Distinguishing electron paramagnetic resonance signature of molecular hydrino

Abstract

Quantum mechanics postulates that the hydrogen atom has a stable ground state from which it can be promoted to excited states by capture of electromagnetic radiation, with the energy of all possible states given by En = -13.598/n2 eV, in which n ≥ 1 is a positive integer. By contrast, it has been proposed that the n = 1 state is not the true ground state, and that so-called ‘hydrino’ states of lower energy can exist, which are characterized by fractional quantum numbers n = 1/p, in which 1 < p ≤ 137 is a limited integer 1,2 . Electron transition to a hydrino state, H(1/p) is non-radiative and requires a quantized amount of energy, 2mE1 (m is an integer), to be transferred to a catalyst 3,4 . Since its inception ⁵ the hydrino hypothesis has remained highly controversial ⁶⁻¹⁷ and laboratory verification studies by its proponents have been criticised 18,19 . Remarkably, no experimental testing by independent researchers has been described in the literature over the past 31 years. Here, we give an account of an independent electron paramagnetic resonance (EPR) study of molecular hydrino H2(1/4) that was produced by a plasma reaction of atomic hydrogen with non-hydrogen bonded water as the catalyst. A sharp, complex, multi-line EPR spectrum is found, whose detailed properties prove to be semi-quantitatively consistent with predictions ²⁰ from hydrino theory with an average error less than 0.09 G (0.2%) over a 39 G span of 37 lines. We have sought but failed to find reasonable alternative, ‘conventional’ interpretations for the detected paramagnetism. Fundamental relevance of the hydrino hypothesis lies in its challenging some of the foundations of the theory of quantum mechanics ¹ . Very high net energy release during hydrino formation signifies technological relevance as a novel method of green energy production with recent validation at the 100 kW continuous power level by measurement of steam production ²⁰⁻²⁷ .

Full text

1
Distinguishing Electron Paramagnetic Resonance Signature of Molecular Hydrino
Wilfred R. Hagen1, Randell L. Mills2
1Department of Biotechnology, Delft University of Technology, Delft, The Netherlands.
2Brilliant Light Power Inc, Cranbury, NJ, USA. e-mail: w.r.hagen@tudelft.nl;
rmills@brilliantlightpower.com
Summary. Quantum mechanics postulates that the hydrogen atom has a stable ground state from
which it can be promoted to excited states by capture of electromagnetic radiation, with the
energy of all possible states given by En = -13.598/n2 eV, in which n  1 is a positive integer. By
contrast, it has been proposed that the n = 1 state is not the true ground state, and that so-called
‘hydrino’ states of lower energy can exist, which are characterized by fractional quantum
numbers n = 1/p, in which 1 < p  137 is a limited integer1,2. Electron transition to a hydrino
state, H(1/p) is non-radiative and requires a quantized amount of energy, 2mE1 (m is an integer),
to be transferred to a catalyst3,4. Since its inception5 the hydrino hypothesis has remained highly
controversial6-17 and laboratory verification studies by its proponents have been criticised18,19.
Remarkably, no experimental testing by independent researchers has been described in the
literature over the past 31 years. Here, we give an account of an independent electron
paramagnetic resonance (EPR) study of molecular hydrino H2(1/4) that was produced by a
plasma reaction of atomic hydrogen with non-hydrogen bonded water as the catalyst. A sharp,
complex, multi-line EPR spectrum is found, whose detailed properties prove to be semi-
quantitatively consistent with predictions20 from hydrino theory with an average error less than
0.09 G (0.2%) over a 39 G span of 37 lines. We have sought but failed to find reasonable
alternative, ‘conventional’ interpretations for the detected paramagnetism. Fundamental
relevance of the hydrino hypothesis lies in its challenging some of the foundations of the theory
of quantum mechanics1. Very high net energy release during hydrino formation signifies
technological relevance as a novel method of green energy production with recent validation at
the 100 kW continuous power level by measurement of steam production20-27.

2
Introduction. The quantized energy levels of the hydrogen atom are En = -13.598/n2 eV, in
which the principal quantum number n is a positive integer. The electronic ground state has n = 1.
Higher states can be populated by absorption of light according to the Rydberg equation  =
RH[(1/n2)-(1/n1)] with RH = 109.677 cm-1. R. Mills has hypothesized and experimentally tested
that the n = 1 state is not the absolute ground state and that lower-energy ‘hydrino’ states
characterized by fractional quantum numbers 1/p, with 2  p  137, can exist1,2,20,28, and,
furthermore, that H(1/p) can be produced from H(n=1) in a non-radiative process whereby a
catalyst reversibly takes up an amount of energy equal to (p-1)27.196 eV such that an total
amount equal to (p2-1)13.598 eV is ultimately released as heat4,20,28. These proposals have been
presented as elements of a contentious theory of much wider coverage, called the grand unified
theory of classical physics (here abbreviated as GUTCP) with the much wider claim to revisit the
foundations of quantum mechanics1,20,29. Although a critical appraisal of this controversial
assertion on theoretical grounds initially developed in journals of established reputation6,7,9,10,
subsequently the works of independent opponents8,11, neutral observers14, and adherents13,15,17
alike have slipped off either to publications with impact factors typically well below unity or to
un-reviewed papers. Thus, the present verdict of the scientific community at large appears to be
one of disregard, if not disdain.
Worthy of note is the fact that all cited evaluations by independent researchers thus far are
exclusively concerned with theoretical arguments, while occasional independent experimental
testing of the theory’s predictions has not been published and has only been indirectly cited14.
Moreover, all of these criticisms are based on the incompatibility of hydrinos with quantum
mechanics with the inherent assumption of the validity of quantum mechanics, which is circular
reasoning. Mills’ prediction of hydrino states of hydrogen is not based on quantum mechanics. It
is based on physical laws, and the resulting physical theory is remarkably predictive over 85
orders of magnitude of scale in exact solutions having fundamental constants only1,20. Since Mills
theory is physical/testable, we consider the bias away from experimental testing and reporting as
undesirable in view of the potentially far reaching fundamental and technological implications of
the GUTCP. The present study is an attempt to break a 30-year independent testing silence by
providing an objective and readily reproducible spectroscopic test on a system allegedly
containing molecular hydrino, H2(1/4). This work is not a test of the GUTCP as a whole; it has a

3
bearing on three sub-aspects: hydrino existence, catalytic hydrino formation, and paramagnetic
properties of H2(1/4) predicted by the theory.
Sample production. A common feature of a hydrino state with an excited H state is that both
comprise an electron, a proton, and a photon. In an excited state, the photon superimposes the
proton field to decrease the central field at the electron to +e/n (e is the fundamental charge) and
creates a radial dipole instability that results in radiation. Conversely, the photon of a hydrino
state increases the central field at the electron to +(1+m)e and creates a radial monopole that is
radiatively stable. According to Mills ground-state (n=1) atomic hydrogen can be converted to
atomic hydrino (n=1/(1+m)) by means of a nonradiative resonant energy transfer to a catalyst
with potential energy = m27.2 eV (that is 2mE1) according to the reaction
m27.2 eV + H(1) + Cat  H*(1/(1+m)) + Cat* + m27.2 eV
in which the energy term on the left is energy absorbed by the catalyst (typically by resonant
ionization) and the term on the right is the energy released by the increase in the potential energy
of the hydrogen atom to form H*(1/(1+m)), an intermediate of the hydrino atom of radius aH.
Subsequently the ionized catalyst, Cat*, regenerates by recombination, with the release of its
previously gained ionization energy, and the hydrino intermediate converts to stable H(1/(m+1))
having a radius of aH/(1 + m) by release of additional energy such that the overall release of
energy is [(m+1)2-1]13.6 eV. By considering quantum state p = m +1 the reaction may be
written
H(1)  H(1/p) + (p2-1)13.6 eV
The Rydberg formula with the inclusion of the hydrino states is
E = 13.6 eV/n2; n = 1/137, 1/136…1/4, 1/3, ½,1,2,3…
The hydrino transition reaction requires atomic H and a single catalyst species which is typically
formed chemically or by a plasma reaction20,28. Further reactivity produces molecular hydrino

