physics, specifically for his success in liquefying helium in 1908 and
discovering superconductivity in 1911. It should be noted that while liquefying helium was
certainly a great triumph, just the logistics of obtaining enough gaseous helium to liquefy required
an equally heroic effort. Initially, he petitioned his friend and rival, James Dewar, for a supply of
helium from the natural gas wells at Bath [England?]. Dewar turned him down, saying that, “It is a
mistake to suppose the Bath supply is so great. I have not been able so far to accumulate
sufficient [helium] for my [Emphasis added.] liquefaction experiments.” Eventually, Kamerlingh
Onnes was able to extract a sufficient quantity of helium from monazite sands found in North
Carolina, USA. See,
Kostas Gavroglu, Yorgos Goudaroulis (Eds.); Through Measurement to Knowledge, The
Selected Papers of Heike Kamerlingh Onnes 1853-1926 (Vol. 124, Boston Studies in the
Philosophy of Science); Kluwer Academic Publishers; 1991; p. lv.
Copyright 1993 Rev. 1.0 47 of 71
Extending his investigation to other metals, Kamerlingh Onnes showed that they
too became superconducting at different temperatures close to absolute zero. In
1914, he showed that it was possible to build a superconducting switch using two
pieces of lead lightly pressed together; when this lead-lead sandwich was
exposed to temperatures below the point where lead becomes superconducting,
the switch closes, i.e., becomes superconducting.
83
In all his earlier electrical
experiments at low temperatures, the ends of the lead (Pb) wire forming the
superconducting coil were joined by simply melting together the ends of the lead
(Pb) wire; the superconducting current was started by magnetic induction, after
the lead (Pb) coil was immersed and thermally equilibrated in liquid helium.
When it was first suggested to him that he should try a lead-lead switch in his
experiments, Kamerlingh Onnes was doubtful it would work given that his own
measurements had shown that, at room temperature, the transition or crossing
resistance (due to a foreign film at the surface of the lead contacts) of such a
switch was relatively large. However, it turned out that at liquid helium
temperatures the switch became superconducting, its room temperature behavior
notwithstanding. Kamerlingh Onnes did not pursue an investigation into why this
switch should exhibit such counterintuitive behavior. These questions and others
were investigated in the late 1920s and early 30s by Holm and Meissner.
Employing two crossed metal rods pressed against one another, Ragnar
Holm and Karl Wilheim Meissner
84
attempted to measure whether or not a
transition or crossing resistance [“…Übergangswiderstand…”] existed at the area
of contact between two metal contacts of the same or different metals, this was
their original purpose, not proving the existence of electron tunneling in this
structure. In the crossed rod experiment, the total resistance was the sum of the
spreading or constriction resistance [“…Ausbreitungswiderstand…”] due to the
fanning out or crowding in of the current as it flows between the crossed rods
through the minuscule circular area of contact, and the transition or crossing
resistance, which was postulated to originate at the surface of each metal contact
83
H. Kamerlingh Onnes; Further experiments with liquid helium. L. The persistence of
currents without electromotive force in supra-conducting circuits. (Continuation of J).;
Communications from the Physical Laboratory at the University of Leiden [sometimes referred to
as the ‘Leiden Communications’]; Vol. 141b; 1914; pp. 15-21. [Note, this journal is not easily
found at most university libraries. However, this particular paper can be found in the following
book, which enjoys a more widespread distribution,
Kostas Gavroglu, Yorgos Goudaroulis (Eds.); Through Measurement to Knowledge, The
Selected Papers of Heike Kamerlingh Onnes 1853-1926 (Vol. 124, Boston Studies in the
Philosophy of Science); Kluwer Academic Publishers; 1991; pp. 356-362.]
84
Karl Wilheim Meissner was best known for his experimental work in 1933, together with R.
Ochsenfeld, which led to the discovery of magnetic flux exclusion from the interior of a
superconductor (a.k.a. perfect diamagnetism), the so-called Meissner effect,
W. Meissner, R. Ochsenfeld; Ein neuer Effekt bei Eintritt der Supraleitfähigkeit [A New Effect
at the Onset of Superconductivity]; Die Naturwissenschaften; Vol. 21; No. 44; November 3, 1933;
pp. 787-788.
It was this work which laid the foundation for the microscopic theory of superconductivity as
enunciated by J. Bardeen, L. Cooper, J. Schreiffer and N. Bogoliubov in 1957, the so-called
BCS theory.
Copyright 1993 Rev. 1.0 48 of 71
due to the inevitable presence of oxides, adsorbed or condensed gases or any
other foreign layer [“…Fremdschicht…”]. Because the sought after transition or
crossing resistance was deduced to be of the same order of magnitude as the
spreading or constriction resistance, the rods were cooled with liquid helium, thus
greatly, and in some cases completely, reducing the spreading or constriction
resistance by reducing the bulk resistivity; measurements were also performed at
room temperature for comparison sake. The measured transition or crossing
resistances were orders of magnitude smaller than one would calculate based on
ohmic conduction of the current through the assumed thickness of oxide. In
addition, the transition resistivity was very insensitive to temperature - they
differed only slightly between liquid helium temperatures and room temperature,
not a result consistent with normal ohmic conduction which was not independent
of temperature, depending as it did on the temperature coefficient of the
resistivity of the conducting material. In some cases, the total resistance of the
contact junction vanished
85
In 1973, the Nobel Prize in physics was awarded to Leo Esaki, Ivar
Giaever and Brian David Josephson “…for the discoveries on the phenomena of
tunneling in solids,…” [“…pour les découvertes sur des phénomènes de
tunneling dans des solides,…” ]. In his Nobel Prize lecture, Esaki said the
following about the work of Holm and Meissner. “These measurements probably
constitute the first correctly interpreted observations of tunneling currents in
solids,…”.
86
This statement implied that Holm and Meissner had the obvious
advantage, over earlier workers such as Earhart, in that at the time they did their
work (circa 1930) quantum theory was firmly established as was the theoretical
basis of tunneling; in other words, Holm and Meissner had a theoretical
framework on which to hang their results - Earhart (circa 1901-1908) did not.
However, it should be noted that the two papers by Holm and Meissner (“Holm,
R. and Meissner, W., Z. Physik 74, 715 (1932), 86, 787 (1933)”) cited by Esaki
did not specifically refer in their respective texts to tunneling. As a matter of fact,
it was Ragnar Holm, alone, who “…first correctly interpreted…” their observations
85
R. Holm, [K.]W. Meissner; Messungen mit Hilfe von flüssigem Helium. XIII.
Kontaktwiderstand zwischen Supraleitern und Nichtsupraleitern [Measurements with the help of
liquid helium. XIII. Contact resistance between superconductors and normal conductors].;
Zeitschrift für Physik; Vol. 74; January-March 1932; pp. 715-735.
Idem; Einige Messungen über den Fließdruck von Metallen in tiefen Temperaturen [Some
measurements on the deformation of metal at low temperatures].; Zeitschrift für Physik; Vol. 74;
January-March 1932; pp. 736-739.
Idem; Einige Kontaktwiderstandsmessungen bei tiefen Temperaturen [Some contact
resistance measurements at low temperatures].; Zeitschrift für Physik; Vol. 86; October-
December 1933; pp. 787-791.
86
L. Esaki; Long Journey into Tunneling; in W. Odelberg (Ed.); Les Prix Nobel; Almqvist &
Wiksell International; 1974; pp. 66-83.
Copyright 1993 Rev. 1.0 49 of 71
as examples of tunneling, and he did it in a 1931 article in the journal, Zeitschrift
für technische Physik.
87
Holm started out his article with a description of the crossed rod contact
junction. The contact junctions were formed in a ‘hard’ vacuum. When the two
rods were of the same material and were both clean, they stuck to one another
after being pressed together in the crossed configuration. In fact, the adhesion
between them was so great that an audible “crack” could be heard as they were
pulled apart. Upon the exposure of the two separated rods to air for a fixed
amount of time and after reformation of the contact junction, it was found during
subsequent separations that the adhesion had disappeared. It was thus
surmised that a foreign layer, presumably an oxide, had formed on the surface of
the two rods. For most metals, the thickness of the foreign layer increased with
the length of exposure to the air. The electrical behavior of the crossed rods
depended greatly on the thickness of this foreign layer. Crossed rods with thick
foreign layers exhibited the expected electrical conduction behavior based on the
contact junction resistance being the sum of the spreading or contraction
resistance and the transition or crossing resistance - the foreign layer invariably
possessed a finite resistivity. On the other hand, contact junctions formed with a
thin foreign layer present (perhaps composed of a monolayer of adsorbed gas),
exhibited an anomalously low resistance, all other things being the same; the
resistivity of the foreign layer, which had been determined beforehand on a
thicker foreign layer, was assumed to be constant. The resistance of the thin
foreign layers was an astounding eight orders of magnitude less than was
expected. Not only that, but for certain metals, e.g., tin, the resistance of the
contact junction vanished, i.e., became immeasurably small, at slightly below the
critical or transition temperature, T
c
, at which the metal - but not the oxide or gas
layer - became superconducting. Since the MOM junction was superconducting
for temperatures ~T
c
, a finite current was observed to flow through the junction
without incurring any voltage drop. Today this current is known as a
‘supercurrent’, and its existence would be predicted in 1962 by Brian David
Josephson - many years after it was unknowingly observed.
