ARTICLES

PUBLISHED ONLINE: 28 AUGUST 2011 | DOI: 10.1038/NNANO.2011.130

Graphene oxide windows for in situ environmental cell photoelectron spectroscopy

Andrei Kolmakov1 *, Dmitriy A. Dikin2, Laura J. Cote2, Jiaxing Huang2, Majid Kazemian Abyaneh3, Matteo Amati3, Luca Gregoratti3, Sebastian Gunther4 and Maya Kiskinova3
The performance of new materials and devices often depends on processes taking place at the interface between an active solid element and the environment (such as air, water or other uids). Understanding and controlling such interfacial processes require surface-specic spectroscopic information acquired under real-world operating conditions, which can be challenging because standard approaches such as X-ray photoelectron spectroscopy generally require high-vacuum conditions. The state-of-the-art approach to this problem relies on unique and expensive apparatus including electron analysers coupled with sophisticated differentially pumped lenses. Here, we develop a simple environmental cell with graphene oxide windows that are transparent to low-energy electrons (down to 400 eV), and demonstrate the feasibility of X-ray photoelectron spectroscopy measurements on model samples such as gold nanoparticles and aqueous salt solution placed on the back side of a window. These proof-of-principle results show the potential of using graphene oxide, graphene and other emerging ultrathin membrane windows for the fabrication of low-cost, single-use environmental cells compatible with commercial X-ray and Auger microprobes as well as scanning or transmission electron microscopes.

he requirement for high-vacuum conditions in conventional electron spectroscopy studies such as X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES) and electron energy loss spectroscopy (EELS) imposes a major obstacle for analysing interfacial physicochemical processes under ambient conditions. Consequently, there has been a persistent trend, mainly in the eld of heterogeneous catalysis and environmental remediation, to overcome this so-called pressure gap and observe processes taking place at surfaces and interfaces under operando conditions, namely elevated or ambient gas pressure or liquid environments (see recent highlights in refs 1 and 2, and references therein). During the last three decades, tremendous progress has been made regarding high-resolution transmission electron microscopy (HRTEM) coupled with the analytical power of EELS and X-ray uorescence spectroscopy, using so-called open environmental cells (E-cells) (see reviews in refs 35). Since the 1930s, elevated pressure has been used in environmental studies using TEM and scanning electron microscopy (SEM) by implementing non-permeable windows (originally made from aluminium6, collodion7,8 or amorphous carbon9). Later, microfabricated SiO2 , Si3N4 or 10100-nm-thick polymer membranes were used, which are highly transparent to the commonly used 10300 keV electrons and 1 103 to 1 104 eV X-rays. This simplied closed-cell design, in combination with a variety of commercially available microfabricated membranes, have boosted activity in environmental electron microscopy1013 and X-ray spectromicroscopy research14,15. However, it still remains a great experimental challenge to adapt powerful, surface sensitive tools such as XPS or AESthe action of which is based on the detection of emitted electrons with relatively low kinetic energy (1 102 to 1 103 eV) to explore objects at ambient conditions. This is mainly due to the rather short electron inelastic mean free path (IMFP), lIMFP, in dense media, which is of the order of a nanometre in condensed matter (Fig. 1) and 1 mm in gases at atmospheric pressure.
1

These main impediments have been successfully resolved through the development of an electron energy analyser with differentially pumped lens systems, thus decoupling the spectrometer from the high-pressure ambient conditions of a sample placed very close to the analyser entrance slit (Fig. 1a; see recent review in ref. 16 for details, and references therein). Alternatively, liquid microjets17 and droplet trains methods18 have also been implemented to probe analytes inside highly volatile solvents. In spite of their tremendous success, these state-of-the-art instruments are currently accessible only in a few laboratories and synchrotron facilities16,17,19, which limits the span of the research to a small number of users. Very recently, protocols for the high-yield fabrication of ultrathin ( 15 nm) suspended membranes made of graphene, graphene oxide (GO) and other two-dimensional materials have been developed2025. Owing to their superior mechanical stability, graphene and GO membranes appear to show great promise as low-atomicnumber Z support materials for HRTEM studies26,27. Together with the pioneering gas impermeability and elastic studies of graphene membranes by Bunch and colleagues28, more recent reports in graphene29,30 and similar tests made on GO31,32 have demonstrated that graphene derivatives (such as GO) are robust enough to be used as membranes for environmental cells, the interior of which can be lled with gas or liquid at atmospheric pressure. Provided such ultrathin membranes are sufciently transparent to 1 102 to 1 103 eV electrons, this will provide a new route to the fabrication of the next generation of E-cells, allowing XPS, AES and possibly photoemission electron microscopy (PEEM) and spectroscopy of fully hydrated samples, liquidgraphene (or GO) interfaces or samples under ambient pressure, as illustrated in Fig. 1b,c. To the best of our knowledge, this Article reports the rst experimental demonstration that even few-monolayer (ML)-thick GO membranes are transparent enough to relatively low-energy (1 102 to 1 103 eV) electrons, and can therefore be used as windows for environmental in situ XPS studies.

