Content uploaded by Xiao-Ying Yu
Author content
All content in this area was uploaded by Xiao-Ying Yu
Content may be subject to copyright.
Probing liquid surfaces under vacuum using SEM and ToF-SIMS†
Li Yang,
a
Xiao-Ying Yu,*
b
Zihua Zhu,
c
Martin J. Iedema
c
and James P. Cowin‡*
a
Received 7th December 2010, Accepted 18th May 2011
DOI: 10.1039/c0lc00676a
We report a newly developed self-contained interface for high-vapor
pressure liquid surfaces to vacuum-based analytical instruments. It
requires no wires or tubing connections to the outside of the
instrument and uses a microfluidic channel with a 3 mm diameter
window into the flowing fluid beneath it. This window supports the
liquid against the vacuum by the liquid’s surface tension and limits
the high-density vapor region traversed by the probe beams to only
a few microns. We demonstrate this microfluidic interface for in situ
liquid surfaces in a time-of-flight secondary ion mass spectrometer
(ToF-SIMS) and a scanning electron microscope (SEM) with
chemical analysis.
Much chemistry takes place at the interface of liquid phases with
gases in environmental, industrial, and biological systems. The
surfaces of these aqueous phases and films (when <1 nm thick) have
unique kinetics and thermodynamics, distinct from the bulk.
1–3
However, many key surface analytical techniques are vacuum-based,
and cannot easily probe these high vapor-pressure interfaces. Our
goal is to permit studies of high vapor pressure liquids for electron
and vacuum-based ion/molecular based techniques. We first discuss
some existing approaches and their limitations.
The liquid jet interface was pioneered by Manfred Faubel
4
for
electron and X-ray analysis of liquids. It directs a 5–30 micron diam-
eter water jet into a cryopumped vacuum chamber.
5
It requires a very
specialized system, and water evaporation from the jet causes extreme
jet supercooling and freezing. The environmental SEM (ESEM),
originally developed by Danilatos
6
then commercialized via Electro-
scan and FEI, and a recent in situ X-ray photoelectron spectrometer
(XPS)
7,8
(not commercially available) both require specially built
instruments. They do look at in situ aqueous samples, but cannot
handle highly reactive gaseous environments, and will not work much
above about 25 Torr. The TEM via its tight geometry and close-
coupled side port has permitted several impressive in situ probes.
9
But
these are inherently tethered to specific instruments. Other techniques
including ‘‘WetStem’’ cells with liquid trapped between two thin silicon
nitride (SiN) films for SEM and STEM
10
and a similar recent TEM
interface (Hummingbird Scientific, Lacey, WA) are not self-contained,
nor are they true vacuum–liquid interfaces, even if the latter does make
flowing liquids directly available for studying TEM systems.
We engineered a unique liquid sample interface to provide
a completely portable and self-contained micro-environment, suitable
as a multimodal platform. That is, it can be potentially used in any
finely focused (<1 mm) analytical device, as well as optical micro-
scopes. Crucial to its design is the small hole size and the flowing
solution, both of which can reduce the effects of solvent evaporation
on sample temperature and concentration to manageable levels
(about 10 C drop and 20% change, respectively, see ESI†). These
effects would be very severe if there were no flow, or if one had the
360exposed geometry of a liquid jet.
Brivio and co-workers
11
took an innovative approach to exam-
ining liquids that shares some similarities to our device. They have
a bulk solution analyzed by passing through a microfluidic channel
which was capped with a 2 mm thick silicon film. Several holes
100–500 nm diameters were drilled via a focused ion beam (FIB) as
‘‘windows’’ into the flowing liquid beneath. A focused laser then
ablates the exposed liquid (and any ablatable surrounding material
that has seeped or diffused out of the hole) to be analyzed in a matrix-
assisted laser desorption ionization mass spectrometer (MALDI-
MS). It probes a true vacuum–liquid interface supported by its
surfacetension,aswedo.Howevertheirsisdesignedtoanalyzethe
bulk fluid, while ours is designed to study the liquid surface. Theirs is
aimed at being a specific MALDI interface, while ours aims to be
a generic interface. Moreover, our new device provides a continuous
flow using an electro-osmotic pump, whereas they use the pressure
difference between an air bubble and vacuum system to move the
liquid. As a result, our flow rates are easily controlled, while theirs
requires much more trial and error to achieve reasonable flows.