The fact that the transition or crossing resistance was, to all intents and
purposes, independent of temperature over an ~300ºK range, and the
occurrence of a DC supercurrent in the case of the tin-tin contact junction for
temperatures below 3.6ºK (N.B., for tin T
c
= 3.729ºK) led Holm to speculate that
he and Meissner were seeing a new phenomenon. Holm pointed out that the
temperature independence of the transition or crossing resistance was a feature
of electron tunneling and that it had been predicted in a recent paper by J.
Frenkel.
88
His tunneling theory not withstanding, Holm did leave open the
87
R. Holm; Vorläufige Mitteilung über Metallkontakte mit sehr dünner Fremdschicht
[Preliminary communication on metal contacts with very thin foreign layers].; Zeitschrift für
technische Physik; Vol. 12; 1931; pp. 663-665.
88
J. Frenkel; On the electrical resistance of contacts between solid conductors.; Physical
Review; Vol. 36 (2nd Series); December 1, 1930; pp. 1604-1618. [Frenkel was concerned with
Copyright 1993 Rev. 1.0 50 of 71
possibility that conduction could be due to something less dramatic such as a
mechanical rupture of the foreign layer during the pressing together of the cross
rods followed by metallic contact via one of the resulting pores. We will return a
little later to this possible artifact as it affected Holm & Meissner and subsequent
workers. Note, while Holm & Meissner were able to explain the electrical
conduction of the MOM junctions for temperatures >T
c
in terms of electron
tunneling, they were still at a loss as to why the whole junction became
superconducting for temperatures ~T
c
? It appeared as if, when the metals on
either side of the oxide or adsorbed gas layer became superconducting, they
caused the insulating layer, in turn, to also become superconducting.
This article by Holm was the first paper, I am aware of, which used the
expression tunnel effect [“…Tunneleffekt…”] to describe the ability of a
microscopic material particle to penetrate into a potential energy region in which
it could not normally exist according to classical mechanics. Holm implied that
the expression ‘tunnel effect’ was out and about at the time he wrote this paper:
“…und die Leitungselektronendie Haut infolge des sogenannten Tunneleffektes
durchqueren […and the conduction electrons traverse the (foreign) layer as a
result of the so-called Tunneleffekt].” If Holm was not the first person to use the
term ‘Tunneleffekt’ in the open literature as he implied, then I am not sure who
was. Note, the word ‘Tunnel’ is an Austrian word meaning tunnel, i.e.,
electrical conduction in granular structures such as conducting powders and thin metallic films
made by cathodic sputtering. It was a well known fact that the resistivity of a sufficiently thin
metallic film was measurably larger than the normal bulk value. By sufficiently thin was usually
meant that the thickness of the film was on the order of the mean free path of the free electrons,
this thickness was called the critical distance, d. One explanation, proposed by J.J. Thomson,
was that the free electrons in the thin film sensed the presence of the nearby surface and, in fact,
were scattered by it; in a normal size object the majority of the free electrons at any one time
were never close enough to any of the surfaces to be influenced (scattered) by them. Frenkel,
however, pointed out that if Thomson’s explanation were true, then the critical distance, d, would
have to have the same temperature dependence as the mean free path, i.e., it would have to vary
as 1/T where T was the temperature. Frenkel then made the observation that the critical distance
was found experimentally to be independent of temperature. From this Frenkel surmised that
sufficiently thin metallic films were probably granular with each separate grain communicating
with its nearest neighbors via electron tunneling. For a somewhat more contemporary
discussion, see,
R.M. Hill; Electrical Conduction in Ultra Thin Metal Films. (Report No. 5232); The Electrical
Research Association; 1967; 56 p.
Note, Hill, who discussed sputtered thin films and screened thick films, mentioned that they
exhibited a negative temperature coefficient of resistance (1/R ∂R/∂T < 0) - in the corresponding
bulk metals it was positive. This assertion probably meant that the temperature coefficient of
resistance could be effectively ‘tuned’ by varying the thickness of the films. Going from positive
with very thick films to negative with very thin films and undoubtedly passing through zero for a
film of intermediate thickness.
This idea of Frenkel’s recalled the ‘electron atmosphere’ idea of Robert Williams Wood, see
APPENDIX B of the 1993 Thesis titled “Coherers, a review” by Thomas Mark Cuff, which is
available on ResearchGate. In 1959, Ragnar Holm resurrected Wood’s idea of an electron
atmosphere, but referred to it as an electron cloud,
R. Holm; Electron cloud outside a metal surface; Journal of Applied Physics; Vol. 30; No. 5;
May 1959; pp. 792-793.
Copyright 1993 Rev. 1.0 51 of 71
‘Tunneleffekt’ is not a Germanized version of the corresponding English
expression.
89
Although, Ragnar Holm was Swedish - born in Skara and
educated in Uppsala, both in Sweden - he spent many years working in Germany
for the Siemens company, and so was quite familiar with the German language.
This makes it highly probable that he knew of the Austrian word ‘Tunnel’, and
may have come up with the expression ‘Tunneleffekt’, himself.
It should not go unmentioned that at the time (circa 1930) that Ragnar
Holm was making these landmark and, for their time, radical discoveries, he was
at the chronological age (50 years old; born on May 6, 1879) where most
scientists are out to pasture - intellectually speaking, that is. In addition, his
unmatched book on electric contacts first appeared in 1941 when he was 61
years old, and over the years it underwent may revisions and eventually
translation into English with still further revisions. The last edition of this book,
that I am aware of, was published in 1967 by Springer-Verlag; Holm would have
been then 87 years old. Also, his output of technical papers continued unabated
up to the early 1960s. After coming to America in 1947, he and his wife, Else,
worked as consulting physicists for the Stackpole Carbon Co., St. Marys,
Pennsylvania until their simultaneous retirement in 1964, at which time Ragnar
Holm would have been 84 years old. I do not know exactly when Holm died, but
the biographical entries for both he and his wife, which appeared in the 1966
edition of American Men of Science (11th Ed.), were absent from the 1972
edition (12th Ed.).
90
As was mentioned previously, Holm and Meissner employed crossed
metal wires pressed together in their work. What was left out of most, if not all,
summaries of their research was their familiarity with a use of coherer behavior.
Holm and Meissner were well aware that their experimental arrangement
constituted a single contact coherer. In preparation for some of their
experiments, they would deliberately cohere [“…gefrittert…”] the contact junction
by applying a voltage greater than the ‘critical voltage’ of the particular junction
materials; to prevent excessive currents from flowing, the greater than ‘critical
voltage’ voltage was applied using a capacitor. The reason why they deliberately
induced some of their contact junctions to cohere was - as near as I can figure
out - so that they would have a control against which they could compare their
uncohered contact junctions. This illustrates the caution with which they viewed
the currents crossing the contact junction. They could not dismiss the possibility
that the currents were crossing the contact junction via one or more inadvertent
metallic bridges. The best way to eliminate this prosaic explanation was to
examine two contact junction pairs which were identical in every way except that
one had been deliberately cohered. If these two contact junctions behaved the
89
Harold T. Betteridge; Cassell’s German-English English-German Dictionary; Macmillan
Publishing Co.; 1978; p. 621.
90
The Jaques Cattell Press (Ed.); American Men of Science, Vol. H-K; R.R. Bowker Co.;
1966; p. 2363.
Idem; American Men and Women of Science, Vol. 3, H-K; R.R. Bowker Co.; 1972; p. 2788.
Copyright 1993 Rev. 1.0 52 of 71
same, this would strongly imply that the conduction was not some new
phenomenon, but simply metallic conduction across a bridge. But to understand
how the contact junctions were supposed to behave, and under what conditions,
one needs to make the acquaintance of the Silsbee hypothesis.
Kamerlingh Onnes knew that a closed coil of, say, mercury
91
became
superconducting at the critical temperature, in this case ~4.2ºK, provided that the
test current was very small. He also knew that the critical temperature increased
in the presence of an externally applied magnetic field. It was also known that
with zero external magnetic field, the critical temperature would rise for
sufficiently large test currents. These two phenomena were thought to be
independent, and no one had an idea why the magnetic effect occurred. The
current effect, however, was thought to be due to ohmic heating. It might appear
obvious to us today that these two effects were related, but it should be
remembered that early workers in the field of superconductivity only knew that
the superconducting resistance was very small - not identically zero. With a finite
but very small value of superconducting resistance, ohmic heating effects could
have still been possible and plausible, hence the separation of the magnetic from
the current effects. In 1918, a physicist at the NBS named Francis B. Silsbee
made the bold assertion that the magnetic effect and the current effect were
really one effect due solely to the presence of a magnetic field, in the one case
externally generated and in the other internally produced, i.e., self generated.
92
Experimental support for Silsbee’s hypothesis was forthcoming. Among the
research groups to provide support for Silsbee’s contention was the research
team of Ragnar Holm and Karl W. Meissner.
Holm and Meissner produced mechanical contact junctions between
metals which were known superconductors. Such junctions, cooled below the
lowest critical temperature of the metals which made up the contact junction,
exhibited supercurrents, i.e., they became superconducting. If a similarly
constructed contact junction were first cohered, it also became superconducting.
The difference was that the critical temperature of the two types of contact
junctions (uncohered and cohered) with the same test current were different.
Specifically, the critical temperature of the cohered contact junction was higher
than that of the uncohered contact junction. This was just what one expected
based on Silsbee’s hypothesis, because the test current in the cohered contact
junction had a higher current density due to its crossing between the two metals
through a small diameter metallic fiber or bridge. The higher current density
produced a larger magnetic field - for the same test current - and so increased
the critical temperature.