Southern Illinois University, Carbondale, Illinois 62901, USA, 2 Northwestern University, Evanston, Illinois 60208, USA, 3 Sincrotrone Trieste 34012 Trieste, Italy, 4 TU Munchen, Chemie Department, Lichtenbergstr. 4, D-85748 Garching, Germany. * e-mail: akolmakov@physics.siu.edu
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLES
a
Electron energy analyser

NATURE NANOTECHNOLOGY
b

DOI: 10.1038/NNANO.2011.130

Photoelectrons

X-rays

Ultrathin membrane Gas Pumping

Sample

c
X-rays IMFP (nm) Sample

100 Energy (eV)

1,000

Figure 1 | Design of ambient-pressure XPS systems. a, Combination of differential pumping stages and advanced transfer electron optics. For details see the recent review in ref. 16. b, An alternative design uses an E-cell with a membrane that is transparent to electrons, but cannot be penetrated by molecules. c, Dependence of inelastic mean free path (IMFP) for electrons in carbon on their kinetic energy, calculated using the NIST SRD-71 database38. The shaded light-yellow area corresponds to the kinetic energy of photoelectrons capable of escaping 1-nm-thick membranes (analogous to GO single-layer membranes).

Fabrication and characterization approaches

Details of GO fabrication, E-cell design, tests and XPS measurements are described in the Methods and Supplementary Section S1. Briey, the GO sheets were produced by a modied Hummers synthesis33 and puried by multiple centrifugation steps. For effective attenuation length (EAL) measurements, GO sheets were deposited on a silicon wafer pre-covered with a 20 nm gold conducting layer. For fabrication of the front window of the E-cells, commercial 50100-nm-thick Si3N4 or SiO2 primary membranes were used (Fig. 2). To ensure the mechanical stability of the suspended GO windows, small 310 mm holes were milled in the primary membranes using a focused ion beam (FIB, Fig. 2a), and individual 1001,000 mm2 GO sheets were deposited over these orices using a LangmuirBlodgett technique (Fig. 2b,c)34. To carry out feasibility tests for the XPS measurements through GO membranes, we used the following samples: (1) 50 nm gold unconjugated colloid nanoparticles drop-cast from a water solution onto the back side of the GO membrane (Fig. 2d,e); (2) a thin 510 nm gold lm deposited via d.c. sputtering directly on the back-side surface of the freestanding GO membrane; (3) a 3 M NaI aqueous solution lling the back side of the membrane. To obtain a vacuum-compatible E-cell, in the third sample, a second silicon wafer was placed over the GO window containing the droplet of NaI solution. The entire assembly was sealed with ultraviolet-curable adhesive (Fig. 2f ). Before
2

ultrahigh-vacuum (UHV) measurements, the mechanical integrity and vacuum compatibility of the sealed cells were tested in an intermediate-vacuum interlock chamber. Owing to the few-micrometre size of the GO windows (Fig. 2g,h), in situ photoelectron spectroscopy measurements with such a cell require microprobe XPS and a sufciently intense photon source provided by the scanning photoelectron microscope (SPEM) operated at ELETTRA35. In SPEM, the X-ray beam is focused onto a spot with a diameter of 100 nm using Fresnel zone plate optics. The local XPS spectra (as well as chemical, topology and conductivity maps) of the (nano-) mesoscopic sample can be obtained by raster-scanning the sample with respect to the focused beam while simultaneously monitoring emitted photoelectrons, transmitted photons and/or the beam-induced sample photocurrent36.

Electron transparency tests of GO layers

For spectroscopic and signal attenuation tests, we selected Au 4f core levels. These levels provide the advantageous combination of appreciable photo-ionization cross-section ( 14 Mbarn) for excitation with soft X-rays (650 and 777 eV were used in the present study) and relatively low binding energy (Eb 83.9 and 87.6 eV for 4f5/2 and 4f7/2 , respectively), which means that the kinetic energy of the emitted Au 4f photoelectrons is high enough for penetration through the GO membrane. The principle and geometry of the experimental setup using GO layers on a gold/silicon substrate