4
H2(1/p) from atomic hydrino H(1/p) when the bond energy is removed by collision with a third
body, which can be a reactor-wall constituent (cf30). A variety of species can resonantly and
nonradiatively accept m27.2 eV from atomic hydrogen to serve as catalyst for hydrino
formation (Ref-1, Chp 5 &20); in the present case we use the nascent (that is, in situ prepared, not
hydrogen-bonded) water molecule with potential energy 327.2 eV wherein the solutions of the
water molecule and hydrogen bonded water molecules were given previously (Ref-1, Chps13 and
16). Details of the sample preparation are given in the METHODS section. Briefly, the reactor is
a closed vessel in which a low-voltage discharge is created between a liquid gallium electrode
and a solid tungsten electrode with water and hydrogen introduced from a supported-Pt H2/O2
recombiner supplied with H2 gas and trace O2 to form trace nascent or non-H-bonded water
catalyst. Either additional oxygen or water vapor are introduced to produce gallium oxide that is
collected. After dissolution of the gallium oxide in 4 M KOH, a unique non-soluble product
comprising Ga(O)OH in the form of an aggregate of micro-spheres containing molecular hydrino
H2(1/4) slowly polymerizes as shown by scanning electron microscopy (SEM) and transmission
electron microscopy (TEM) in Fig. 1. Energy dispersive X-ray spectroscopy (EDS) showed an
elemental composition of GaO2.1 (Extended Data Fig. 1). Rutherford backscattering spectrometry
(RBS) performed on the GaOOH:H2(1/4) identified the composition as GaO1.68H1.32 with a
density of 8.56 X 1022 atoms/cm2 corresponding to an excess H content, some of which is
hydrino hydrogen based on the results of EPR reported herein and other analytical tests reported
by Mills et al.20,28. Time-of-flight secondary ion mass spectrometry (ToF-SIMS), presented in
Extended Data Fig. 2, showed Ga in the positive ion spectrum and O and H as dominant ions in
the negative ion spectrum wherein the hydride ion was elevated compared to control GaOOH. No
hydrocarbons above adventitious levels were present and no nitrogen was found indicating the
unlikeliness for EPR signals to originate from organic radicals. Equally, in the positive spectrum
no potentially paramagnetic transition ions were present. Selected area electron diffraction
(SAED) with the transmission electron microscope (Extended Data Fig. 3) revealed the samples
to comprise two different morphological and crystalline forms of GaOOH: rods with
orthorhombic diffraction pattern matched control GaOOH, which lacks molecular hydrino, in
morphology and crystalline structure31, and were not sensitive to the TEM electron beam; on the
other hand, morphologically polymeric crystals comprising hexagonal crystalline structure were
very electron-beam sensitive, and were assigned to novel GaOOH:H2(1/4). X-ray diffraction

5
(XRD) showed a phase shift from the GaOOH control lines with different deviations between
NaOH and KOH formed GaOOH:H2(1/4) as illustrated in Extended Data Fig. 4.
Paramagnetism. Extensive background information for this paragraph is provided in
Supplementary information20. Alternative to the probabilistic matter waves of quantum
mechanics, the electron in a hydrogen atom is modelled in GUTCP as a two-dimensional
spherical membrane of infinitesimal thickness in which current flows along two infinite, nested
rotation sets of great circle filaments. This current pattern naturally gives rise to both orbital and
spin angular momentum wherein the latter defines a g factor equal to29 2 + 0.0023193. In the
hydrogen molecule the spherical current pattern becomes a prolate spheroid in which the pairing
of two electrons leads to a diamagnetic ground state. Atomic hydrino differs from H(n>1) states
in that rather than the absorption of a photon to form an excited state, H(n=1/p) it is formed by a
non-radiative energy transfer to a resonant energy acceptor followed by continuum extreme
ultraviolet radiation to the final stable hydrino atomic state. The continuum EUV radiation was
recorded in the laboratory at the 20 MW optical power level with a predicted 10.1 shortwave cut-
off, and this radiation is observed astrophysically over all space20,32,33. Two hydrino atoms react
to form molecular hydrino having two photons that are phase-locked to the electron current and
circulate in opposite directions. Consequently, the molecule has a diamagnetic and a
paramagnetic electron, the latter with g equal to 2 + 20.0023193 = 2.0046386 (Ref-1, Chp16
&20,28). This fundamental prediction from first principles provides a simple and accurate testing
criterion for the existence of molecular hydrino.
EPR spectroscopy. A wide magnetic field scan EPR spectrum of the Ga(O)OH solid powder
taken at ambient temperature, exhibits a single derivative feature only against an essentially flat
background, and with g value close to the free-electron value (Fig. 2a). Zooming-in on this
feature (Fig. 2b) shows it to consist of two separate lines plus multiple weak signals in the low-
and high-field wings. The center of the two main lines corresponds to an apparent g value of
2.0045(6) which is close to the value of 2.00464 predicted for the H2(1/4) S = 1/2 spin-only
doublet system. The two lines are separated by circa 4 Gauss and are of equal intensity.

6
Microwave power saturation plots (Extended Data Fig. 5) are very similar for the two peaks and
are consistent with inhomogeneous broadening34 (see below).
Concentrating on resolving fine structure in the main peaks we reduce the magnetic-field
modulation amplitude to 25 mGauss (that is, below the bandwidth of the 100 kHz modulation
frequency). The spectral amplitude in a single scan drops to below a signal-to-noise ratio of unity
and extensive averaging over six hours and filtering is required to afford the high-resolution
pattern in Fig. 2c. Each line has resolved in an isotropic equidistant beat pattern with sub-line
separation of circa 0.32 Gauss.
For individual sub-lines we observe an apparent peak-to-peak width of circa 170 mGauss
which is highly unusual for solid-state samples. Such narrow lines have been found for (i)
organic radicals in organic solvents at ambient temperature35; (ii) small paramagnetic molecules
in matrices of noble gasses solidified at cryogenic temperatures36; (iii) single hydrogen atoms
encapsulated in molecular cages37; and (iv) paramagnetic molecules in the gas phase at low
pressure38. Excluding the first two options on obvious grounds (no organic solvents and no
cryogenic temperatures) and the third one on spectroscopic grounds (atomic hydrogen EPR is a
single line at the free electron g value split widely by proton hyperfine interaction), the narrow
line width that we observe would be consistent with the detection of a low-pressure paramagnetic
gas occluded in a solid.
We recorded the spectrum in Fig. 2d under optimized conditions for the detection and
resolution of satellite lines whose existence was indicated by the small periodic peaks in the
wings of the spectrum in Fig. 1b. Thus, the fine structure of the two central lines was slightly
deformed by over-modulation and by mild microwave power saturation, and the data collection
was extended to 40 hours with constant frequency monitoring for subsequent correction of
individual 4-min traces for minor frequency drift.
In spin quantification (METHODS) we find that a complete spectrum of high resolution,
such as in Fig. 2d, represents an S = 1/2 concentration of circa 2.6 M if the paramagnet would
be homogeneously distributed over the sample volume. Transmission electron microscopy (Fig 1)
and XRD show the Ga(O)OH polymer to comprise micro-spherical particles of the order of 100
nm diameter with an estimated spatial occupancy of roughly 10%. This would make the actual
concentration of the H2(1/4) gas in occlusion very approximately 26 M, which is equivalent to a
partial pressure of circa 6  10-4 bar, qualitatively consistent with the observed narrow EPR line