93
91
The mercury was formed into a coil by the expedient of putting it inside a glass tube
previously formed into the shape of a coil.
92
F.B. Silsbee; Note on electrical conduction in metals at low temperatures; Bulletin of the
Bureau of Standards; Vol. 14; 1918-1919; pp. 301-306.
93
Note, while evidence did accrue in favor of Silsbee’s hypothesis, it was still not an open-
and-shut case. That is, there would still be many instances wherein Silsbee’s hypothesis would
Copyright 1993 Rev. 1.0 53 of 71
Being conscious of the coherer phenomenon, Holm and Meissner were
also careful to limit the test voltage applied to contact junctions, which had not
been cohered, lest the applied test voltage inadvertently caused the junction to
cohere during the actual I-V (current-voltage) measurements. More than a
quarter century later, in 1958, when Hans W. Meissner duplicated the
experiments of Holm & Meissner, the same proscription regarding coherer
behavior was likewise voiced anew: “Any possibility of a formation of a [metallic]
bridge due coherer action is therefore excluded (see reference 3, p. 30).”
94
Although, I would be willing to bet that very few people - technical or otherwise -
in 1958 would truly have understood the previous quotation, especially with
regards to knowing what a ‘coherer’ was. The amazing thing about the renewed
efforts at examining superconducting contact junctions was that they were still
dogged by the uncertainty of whether the experimentally measured currents were
real tunnel currents or just artifacts of metallic bridge conduction. This
uncertainly became most acute upon the observation of DC supercurrents
(significant current flows across the contact junction accompanied by an
immeasurably small voltage difference). Ivar Giaever’s 1973 Nobel Prize lecture
was a good case in point:
In our first paper Megerle and I published a curve, which is
shown in Figure 13, demonstrating such a supercurrent and also
that it depended strongly on a magnetic field. However, I had a
ready-made explanation for this supercurrent-it came from a
metallic short or bridge. I was puzzled at the time because of the
sensitivity to the magnetic field which is unexpected for a small
bridge, but no one knew how a 20Å long and 20Å wide bridge
would behave anyway.
95
be contradicted. However, so great was the intellectual appeal of Silsbee’s hypothesis, that the
negative evidence would usually be explained away in terms of some presumed nonideal
characteristics of the material under investigation. A good example of this can be found in the
1960 paper by Jacques I. Pankove,
J.I. Pankove; Superconducting Contacts; IRE Transactions on Electron Devices; Vol. ED-7;
1960; pp. 137-141.
In this paper, Pankove measured the critical magnetic intensity, H
c
, needed to quench (destroy)
the superconducting behavior of a superconducting MOM structure made of crossed niobium
wires, and found it to be 1/10 the published value for niobium. This discrepancy, he attributed to
“…impurities or strains in the wire.”
94
H. Meissner; Measurements on Superconducting Contacts; Physical Review; Vol. 109; No.
3; February 1, 1958; pp. 686-694. [The coherer quote can be found on p. 687. A similar
proscription can also be found on p. 673 of Meissner’s 1960 followup paper,
Idem; Superconductivity of Contacts with Interposed Barriers; Physical Review; Vol. 117; No.
3; February 1, 1960; pp. 672-680.
Note, I have not been able to find any indication that Karl Wilheim Meissner (1891-1959) and
Hans Walter Meissner (1922 - ) were related to one another.]
95
Ivar Giaever; Electron Tunneling and Superconductivity; in W. Odelberg (Ed.); Les Prix
Nobel; Almqvist & Wiksell International; 1974; pp. 86-102.
Copyright 1993 Rev. 1.0 54 of 71
Note, in the case of a normal tunnel current, i.e., a current which did require a
measurable finite voltage to drive it across the contact junction, Giaever was far
more certain that they were not a metallic bridge artifact, because these normal
tunnel currents were, for the same voltage, proportional to the active area of the
MOM structures. Since Giaever formed his junctions by Joule evaporating one
metal layer on top of the other through a mask, he had very direct control over
the contact area; after the first metal layer was deposited, it was usually either
exposed to room air for a fixed time to grow a layer of native oxide or heated in
an air furnace to force grow a thicker oxide layer.
Eventually, the general consensus was that the experimentally measured
supercurrents were not artifacts due to metallic bridges. Rather, that they were a
bona fide phenomenon predicted by Brian D. Josephson and called the DC
Josephson effect. That this explanation of the observed supercurrents was still
not the whole story took some time to come out.
§10. THE ORIGINS OF DC SUPERCURRENTS: TUNNELING vs.
METALLIC BRIDGES. - The unwanted metallic bridges, that superconducting
MOM structures were heir to, were eventually studied for themselves. One can
trace this effort back to Jacques I. Pankove’s papers on the nature of the
supercurrents in squeezed superconducting MOM structures.
96
Note,
Pankove’s first two papers cited as their first reference the 1958 Physical Review
paper by Hans Meissner, thus maintaining a clear paper trail all the way back to
the ground breaking work of Heike Kamerlingh Onnes. Pankove utilized the
crossed wire technique, made famous by Holm and Meissner, to produce the
squeezed MOM structures; the squeezing was simply produced by pressing the
two crossed wires against one another with a known amount of force. Pankove’s
interest in these squeezed superconducting MOM structures arose from the fact
that such devices were bistable and could be made to switch quickly from a low
resistance to a high resistance state by a suitable control current, which, it was
thought, caused the metals to go from the superconducting to the normal state.
96
J.I.Pankove; Superconducting Contacts; IRE Transactions on Electron Devices; ED-7;
1960; pp. 137-141.
Idem; New Effect at Superconducting Contacts; Physics Letters; Vol. 21; No. 4; June 1, 1966;
pp. 406-407. [Caution, there are errors in two of the citations contained in the reference section
of this paper: “J.I. Pankove, IRE Transactions PGED-7 (1960) 137.” should read “J.I. Pankove,
IRE Transactions ED-7 (1960) 137.”, and “B.J. Josephson, Physics Letters 1 (1962) 261.” should
read “B.D. Josephson, Physics Letters 1 (1962) 251.” In addition, the reader with average
patience and eyesight should be forewarned that Fig. 1 of this paper is both incredibly busy and
ridiculously small (especially Fig. 1b).]
Idem; Superconducting Contacts in High Magnetic Fields; Physics Letters; Vol. 22; No. 5;
September 15, 1966; pp. 557-558. [Note, these appeared to be the only three papers on
superconducting devices that Pankove wrote. Concurrent with his transient interest in
superconducting devices, there developed a more long term interest in semiconductor devices,
especially electro-optical solid state devices such as GaAs injection lasers. This shift in interest
can be clearly seen by simply perusing the entries under ‘Pankove’ in the author index of Science
Abstracts, Series A from the years 1966 to 1969.]
Copyright 1993 Rev. 1.0 55 of 71
The hope was that such contrivances could be employed in both analog and
digital circuits. In this respect, Pankove’s work on these components at RCA
Labs, Princeton, New Jersey, in the 1960s presaged the massive, but ultimately
unsuccessful effort, launched by IBM to harness the Josephson junction as a
commercially viable product; the IBM effort was effectively terminated in 1983.
97
That both the RCA and IBM efforts failed to yield commercially competitive
devices based on Josephson junctions was a testament to the immense chasm
separating the research laboratory from the shop floor and the fact that new does
not necessarily imply better as has been adequately demonstrated by the
dominance of silicon semiconductors devices in the face of all comers.
In the introduction to his 1960 paper, Pankove listed the advantages of a
junction made of crossed superconducting wires: 1) Because of the smallness of
the junction, the current density through the junction will be very high, even for
minute currents. This high current density implied that the critical current density
could be reached easily with the normal currents that flowed through the junction.
At the critical current density, the superconducting junction will be quenched, i.e.,
it will behave as if the two wires were normal, as opposed to superconducting,
metals. 2) Since the junction was so small, the transition between the
superconducting and the normal states, and vice versa, should be very rapid. 3)
And finally, due to the superconducting nature of the junction and its minute size
with its commensurately small current, the power dissipation should be very low.
Pankove constructed his crossed wire junctions using “…two freshly
etched 3-mil-diameter niobium wires…” maintained at 4.2ºK (the critical or
transition temperature, T
c
, at which niobium becomes superconducting, being
8.70ºK). One might reasonably ask why the wires needed to be “…freshly
etched…”, and how one would accomplish such etching? However, one would
find oneself rebuffed by the lack of any explicit explanation in the paper. A hint
can be found, though, in Brian D. Josephson’s 1973 Nobel Prize lecture (p. 112),
where he tells the story of how Paul Wraight related the fact that one cannot
solder niobium using ordinary solder, to the existence of an oxide layer on the
surface of the niobium wire.
98
Thus, I believe that Pankove’s use of “…freshly
etched…” niobium wire had to do with the fact that niobium wire, which had sat
around, probably had too thick an oxide layer to allow electron tunneling and so
had to be etched to reduce the thickness of the oxide.
It was apparently while observing the I-V plots of some of these crossed
wire junctions that Pankove discovered an anomaly which led him to deduce the
existence of a new type of DC supercurrent, carried by what what he
amorphously referred to in his June 1966 paper as a “…saturable
superconductor bridge…”. That he was actually referring to a metallic bridge
97
E.W. Pugh; Technology Assessment; Proceedings of the IEEE; Vol. 73; No. 12; December
1985; pp. 1756-1763.