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

NATURE NANOTECHNOLOGY
a
FIB

DOI: 10.1038/NNANO.2011.130

ARTICLES
c

SiO2 GO

Si H2O

e
Au Intensity (a.u.) 100 Au

SiO2 membrane GO membrane

50

C O

Sealant 0 2 Si 4 (keV) 6

3 m 8

Au NP

10

i g Sealed membrane h
GO GO

SiO2

Figure 2 | Fabrication of GO windows for E-cells. a, Ion-beam milling of the microhole in the primary Si3N4 (or SiO2) membrane. b, Deposition of the GO suspended secondary membrane over the ion-milled hole using the LangmuirBlodgett approach. c, SEM image of the resultant deposit (scale bar, 20 mm). d, Deposition of the metal colloid nanoparticles on the back side of the membrane. e, SEM image and EDX of gold nanoparticle aggregates made through the GO membrane (scale bar, 3 mm). f, Enclosed E-cell with ultraviolet-curable sealant placed along the perimeter of the silicon wafer containing the primary Si3N4 and secondary GO membrane. g, Magnied (50) optical image of the front membrane sealing the E-cell lled with water solution. h, Magnied (1,000) optical image showing the GO membrane sealing the window in the lled E-cell. i, TEM image (before sealing) of the same GO membrane as in g and h. Areas with increased numbers of GO sheets can be seen as darker patches (scale bar, 1 mm).

is presented in Fig. 3a. The general morphology of the Langmuir Blodgett-deposited GO layers on the gold/silicon substrate was inspected optically, with SEM and atomic force microscopy (AFM). The average lateral dimension of the individual GO akes was of the order of 20 mm, and folding, overlapping and wrinkle formation were commonly observed. Analysis of the AFM images (similar to Fig. 3b) indicates that the majority of the akes are single-layer-thick individual GO sheets. From the Z-height histogram (Fig. 3c) taken for the representative square area in Fig. 3b, the thickness of the individual GO layer can be inferred to be d1 1.0+0.07 nm. The discrete electron transparency for GO sheets of different thicknesses is illustrated by the three-dimensional (3D) Au 4f intensity map in Fig. 3d, in which patches of single, double, triple (and so on) GO layers can be identied easily. Figure 4a depicts the same 25.6 25.6 mm2 Au 4f image in a two-dimensional (2D) representation, where the variations of the grey scale reect the Au 4f signal attenuation by the multiply folded and overlapping GO layers covering the gold thin lm. The Au 4f spectra in Fig. 4b, collected at selected locations with known thickness (number of GO overlayers, n 0, 1, 2, 3. . .), quantitatively support the information obtained by the Au 4f images. The Au 4f spectra line shape corresponds to metallic gold and, as expected, the Au 4f intensity drops exponentially with increasing number of covering GO sheets due to inelastic photoelectron scattering. The quantication of the attenuation effect is

commonly obtained from the ratio In/I0 , where In and I0 are the Au 4f intensity attenuated by n GO layers and from a bare gold substrate (n 0), respectively. The log In/I0 plot versus the number of GO layers n on top of the gold substrate in Fig. 4c shows an exponential dependence with rather small standard deviation. Using the standard relationship of the straight-line approximation (SLA)37 for the attenuated Au 4f signal in n GO layers due to inelastic electron scattering we obtain In = exp d/lIMFP cos u I0 (1)

where d is the thickness of the GO overlayer, lIMFP is the inelastic mean free path of the Au 4f electrons in GO and u is the emission angle with respect to the sample normal. By substituting lIMFP with the effective attenuation length lEAL to account for elastic scattering effects (see denitions in ref. 38), one can obtain the lEAL value for GO. However, unlike graphene, the reported experimental data for the GO monolayer thickness dGO scatter signicantly from 0.35 nm to 1.7 nm. The major reasons for this scattering are dGO variations likely resulting from different degrees of thermal or chemical reduction of the lm25,39 and intrinsic surface roughness40. In addition, the X-ray beam could partially reduce GO as it can undergo a de-oxygenation reaction upon gentle heating41. Therefore, instead of dGO , it is more feasible to measure the ratio
3

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLES
a
XPS analyser 30 C 1s Au 4f X-rays GO

NATURE NANOTECHNOLOGY
b

DOI: 10.1038/NNANO.2011.130

AFM map

Au

30

c
Number of events 200

1.02 0.00 1.04

1.86

2 3 1 0

100

Au substrate

2 layers

XPS Au 4f map

0 2 1 0 1 Z height (nm) 2 3

Figure 3 | Experimental setup and GO overlayer characterization. a, Sketch of the experimental setup for photoelectron imaging and electron attenuation experiments. b, Non-contact AFM topology image of GO deposit on the gold/silicon substrate. Scale bar, 2.5 mm. c, Z-height histogram taken for the square area in b. d, 3D Au 4f map of the gold sample, where discrete colour (Au 4f signal intensity) changes correspond to different local GO thicknesses.