7
width38,39. Even if regular H2 would co-occlude at, say, atmospheric pressure the small cross
section for collision of molecular hydrino H2(1/4) would ensure a low collision frequency in
agreement with the observed line width.
Simulation of the fine structure in spectrum Fig. 2c indicated the line shape to be
Gaussian within the limitation set by the overlap of individual lines. The sets of satellite lines are
better separated, and analysis of the first down-field triplet clearly shows the line shape to be
essentially Gaussian (Extended Data Fig 2). This implies inhomogeneous broadening, consistent
with the power-saturation analysis above, and could be caused by interaction of hydrino
molecules with the inner ‘wall’ of the inorganic polymer cage. In turn, this would imply the real
lifetime line width from gas collision to be significantly less than the observed 170 mGauss
inhomogeneous line width.
EPR interpretation. Supplementary information provides detailed theoretical background on
EPR line assignments20. Molecular hydrino comprises two protons at the foci of a two-electron
prolate spheroid molecular orbital membrane, and an absorbed photon. The latter splits into two
photons that are phase locked with the oppositely directed current patterns of the two electrons
each consisting of an angularly distributed infinite ensemble of closed grand ellipse filaments of
moving charge1 of an equipotential, minimum energy membrane surface. Under this model exact
solutions of a fine structure in the EPR ensues with parameters whose predicted magnitudes (Ref-
1, Chp 16 and20,28) can be tested against experimental values.
The unique electronic structure results in one paramagnetic and one diamagnetic electron.
The former induces a current in the latter by means of spin-orbit coupling resulting in a split of
the original resonance into two lines separated by a frequency-independent interaction, which is
for H2(1/4) predicted to be of magnitude 3.9943 Gauss with the field center of the two lines
corresponding to the original g value of20 2.00464. Experimentally we observe two lines of equal
intensity separated by 3.9 Gauss whose center is found at g = 2.0045(6).
Linkage of magnetic flux by the electron membrane is quantised in units of the magnetic
flux quantum φ0 = h/2e, which results in a sub-line pattern of each of the two main lines with a
predicted separation of 0.311 Gauss20. The observed separation is 0.32 Gauss. Similar to the case
of excited-states of the regular H2 molecule, the two electrons in H2(1/4) may rotate relative to

8
each other along the semimajor axis during a spin transition. The relative rotation is quantized in
terms of m integer units of ħ in opposite directions with the spin-orbit splitting in frequency-
independent field units equal to m times twice the splitting between the two main lines, that is
m  7.9885 Gauss. Additionally, the unpaired electron must link the magnetic flux component
corresponding to spin-orbit coupling. This flux contribution increases the magnetic energy and
the energy of the combined spin flip and spin-orbit coupling transition energy for a given spin-
orbital quantum number m. Thus the downfield spin-orbital splitting peaks are shifted further
downfield by the corresponding magnetic energies, whereas the upfield spin-orbital splitting
peaks are not shifted since they correspond to emission of the spin-orbital coupling transition
energies alone.
Furthermore, each of these satellite lines is split through the linkage of magnetic flux
during a spin transition, and the exact solution of the splitting is circa 0.62 Gauss for |m| = 1 and
circa 0.93 for |m| > 1 where the latter lines follow an intensity pattern20 Im+1/Im = m / (m+2). The
predicted details20 of this complex pattern of split satellite lines asymmetrically grouped around a
g value of 2.00464 make up a stick spectrum that, when convoluted with a Gaussian derivative,
forms a semi-quantitative reproduction (Fig. 2e) of the experimental spectroscopy (see also
Supplementary information Table-4).
Consistency controls. Unequivocal interpretation of complex EPR spectra typically requires
analysis of data taken at more than one microwave frequency. The magnetic model of molecular
hydrino H2(1/4), providing a basis for interpretation of the EPR, predicts a number of features to
be either dependent or independent of microwave frequency. These predictions can be checked in
separate experiments as consistency tests. The g value of 2.00464 in between the two main lines
is a real g value and thus its field position should be linear in the microwave frequency.
Contrarily, all fine structure splittings are predicted to be constant in field units and thus
independent of the frequency.
As a check we have taken data in Q-band at circa 35 GHz. Here, practical complications
arise resulting in reduced signal-to-noise ratios. For S = 1/2 systems, any spectrometer operating
in a frequency band different from X-band is generally found to exhibit a significantly lower
concentration sensitivity. Furthermore, the maximal applicable intensity of the microwave is

9
found to be limited (that is, the spectrometer is not tunable at higher microwave powers)
apparently due to a relatively high dielectric permittivity of the Ga(O)OH samples. Fig 3a shows
two traces resulting from extensive averaging, one taken under over-modulating conditions to
emphasize the main two-line pattern, and one taken at lower modulation amplitude in an attempt
to resolve fine structure. Consistent with the interpretation of the X-band spectrum we find a
doublet of lines whose spectral center has a real g value of 2.0046 and with a frequency-
independent splitting of circa 4 Gauss. Under the employed conditions, the underlying broad
signal has turned dispersive and thus shows up as an absorption-shape feature. A lower
modulation amplitude does not afford resolution of the two-lines’ fluxonal fine structure, which
indicates that the spectral line width has increased with frequency. This is in fact consistent with
our previous conclusion (cf. Fig 2c and Extended Data Fig. 6) that the line shape is Gaussian due
to inhomogeneous broadening, which implies a line width in field units linear in the frequency40.
Since the signal-to-noise ratio in Q-band was insufficient to detect the satellite lines, and
since attempts to measure the samples in other frequency bands were hitherto unsuccessful (not
shown), we took data at two, well-separated frequencies within the X-band thus allowing for
comparison of high-resolution spectra with the trade-off of reduced frequency resolution (Fig.
3bc). Data taken at 9.46 GHz were transformed for comparison with data taken at 9.85 GHz in
two ways: (1) frequency-ratio conversion of every digital point of the field axis, and (2) single-
valued overall field shift to create maximal overlap of the two spectra. In the first method all real
g values will overlay while features constant in the field will mismatch. In the second method all
features of a fine-structure pattern constant in the field will overlay when the selected field point
of conversion corresponds to the g value of that pattern. Fig. 3b gives the result of the first
method: all features mismatch except for the spectral center at g = 2.0046, therefore the latter is
the only real g value and all other features are from frequency-independent hyperfine
interactions. Fig 3c gives the result of the second method: all lines match, including all satellite
lines and all fluxon sub-lines of the two main lines, therefore all features are from frequency-
independent hyperfine interactions and they all share a single, common g value. The spectral
features assigned to H2(1/4) were repeated in duplicate at the spectrometer manufacturer location
(Bruker Scientific LLC, Bileria, MA) using two stations, EMXnano and EMXplus instruments20.
Moreover, Raman, and electron-beam excitation spectra show the same spin-orbital
coupling and fluxon linkage splittings as EPR in energy ranges that differ by reciprocal of the