The design of the liquid vacuum module is shown in Fig. 1A,
mainly consisting of a PDMS microfluidic block with microchannel
on the top made by soft lithography (Sylgard 184, Dow Corning
Co., Midland, MI) (a), an electro-osmotic pump (Model number:
3000126, Dolomite) (b), a battery (Saft, Li-SOCl
2
)(c),andValco
PTFE connecting tubes (d). Although PDMS and PTFE are
permeable to water vapor, we calculate that only 3.8 10
8
and 2.0
10
7
g water diffuse through the tubing and PDMS block during
the 8 hours of run per fill (calculation in ESI†). These are small
compared with the 0.2 g of water in the device. The PDMS
a
Chemical and Materials Sciences Division, Pacific Northwest National
Laboratory, Richland, WA, 99354, USA. E-mail: jpcowin@charter.net
b
Atmospheric Sciences and Global Climate Change Division, Pacific
Northwest National Laboratory, Richland, WA, 99354, USA. E-mail:
xiaoying.yu@pnl.gov
c
Scientific Resources Division, W. R. Wiley Environmental Molecular
Science Laboratory, Pacific Northwest National Laboratory, Richland,
WA, 99354, USA
† Electronic supplementary information (ESI) available. See DOI:
10.1039/c0lc00676a
‡ Current address: Cowin In-Situ Science, L.L.C., Richland, WA 99354,
USA.
This journal is ªThe Royal Society of Chemistry 2011 Lab Chip, 2011, 11, 2481–2484 | 2481
Dynamic Article LinksC
<
Lab on a Chip
Cite this: Lab Chip, 2011, 11, 2481
www.rsc.org/loc COMMUNICATION
microfluidic block was coated with a gold film to further reduce gas
permeation
12,13
and prevent charging. The continuous liquid flow
was obtained by an electro-osmotic pump driven by a battery. A tee
on the line (a small viton tube (f) intersected with a short glass
capillary) was used during device filling to prevent overpressure
events. The flow channel (Fig. 1B) on the PDMS block was covered
by a silicon nitride (SiN) film,
14
which subsequently had a 2–3 mm
hole (Fig. 1C) drilled through by a focused ion beam (FIB, FEI
Helios, FIB/SEM).
Our microchip-based vacuum interface was deployed in an SEM
(FEI XL30) and a ToF-SIMS (IONTOF GmbH, M€
unster, Ger-
many). The ToF-SIMS rasters a focused (0.25 mm) primary ion beam
over the sample. Ejected secondary ions are mass-analyzed to
form surface chemical maps with about 0.25 mm resolution. The
ToF-SIMS provides molecular information on the top nm of the
liquid. The SEM, via the energy dispersed X-rays generated by the
rastering electron beam, provides images and elemental maps for
elements from B and up. The depth resolution is 0.1 to 1 micrometre
for the elemental mapping. The aperture size needed to be large
enough to permit the analytical methods (SEM/EDX and ToF-
SIMS) to cleanly probe within the aperture, yet small enough to limit
mean-free path issues and fluid loss. Evaporating vapor will be dense
only around the distance of the aperture diameter away from the
device. For electron based SEM, the mean free path through a gas for
electron energies of 20 keV is millimetres,
15
much longer than the
aperture size. The ToF-SIMS uses 25 keV Bi ions as probes, while the
detected ions typically range from 0 to 10 eV. For simplicity, we
assume the probability of the ion mean free path is the same as that of
a water molecule. The mean free path of water molecules is 13.5 mm
at 24 Torr.
16
For the 3 mm aperture, the radius (R)is1.5mm, making
the ion collision approximately 0.11 (1.5/13.5), acceptably small.
The fluid was mechanically supported by its surface tension (s)
across the opening, against the pressure difference inside to outside.