120 103 100 Intensity (counts s1) 80 Au 4f5/2 0 4f7/2

Au 2 3 1

1 ln(In /I0)

1 60 40 20 0 90 88 86 84 Binding energy (eV) 82 2 3 9L

2 Slope: 0.457 3

4 0 2 4 6 8 GO thickness (ML) 10

Figure 4 | Principle of effective attenuation length measurements of GO sheets. a, 2D Au 4f map of the gold sample, with the grey scale contrast (Au 4f signal intensity) changes corresponding to different local GO thicknesses. b, Microspot Au 4f spectra taken at different locations depicted as white circles in a. The spot corresponding to 9-ML-thick GO is beyond the perimeter of a. c, Log plot of Au 4f intensity attenuation as a function of the number of GO overlayers. Data point sets were taken from different areas of independently prepared samples, and the slope of the tting line denes (lEAL.cosu )21.

lEAL/dGO , which can be obtained from the experimental values of I0 /In from the SLA-derived expression lEAL I = n cos u ln 0 dGO In
1

(2)

The best t gives lEAL/dGO 2.53+0.05, where only the unambiguous number of the GO layers (n , 5) is used (see red line in Fig. 3c). Taking the AFM estimated single GO layer thickness, dGO 1 nm, the effective attenuation length for 567 eV photoelectrons and u 308 is estimated as lEAL 2.5 nm.

Probing the model samples through the GO membrane

The enhanced transparency of the GO for photoelectrons has been conrmed by the ability to collect XPS spectra and images from samples placed on the back side of suspended GO membranes. Owing to the appreciable lEAL value, this shows that GO membrane windows consisting of one to four layers can be used for quantitative
4

in situ XPS. Comparing the data obtained with the two types of gold samples, the photoelectron signal from the percolating gold nanoparticle layer evaporated directly on the back side of the GO membrane is much higher than one from the drop-cast colloidal gold, most likely due to the presence of citrate molecular layers at the interface between the gold colloid nanoparticles and membrane in the latter case. These comparative experiments indicate that for this technique the intimate proximity of the tested sample to the membrane surface is critical. The SEM image in Fig. 5a shows an individual GO ake covering a hole in a SiO2 primary membrane. The TEM/HRTEM images in Fig. 5b,c depict the same membrane after deposition of a few nanometres of gold onto the back side of the membrane. The morphological electron microscopy data were complemented with SPEM chemical mapping of the same membrane, shown in Fig. 5d,e. Comparing the TEM/HRTEM and SPEM images before and after gold deposition indicates that (1) similar to graphene, a few-ML-thick GO membrane can be considered a promising sample support for TEM/HRTEM imaging;

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

3 layers

1 layer

NATURE NANOTECHNOLOGY
a

DOI: 10.1038/NNANO.2011.130

ARTICLES
c

10 1/nm

10 m
1 m 200 nm

d Au 4f 7/2

e C 1s
B

A B Au 4f C 5/2

7/2

90 88 86 84 82 Binding energy (eV)

A B C

C 1s

3.0 104 2.5 Intensity (s1) 2.0 1.5 1.0 0.5 0.0 300 h = 650 eV 250 200 150 Binding energy (eV) GO Au

A B C

COH O=COH C=O

CC, C=C

290

288 286 284 Binding energy (eV)

100

Figure 5 | Suspended GO membranes as windows for an E-cell. a, SEM image of the suspended membrane before gold deposition. b,c, TEM/HRTEM images after deposition of a few nanometres of gold on the membrane back side. The SAED pattern in c shows a typical diffraction pattern from gold deposit, characteristic for randomly oriented fcc gold nanoparticles. d,e, Au 4f7/2 and C 1s chemical maps of the GO membrane with percolating gold nanoparticles on the back side. f, Survey (bottom right), C 1s (bottom left) and Au 4f (top) photoelectron spectra taken at locations A, B and C in e.

(2) the GO membrane preserves its mechanical integrity even after being loaded by up to few-nanometre-thick metal layers or prefabricated nanostructures. The TEM (Fig. 5b) and HRTEM (Fig. 5c) images reveal the morphology of a gold deposit composed of percolating single-crystal nanoparticles, typical for a VolmerWeber-type growth mode of metals on weakly interacting substrates, such as graphite or metal oxides. The HRTEM image and selected area electron diffraction (SAED) pattern of the GO/gold sandwich in Fig. 5c are typical for randomly oriented gold nanoparticles with major lattice fringes indexed to a polycrystalline gold sample with face-centred cubic (fcc) crystallites. SPEM data in Fig. 5df provide complementary information about the chemical status of the gold specimen surface and the interface with the GO window. For this membrane, an effective thickness of 34 ML was estimated, based on the IC1s/IAu4f ratio recorded for GO layers with known thickness. The quality of the XPS survey and Au 4f and C 1s spectra taken at different locations (Fig. 5f) is sufcient to identify the chemical status of the sample on the back side of the membrane. As expected, the t of the Au 4f doublet corresponds to unreacted metallic gold. The deconvoluted C 1s spectrum of the membrane contains low weight of the C O, COH components. They resemble the C 1s spectra of thermally reduced GO42, presumably as a result of oxygen desorption induced by prolonged electron or X-ray irradiation during the measurements. The Au 4f7/2 (Fig. 5d) and C 1s (Fig. 5e) chemical maps of the membrane indicate certain inhomogeneities in the