98
B.D. Josephson; The Discovery of Tunnelling Supercurrents; in W. Odelberg (Ed.); Les
Prix Nobel; Almqvist & Wiksell International; 1974; pp. 106-113.
Copyright 1993 Rev. 1.0 56 of 71
connecting the two metals was also explicitly asserted by N. I. Bogatina and I. K.
Yanson in the first paragraph of their 1973 paper in which they investigated the I-
V characteristics of such metallic bridges.
99
Pankove’s assertion that the
supercurrents he observed were anomalous was based on the following
observations: 1) The DC supercurrents persisted for bias voltages at least a
decade larger than 2∆.
100
2) While a portion of the DC supercurrent could be
eliminated by quenching the junction, there existed an unquenchable (by
current)
101
DC supercurrent in some superconducting crossed wire junctions. 3)
This unquenchable (by current) DC supercurrent saturated at high currents. To
understand why DC supercurrents having these three characteristics were
considered to be anomalous, we need to digress for a brief moment and talk
about the DC Josephson effect.
In 1962, a British graduate student in physics wrote a very terse,
mathematically intensive, and impenetrable theoretical paper on tunneling in
superconducting structures.
102
Among the effects predicted in this paper was
the existence of a DC supercurrent in superconducting structures composed of
essentially two superconductors separated from one another in such a way that
they were loosely coupled. It must be noted that by defining a Josephson
junction as two superconducting reservoirs loosely coupled together, the number
of admissible structures becomes quite large and is not limited to just MOM
structures composed of superconducting metals: 1) MOM structures, in which
both metals were superconductors - the traditional, vanilla-flavored Josephson
structure; 2) MnMM structures formed by sandwiching a thin layer of a
nonsuperconducting metal, say copper or gold, between two superconducting
metals (extensively studied by Hans Meissner); 3) Dayem bridges, produced by
99
N.I. Bogatina, I.K. Yanson; The nonlinearity mechanism of the volt-ampere characteristics
of point junctions; Soviet Physics JETP; Vol. 36; No. 4; April 1973; pp. 692-696.
100
2∆ represents the energy gap found in the ground state of a superconducting metal as
predicted by the BCS (Bardeen, Cooper and Schrieffer) theory of superconductivity. In a metal in
the normal, i.e., nonsuperconducting, state at T = 0ºK, there is no energy gap between the
electrons at the highest occupied energy state (the Fermi level) and the next highest unoccupied
energy level. In this situation, an infinitesimal amount of excess energy will be able to promote an
electron into the next highest unoccupied state. Once the same metal becomes superconducting,
though, a finite energy gap is created between the highest occupied energy state and the rest of
the occupied states. As usual there are exceptions, i.e., the so-called gapless superconductors,
which are differentiated from the ordinary superconductors by not having an energy gap, 2∆, in
the superconducting state.
101
The superconducting state of a MOM junction can be destroyed (quenched) by either
applying an external magnetic field or, according to Silsbee’s hypothesis, by the magnetic field
arising from the current through the junction, itself. However, Pankove’s use of the term
‘quenched by current’ refers not to current going through the junction, but to current passing near
the junction.
102
B.D. Josephson; Possible New Effects in Superconductive Tunnelling; Physics Letters;
Vol. 1; No. 7; July 1, 1962; pp. 251-253. [That reading this paper was an extremely arduous task
for all but the hardiest of souls has been attested to by numerous technical personae including
Ivar Giaever and Charles Kittel.]
Copyright 1993 Rev. 1.0 57 of 71
linking two large pads of the same superconducting metal via a thin narrow
bridge (“…2µ long 1/10µ wide”) of the same superconducting metal; ….
103
Josephson, himself, attempted to experimentally measure his predicted DC
supercurrents, but he was unsuccessful.
104
According to Josephson’s theory, DC supercurrents flowed only when the
voltage across the Josephson junction was identically zero. In addition, these
current were also predicted to be very sensitive to the presence of external
magnetic fields, so sensitive that even the Earth’s magnetic field could cause
problems if it was not neutralized by a Helmholtz coil. Because of time and
space constraints I will not be able to adequately cover the Josephson effects.
The metal filaments, which caused such vexation for the early researchers
working on electron tunneling, were both studied and used to advantage only a
few years later. In 1975, I. K. Yanson published a conference paper in which he
attempted to confirm a theoretical expression for the resistance of such small
diameter structures and at the same time used the current passing through these
types of metal filaments to study the electron-phonon interaction.
105
Briefly,
Yanson constructed the same type of metal-dielectric-metal planar sandwich as
Giaever; the dielectric layer thickness was from ~10 to 100 Å and the metal
layers were some several thousands of Ångströms thick. In order to produce the
this-time-desired metal filaments, the sandwiches were cohered at liquid helium
temperatures by one of three methods: 1) applying a voltage, greater than the
critical voltage, across the sandwich using a capacitor in series with a current
limiting resistor; 2) applying a steady DC voltage across the sandwich using a
voltage source with a large internal resistance; or 3) simply using a sandwich
which had spontaneously cohered. Note, the nuts-and-bolts experimental details
on how the sandwiches were prepared and tested was not to be found in this
1975 conference paper, which was an extremely terse and fragmentary retelling
of the events. Rather, this information was gathered by daisy chaining
backwards through Yanson’s earlier papers.
103
J. Clarke; The Josephson Effect and e/h; American Journal of Physics; Vol. 38; No. 9;
September 1970; pp. 1071-1095.
104
B.D. Josephson; The Discovery of Tunnelling Supercurrents; in W. Odelberg (Ed.); Les
Prix Nobel; Almqvist & Wiksell International; 1974; pp. 106-113. [As an example of an I-V curve
containing a DC supercurrent, Josephson presented a curve by Nicol, Shapiro and Smith taken
using a MOM superconducting structure. Warning, in citing the paper from which this curve was
copied, Josephson incorrectly identified it as “…reference 6).”. The correct citation should have
read “reference 12).”, which was the following paper,
J. Nicol, S. Shapiro, P.H. Smith; Direct Measurement of the Superconducting Energy Gap;
Physical Review Letters; Vol. 5; No. 10; November 15, 1960; pp. 461-464.]
105
I.K. Yanson; Non-linear effects in electrical conductivity of point contacts and electron-
phonon interaction in normal metals.; in Matti Krusius, Matti Vuorio (Eds.); Proceedings of the
14th International Conference on Low Temperature Physics, Otaniemi, Finland, August 14-
20,1975, Vol. 3, Low Temperature Properties of Solids; North-Holland Publishing Co.; 1975; pp.
506-509. [The Library of Congress call number is QC 278 I498.]
Copyright 1993 Rev. 1.0 58 of 71
Yanson’s first paper on this subject appeared in 1973.
106
By way of an
introduction to this paper, Yanson mentioned that various other researchers had
been studying these “…narrow conducting bridge[s]…” formed when MOM
(Metal-Oxide-Metal) sandwiches were broken down due to the application of
excessive voltage. Of particular interest to these researchers was the resistance
of the metallic bridges as a function of the applied voltage used to measure their
resistance. Yanson mentioned that the bridge current versus voltage (I-V) plot
was nonlinear if one or both of the metals making up the MOM structure were in
its superconducting state as opposed to its normal or nonsuperconducting state,
where the I-V plots reverted to being linear. Before continuing, it is incumbent
upon me to remark that Yanson made no mention of the fact that the
phenomenon he was studying was an example of a coherer; the word ‘coherer’
did not appear in this 1973 paper, the 1974 paper we will review shortly or in the
previously cited 1975 conference paper. The usual explanation for the
nonlinearity in such I-V plots was, as one might expect, of a thermal nature. As a
crude analogy, consider the tungsten filament of the common incandescent light
bulb. For small voltages («110 VAC) the resulting I-V plot would be visibly linear.
However, as the applied voltage was increased the resulting increase in the
current would cause the filament to heat up. Since like most metals, tungsten
has a positive temperature coefficient of resistance (1/R ∂R/∂T > 0), its hot
resistance will be greater than its its cold resistance, and this will cause the
resulting currents to be less than they would be if there was no self-heating or if
the filament was externally cooled in such a way as to maintain its temperature at
ambient.
Yanson’s next paper in 1974 dealt almost exclusively with the use of the
cohered sandwiches being used to study electron-phonon interactions.
107
It
was in this paper that Yanson provided the reader with the only detailed récipé
of how he created the conducting bridges (see section 3. EXPERIMENTAL
PROCEDURE of this 1974 paper). Among the details made public by Yanson
was his use of the organic compound adenine in the formation of the oxide layer
of the MOM structure. A small amount of adenine was apparently always
included in the oxide layer to prevent the formation of spontaneous shorts
(bridges). The startled reader might well ask why adenine (C
5
H
5
N
5
), an aromatic
heterocyclic, almost planar, organic molecule of the class purine and most
famous as one of the five building blocks (uracil, thymine, cytosine, adenine and
guanine) of nucleic acids such as DNA and RNA, had this ability to prevent
shorts in MOM structures and who first discovered this useful property? A
question which Yanson conveniently forgot to answer. The nature of these
106
N.I. Bogatina, I.K. Yanson; The nonlinearity mechanism of the volt-ampere characteristics
of point junctions; Soviet Physics JETP; Vol. 36; No. 4; April 1973; pp. 692-696. [The journal
Soviet Physics JETP is a direct translation of the Russian language journal Zhurnal
Éksperimental’noi i Teoreticheskoi Fiziki.]