This maximum pressure is P
max
¼2s/R.
17
For sof pure water
(0.073 N m
1
)andRof 1.5 10
6
m, P
max
is 97 000 Pa, just short of
1 bar. This can easily hold off the vapor pressure of water (about 0.03
bar), and the pressure needed to push the liquid through the channel
(about 0.1–0.2 bar). Flowing fluids limit concentration changes from
evaporation, temperature drop, and beam damage to the solution
compared to having a static fluid. In order to maintain higher linear
flow rates directly behind the aperture and yet prevent too high of
a pressure drop, the designed channel had two widths (80 mmwide,
1.97 mm long and 10 mmwide,30mm long). This makes for an
average linear flow rate at the aperture of 3.5 cm s
1
(at 208 nL min
1
fluid flow) at only 0.13 bar of pressure drop. The temperature drop is
about 17 K, according to our calculations
18
(see ESI†) by the time the
fluid crosses the 3 mm aperture.
Fig. 2A shows the SEM image of the microchannel with DI water
flowing through the channel. The hole in the SiN window is shown in
Fig. 2B, taken at a high energy (30 keV) electron beam. The liquid is
seen as a low contrast blurry region in the hole with a dark band near
the edge of the hole. The liquid presence is confirmed by the EDX
results. Fig. 2C shows the intensity of observed elements at different
spots (with the e beam parked), including the spot outside the channel
(S1), outside the hole but in the channel (S2), and in the hole (S3),
taken at a low energy beam (10 keV). The atomic percentages at the
different locations are summarized in Table 1. The lower energy
beam was chosen, as it barely penetrates the SiN film. Indeed the
much higher oxygen atom percent inside the hole demonstrates the
presence of the bare water liquid surface compared with the other two
locations. Similarly, the low atom percent of C, Si, and N for the
beam over the aperture is consistent with the bare liquid interface.
Images of holes with no water behind them (shown in the ESI†) have
very dark secondary electron and EDX signals.
This is the first time that highly volatile liquid surfaces have been
investigated using ToF-SIMS to the best of our knowledge. The main
chamber operating vacuum pressure in the experiments was 2.5 to 5.5
10
7
mbar. Fig. 3A shows the secondary ion images around the
aperture, with the channel unfilled. The primary ion beam was 25 keV
Bi
+
(beam size: 250 nm). The beam current is 1.0 pA instantaneously,
chopped at 20 kHz, with a beam width of 130 ns. Data shown in
Fig. 3A was taken for a total integration time of 65.5 s, over a scan area
of 10 mm
2
square, with 256 256 pixels. A clear hole can be found at
the center of the Si
+
image with a diameter about 2mm, consistent
Fig. 1 (A) Assembly of the liquid interface. The parts are: (a) channel,
(b) electro-osmotic pump, (c) battery, (d) 1 of 2 Teflon tube fluid reser-
voirs, (e) connecting wires to battery, (f) pressure relief tee (see text). (B)
Optical micrograph of channel. Water flow is from left to right, and the
narrow section is where the aperture is. (C) Optical micrograph of the 2
micron hole made by FIB.
2482 | Lab Chip, 2011, 11, 2481–2484 This journal is ªThe Royal Society of Chemistry 2011
with the SEM results. In addition, low H
,Na
+
and I
signals are
observed in the hole. When we flow DI water through the channel
(Fig. 3B), a hole is also found in the Si
+
image. The Na
+
signal origi-
nates from an area considerably larger than the hole (4mmdiameter),
and is not uniform around the hole. This Na signal appears to have
been brought to the aperture by the DI water, and may come from
fairly low trace impurities (the sensitivity of ToF-SIMS to Na is often
orders of magnitude higher than that for other species). Mobile ions on
surfaces have been observed in the presence of monolayer amounts of
adsorbed water.
19
There may be a partial monolayer of hydrated Na
impurities on the surface, at equilibrium with the DI water.
The high H
signal in the hole is observed as expected when water
is present. The H
signal comes from a region a little larger than the
hole seen in the Si
+
signal. This can be partly explained by a beam size
of around 250 nm. The H
signal may also be a little larger than the
actual hole size, as there is a signal originating from monolayer
amounts of surface water.