gold and carbon distribution across the membrane area, presumably due to radiation damage. As is shown in Supplementary Section S2, the long-term radiation effects are not of crucial importance provided the ultrathin membranes are designed as inexpensive single-use E-cells for in situ XPS characterization. The feasibility of conducting photoelectron spectromicroscopy of liquid media was tested by lling the E-cell compartment with a 3 M NaI aqueous solution, and subsequently vacuum sealing it using the procedure described in the Methods. It was found that few-nanometre-thick membranes are capable of maintaining the liquid and/or elevated-pressure water (vapour) for an extended period of time, sufcient for routine XPS studies (Supplementary Section S1). Some interfacial diffusion between the GO sheets and the primary silicon nitride membrane was observed in only a few cells after long experiments. Similar phenomena, namely interfacial mass transport and trapping of water layers, have been reported recently30,43. Therefore, the vacuum tightness of the E-cell front side is determined mainly by the initial adhesion of the GO sheet to the Si3N4 window, which can be improved by moderate thermal annealing. These samples sustained UHV conditions for at least a few hours of continuous measurements. Figure 6 shows representative O 1s spectra taken at the points indicated in the Au 4f7/2 images, namely in the centre of the wet hole (point A), where the GO membrane isolates the 3 M NaI aqueous solution from the vacuum, and in the dry GO covered area 30 mm away from the hole (point B). The O 1s spectrum from the dry GO on
5

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLES
B O 1s

NATURE NANOTECHNOLOGY

DOI: 10.1038/NNANO.2011.130

Point B

H2OVAP
Au 4f

H2OLIQ

C=O, COH, SiO2 O=COH

Point A

540

535 Binding energy (eV)

530

kinetic energy) due to one- to nine-layered GO membranes covering the evaporated gold lm, the effective attenuation length in the GO matrix has been estimated as 2.5 nm for an emission angle of 308. The results in this Article indicate that, in addition to their unique electronic, thermal, magnetic, mechanical and other properties, the recently demonstrated 2D and quasi-2D crystals, such as graphene, GO, BN, MoS2 , NbSe2 , Bi2Sr2CaCu2Ox , Bi2Te3 (and many others), appear to be an ideal platform for experimental XPS tests of interfacial, electron transport and inelastic phenomena in solids, because the morphology and the thickness of such an overlayer can be precisely determined. From the broader application prospective, ultra-transparent and robust graphene/GO windows open the door to unprecedented opportunities for in situ electron spectroscopy in operando conditions. The development of large-yield protocols for the fabrication of suspended graphene and GO membranes is a current agenda for many groups44 and has already reached a commercialization stage45. In contrast to the current environmental XPS methodology, based on sophisticated instrumentation using synchrotron radiation sources, these low-cost single-use E-cells can potentially be used with any commercial micro-XPS or AES instrument, routine SEM/TEM or optical microscopes.

Intensity (a.u.)

Figure 6 | XPS on wet samples. Top inset: 40 50 mm2 SPEM image of the GO membrane with 3 M NaI aqueous solution on the back side taken with Au 4f7/2 photoelectrons. The bottom inset is an enlarged 12 12 mm2 area around the membrane. Point A is at the centre of the GO membrane, and point B is 30 mm away. Top and bottom spectra were taken at points B and A, respectively. In addition to the common O 1s spectrum from the dry area, the spectrum from the membrane contains H2O vapour and liquid contributions from the enclosed cell compartment. The red dots indicate the raw data points of the O 1s spectra, and the small dark spots represent the interpolated spectrum used for peak deconvolution analysis. The green curve depicts a sum of the tted peaks. The tting details can be found in the Methods.