10
H2(1/4) diamagnetic susceptibility coefficient: 1/710-7 = 1.4106, wherein the induced
diamagnetic orbital magnetic moment active during EPR was replaced by the orbital molecular
rotational magnetic moment active during Raman and electron-beam excitation of rotational
transitions20. It is also remarkable, that ten Raman lines recorded on GaOOH:H2(1/4):H2O match
those of Diffuse Interstellar Bands (DIBS)20. Hydrino is consistent with observations that
implicate that the identity of dark matter is an unusual state of hydrogen20,33.
Alternative interpretation. Alternative to the hydrino analysis in Fig 2e the spectrum in Fig 2d
can also be approximately reproduced under a conventional phenomenological spin Hamiltonian
assuming an unusual combination of two isotropic radicals of unequal intensity each with a g
value of 2.0046. This model would require the two main lines to be due to an isotropic S = 1/2
system split by an I= 1/2 nucleus with Aiso  3.9 Gauss with additional hyperfine structure form a
combination of some 5 nuclei the majority of which has also I = 1/2. A second S = 1/2 system
should give rise to the satellite lines due to a different combination of some 5 nuclei, one of
which should have I = 1 (e.g. 14N) to account for the repeating triplet pattern (see Extended Data
Fig. 7 for a detailed analysis).
We consider this alternative explanation of the EPR highly unlikely on the following
grounds. The reaction mixture only contains H2, O2, H2O, and Ga. Even in the presence of trace
contaminants of, e.g., C, N, we cannot envision how the high-temperature plasma reaction
conditions and sample formation in strong aqueous base could lead to the formation of stable
radical structures of considerable complexity. The ToF-SIMs, EDS, and XRD analyses also
eliminate alternatives. Furthermore, since the sample is a solid, for complex radicals one would
expect to see anisotropy in the spectra. In particular absorption-shaped peaks that come with axial
or rhombic symmetry of the spin Hamiltonian are not observed.
Unsolved problems. The GUTCP fit to the X-band spectrum of H2(1/4) is semi-quantitative with
an average error of 0.096 G over the 11 lines assigned to spin-orbital coupling splitting. The
actual positions of the satellite lines slightly deviate from their predicted values (Fig. 2de and
Supplementary information Table-4). Also, the fluxon separation for any given position does not
quantitatively fit the predicted value; in particular the separation is not a constant. Also, the

11
number and relative intensities of fluxon lines for a given satellite line are presently not
understood. Possibly these ‘irregularities’ are caused by interactions of the gaseous H2(1/4) with
the wall of the polymeric Ga(O)OH microspheres.
A broad signal underlies the molecular-hydrino assigned spectrum. Its spectral center
corresponds to the g value of 2.0046 within experimental error. Its temperature behaviour is very
different from that of the hydrino-assigned spectrum (Extended Data Fig. 8). The origin and nature
of the broad signal are presently unknown, however, a reasonable hypothesis would be to assume
that there are two phases of GaOOH that encapsulate H2(1/4) wherein H2(1/4) is a near free gas in
only one phase20,28. A scanning/transmission electron microscope (SEM/TEM) used for imaging
and selected area electron diffraction (SAED) (Extended Data Fig. 3) showed that the
GaOOH:H2(1/4) sample comprised two different morphologically polymeric crystals of GaOOH,
a hexagonal crystalline structure that was very sensitive to the TEM electron beam, and rods having
orthorhombic crystalline structure that were not electron beam sensitive. The rod crystal
morphology and crystalline structure match those of the literature for control GaOOH that lacks
gaseous molecular hydrino inclusion31. The XRD crystal system for Tsumgallite (control GaOOH)
is orthorhombic. The hexagonal phase is likely the source of the fine structure EPR spectrum, and
the orthorhombic phase is likely the source of the broad background EPR feature. Cooling may
selectively eliminate, e.g., by microwave power saturation, the observed near free-gas-like EPR
spectral behavior of H2(1/4) trapped in the hexagonal crystalline matrix. In addition to wall
interactions, deviations from theory could be due to the influence of the proton of GaOOH and
those of absorbed water. Also, matrix orientation in the magnetic field, matrix interactions, and
interactions between one or more H2(1/4) could cause some shifts.
Deuterium substitution was performed to eliminate an alternative assignment of any EPR
spectral lines as being nuclear split lines. The deuterated analog of GaOOH:H2(1/4),
GaOOH:HD(1/4), was confirmed by Raman spectroscopy20,28. The EPR spectrum of the deuterated
analog showed a singlet with no fine structure; thus, eliminating any possible nuclear splitting
assignment. The g factor and profile matched that of the singlet of GaOOH:H2(1/4) wherein the
singlet in both cases was assigned to the orthorhombic phase. The XRD of the deuterated analog
matched that of the hydrogen analog, both comprising gallium oxyhydroxide. TEM confirmed that
the deuterated analog comprised 100% orthorhombic phase (cf31). The phase preference of the

12
deuterated analog may be due to a different hydrino concentration and kinetic isotope effect which
could have also reduced the concentration.
Lastly, further investigation is warranted to assign a peak slightly downfield from the
central g value of 2.0046 the X-band spectra having a small signal of apparent axial symmetry (cf
Fig. 2d). This peak is likely the pure spin-flip (no spin orbital coupling or fluxon linkage
splitting) transition peak of the orthorhombic GaOOH phase with entrapped H2(1/4) molecules
that are constrained relative to the free gas state or near free gas state of the hexagonal phase.
Conclusions. A plasma reaction has been carried out intended to produce molecular hydrino
using non hydrogen bonded water as the catalyst and with liquid gallium as one of the electrodes.
Polymeric Ga(O)OH with a spherical particle structure, presumably containing H2(1/4), was
purified from the reaction mixture. H2(1/4) is proposed to be an S = 1/2 paramagnet with
complex fluxonal and spin-orbital coupling level structure. The solid Ga(O)OH compound
exhibits a complex gas-phase-like X-band EPR spectrum at ambient temperature whose fine
structure semi-quantitatively agrees with hydrino-theory predictions. This analysis is consistent
with frequency-dependent studies, while alternative, conventional interpretations are judged to be
extremely unlikely. In summary, the present study provides compelling EPR spectroscopic
evidence for the existence of hydrino. In view of the possible far-reaching implications of this
conclusion for the theory of quantum mechanics, for hydrogen-related chemistry, for astronomy,
and for energy transduction and production technology (Refs 1,20 and references therein), it is
also offered as an urgent invitation to academia at large to repeat and extend the described
experiments in lieu of refutation on quantum mechanical theoretical grounds.