107
I.K. Yanson; Nonlinear effects in the electric conductivity of point junctions and the
electron-phonon interaction in normal metals; Soviet Physics JETP; Vol. 39; No. 3; September
1974; pp. 506-513.
Copyright 1993 Rev. 1.0 59 of 71
spontaneous bridges was also not made clear. That is, were they the result of
the MOM junction being inadvertently cohered by perhaps accumulate static
charges, or were these bridges the result of spontaneous growth of metal
whiskers?
During the Second World War, the makers of military radios experienced
failures in their equipment of a novel type: the air dielectric, variable capacitors
used for tuning the radios to different frequencies were found to have
spontaneously shorted out. After much head scratching, it was discovered that
microscopically thin metal whiskers were growing out perpendicularly from the
surface of the capacitor plates. Eventually these whisker grew long enough that
they reached the opposite plate and so shorted out the capacitor. The whiskers
originated from and were composed of cadmium, which had been electroplated
onto the capacitor plates to prevent corrosion. Physically, a typical whisker was
between 0.00001 to 0.00005 inches in diameter, could be as long as 0.040
inches, and were remarkably straight. Electrically, “…the resistance of one
crystal [whisker], approximately .020 in. long and .00003 in. diameter, is of the
order of magnitude of 300 ohm.”
108
Howard L. Cobb, who first chronicled these
events and observations, mentioned that, at the time, there were precious few
facts to be had on the subject of the growth of metal whiskers. In fact, research
on this subject did not start in earnest until the channel filters, used in the Bell
Telephones multi-channel telephone line, began to fail due to shorts caused by
the zinc whiskers growing out of the zinc plating used on the electronic
components in the filters as related in a 1951 article coauthored by Karl G.
Compton.
109
Compton et al., as a result of their literature search, found that the
only other article on the subject of metal whisker growth was the 1946 article by
Cobb. With such a dearth of background information, Compton et al. decided
that they had no choice but to initiate the study of the topic themselves. Their
results, while interesting, were equivocal. If these metal whiskers had simply
remained only a nuisance to electronic equipment, probably not much more basic
research would have been done on them. But in 1952, W. Conyers Herring and
John K. Galt published the results of their investigation into the mechanical
strength of tin whiskers “…grown from a tin-plated surface…”. What they found
was that the tensile strength of these whiskers was over 1000 time greater than
that of ordinary bulk tin crystals.
110
The rest, as they say, is history.
111
An objection to assuming that whisker growth was the mechanism behind
spontaneous cohering of MOM structures might go along the following lines. The
108
H.L. Cobb; Cadmium Whiskers; The Monthly Review - American Electroplaters’ Society;
Vol. 33; January 1946; pp. 28-30, 95.
109
K.G. Compton, A. Mendizza, S.M. Arnold; Filamentary Growths On Metal Surfaces -
”Whiskers”; Corrosion - National Association of Corrosion Engineers; Vol. 7; October 1951; pp.
327-334.
110
[W.]C. Herring, J.K. Galt; Elastic and Plastic Properties of Very Small Metal Specimens;
Physical Review; Vol. 85 (2nd Series);1952; pp. 1060-1061.
111
S.S. Brenner; Metal “Whiskers”; Scientific American; Vol. 203; No. 1; 1960; pp. 64-72.
Copyright 1993 Rev. 1.0 60 of 71
growth of metal whiskers found in radio and telephone equipment was found to
take an amount of time on the order of weeks or months. The problem with this
argument is that the length of the whiskers needed to short out radio or
telephone equipment is on the order of millimeters. Bridging the oxide in the
case of MOM structures, on the other hand, would only require whiskers on the
order of 100 Å, and so the time for this growth would likewise be significantly
less.
Another question left unanswered by Yanson had to do with how the width
(diameter) of the bridges depended on the voltage and current used to coherer
the MOM sandwiches? A partial answer to this question can be found in a paper
by Angelika Székely.
112
As much as I would like to discuss the contents of this
paper - having spent the time translating it - I have run out of time and space to
do so.
§11. FIELD EMISSION X-RAY TUBES. - At the same time the theoretical
work was progressing, experimental work on field emission moved away from the
use of low voltages applied to air gaps in the 100-1000 Å range, and began to
focus more on macroscopic vacuum gaps across which tens or hundreds of
thousands of volts could be applied. The main impetus for this shift in
experimental system was the ready availability of commercially made high quality
x-ray tubes. In addition, the accessibility of the high vacuum technology used in
making these high vacuum x-ray tubes, allowed researchers to construct
modified x-ray tubes specifically for investigating field emission. Until the advent
of readily available high vacuum systems, any electrical conduction through a
vacuum between electrodes of large separation could just as well have been
ascribed to gaseous discharge due to incomplete evacuation of the gas as to
field emission. These newer field emission systems were also obviously much
easier to handle than the Michelson interferometers, precision lead screws, etc.
of the Earhart type of contrivance.
Since we are on the topic of field emission x-ray tubes, I should like to
backtrack for a moment and discuss briefly the work of Julius Edgar Lilienfeld on
the field emission x-ray tube. As was stated earlier, the Coolidge x-ray tube,
introduced in 1913, quickly usurped the up-to-then sacred position occupied by
the Hittorf-Crookes gas x-ray tube. What I neglected to mention was that there
was a third x-ray tube candidate, the field emission x-ray tube of J. E. Lilienfeld.
Lilienfeld’s original research interest appeared to be thermionic vacuum tubes;
his having been a student of Emil G. Warburg, no doubt, contributed to his
interest in conduction in rarefied gases. His later work on the field emission x-ray
tube developed as a natural offshoot of his earlier electron tube work, his first
112
A. Székely; Über die Art des Elektrizitätsüberganges zwischen Metallen, die sich lose
berühren [On the types of electrical conduction between metals which are in loose contact].;
Zeitschrift für Physik; Vol. 22; February-March 1924; pp. 51-69.
Copyright 1993 Rev. 1.0 61 of 71
attempts (circa 1912) at designing a thermionic x-ray tube and last, but not least,
his association with Franz Rother.
In 1922, Lilienfeld published a paper describing his field emission x-ray
tube.
113
As previously mentioned, the modern version of the diffusion pump
was extant by 1920, and so really ‘hard’ vacuums were available to researchers
from that time on. Lilienfeld took advantage of this technological breakthrough to
build cold cathode x-ray tubes. Besides pumping down the tubes, Lilienfeld also
baked the electrodes - and probably the glass walls of the vessel, too - during the
course of the evacuation until the vacuum stabilized. As an added precaution, he
also never operated the tubes in any manner that would cause the temperature
of their components parts to becomes equal to or greater than the bake out
temperature. Tubes prepared in this manner were found to exhibit conduction
when a large DC voltage was applied across their anode and cathode. The size
of the field emission current depended on the shape of the electrodes, the
distance between the electrodes and the metal out of which the electrodes were
constructed. The field emission current was essentially independent of both the
temperature, provided it did not approach the point at which thermionic emission
began, and the pressure, provided it was below the point of Paschen type of
discharges. Note, Lilienfeld did not refer to the process as field emission, instead
he called this process “…autoelektronische [autoelectronic]…”. Even though he
was using state of the art vacuum techniques, Lilienfeld still felt that he needed to
prove that the discharge he was seeing was qualitatively different from residual
gas ionization. To this end, he pointed out that the cathode did not appear to
undergo any heating nor did it suffer any type of disintegration (sputtering) even
after a hundred hours of operation. The absence of both these effects implied
the absence of positive ions (from residual gases).
Lilienfeld’s enthusiasm for his field emission x-ray tube notwithstanding, it
was not commercially viable. Perhaps one of its main problems was
reproducibility. Four years after Lilienfeld’s paper, a General Electric researcher,
B. S. Gossling, reinvestigated these tubes.
114
His results demonstrated that
the tube exhibited irregular behavior coupled with large spontaneous excursions
at high field emission currents and/or voltages. Most of these irregularities came
from deficiencies in materials processing, especially of the electrodes, the field
emission current being particularly susceptible to surface conditions. In fact, field
emission x-ray tubes would not make their way onto the commercial scene until
the 1960s when they were adapted to flash radiography work (see APPENDIX F
of the 1993 Thesis titled “Coherers, a review” by Thomas Mark Cuff, which is
available on ResearchGate). Lilienfeld’s research interest in field emission
113
J.E. Lilienfeld; Die Röntgenstrahlung der Kathode bei der autoelektronischen Entladung
[X-rays from the cathode during the autoelectronic discharge].; Physikalische Zeitschrift; Vol. 23;
1922; pp. 506-511.
114
B.S. Gossling; The Emission of Electrons under the Influence of Intense Electric Fields;
Philosophical Magazine [and Journal of Science]; Vol. 1 (7th Series); No. 3; March 1926; pp. 609-
635.
Copyright 1993 Rev. 1.0 62 of 71
waned after the 1920s, and his later research, after he came to the American
Virgin Islands in 1935, was concerned with the study of anodically formed
dielectric layers on aluminum. These insulating or blocking layers were essential
to the operation of electrolytic rectifiers and more importantly electrolytic
capacitors.
It also must be pointed out that in 1930 and 1933, Lilienfeld was issued
two US patents for a solid state device resembling the modern field effect
transistor.