Methods
GO synthesis and purication. GO was prepared from graphite powders (Bay carbon, SP-1) using a modication of Hummers method33. Graphite (2 g), NaNO3 (1 g) and 46 ml of H2SO4 were stirred together in an ice bath. KMnO4 (6 g) was then added slowly. All chemicals were purchased from Sigma-Aldrich and were used as received. Once mixed, the solution was transferred to a 35 8C water bath and stirred for 1 h, forming a thick paste. Water (80 ml) was then added, and the solution was stirred for 1 h while the temperature was raised to 90 8C. Finally, 200 ml of water was added, followed by the slow addition of 6 ml H2O2 (30%), causing the solution to turn from dark brown to yellow. The warm solution was then ltered and washed with three aliquots of 200 ml 10% HCl, followed by 200 ml water. The lter cake was dispersed in water by mechanical agitation and stirred overnight. The dispersion was allowed to settle and the top clear, yellow dispersion was subjected to dialysis for 1 month. LangmuirBlodgett assembly of GO. See ref. 34. The as-prepared GO was diluted to a 5:1 methanol/water solution. The trough (Nima Technology, model 116) was carefully cleaned with chloroform and then lled with deionized water (DI). GO solution was spread dropwise onto the water surface using a glass syringe, to a total volume of 1015 ml, and the lm was then allowed to equilibrate for at least 20 min. The substrates were vertically dipped into the trough and slowly pulled up (2 mm min21) after compressing the lm to form a dense GO monolayer at the airwater interface. Samples and cells preparation. Most of the experiments were carried out using commercial (SiMPore Inc/TEMwindows) 200-mm-thick frames with 50-nm-thick 100 100 mm2 Si3N4 (or 40 nm SiO2) windows as primary membranes. Windows for GO deposition were fabricated by ion beam milling of the primary membrane using an FEI Helios NanoLab Dual Beam SEM/FIB. To eliminate charging, the surface of the primary membranes was coated with 10 nm of gold before GO deposition. Three different types of samples were used in the study. The rst had unconjugated monodispersed 50 nm gold colloids (Tedd Pella) drop-cast onto the back of the membrane and dried. In the second, to evaluate the transparency of the freestanding GO membranes, 510 nm of gold was d.c. sputtered onto the backside of the GO membrane. To protect the front side of the membrane from unintentional deposition of gold during sputtering, the periphery of the front side of the membrane was sealed using a ring made of Gel-Film. In the third sample, 3 M NaI aqueous solution in double-distilled (DD) water: 1 1022 ml of analyte was lled in the compartment of the primary membrane and a second silicon wafer was placed over the lled compartment to create a sealed E-cell. The excess water was removed and the entire assembly sealed using ultraviolet-curable Norland optical adhesive NOA-81. A strip of conducting silver paste was drawn between the front side of the cell and the metal support to ground the front part of the primary membrane. Such a cell was compatible with the typical vacuum (1 1029 torr) of the main chamber. Imaging and spectroscopy. SEM imaging was performed with an FEI Nova 600 NanoSEM in reection mode by collecting secondary electrons, and also in transmission mode using a two-segment solid-state STEM detector with either bright- or dark-eld electron collection. The routine imaging conditions were an acceleration voltage of 10 kV, beam current of 1 10211 A, and an electron dose of 1 mC cm22. TEM imaging was carried out with a JEOL JEM-2100F FAST TEM using an acceleration voltage of 200 kV. AFM images were recorded using a Park-XE100 (Park Systems Corp) microscope. The scanning probe microscope (SPM) was equipped with a silicon

gold-covered SiN is similar to the standard O 1s XPS spectrum of partially reduced GO39,42, with a small contribution from H2O presumably trapped at the interface or between the GO sheets and the substrate43. The O 1s spectrum of the wet area lled with the solution is very different, apparently due to pronounced components corresponding to the liquid and vapour phases of water. The concomitant XPS features in the region of iodine 4d, sodium 2s and sodium 2p corresponding to the presence of I2 and Na solvated ions (not shown here) appears to be too weak to be unambiguously assigned, presumably due to the low ion concentrations and a relatively thick GO membrane. However, the presented data are a proof of concept for environmental electron spectroscopy in membranebased E-cells. Further developments are under consideration to implement the proposed technique to study the interfaces under operando conditions. The feasibility of using the thinnest (5 nm) commercially available SiN membranes as windows for XPS-compatible E-cells is described in Supplementary Section S3.

Conclusions
In conclusion, we have demonstrated that GO membranes are a very promising window material for E-cells to enable XPS studies of samples immersed in liquid or dense gaseous media. Our proofof-concept experiments proved that up to several layers suspended GO membranes have sufcient transparency for photoemitted electrons with kinetic energies .450500 eV. Good-quality XPS spectra can be obtained from aqueous solutions and nanoparticles deposited on the back side of the GO membranes. Monitoring the attenuation in the Au 4f electron emission (567 eV electron
6

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.

NATURE NANOTECHNOLOGY

DOI: 10.1038/NNANO.2011.130

ARTICLES

cantilever (DP15 from MikroMasch) with a force constant k 40 N m21 and a curvature radius r 10 nm. The AFM was operated in non-contact mode at a resonance frequency of f 270 kHz. Image scanning was performed with a speed of 0.5 Hz, which corresponds to one line per second. SPEM measurements were carried out at the ESCA microscopy beamline at the ELETTRA synchrotron facilities. The incident soft X-ray photon beam (2001,400 eV) was focused to a small spot with a diameter of 100 nm at the sample surface. The sample could be scanned with an accuracy of 10 nm in front of the beam. A Fresnel zone plate was used to demagnify the source into a microprobe. The measured ux in the probe was 1 109 to 1 1010 ph s21. For XPS spectra and imaging, a hemispherical 100 mm energy analyser with a 48-channel detector and an energy resolution better than 200 meV was used. XPS spectra analysis and image data were processed using IGOR Pro software. For quantitative intensity comparisons, peak areas were determined after Shirley background subtraction. The O 1s spectra were tted assuming a Doniach Sunjic line shape of low asymmetry (a 0.02), a Lorentzian linewidth of 0.25 eV and a Gaussian linewidth ranging between 1.8 and 2.6 eV and subtracting a linear background. The peak positions of the indicated species were 532.0532.5 eV (C O, COH, SiO2 , O COH; that is, species belonging to the membrane), 533.6 eV (H2O liquid) and 536.8 eV (H2O vapour).