13
REFERENCES
1. Mills, R.L. The grand unified theory of classical physics, three volumes.
brilliantlightpower.com/book-download-and-streaming / (2020).
2. Mill, R.L. The hydrogen atom revisited. Int. J. Hydrogen Energy 25, 1171-1183 (2000).
3. Mills, R.L., et al. Commercializable power source using heterogeneous hydrino catalysts.
Int. J. Hydrogen Energy 35, 395-419 (2010).
4. Mills, R., Lotoski, J., Good, W. & He, J. Solid fuels that form HOH catalyst. Int. J.
Hydrogen Energy 39, 11930-11944 (2014).
5. R. L. Mills and S. Kneizys, “Excess heat production by the electrolysis of an aqueous
potassium carbonate electrolyte and the implications for cold fusion”, Fusion Technol. 20,
65-81, (1991).
6. Reichhardt, T. Out of this world. Nature 420, 10-11 (2002).
7. Rathke, A. A critical analysis of the hydrino model. New J. Phys. 7, 127 pp 1-9 (2005).
8. Connett, J.E. & Zimmerman, P.D. Against hydrinos. Phys. World 18, 21 (2005).
9. Dombey, N. The hydrino and other unlikely states. Phys Lett. A 360, 62-65 (2006).
10. de Castro, A.S. Orthogonality criterion for banishing hydrino states from standard
quantum mechanics. Phys Lett. A 369, 380-383 (2007).
11. Guizzo, E. Hot or not? IEEE Spectrum 46, 36-38 (2009).
12. Loureiro, J, & Amorim, J. Possibility of nonexistence of hot and superhot hydrogen atoms
in electrical discharges. Phys. Rev. E 82, 035401 (2010).
13. Naudts, J. On the hydrino state of the relativistic hydrogen atom.
arXiv:physics/0507193v2 (2005).
14. Rodgers, P. Hydrogen results cause controversy. Phys. World 18, 12-13 (2005).
15. Bourgoin, R.C. Inverse quantum mechanics of the hydrogen atom: a general solution.
Adv. Studies. Theor. Phys. 1, 381-393 (2007).
16. Phillips, J. Selective atomic hydrogen heating in plasmas: implications for quantum
theory. Int. J. Hydrogen Energy 34, 9816-9823 (2009).
17. Selke, D.L. Against point charges. Appl. Phys. Res. 7, 138-139 (2015).
18. Phelps, A.V. & Clementson, J. Interpretation of EUV emissions observed by Mills et al.
Eur J. Phys. D 66, 120 pp 1-4 (2012).

14
19. Lawler, J.E. & Goebel, C.J. Comment on “Time-resolved hydrino continuum transitions
with cutoffs at 22.8 nm and 10.1 nm”. Eur. J. Phys. D 66, 29 pp 1-2 (2012).
20. Mills, R. L., Dong, Z., Jenkins, J. Gandhi, R., Mehta, N.S., Mhatre, S, Sharma, P.,
“Hydrino states of hydrogen”, Supplementary information to the present paper,
https://brilliantlightpower.com/hydrino-states-of-hydrogen/.
21. Nansteel, M.W., “Water bath calorimetry (120420): report”,
https://brilliantlightpower.com/pdf/Report_on_Water_Bath_Calorimetry_12.04.20.pdf.
22. Nansteel, M.W., “Plasma cell water bath calorimetry: data and analysis (March 11,
2020)”,
https://brilliantlightpower.com/pdf/Waterbath_Calorimetry_Data_and_Analysis_031120.
pdf.
23. Nansteel, M.W., “Molten metal plasma cell calorimetry: data and analysis (January,
2020)”,
https://brilliantlightpower.com/pdf/Nansteel_Molten_Metal_Calorimetry_Data_and_Anal
ysis_Jan_2020.pdf.
24. Booker, R., “Report on the power output of Liquid gallium SunCells at Brilliant Light
Power”, https://brilliantlightpower.com/pdf/Randy_Booker_Report.pdf.
25. Tse, S.D., “Consultant report on onsite molten gallium metal and water bath calorimetry”,
https://brilliantlightpower.com/pdf/Tse-Validation-Report-Brilliant-Light-Power.pdf.
26. Nansteel, M.W, “Report on Parr calorimetry experiments conducted February-March,
2019, https://brilliantlightpower.com/wp-
content/uploads/pdf/Calorimetry_Validation_Report-3.2019.pdf.
27. Nansteel, M.W. “Report on BLP Spectroscopy Experiments Conducted on October 6,
2017”, https://brilliantlightpower.com/pdf/Spectroscopy_Nansteel_Report_040219.pdf.
28. Mills, R.L., “Techniques and unique and characteristic signatures to identify Hydrino are
predicted from exact closed-form solutions of atoms and molecules”,
https://brilliantlightpower.com/pdf/Analytical_Presentation.pdf.
29. Mills, R.L. The grand unified theory of classical quantum mechanics. Int J. Hydrogen
Theory 27, 565-590 (2002).
30. Poole, H.H. Atomic hydrogen III-The energy efficiency of atom production in a glow
discharge. Proc. Roy Soc. 163, 424-454 (1937).

15
31. Li, S.-J., Zheng, C. & Lobring, K.C. Z. Kristallogr. NCS 218, 11-12 (2003).
32. R. Mills, Y. Lu, R. Frazer, “Power Determination and Hydrino Product Characterization
of Ultra-low Field Ignition of Hydrated Silver Shots”, Chinese Journal of Physics, Vol.
56, (2018), pp. 1667-1717.
33. R. Mills, J. Lotoski, Y. Lu, “Mechanism of soft X-ray continuum radiation from low-
energy pinch discharges of hydrogen and ultra-low field ignition of solid fuels”, Plasma
Science and Technology, Vol. 19, (2017), pp. 1-28.
34. Portis, A.M. Electronic structure of F centers: saturation of the electron spin resonance.
Phys. Rev. 91, 1071-1078 (1953).
35. Jones, M.T. Electron spin resonance absorption of tris-p-nitrophenylmethyl. J. Chem.
Phys. 35, 1146 (1961).
36. Feldman, V.I., Sukhov, F.F. & Orlov, A.Y. Hydrogen atoms in solid xenon: trapping site
structure, distribution, and stability as revealed by EPR studies in monoisotopic and
isotopically enriched xenon matrices. J. Chem. Phys, 128, 214511 (2008).
37. Päch, M. & Stösser, R. Scavenger assisted trapping of atomic hydrogen in Si8O12 cages. J.
Phys. Chem. A 101, 8360-8365 (1997).
38. Beringer, R. & Castle, J.G. Microwave magnetic resonance spectrum of oxygen. Phys.
Rev. 81, 82-88 (1951).
39. McDonald, C.C. Multiple-quantum transitions in EPR spectra of atomic oxygen. J. Chem.
Phys. 39, 3159-3160 (1963).
40. Hagen, W.R. Dislocation strain broadening as s source of anisotropic linewidth and
asymmetrical line shape on the electron paramagnetic resonance spectrum of
metalloproteins and related systems. J. Magn. Reson. 44, 447-469 (1981).