Fig. 3C shows the case where a 5 mM sodium iodide aqueous
solution is flowing through the channel, for the same conditions as
those of Fig. 3A and B. The Si
+
image shows a hole as expected. The
Na
+
signal has a very bright core, about the same size as the Si
+
hole.
The I
signal originates from a region about the size of the Si
+
hole.
This is compatible with what is expected for imaging the liquid
surface. The H
signal comes from a region a little larger than the
hole, as does the less-intense halo of the Na
+
,whichmayindicate
a small amount of hydrated Na
+
adsorbed on the surface of the SiN
film. Our results show the aqueous solution is exposed to the vacuum,
and its composition can be probed by the ToF-SIMS.
When an amino acid solution, 1% glutamic acid, was flown
through the channel, the anions H
and [M H]
were strongly seen
in the liquid exposed at the hole, as expected. The Na
+
signal showed
just trace amounts, also as expected (See Fig. 3D). Understanding
what biological molecules are exposed at the surface of a solution
could be important in many applications. This interface enables the
exploration of those systems.
The ToF-SIMS can directly be used to drill the aperture, instead of
using the FIB/SEM. It is done by turning up the ion beam current. This
reduces the handling of the whole device prior to usage and possible
contamination, compared to making the hole in advance with the FIB.
In this experiment, the channel contained flowing D
2
O. The focused
250 nm Bi
+
beam was rastered over a circular area with a diameter of
3mm. To get a rapid sputtering rate, a high average current was
required. This was achieved by lengthening the pulse width to 800 ms,
making the average current 730 times what was used for Fig. 3. In this
mode the mass resolution is much reduced, so only H
and D
could
be cleanly separated in the spectra. High mass resolution spectra
(narrow pulse width) show that H
2
peak is very weak (<1% of D
),
and thus the signal around 2.0 amu range can be regarded as pure D
.
We monitored the ToF-SIMS signal while the intense sputtering was
ongoing, yielding a depth profile. Fig. 4 shows that the Bi
+
beam can
drill a hole through the 100 nm silicon nitride layer in about 42 seconds.
The D
signal shows a dramatic jump (30 times in signal amplitude)
Fig. 2 SEM image of the microchannel (A) and the micron hole above
the channel (B). EDX spectra of different spots (including outside
channel, outside hole but in channel, and in hole) (C).
Table 1 Atomic percent of different spots (S1, S2 and S3 in Fig. 2)
Location
Atomic percent (%)
C O Si N Na Cl
Outside channel (S1) 38 6 14 41 1 0
Outside hole but in channel (S2) 36 15 12 37 1 0
In hole (S3) 9 86 2 3 0 0
Fig. 3 ToF-SIMS imaging (10 10 mm
2
) of the silicon nitride
membrane surface around the hole. Microchannel was (A) empty, (B)
holds flowing H
2
O, (C) holds a flowing 5 mM sodium iodide aqueous
solution, and (D) holds a flowing 1% glutamic acid solution.
This journal is ªThe Royal Society of Chemistry 2011 Lab Chip, 2011, 11, 2481–2484 | 2483
as soon as the Bi
+
beam pass through silicon nitride layer, and H
signal only shows a small jump, as low as 80%. At the initial punch-
through, probably only a small and irregular hole is formed, and the
size of the hole becomes larger and larger with additional doses of Bi
+
ions, and finally the hole size becomes about 3 mmsizeandtheD
and
H
signals become constant. Positive ion spectra (not shown here)
show similar behavior. Fig. 4 (inset) shows the optical image of a ToF-
SIMS-drilled hole after Bi
+
bombardment. It is nearly as round as that
made by FIB.
To conclude, we demonstrated that liquid surfaces can be studied
in situ by advanced vacuum based techniques such as SEM/EDX and
ToF-SIMS using the microfluidic interface assembly developed in our
group. The ToF-SIMS has been applied to study room temperature
aqueous surfaces for the first time. It is a significant technical
breakthrough to broaden the applications of vacuum based surface
techniques to study liquid surfaces of importance in various areas.