Received 21 April 2011; accepted 14 July 2011; published online 28 August 2011

References
1. Bluhm, H. et al. In situ X-ray photoelectron spectroscopy studies of gassolid interfaces at near-ambient conditions. MRS Bull. 32, 10221030 (2007). 2. Somorjai, G. A. & Park, J. Y. Molecular surface chemistry by metal single crystals and nanoparticles from vacuum to high pressure. Chem. Soc. Rev. 37, 21552162 (2008). 3. Parsons, D. F. Structure of wet specimens in electron microscopy. Science 186, 407414 (1974). 4. Boyes, E. D. & Gai, P. L. Environmental high resolution electron microscopy and applications to chemical science. Ultramicroscopy 67, 219232 (1997). 5. Sharma, R. & Weiss, K. Development of a TEM to study in situ structural and chemical changes at an atomic level during gassolid interactions at elevated temperatures. Microsc. Res. Tech. 42, 270280 (1998). 6. Marton, L. La microscopie electronique des objects biologiques. Bull. de LAcad. Royale de Belgique 21, 553560 (1935). 7. Abrams, I. & McBain, J. A closed cell for electron microscopy. J. Appl. Phys. 15, 607609 (1944). 8. Swift, J. & Brown, A. An environmental cell for the examination of wet biological specimens at atmospheric pressure by transmission scanning electron microscopy. J. Phys. E 3, 924926 (1970). 9. Kohyama, N., Fukushima, K. & Fukami, A. Observation of hydrated form of tubular halloysite by an electron microscope equipped with an environmental cell. Clays Clay Miner. 26, 2540 (1978). 10. Daulton, T. L., Little, B. J., Lowe, K. & Jones-Meehan, J. In situ environmental cell transmission electron microscopy study of microbial reduction of chromium(VI) using electron energy loss spectroscopy. Microsc. Microanal. 7, 470485 (2001). 11. Gai, P. L. & Calvino, J. J. Electron microscopy in the catalysis of alkane oxidation, environmental control, and alternative energy sources. Ann. Rev. Mater. Res. 35, 465504 (2005). 12. Thiberge, S., Zik, O. & Moses, E. An apparatus for imaging liquids, cells, and other wet samples in the scanning electron microscopy. Rev. Sci. Instrum. 75, 22802289 (2004). 13. de Jonge, N., Peckys, D. B., Kremers, G. J. & Piston, D. W. Electron microscopy of whole cells in liquid with nanometer resolution. Proc. Natl Acad. Sci. USA 106, 21592164 (2009). 14. Hitchcock, A. P., Morin, C., Heng, Y. M., Cornelius, R. M. & Brash, J. L. Towards practical soft X-ray spectromicroscopy of biomaterials. J. Biomat. Sci. Polymer Edition 13, 919937 (2002). 15. Yoon, T. H. Applications of soft X-ray spectromicroscopy in material and environmental sciences. Appl. Spectrosc. Rev. 44, 91122 (2009). 16. Salmeron, M. & Schlogl, R. Ambient pressure photoelectron spectroscopy: a new tool for surface science and nanotechnology. Surf. Sci. Rep. 63, 169199 (2008). 17. Winter, B. et al. Full valence band photoemission from liquid water using EUV synchrotron radiation. J. Phys. Chem. A 108, 26252632 (2004). 18. Starr, D. E., Wong, E. K., Worsnop, D. R., Wilson, K. R. & Bluhm, H. A combined droplet train and ambient pressure photoemission spectrometer for the investigation of liquid/vapor interfaces. Phys. Chem. Chem. Phys. 10, 30933098 (2008). 19. Gunther, S. et al. In situ X-ray photoelectron spectroscopy of catalytic ammonia oxidation over a Pt(533) surface. J. Phys. Chem. C 112, 1538215393 (2008). 20. Novoselov, K. S. et al. Two-dimensional atomic crystals. Proc. Natl Acad. Sci. USA 102, 1045110453 (2005). 21. Pacile, D., Meyer, J. C., Girit, C. O. & Zettl, A. The two-dimensional phase of boron nitride: few-atomic-layer sheets and suspended membranes. Appl. Phys. Lett. 92, 133107 (2008).