115
The first patent gave a very crude description of the proposed
field effect device. For this reason, I shall concentrate on the description
provided by the second patent, which was more detailed in both the origins and
the construction of this device. This device consisted of an aluminum base, the
top surface of which had had a layer of aluminum oxide (Al
2
O
3
) grown on it to a
thickness of ~10
-4
mm (1000 Å). On top of the aluminum oxide layer there was
subsequently deposited a thin layer of a compound semiconductor such as
cuprous sulfide (Cu
2
S) or cuprous oxide (Cu
2
0); a metal pad was attached to
either end of the semiconductor stratum so that current could be made to flow
through it. If the semiconductor channel was thin enough, then its conductivity,
which was modulated by applying the proper potential to the aluminum base
electrode, could be controlled and along with it the current flow.
Lilienfeld mentioned in his second patent that the aluminum oxide layer
acted as a very high quality insulator, which had the added advantage that it was
self healing in the event of a puncture due to too great an electric field being
applied across it. All of this insight into the properties of aluminum oxide was
garnered, no doubt, during Lilienfeld’s tenure working on electrolytic capacitors.
With this insulating layer interposed between the aluminum base (gate) electrode
and the semiconductor channel, one could apply large enough voltages to the
gate so that the resulting electric field would modulate the channel conductivity.
As pointed out in a 1964 article reviewing Lilienfeld’s work, the cuprous sulfide
layer most certainly behaved as a P-type semiconductor due to the inevitable
presence of trace amounts of cupric sulfide (CuS): pure cuprous sulfide being an
insulator, while cupric sulfide was almost metallic in its conductivity.
116
Cuprous
115
Julius E. Lilienfeld; Method and Apparatus for Controlling Electric Currents; US Patent
No. 1,745,175; January 28, 1930.
Julius E. Lilienfeld; Device for Controlling Electric Current; US Patent No. 1,900,018; March 7,
1933.
116
Anon.; Obituary: Julius E. Lillienfeld [sic]; Physics Today; Vol. 16; No. 11; November
1963; p. 104.
V.E. Bottom; Invention of the Solid State Amplifier; Physics Today; Vol. 17; No. 2; February
1964; pp. 60-62.
J.B. Johnson; More on the Solid-State Amplifier and Dr. Lilienfeld; Physics Today; Vol. 17;
No. 5; May 1964; pp. 60-62. [J.B. Johnson was, of course, the person responsible for the
concept of Johnson noise. As he mentioned in his article, he also did work with field emission
early in his professional career before his landmark work on noise in active and passive
components. More to the point with respect to the subject of this appendix, he gave a brief
history of the development of field emission which corroborated much of what I have been
saying. Consider the following examples: “The discovery of field emission of electrons has been
Copyright 1993 Rev. 1.0 63 of 71
oxide, on the other hand, would have acted as a defect semiconductor, see
APPENDIX G of the 1993 Thesis titled “Coherers, a review” by Thomas Mark
Cuff, which is available on ResearchGate. While Lilienfeld’s device saw no
commercial application, his patents precluded John Bardeen and also Walter H.
Brattain & Robert B. Gibney from claiming “no prior art” in the their respective
patents on their versions of the field effect transistor granted in 1950.
117
The theoretical work on tunneling and field emission attained its most
persistent explanatory form in the Fowler-Nordheim theory.
118
Note, in addition
to being the source of the current paradigm on tunneling, the aforementioned
paper was my starting point in rediscovering the contributions of Robert Francis
Earhart with respect to tunneling. Specifically, this paper, which was basically a
theoretical look at the process of tunneling, contained in its first page, references
to a number experimental and theoretical articles on tunneling. When I consulted
two of these papers,
119
I found that both introduced their respective histories of
electron tunneling with Earhart’s 1901 Philosophical Magazine article.
§12. THE UNREMEMBERED. - It should be clear by now that Earhart and
Rother can be legitimately claimed to be the ‘Fathers’ of the STM, but for reasons
unclear to me their names and accomplishments do not grace any books, papers
or monographs on the subject of STMs to my knowledge.
120
Earhart’s method
attributed by some [Good & Müller] to R.W. Wood (1897), I think with very scant justification. The
real discovers were probably three men [Earhart, Kinsley and Hobbs - see, his footnote #9] who
worked with Michelson at Chicago, on problems suggested by Millikan, in the early 1900’s. They
studied the starting voltage of current flow across very small air gaps measured with
interferometers.” Johnson also gave credit to J.J. Thomson for correctly interpreting the
experimental work of Earhart et al. in terms of the electrons being dragged out of the metal
electrodes against the image force. What Johnson failed to do, however, was to realize that what
Earhart et al. actually observed were ‘vacuum sparks’, and that not until 1911 did Rother
postulate and then experimentally prove that by reducing the voltage across the electrodes, one
could suppress the ‘vacuum arcs’ and measure the resulting constant field emission currents.]
117
John Bardeen, Bell Telephone Laboratories, Inc.; Three-Electrode Circuit Element
Utilizing Semiconductive Materials; US Patent No. 2,524,033; October 3, 1950.
Walter H. Brattain, Robert B. Gibney, Bell Telephone Laboratories, Inc.; Three-Electrode
Circuit Element Utilizing Semiconductive Materials; US Patent No. 2,524,034; October 3, 1950.
118
R.H. Fowler, L. Nordheim; Electron Emission in Intense Electric Fields; Proceedings of
the Royal Society (London); Vol. 119 (Series A); 1928; pp. 173-181. [For a very informative
exposition on the history, successes and shortcomings of the Fowler-Nordheim theory see,
Leonard B. Loeb; Fundamental Processes of Electrical Discharge in Gases; John Wiley &
Sons, Inc.; 1939; pp. 471-475.]
119
O.W. Richardson; On the Extraction of Electrons from Cold Conductors in Intense Electric
Fields; Proceedings of the Royal Society (London); Vol. 117 (Series A); 1928; pp. 719-730.
R.A. Millikan, C.F. Eyring; Laws Governing the Pulling of Electrons Out of Metals by Intense
Electrical Fields; Physical Review; Vol. 27; January 1926; pp. 51-67.
120
See, for example,
C.B. Duke; Tunneling in Solids; Academic Press; 1969. [This book was an exhaustive
compendium of all known tunneling phenomenon up to and including 1969, but there was no
mention of R.F. Earhart, P.E. Shaw, C. Kinsley or G.M. Hobbs. Note, this book was actually Vol.
10, Supplement 1 of: Frederick Seitz, David Turnbull, Harry Ehrenreich (Eds.); Solid State
Copyright 1993 Rev. 1.0 64 of 71
of controlling and measuring the separation between closely spaced electrodes,
together with Rother’s realization that there was a phenomenon (field emission)
‘underneath’ the rather annoying ‘vacuum sparks’, surely deserved much wider
dissemination that they have heretofore received. To complete this process of
familiarization I will now present the biographies, such as I could find, of these
two forgotten scientists.
Earhart’s name did not appear in such biographical compendiums as
World Who’s Who in Science or American Men & Women of Science.
121
The
paucity of biographical information about him in the open literature forced me to
look elsewhere for biographical information on his life.
122
Robert Francis Earhart was born in Toledo, Iowa on February 2, 1873.
123
His higher education consisted of Northwestern University (18??-1893), BS;
Johns Hopkins University (1897-98), degree conferred, unknown; University of
Chicago (1898-1900), PhD; between 1893 and 1897 he worked as an electrician
for the Peoples Power Co., Illinois. After attaining his doctorate, Earhart held a
couple of teaching jobs before finally settling down to a permanent teaching post
at the Ohio State University (OSU), Columbus, Ohio. He remained at OSU from
1903 to 1931, when he retired, rising to the rank of full professor by 1912. He
Physics; Academic Press. Many libraries list Duke’s book under the title Solid State Physics and
not Tunneling in Solids.
While the names of Earhart and Rother did not grace the pages of specialized texts on
tunneling, Earhart’s name did appear - albeit only in a brief footnote - in a well regarded history of
physics,
Sir Edmund [Taylor] Whittaker; A History of the Theories of Aether and Electricity, Vol. II, The
Modern Theories, 1900-1926; Humanities Press; 1973; p. 236, footnote 7.
In the footnote, Whittaker mentioned that the study of conduction between closely spaced
electrodes originated with Earhart’s examination of short sparks. In footnote 8 of the same page,
the credit for the first theoretical explanation of this phenomena was attributed in the following
curious manner, “A first approximate theory was given by W. Schottky, ZS. f. P. xiv (1923), p. 63,
following on J.E. Lilienfeld, Phys. ZS. xxiii (1922), p. 306.” Warning, this citation contains an
error: the starting page should be 506 not 306. J.J. Thomson’s priority in this matter seemed also
to have been overlooked by Whittaker. In addition, no attention was directed to fact that what
Earhart actually observed were ‘vacuum sparks’, and that this particular phenomenon has not yet
been accounted for theoretically.]
121
Allen G. Debus (Ed.); World Who’s Who in Science; A.N. Marquis Co.; 1968.
Jaques Cattell Press (Ed.); American Men & Women of Science; R.R. Bowker Co.; 1986.
122
All the biographical information on Earhart was obtained from the Ohio State University’s
University Archives. I wish thank Miss Bertha L. Ihnat [pronounced, ee-not] of their University
Archives for looking up the information, xeroxing it and sending it to me in a most efficacious
manner.