This development opens new avenues for understanding interfacial
phenomena occurring on liquid surfaces in the future.
Acknowledgements
We are grateful for help from Bruce Arey of the EMSL, in FIB
fabrication. Support from a Department of Energy (DOE) Division
of Chemical Sciences, Geosciences, and Biosciences (BES Chemical
Sciences grant, KC-0301020-16248) is gratefully acknowledged. The
research was performed in the W. R. Wiley Environmental Molec-
ular Sciences Laboratory (EMSL), a national scientific user facility
sponsored by the DOE’s Office of Biological and Environmental
Research (OBER) and located at the Pacific Northwest National
Laboratory. Pacific Northwest National Laboratory is operated for
DOE by Battelle.
References
1 N. Altimir, P. Kolari, J.-P. Tuovinen, T. Vesala, J. Back, T. Suni,
M. Kulmala and P. Hari, Biogeosciences, 2006, 3, 209–228.
2 A. L. Sumner, E. J. Menke, Y. Dubowski, J. T. Newberg,
R. M. Penner, J. C. Hemminger, L. M. Wingen, T. Brauers and
B. J. Finlayson-Pitts, Phys. Chem. Chem. Phys., 2004, 6, 604–613.
3 S. C. Ringwald and J. E. Permberton, Environ. Sci. Technol., 2000, 34,
259–265.
4 M. Faubel, S. Schlemmer and J. P. Toennies, Z. Phys. D: At., Mol.
Clusters, 1988, 10, 269–277.
5 B. Winter and M. Faubel, Chem. Rev., 2006, 106, 1176–1211.
6 G. D. Danilatos, Scanning, 1981, 4, 9–20.
7 J. Pantfoerder, S. Poellmann, J. F. Zhu, D. Borgmann, R. Denecke
and H.-P. Steinrueck, Rev. Sci. Instrum., 2005, 76, 014102–014109.
8 S. Ghosal, J. C. Hemminger, H. Bluhm, B. S. Mun,
E. L. D. Hebenstreit, G. Ketteler, D. F. Ogletree, F. G. Requejo
and M. Salmeron, Science, 2005, 307, 563–566.
9 F. Banhart, In situ Electron Microscopy at High Resolution, World
Scientific Publishing Co. Pte. Ltd., Singapore, 2008, pp. 229–230.
10 A. Bogner, G. Thollet, D. Basset, P.-H. Jouneau and C. Gauthier,
Micron, 2007, 38, 390–401.
11 M. N. R. Brivio, M. H. Tas, H. J. G. E. Goedbloed, W. Gardeneiers,
A. V. D. Verboom and D. N. R. Berg, Lab Chip, 2005, 5, 378–
381.
12 S. J. Metz, W. J. C. Ven van de, J. Potreck, M. H. V. Mulder and
M. Wessling, J. Membr. Sci., 2005, 251, 29–41.
13 G. C. Randall and P. S. Doyle, Proc. Natl. Acad. Sci. U. S. A., 2005,
102, 10813–10818.
14 J. C. Mcdonald and G. M. Whitesides, Acc. Chem. Res., 2002, 35,
491–499.
15 B. L. Thiel and M. Toth, J. Appl. Phys., 2005, 97, 051101–051118.
16 E. J. Davis, Atmos. Res., 2006, 82, 561–578.
17 M. C. Porter, in Handbook of Industrial Membrane Technology, Noyes
Publications, USA, 1990, p. 71.
18 J. Crank, The Mathematics of Diffusion, Clarendon Press, Oxford,
2nd edn, 1979, p. 102.
19 L. Xu and M. Salmeron, Langmuir, 1998, 14, 5841–5844.
Fig. 4 H
and D
signals vs. Bi
+
ion erosion time, while making a hole
into the channel in the ToF-SIMS. An optical image (10 10 mm
2
) of the
hole created is also shown (inset).
2484 | Lab Chip, 2011, 11, 2481–2484 This journal is ªThe Royal Society of Chemistry 2011