22. Hernandez, Y. et al. High-yield production of graphene by liquid-phase exfoliation of graphite. Nature Nanotech. 3, 563568 (2008). 23. Emtsev, K. V. et al. Towards wafer-size graphene layers by atmospheric pressure graphitization of silicon carbide. Nature Mater. 8, 203207 (2009). 24. Kim, K. S. et al. Large-scale pattern growth of graphene lms for stretchable transparent electrodes. Nature 475, 706710 (2009). 25. Stankovich, S. et al. Synthesis of graphene-based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon 45, 15581565 (2007). 26. Meyer, J. C., Girit, C. O., Crommie, M. F. & Zettl, A. Imaging and dynamics of light atoms and molecules on graphene. Nature 454, 319322 (2008). 27. Wilson, N. R. et al. Graphene oxide: structural analysis and application as a highly transparent support for electron microscopy. ACS Nano 3, 25472556 (2009). 28. Bunch, J. S. et al. Impermeable atomic membranes from graphene sheets. Nano Lett. 8, 24582462 (2008). 29. Booth, T. J. et al. Macroscopic graphene membranes and their extraordinary stiffness. Nano Lett. 8, 24422446 (2008). 30. Stolyarova, E. et al. Observation of graphene bubbles and effective mass transport under graphene lms. Nano Lett. 9, 332337 (2009). 31. Dikin, D. A. et al. Preparation and characterization of graphene oxide paper. Nature 448, 457460 (2007). 32. Chen, C. M. et al. Self-assembled free-standing graphite oxide membrane. Adv. Mater. 21, 30073011 (2009). 33. Hummers, W. S. & Offeman, R. E. Preparation of graphitic oxide. J. Am. Chem. Soc. 80, 1339 (1958). 34. Cote, L. J., Kim, F. & Huang, J. X. LangmuirBlodgett assembly of graphite oxide single layers. J. Am. Chem. Soc. 131, 10431049 (2009). 35. Gregoratti, L. et al. 48-Channel electron detector for photoemission spectroscopy and microscopy. Rev. Sci. Instrum. 75, 6469 (2004). 36. Kolmakov, A. et al. Spectromicroscopy for addressing the surface and electron transport properties of individual 1-D nanostructures and their networks. ACS Nano 2, 19932000 (2008). 37. Cumpson, P. J. & Seah, M. P. Elastic scattering corrections in AES and XPS. 2. Estimating attenuation lengths and conditions required for their valid use in overlayer/substrate experiments. Surf. Interface Anal. 25, 430446 (1997). 38. Powell, C. J. & Jablonski, A. Surface sensitivity of X-ray photoelectron spectroscopy. Nucl. Instrum. Methods A 601, 5465 (2009). 39. Akhavan, O. The effect of heat treatment on formation of graphene thin lms from graphene oxide nanosheets. Carbon 48, 509519 (2010). 40. Mkhoyan, K. A. et al. Atomic and electronic structure of graphene-oxide. Nano Lett. 9, 10581063 (2009). 41. Cote, L. J., Cruz-Silva, R. & Huang, J. X. Flash reduction and patterning of graphite oxide and its polymer composite. J. Am. Chem. Soc. 131, 1102711032 (2009). 42. Yang, D. et al. Chemical analysis of graphene oxide lms after heat and chemical treatments by X-ray photoelectron and micro-Raman spectroscopy. Carbon 47, 145152 (2009). 43. Xu, K., Cao, P. G. & Heath, J. R. Graphene visualizes the rst water adlayers on mica at ambient conditions. Science 329, 11881191 (2010). 44. Aleman, B. et al. Transfer-free batch fabrication of large-area suspended graphene membranes. ACS Nano 4, 47624768 (2010). 45. EMS & GF1200 Graphene Support Films for TEM (Electron Microscopy Supplies, 2010); available via www.emsdiasum.com.

Acknowledgements
A.K. thanks E. Strelcov, C. Watts and J. Bozzola (SIUC) for their help in the preparation of the experiment. The work at ELETTRA was partly supported by AMBIOSEN Friuli Venezia-Giulia regional grant 47/78. M.K.A. thanks P. Parisse for AFM measurements. The TEM and FIB work was performed in the EPIC facility of the NUANCE Center at Northwestern University. The NUANCE Center is supported by NSF-NSEC, NSF-MRSEC, the Keck Foundation, the State of Illinois and Northwestern University. The SIUC part of the research was supported by a NSF ECCS-0925837 grant. L.J.C and J.H. were supported by the NSF through a CAREER award (DMR 0955612).

Author contributions
A.K. conceived the project, designed and tested the E-cell prototypes, and assembled the manuscript, with contributions from all co-authors. D.D., L.C. and J.H. developed the methods of GO synthesis, processing and LangmuirBlodgett deposition onto SiO2/Si3N4 membrane samples. D.D. performed all micromachining and carried out SEM, TEM and HRTEM characterization of the GO overlayers and suspended membranes. M.K.A., M.A., L.G., S.G. and M.K conducted the SPEM experiments and the corresponding data analysis of the photoelectron images and spectra. A.K. and S.G. participated in spectromicroscopy tests as users of the ELETTRA ESCA microscopy beamline.

Additional information
The authors declare no competing nancial interests. Supplementary information accompanies this paper at www.nature.com/naturenanotechnology. Reprints and permission information is available online at http://www.nature.com/reprints. Correspondence and requests for materials should be addressed to A.K.

NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology

2011 Macmillan Publishers Limited. All rights reserved.