123
There seems to be some confusion as whether Earhart was born in Toledo, Iowa or
Toledo, Indiana. The biographical information from the OSU University Archives consisted of the
following: 1) a June 5, 1965 questionnaire from the editor, Hans Salié, of J.C. Poggendorff’s
Biographisch-Literarisches Handwörterbuch, Vol. VIIb [Summary of the Years 1932 to 1962]
which was filled out by one of Earhart’s sons, Daniel S. Earhart, 2) an obituary notice from p. 7 of
the May 1946 edition of The Ohio State University Monthly, and 3) p. 83 from The Ohio State
Centennial History of the College of Mathematics and Physical Sciences. Two of these items say
he was born in Toledo, Iowa and one says Toledo, Indiana.
Copyright 1993 Rev. 1.0 65 of 71
married [date unknown] and had one daughter and four sons, all of whom
graduated from OSU. An examination of the Science Abstracts, Series A
revealed that his scientific papers spanned the time period, 1901 - 1933, and
were mostly concerned with gas discharges. During his tenure at OSU, Earhart
kept in contact with prominent members of the scientific community both at home
and abroad; he made a number of trips to Europe for this purpose at his own
expense. The following table contains a list of the twenty-four citations to
Earhart’s work in Science Abstracts, Series A - not all these citations were
papers by Earhart, some were just abstracts of papers by other workers in the
field who happened to mention Earhart’s name. Papers by Earhart are
designated by an asterisk next to the Abstract No.
TABLE 1E
YEAR SCI. ABSTS. Vol. ABST. No.
1901 4A 514*
1903 6A 248*
1904 7A 2965*
1906 9A 355, 931, 1106
1907 10A 98*,461,924
1908 11A 1467*,2008
1909 12A 1741*
1910 13A 971*
1911 14A 382*
1912 15A 1087*
1913 16A 853*
1914 17A 1099*,1719*
1915 18A 1096
1916 19A 934*
1917 20A 1367*
1919 22A 1537
1929 32A 1623*
1933 36A 1989*
After retiring in 1931, Earhart devoted himself to writing a general physics
book for engineering students, this subject was apparently his specialty having
established by his own dint the curriculum for the degree of Bachelor of
Engineering Physics while tenured at OSU. Unfortunately, the economic climate
at the time was not very good, this being the era of the Great Depression, as a
result this book, which he coauthored with Alvin H. Nielsen of the University of
Tennessee, was not published until 1941, and even then it fell on hard times due
Copyright 1993 Rev. 1.0 66 of 71
to the nation’s preoccupation with the Second World War.
124
Undaunted by all
this, Earhart did his part to help the war effort by writing technical manuals for the
Armed Forces. On April 4, 1946, Dr. Robert Francis Earhart died in Columbus,
Ohio after a heart attack suffered in the hotel where he lived; his wife had died
the previous August (8/24/45).
The biographical sketch I shall present of Franz Rother will be decidedly
emaciated. My only sources of information on Rother are Science Abstracts,
Series A for his scientific achievements and J. C. Poggendorff’s Biographisch-
Literarisches Handwörterbuch Der Exakten Naturwissenschaften [née J. C.
Poggendorff’s Biographisch-Literarisches Handwörterbuch für Mathematik,
Astronomie, Physik, Chemie und verwandte Wissenschaftsgebiete] for both
scientific and biographical information. Examination of Science Abstracts, Series
A from 1911 (Vol. 14A) to 1934 (Vol. 37A) revealed that the first twenty years of
Rother professional life were devoted to the study of emission of electrons from
cold metals, especially between closely spaced electrodes à la Earhart’s
apparatus. At the same time, Rother also studied the properties of the field
emission x-ray tube designed by J. E. Lilienfeld.
It is interesting to note that both Earhart and Rother wrote their respective
PhDs on the conduction between closely spaced cold metal electrodes.
Earhart’s thesis defined the apparatus used to control and measure the spacing
between the electrodes, but the conduction phenomenon he observed at the
smallest separations were due to ‘vacuum sparks’. Rother’s thesis verifies
Earhart’s original work, but went on to uncover the electron tunneling and/or field
emission preceding the ‘vacuum spark’; Rother, in fact, was the first researcher
to measure the field emission current flowing between closely spaced cold metal
electrodes. The conduction currents indirectly observed (via the voltage drop
across the electrodes) by Earhart, Shaw, Kinsley, Hobbs and Hoffmann were
124
According to,
Hans Salié (Ed.); J.C. Poggendorff’s Biographisch-Literarisches Handwörterbuch Der
Exakten Naturwissenschaften, Vol. VIIb [Summary of the Years 1932 to 1962], Part 2; Akademie-
Verlag; 1968; p. 1181.
Earhart’s textbook was not published until 1941. However, according to a letter from one of
Earhart’s sons, Daniel S. Earhart, to Hans Salié, in reference to a biographical questionnaire
about Robert F. Earhart, the book was actually published in 1939. In an effort to clear up this
point, I obtained a copy of the book via interlibrary loan from the Ohio State University main
library. Due to the difficulty in finding this book, let me include its Library of Congress call
number: QC 21 .E12.
Robert Francis Earhart, Alvin H. Nielsen; College Physics for Engineers; Edwards Brothers,
Inc.; 1941; 428 p. [A cursory examination of this text revealed five references to quantum
mechanical phenomena: Planck’s discretization of the black-body oscillators, the photoelectric
effect, the Bohr atom, an exceedingly brief mention of matrix and wave mechanics, and lastly the
Gurney, Condon & Gamow theory of (alpha) radioactive decay - in fact, Earhart erroneously
referred to it as solely Gamow’s theory. Although this theory was cited, there was no attempt
made to contrast the nonclassical nature of quantum mechanical tunneling, nor was the
relationship between tunneling and field emission commented on. In short, it was difficult to tell
whether or not Earhart, himself, accepted the idea of tunneling, which played such a major part in
the results of his work.]
Copyright 1993 Rev. 1.0 67 of 71
large in magnitude and transient, just what one would expect from a ‘vacuum
spark’. Rother’s intuitive leap was to realize that if he reduced the applied
voltage across the electrodes, the ‘vacuum sparks’ would cease, and only the
underlying steady field emission currents would remain. By this simple sounding
artifice, together with a cleverly designed current measuring contrivance able to
measure down to ~10
-14
A, Rother was able to be the first person to actually
obtain an I-V curve of this conduction process. Later refinements of his whole
apparatus enabled him to produce the complete I-V curves having the
exponential dependence expected from a field emission process.
From about 1933 onwards, Rother’s interest in the emission of electrons
from cold metals was redirected towards the photoelectric effect. The late 1920s
and the early to mid 1930s witnessed a renaissance of sorts with respect to the
photoelectric effect. Most physics textbooks give the erroneous impression that
this phenomenon was bludgeoned into submission by the combined efforts of
Albert Einstein’s theoretical virtuosity and Robert Andrews Millikan’s
experimental acumen. The truth of the matter was that the photoelectric effect
was far from being wrung dry of mystery, even by the likes of Einstein and
Millikan, see APPENDIX D of the 1993 Thesis titled “Coherers, a review” by
Thomas Mark Cuff, which is available on ResearchGate.
The following table contains a complete enumeration of entries under
Franz Rother’s name in the Science Abstracts, Series A,
TABLE 2E
YEAR SCI. ABSTS. Vol. ABST. No.
1911 14A 1385
1915 18A 341
1920 23A 1642
1923 26A 741
1924 27A 1123
1926 29A 2940
1929 32A 1958
1931 34A 1287
1933 36A 3258
1934 37A 391-2,1757,4748
Franz Rother was born in Nürnberg (a.k.a. Nuremberg) on May 15, 1887.
He received his PhD in 1914 from the University of Leipzig under the direction O.
Wiener and H. Scholl. From the time he graduated until 1923, he held various
Copyright 1993 Rev. 1.0 68 of 71
teaching positions at the University of Leipzig. In 1923, he entered industry as a
Technical Director at Reiniger, Gebbert & Schall A. G. in Erlangen, where
Erlangen is a city just north of Nürnberg. His stay in industry seemed to have
been rather abbreviated, since in 1926 he apparently returned to the Physics
Institute of the University of Leipzig. From there he journeyed to Berlin in 1928
and became a guest at the Institute for Radiology at the University of Berlin.
Although he continued to publish scientific papers up to and including 1934, after
1934, “Weitere Daten nicht bekannt [Further information is not known].”
125
§13. MODERN TIMES, TUNNELING HITS ITS STRIDE. - As long as the
study of tunneling was an end in itself, the model systems utilized by researchers
remained of the x-ray tube and MOM structure type. But after the advent of the
transistor in 1948 and the concomitant rise of solid state physics as a separate,
distinct and commercially important field of scholarship, people’s attention started
to focus on the problems related to the scaling of devices to smaller and smaller
sizes. With the insatiable striving for miniaturization came the need to be able to
quantify the properties of these small pieces of metal or semiconductor. In
particular, the quantification of their surfaces became of paramount importance
due to the increasing influence of the surface on the behavior of small volume
electronic devices. At the same time as the semiconductor revolution was taking
place, the field of tribology was also coming of age, and people were looking for
better ways to quantify, among other things, the roughness of machined
surfaces.
These problems and others like them inspired Russell D. Young in 1966 to
try to design a noncontact surface probe. In order to sense the surface without
touching it, Young decided to have his probe operate by sensing tunnel current,
this approach, of course, restricted the composition of the surfaces his probe
could examine to those made of metal or semiconductor. His first design attempt
was imminently impractical, but it did serve as a test bed and valuable learning
experience.