16
METHODS
Reactor setup
The plasma reactor (SunCell®)20 comprised a reactor cell, a reaction cell chamber, a molten
metal injector system with an electromagnetic pump driven by a DC power supply and a gallium
reservoir that served as an electrode, a counter electrode, gas flow systems, bus bars to the
electrodes, an ignition power source, and voltage, current, and temperature sensors. The reactor
cell was a Type 347 stainless steel (SS) cylindrical tube measuring 7.3 cm ID, 19.7 cm in height,
and 0.635 cm thick with 3.17 mm thick boron nitride (99%) liner to provide an electrical
insulation barrier and a physical barrier to prevent the internal gallium inventory from alloying
with the stainless steel at temperatures above 500 °C. The cell was pressure leak checked at the
shop following fabrication, and the high-vacuum integrity of the cell and gas and vacuum
connections was confirmed by mass spectroscopy using a residual gas analyzer (Ametek Dycor
Q100M). The cylindrical reaction cell chamber of about 50 ml plasma volume was between the
electrodes and confined internal to the BN liner. The molten metal injection system comprised
0.9 kg of molten gallium in the bottom of the rector that served as a reservoir of gallium and an
electrode, a Type 304 SS injection tube with a W injector nozzle submerged in the gallium by 0.7
cm, and a DC electromagnetic pump. An electrical bus bar (W solid rod, 1 cm OD) penetrated
the bottom of the reservoir through a Swagelok fitting (SS-10MO-1-6W) and was submerged in
the gallium by at least 2.54 cm such that at least one of the injector nozzle and the molten gallium
served as an electrode. The opposing counter electrode oriented along the negative z (vertical)
axis (Extended Data Fig 5) received injected gallium from the injector nozzle to create an
electrical connection between the two electrodes. The counter electrode comprised a feed
through (solid sealing technology, part number FA10775) in a flange (Kurt Lesker 4.5 inch CF
flange) sealed at the top of the reaction cell chamber by a gasket (silver plated Cu), a W bus bar
1.37 cm diameter with male threads on each end to screw-in collect to a 2.54 cm diameter copper
bus bar on the internal side of the feed through on the top end and the W electrode on the
opposite end. The bus bar was covered by an electrically insulating fused quartz sheath
(Technical Glass Products, 2.7 cm ID x 5.1 cm long) sealed with an inner quartz collar that
penetrated into the feed through and was further sealed with an alumina-based cement (Resbond
989) at the feed-through end. The counter electrode, a concave refractory metal electrode (W 3.8

17
cm OD, 1.37 cm height, with a concavity of 2 cm at the apex), screwed onto a threaded end of the
W bus bar and pressed the quartz sheath against the top flange at the opposite end of the sheath.
Production of reactants
Reactants comprising nascent H2O catalyst and a source of atomic hydrogen were provided by
separately controlled H2 (Praxair UHP, 99.999%) and 8% O2 (Praxair industrial grade) flows
using two mass flow controllers (MKS Model PR4000-F2V1N with MKS Model
1179A53CR1BVS for 2500 SCCM of H2 flow and MKS Model M100B12R1BB for 200 SCCM
of O2 flow) that were each calibrated with a rotameter (Dwyer Instruments VA10423, accurate to
+/- 2%). The gases were mixed in an oxyhydrogen torch and flowed through a recombiner
chamber comprising 1 g of a granular platinum catalyst (10% Pt/Al2O3 beads from Alfa Aesar)
heated to greater than 90 °C by the H2+1/2O2 recombination reaction before flowing into the
reaction cell chamber using Swagelok fittings (SS-400-6-4W). The ignition system comprised
either a switch-mode rectifier or a capacitor bank that supplied high-current DC electrical power
sufficient to cause the reactants to react to form plasma. The current was measured with a Hall
sensor and the voltage was measured with a PicoScope. The temperature of the molten gallium
reservoir was measured using 2 K-Type ungrounded thermocouples rated to 1335 °C. The
thermocouples penetrated the side section of the reaction cell chamber through Swagelok fittings
(SS-200-6-2W) extended about 1 cm into the gallium.
Prior to the start of the reaction, the cell was connected to a scroll vacuum pump (Anest
Iwata Model ISP-250) by a 2.54 cm OD stainless steel vacuum line with an intervening liquid
nitrogen cryotrap. All unwanted gases were removed down to a pressure of approximately 40
mTorr (MKS Model 626A11TBE 10 Torr gauge and MKS Model 626A13TEE 1000 Torr
gauge). During operation with flowing gas reactants, the pressure was maintained under 5 Torr.
Prior to operation, the pressure gauges were verified for accuracy within +/- 1% using the same
unit that was vendor calibrated.
Reaction control
The reaction within the cell was maintained using two separate electrical systems: an
electromagnetic (EM) pump system to complete the circuit between the two electrodes within the
cell, and an ignition system to supply electrical input energy to initiate the reaction. The EM

18
pump was powered by a programmable DC power supply (Model: Matsusada Precision REK10-
1200) set to current control mode wherein the current output directly controlled the flow rate of
liquid Ga through the pump tube assembly. During typical operation, the Ga flow rate was
measured to be approximately 40 cm3/sec wherein the voltage and current across the EM pump
was about 0.1 V and 200 A, respectively. The ignition circuit was powered by a LabVIEW-
controlled (National Instruments) switch-mode rectifier (Model: American CRS Q500 IP32)
rated to a maximum of 50V/1500A. The negative and positive terminals of the rectifier were
connected to a solid tungsten (W) rod anode and a liquid Ga cathode, respectively. The excess
power was observed to be strongly dependent on the applied current. So, tests were also
performed using an initial and peak current in the range of 3000-6000 A that was supplied by a
capacitor bank charged to 48 V by the switch-mode rectifier. The capacitor bank comprised
either four or eight Maxwell Technologies modules (Model BMOD0165 P048 C01) connected in
parallel. Each module was rated to 48 V with a capacitance of 165 F wherein four or eight
modules in parallel increased the capacitance to 660 F or 1320 F, respectively. For the ignition
circuit, the electrical response was recorded on a high sampling rate and high-resolution
oscilloscope (Model: PicoScope 5000 Series) using a voltage differential probe (Model: PicoTech
TA041, ±70V) and a DC Hall effect sensor (Model: GMW CPCO-4000-77-BP10, ±4 kA). The
Hall sensor was redundantly calibrated with three Matsusada DC power supplies (Model:
Matsusada Precision REK10-1200) that were current calibrated by the manufacturer. The voltage
probe was calibrated using a standardized voltage source (Model: Agilent E3631A +/- 0.01 V).
The independently validated power developed was in the range of 200,000W-340,000W20-27.
Product processing
Either water or additional oxygen was flowed into the reaction cell chamber to form gallium
oxide to entrap H2(1/4) gas formed in the cell wherein the production of the H2(1/4) gas was
confirmed by gas chromatography following cryogenic collection as well as thermal release of
gas from gallium-oxide trapped H2(1/4) product20. Gallium oxide material was collected from a
hydrino reaction run in the SunCell®, and the gallium oxide material (50 g) was dissolved in 4 M
KOH solution (500 ml). After 0.5-1 hour, the solution was filtered to remove any insoluble solid
phases. A white polymeric material began to nucleate from the clear filtrate after 24 hours.
Using a Buchner funnel, side-arm flask, and filter paper (WhatmanTM, Grade 50, 09-865C), the