126
His next attempt, unveiled in 1972, yielded a practical device,
which again sensed the surface via electron tunneling and in addition could scan
the surface in x and y directions; the sample was brought into proximity of the
metal field emission tip by means of a differential screw, the fine adjustment of
the tip in the x, y and z directions was accomplished by three separate
piezodrivers. Young christened this surface probe, the Topografiner.
127
In
normal use, the Topografiner was operated under high vacuum and the
125
Rudolph Zaunick, Hans Salié (Eds.); J. C. Poggendorff’s Biographisch-Literarisches
Handwörterbuch Der Exakten Naturwissenschaften, Vol. 7a, Part 3: L-R; Akademie-Verlag; 1959;
p. 827.
126
R.D. Young; Field Emission Ultramicrometer; Review of Scientific Instruments; Vol. 37;
No. 3; March 1966; pp. 275-278.
127
R. Young, J. Ward, F. Scire; The Topografiner: An Instrument for Measuring Surface
Microtopography; Review of Scientific Instruments; Vol. 43; No. 7; July 1972; pp. 999-1011.
Idem; Observation of Metal-Vacuum-Metal Tunneling, Field Emission, and the Transition
Region; Physical Review Letters; Vol. 27; No. 14; October 4, 1971; pp. 922-924.
Copyright 1993 Rev. 1.0 69 of 71
electrolytically etched field emission tip was ‘flown’ over the surface to be
examined at 200 Å with the tip bias at -50 to -100 V.
Young’s main improvement over Earhart’s apparatus was his use of
piezodrivers in place of the lead screw arrangement favored by Earhart. Not only
are piezodrivers smaller and cheaper, but they offer better spatial resolution.
And, of course, Earhart’s device could not scan.
It should be noted that Young’s accomplishment using piezodrivers, great
though it was, paled in comparison to some slightly earlier work in which the
piezoelectric crystals were used to sense small displacements rather than to
affect such movements. In the mid to late 1950s, Joseph Weber began to
analyze, design and construct gravity wave detectors. In the end, the detector he
decided on consisted of a 1.5 ton aluminum cylinder, 5 ft long X 2 ft in diameter;
the cylinder had quartz strain gauges (piezoelectric crystals) cemented around its
equator; and the cylinder with its gauges was suspended about its equator via a
loop of wire, the ends of the wire were attached to a cross piece which rested on
two vibration isolation columns to attenuate any environmental vibrations. The
vibration isolation columns, which consisted of alternating layers of steel and
rubber plates, is employed today in STMs for this same purpose - for STMs the
vibration isolation column usually consists of alternating layers of steel plates and
Viton® O-rings, the Viton® O-rings are UHV (Ultra High Vacuum) compatible.
128
The cylinder + strain gauges + cross piece + vibration isolation columns + rolling
base was housed inside a vacuum chamber to protect it from acoustical
vibrations in the air. The aluminum cylinder acted as an antenna to detect gravity
waves by means of its change in length under the action of such waves. This
detector together with its electronics was supposed to be able to sense
displacements on the order of 10
-14
cm. The details of exactly how all this was
done is outside the scope of this thesis, but the interested reader can consult the
U.S. patents taken out by Weber, Forward and Zipoy.
129
128
The idea of making a sandwich out of alternating layers of rigid plates and rubber rings or
plates, for the purpose of vibration isolation, can actually be traced back to, at least, 1900, see,
P.E. Shaw; The Improved Electric Micrometer; Proceedings of the Royal Society (London);
Vol. 76 (Series A); August 4, 1905; pp. 350-359. [In particular, see Figure 6, p. 352 and its
accompanying text.]
129
Although Weber’s detector was constructed around 1960, most of the corresponding US
patents were not submitted (filed) until the mid-1960s and the actual patents were not granted
until the early-1970s. See,
Joseph Weber, Hyman Hurvitz; Amplifier with Feedback Particularly Useful with a Gravity
Wave Detector; US Patent No. 3,554,033; January 12, 1971.
Joseph Weber, David M. Zipoy, Robert L. Forward, Hughes Aircraft Co.; Detector of Dynamic
Gravitational Force Gradient Fields; US Patent Nos. 3,722,284, 3,722,289 & 3,722,290; March
27, 1973.
Note, besides the gravity wave detectors, Weber was also responsible for ground breaking
research into the use of stimulated emission for the purposes of amplification. See,
Joan Lisa Bromberg; The Laser in America, 1950-1970; MIT Press; 1991.
Copyright 1993 Rev. 1.0 70 of 71
The piezo-driven apparatus of Young et al. inspired other workers to try
achieve tunneling at even closer distances. Most notable among this group was
Edgar Clayton Teague, who studied tunneling taking place between gold
electrodes in a vacuum at various electrode spacings down to about 20 Å.
130
While the piezodriver was becoming the method of choice to affect motion
requiring Ångström resolution, it was by no means the only way of achieving
these incredibly minute motions. W. A. Thompson and S. F. Hanrahan used the
nonzero coefficient of thermal expansion or contraction of metals to achieve
stable and repeatable motions with Ångström resolution;
131
and R. J. Walko
utilized magnetic attraction coupled with a restoring forced provided by a metal
spring to study the damage to field emission tips just touching a ‘flat’ metal
surface.
132
As I was completing this thesis, I stumbled across yet another example of
‘parallel evolution’ in the field of precursors to the STM. In the mid to late 1970s,
a group of researchers in the Netherlands was investigating electron-phonon
interactions via an apparatus which had a startling resemblance to the modern
STM of Binnig and Rohrer. The apparatus of Jansen, Mueller and Wyder was
used to bring an electrolytically sharpened “…spear…” into the lightest contact
possible with a larger piece of the same or a different metal called the
“…anvil…”. The approach mechanism, which had to have a resolution
somewhat better than 100 Å consisted of a differential screw for coarse
adjustment and a piezoelectric stack for the final approach; the whole apparatus
130
E.C. Teague; Room temperature Gold-Vacuum-Gold Tunneling Experiments; American
Physical Society Bulletin; Vol. 23; No. 3; March 1978; p. 290. [Note, this was only an abstract of
a poster session and as such did not contain much information beyond a very abbreviated
explanation of what was done. More detailed information can be found in Teague’s 1978 PhD
Thesis,
E. Clayton Teague; Room Temperature Gold-Vacuum-Gold Tunneling Experiments; North
Texas State University; 1978; 229 p. Note, Teague revealed that, when the voltage difference
across the two gold electrodes approached ~2-3 V @ 30 Å electrode separation, breakdown
occurred, i.e., the current increased seemingly without bound. Although Teague did not mention
it, this was coherer-like behavior and the system appeared to have a ‘critical voltage’ of around 2-
3 V, see Fig. 37, p. 140 in his thesis.
131
W.A. Thompson, S.F. Hanrahan; Thermal drive apparatus for direct vacuum tunneling
experiments; The Review of Scientific Instruments; Vol. 47; 1976; pp. 1303-1304.]
132
Robert J. Walko; Lattice Damage by Mechanical and Electrical Contact Investigated by
Field Ion Microscopy; Pennsylvania State University; 1974. [Note, Walko was not the first person
to use a magnetic field coupled to a spring for use in moving contacts smoothly and precisely
over distance measured in Ångströms. As earlier examples of this approach, see,
G. Hoffmann; Elektrizitätsübergang durch sehr kurze Trennungastrecken [Passage of
electricity across very minute air-gaps]; Physikalische Zeitschrift; Vol. 11; 1910; pp. 961-967. Or
see Science Abstracts, Series A; Vol. 14A; 1911; Abstract No. 122.
Idem; Der Elektronenaustritt aus Metallen unter Wirkung hoher Feldstärken [Electron
emission from metals under the influence of high electric fields]; Zeitschrift für Physik; Vol. 4;
1921; pp. 363-382.
R. Deaglio; Einfluß der Wasserhäute auf den Voltaeffekt [Influence of Water Films on Contact
Potentials (Volta effect)]; Zeitschrift für Physik; Vol. 51; 1928; pp. 279-286. Or see Science
Abstracts, Series A; Vol. 32A; 1929; Abstract No. 594.]
Copyright 1993 Rev. 1.0 71 of 71
could be immersed in a Dewar flask so as to bring its temperature down around
1.2ºK. The use of vibration isolation coupled with operating the apparatus at
night to further reduce mechanical and acoustical noise pickup completed the
analogy between this work and that of Binnig and Rohrer on the STM.
133
A
detailed discussion of what Jansen et al. were doing and why can be found in
this appendix and in APPENDIX A of the 1993 Thesis titled “Coherers, a review”
by Thomas Mark Cuff, which is available on ResearchGate. It turns out that their
results may help solve the question of how MOM diodes function?
133
A.G.M. Jansen, F.M. Mueller, P. Wyder; Direct Measurement of α
2
F in Normal Metals
using Point-Contacts: Noble Metals; in D.H. Douglass (Ed.); Superconductivity in d- and f-Band
Metals; Plenum Press; 1976; pp. 607-623.
Idem; Direct measurement of electron-phonon coupling α
2
F(ω) using point contacts: Noble
metals; Physical Review B, Solid State; Vol. 16 (3rd Series); No. 4; August 15, 1977; pp. 1325-
1328.
Idem; Normal Metallic Point Contacts; Science; Vol. 199; March 10, 1978; pp. 1037-1040. [I
wish to thank Dr. Vallorie Peridier and Dr. Thomas E. Sullivan for having encouraged me to read
a series of articles on heat transfer in micro-scale objects, which ultimately led me to the articles
of Jansen et al.]