19
ultrafine precipitate was suction filtered from the solution. The filtered compound was carefully
removed from the filter paper with a spatula without contacting the filter paper. To wash the
recovered compound, it was suspended in deionized (DI) water, and filtered out a second time
using the prior procedure while applying additional DI water while filtering. The compound was
dried in air at 60-80 °C for 12 hours. XRD showed that the resulting white polymeric material
comprised GaOOH with an average particle size of 111 nm.
Product analysis
Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS)
performed on polycrystalline powder samples mounted on a grid using an XL30 FEG-SEM
equipped with an EVEX EDS, with an in-situ Tensile Stage, and a Gatan MiniCL imaging
system. The lens magnification is 800x and secondary electrons are detected with resolution of 2
nm. Electrons from the gun were accelerated at 10 kV with working distance between the gun
and sample adjusted to 24.3 mm within the low-pressure chamber.
Rutherford backscattering spectrometry (RBS) was performed on polycrystalline powder
samples by EAG Laboratories, Sunnyvale CA using a NEC Pelletron, Model 3SDH particle
accelerator with He++ ions having an energy of 2.275 MeV, a normal detector angle of 160°, and
grazing detector angle of ~100°. The crystal channeling rotating random (CC RR) analysis mode
was used to process the data.
Time-of-flight secondary ion mass spectrometry (ToF-SIMS) was performed on powder
samples sprinkled onto the surface of a double-sided adhesive tape using a Physical Electronics
nanoTOF TRIFT V ToF-SIMS instrument. The primary ion source was 69Ga+ at 30 kV, 1 nA DC,
and bunched to 1 ns. The scan parameters were a scan area of 250 µm  250 µm, a mass range of
0-1850 amu), and a post acceleration voltage of 5000 V with acquisition modes of positive and
negative. Charge compensation was applied to the electron gun and gas gun during position
acquisition and applied to the electron gun only during negative acquisition. The aperture size
was 100
m
m. In order to remove surface contaminants and expose a fresh surface, the samples
were sputter-cleaned for 800 seconds prior to data collection using a 1 mm  1 mm raster of the
69Ga+ gun at a total ion dose of 1015 ions/cm2. Both the positive and negative spectra were
obtained with a total ion dose of 1012 ions/cm2.

20
Transmission electron microscopy (TEM), selected area electron diffraction pattern
(SAED), and scanning transmission electron microscopy (STEM) were recorded on solid
samples, each mounted on a copper grid using a Thermo Scientific Talos F200X
Scanning/Transmission Electron Microscope (S/TEM) equipped with an X-FEG source and a
super-X energy dispersive spectrometer (super-X EDS). The system was operated at 200 kV, and
the electron source brightness was 1.8  109 A/cm2 srad at 200 kV which provided a point
resolution of 0.12 nm for TEM and an information resolution of 0.16 nm for STEM. High
vacuum of less than 10-7 Torr and high-resolution electron tension were maintained at 200 kV.
X-ray diffraction (XRD) was performed on powder samples ground with a mortar and
pestle prior to testing. The samples were then loaded into standard sample holders and placed into
a Panalytical X'pert diffractometer using Cu radiation at 45 kV/40 mA, and the scans were run
over the range of 6° to 80° with a step size of 0.0167° and an accumulated counting time of 250
seconds per step. Once the patterns had been collected, the crystalline phases were identified with
the aid of the Powder Diffraction File published by the International Centre for Diffraction Data
or the Inorganic Crystal Structure Database.
EPR spectroscopy
X-band spectra were recorded at 9.4-9.9 GHz with a Bruker EMX-plus spectrometer using the
high-sensitivity ER4119HS resonator. The microwave frequency was monitored with a 20 GHz
Hewlett Packard 5350B frequency counter. Q-band data were taken at ca 35 GHz with a Varian
E-line spectrometer with the recorder y-axis amplitude and the x-axis micro switches connected
to a NI 6001 data acquisition interface (National Instruments) for 20 kS/s digital storage using a
LabVIEW program. A WR-28 waveguide 566-series cross guide directional coupler combined
with a 410-series waveguide to coax transition (Millimeter Wave Products) were built into the
main microwave path of the Q-band bridge to monitor the frequency with a 46 GHz Hewlett
Packard 5352B frequency counter. Frequency counters were connected to PC’s via USB IEEE
488 GPIB interfaces (National Instruments) to log frequency drift over extended data-collection
times with a LabVIEW program for later off-line normalization. For Q-band the static magnetic
field value was permanently monitored in repetitive sans with a F55 field meter (Magnet-Physik)
equipped with an extra-long axial probe, and data were logged with a LabVIEW program.

21
Individual scans for signal averaging were normalized for drifts in field and/or frequency with a
LabVIEW program before averaging.
EPR data analysis
The Bruker spectrometer can be set to automatically collect 2D data sets for varying microwave
power intensity (that is, EPR amplitude versus magnetic field and microwave power), but the
manufacturer’s software lacks an option to analyze these data in terms of inhomogeneous
broadening. Therefore, a LabVIEW program was written for non-linear Levenberg-Marquardt
fitting to Portis’ theory34.
EPR spectra taken at two different frequencies were compared in a LabVIEW program
that afforded frequency normalization versus field scan shift normalization in order to separate
frequency-dependent from frequency-independent spectral components.
To generate simulations of EPR spectra two programs were written in Intel FORTRAN
with a GUI written in LabVIEW. The first program simply uses the field values predicted by
hydrino theory (Supplementary Information) in combination with symmetric Gaussians for the
individual m =1/2 fluxon lines (standard deviation,  = 130 mGauss), for their distribution ( =
1.30 Gauss), for the m ≥ 1 satellite spin-orbital fluxon lines ( = 124 mGauss), and for the broad
underlying feature ( = 6.0 Gauss) with relative amplitudes for the m = 1/2 spectrum, the m ≥ 1
spectrum, and the broad spectrum fitted as 100 : 8 : 19. The second program is a classical spin-
Hamiltonian simulator in which hyperfine interactions are taken to second-order perturbation of
the electronic Zeeman interaction41,42.
Spin quantification43 was done with respect to an external standard of known S = 1/2
concentration. For this purpose, the X-band spectrum of a powder of 0.5 % (metal/metal) of44
Cu(II) in Zn(II)SO4·1H2O was taken, which has essentially the same density (weight/volume) as
the hydrino-containing Ga(O)OH powder.

22
Data availability
All data necessary to evaluate the claims of this paper are provided in the main manuscript and
Supplementary information. Raw data files, instrument controlling code, data analysis code and
simulation code are freely available upon request.
41. Pake, G.E. & Estle, T.L. The physical principles of electron paramagnetic resonance 2nd
edn (W.A. Benjamin, 1973).
42. Hagen, W.R. EPR spectroscopy as a probe of metal centers in biological systems. Dalton
Trans. 2006, 4415-4434.
43. Hagen, W.R. Biomolecular EPR spectroscopy (Taylor & Francis 2009).
44. Hagen, W.R. Broadband tunable electron paramagnetic resonance spectroscopy of dilute
metal complexes. J. Phys. Chem. 123, 6986-6995.
Acknowledgements
We are grateful to Dr Peter van Noorden for creating the liaison between the authors.
Author contributions
RLM developed the theory and supervised the production and analysis of the samples (with
technical assistance by the authors of Supplementary information, ZD, JJ, RG, NSM, SM & PS);
WRH did the EPR experiments and wrote the dedicated software; WRH and RLM wrote the
manuscript; RLM wrote the Supplementary information.
Competing interests The authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to WRH or RLM

23
Fig. 1 Scanning electron microscopy and transmission electron microscopy of
Ga(O)OH:H2(1/4). Trace a: SEM at 800 magnification showing chains of microspherical
particles; trace b: SEM showing 5 µm width of the particles, each comprising very fine fibers;
trace c: TEM imaging of morphologically polymeric crystals of hexagonal structure
(Extended Data Fig. 2), which were very sensitive to the TEM electron beam. Observed
spherical particles have approximately 100 nm average diameter.

24
Fig. 2 EPR of postulated molecular
hydrino H2(1/4) caged in solid
Ga(O)OH polymer. Trace a: wide-scan
overview spectrum showing a single
feature only close to the free electron g
value. Trace b: zoom-in of the single
feature in trace a shows two main lines
of equal intensity, separated by circa 4
Gauss, and whose center is distinctly
shifted from the free electron value to g
= 2.0045. Trace c: further zoom-in on
one of the lines now recorded with a
very small modulation amplitude reveals
a fine structure of multiple lines with
apparent peak-to-peak derivative line
width of 0.17 Gauss and separated by
circa 0.32 Gauss. Trace d: extensively
averaged spectrum taken under
conditions optimized for maximal
signal-to-noise ratio at the expense of
minor over-modulation and power
saturation, exhibits a complex pattern of
triplet satellite lines. Data collection
times for traces a-d were 10, 16, 375,
and 2400 min, respectively. All spectra
were taken at ambient temperature.
Trace e is a simulation using field
positions predicted by hydrino theory.
Other experimental conditions and
simulation parameters are given in the
METHODS section.

25
Fig.3 Multi-frequency experiments as checks on consistency of the EPR interpretation.
Trace a: extensively averaged Q-band spectra taken at two different modulation amplitudes of
1 Gauss (red) or 250 mGauss (blue). No fine structure is resolved in addition to the two main
lines consistent with an inhomogeneous line width linear in the microwave frequency. The
central g value and the splitting between the two lines in field units are identical to those
observed in X-band. Trace b: extensively averaged intra X-band experiment at two
frequencies, 9.4629 GHz (red) and 9.8209 GHz (black). Each field point of the red spectrum
is frequency transformed to that of the black spectrum where the overlay shows that only the
center of the two main lines is a real g value. In trace c the red spectrum is shifted in its
entirety to a higher field for maximal overlap with the black spectrum. Here the overlay
proves that there is only a single real g value and that all other features are constant in the
field. See METHODS for experimental conditions.

26
Atom%
O 67.0
Cu 1.3
Ga 31.7
Extended Data Fig. 1 Energy dispersive X-ray spectroscopy of Ga(O)OH:H2(1/4).
Only Ga and O are detected in a stoichiometry of GaO2.1. Cu signals are from the TEM
grid.

27
Negative Spectrum
Positive Spectrum
Extended Data Fig. 2 Ga ion time-of-flight secondary ion mass spectrometry of
Ga(O)OH:H2(1/4). The negative spectrum shows hydride ion as a dominant ion
fragment due to the stability of hydrino hydride ion. The result confirms the H content
supporting the molecular hydrino component. No hydrocarbons above adventitious HC
and no N containing fragments were present that could give rise to a radical EPR
spectrum. In the positive spectrum only K and Ga were observed. No transition metals
were present that could give rise to an EPR spectrum.

28
Extended Data Fig. 3 Transmission electron microscopy and selected area electron
diffraction analysis for Ga(O)OH:H2(1/4). The sample was observed to comprise two
different morphological and crystalline forms of Ga(O)OH: polymeric crystals with hexagonal
structure (upper panels) and rods with orthorhombic crystalline structure (lower panels). The
polymeric crystals were very sensitive to the TEM electron beam, whereas the rods were not
electron-beam sensitive. Properties of the rods correspond to literature values for Ga(O)OH
that lacks gaseous molecular hydrino inclusion.

29
0
50
100
150
200
250
300
350
400
10.00 20.00 30.00 40.00 50.00 60.00 70.00
Intensity/ a.u.
2 Theta/ degree
KOH-formed
NaOH-formed
Extended Data Fig. 4 X-ray diffraction of Ga(O)OH:H2(1/4). Samples were formed
by dissolving Ga2O3 collected from a hydrino reaction run in 4 M aqueous KOH or 4 M
NaOH, allowing fibers to grow, and float to the surface where they were collected by
filtration. The KOH pattern is shifted to lower 2 relative to that of NaOH and both
patterns are shifted relative to the standard pattern of Ga(O)OH.

30
Extended Data Fig. 5 The power-saturation characteristics of the two main peaks and their
fluxonal sub-peaks. Upper panels are example spectra taken at ambient temperature under proper
modulation (modulation amplitude = 0.1 Gauss; power = 0.13 mW) or over-modulation (MA = 0.5
Gauss; power = 2 mW). The lower panel gives power plots of the left peak (magenta) and right peak
(green) as well as for fluxonal sub-peaks of the left peak (red) or the right peak (blue). The fit is
according to Portis, case-3, that is for inhomogeneous broadening with relatively slow T1 relaxation
time34. The data indicate full inhomogeneous broadening, therefore, the observed line width should
be linear in the frequency and the homogeneous line width should be significantly less than the
sharpest observed line (155 mGauss).

31
Extended Data Fig. 6 The line shape is Gaussian. The upper panels indicate the
origin of the resonance line, which in the lower panel has been integrated and fitted to
different line shapes. The optimal Lorentzian fit has 0.1405 Gauss half-width-at-half-
height and coefficient of determination R2 = 0.934; the optimal Gaussian fit has 0.1578
gauss standard deviation and R2 = 0.998.The observed purely Gaussian line shape is
taken to imply linearity of the line width with microwave frequency40.

32
Extended Data Fig. 7 Alternative interpretation of the EPR spectrum assuming a
complex radical pattern. The simulation assumes two radicals each with an isotropic
g value of 2.045 and in a 1:0.08 concentration ratio. The first radical gives rise to the
two main lines by means of isotropic hyperfine coupling Aiso = 0.3 Gauss to a single I
= 3/2 nucleus, an Aiso = 0.3 Gauss coupling to four I = 0.5 nuclei and an Aiso = 3.88
Gauss coupling to another I = 0.5 nucleus. Alternatively, the latter splitting can also be
generated by two identical radials at a mutual distance of circa 11 Å in isotropic
dipolar interaction equal to 3.88 Gauss. The second radical is subject to hyperfine
coupling with an I = 2 nucleus with Aiso = 4 Gauss, three I = 1 nuclei with Aiso = 4
Gauss and another I = 1 nucleus with Aiso = 1 Gauss. All lines have a half-width-at-
half-height of W = 160 mGauss. A broad underlying signal is also simulated as a
single line with g = 2.0045 and linewidth W = 6 Gauss. In view of the number of
fitting parameters required, the fit may not be unique; it is presented to illustrate the
necessity to assume a very complex – and therefore highly unlikely – radical structure
to reproduce the experimental spectrum under the assumption of a conventional spin
Hamiltonian encompassing isotropic Zeeman interaction plus hyperfine interaction
with multiple nuclei.

33
Extended Data Fig. 8 Differential temperature dependence of the putative molecular
hydrino signal and a broad underlying signal. Upon lowering the measuring temperature
from ambient to 160 K the hydrino spectrum completely disappears by saturation. The
underlying broad signal is not affected clearly indicating that it is not part of the hydrino
fine-structure spectrum. Its origin is yet to be identified, but is likely due to the
orthorhombic phase of GaOOH. The blue spectrum has been transformed to the frequency
of the red spectrum; its amplitude has been arbitrarily adjusted for easy visual comparison.
EPR conditions: frequency, 9.4011 GHz (blue) or 9.7013 GHz (red); microwave power, 2
mW (blue) or 0.8 mW (red); modulation frequency, 100 kHz; modulation amplitude, 0.5
Gauss.