Advances in gas-mediated electron beam induced etching and ... · Applied Physics A manuscript No....
Applied Physics A manuscript No. (will be inserted by the editor) Advances in gas-mediated electron beam induced etching and related material processing techniques Milos Toth Received: July 5, 2014/ Accepted: Insert Date Abstract Electron beam induced etching (EBIE) has traditionally been used for top-down, direct-write, chem- ical dry etching and iterative editing of materials. The present article reviews recent advances in EBIE mod- eling and emerging applications, with an emphasis on use cases in which the approaches that have convention- ally been used to realize EBIE are instead used for ma- terial analysis, surface functionalization, or bottom-up growth of nanostructured materials. Such applications are used to highlight the shortcomings of existing quan- titative EBIE models, and to identify physico-chemical phenomena that must be accounted for in order to en- able full exploitation and predictive modeling of EBIE and related electron beam fabrication techniques. Keywords: electron beam induced etching, direct-write nanofabrication, nanostructures, surface functionaliza- tion, self-assembly, radiation eﬀects, nanomaterials PACS codes: 81.65.Cf, 81.07.-b, 79.60.Dp, 68.43.Mn, 68.43.-h, 82.30.Lp, 61.80.Fe, 61.80.-x, 81.16.Dn 1 Introduction Gas-mediated electron beam induced etching (EBIE) is a direct-write, subtractive nanofabrication technique in which an electron beam and a precursor gas are used to realize chemical dry etching with a spatial resolu- tion of ∼ 10 nm [1–5]. EBIE is typically performed us- ing electron microscopes that are equipped with gas M. Toth* School of Physics and Advanced Materials University of Technology, Sydney PO Box 123, Broadway, NSW 2007, Australia E-mail: [email protected]injectors, and enable in-situ imaging and analysis of the features fabricated by an electron beam. The tech- nique is analogous to focused ion beam processing [6– 11], but avoids damage, staining and redeposition ar- tifacts caused by ion bombardment. EBIE is realized using gaseous precursors such as H 2 O, O 2 , NH 3 , XeF 2 , Cl 2 and SF 6 , which have been used to volatilize a wide range of materials, including graphene , single  and multi-walled  C nanotubes, amorphous carbon [15–18], single crystal [19–22] and nano-crystalline  diamond, Si, SiO 2 , Si 3 N 4 , Cr, Ti, TaN and photoresist [24–38]. Historical overviews and reviews of the EBIE technique and the underlying chemical pathways can be found in references [1–5]. The present article is fo- cused on recent advances in EBIE modeling methods, and emerging applications of EBIE and related electron beam material restructuring, fabrication and analysis methods. 2 Mechanisms EBIE precursor gases are injected into an electron mi- croscope specimen chamber using one of two methods. Either a capillary is positioned near the electron beam impact point at the substrate surface, and used to in- ject the gas into a chamber that is pumped continu- ously using a high vacuum pumping system [39,40]. Alternatively, the entire vacuum chamber, or a sub- chamber  is ﬁlled with a precursor gas, as is done in environmental electron microscopy [41–44]. Ideally, the precursor gas does not etch the substrate sponta- neously 1 . Instead, the chemical reactions that give rise to etching are driven by interactions between the inci- 1 This is, however, not true in some cases, such as when XeF 2 is used for etching of Si [45,46,31], TaN , or TaBN
Advances in gas-mediated electron beam induced etching and ... · Applied Physics A manuscript No. (will be inserted by the editor) Advances in gas-mediated electron beam induced
Gas-mediated electron beam induced etching (EBIE) is
a direct-write, subtractive nanofabrication technique in
which an electron beam and a precursor gas are used
to realize chemical dry etching with a spatial resolu-
tion of ∼ 10 nm [1–5]. EBIE is typically performed us-
ing electron microscopes that are equipped with gas
M. Toth*School of Physics and Advanced MaterialsUniversity of Technology, SydneyPO Box 123, Broadway, NSW 2007, AustraliaE-mail: [email protected]
injectors, and enable in-situ imaging and analysis of
the features fabricated by an electron beam. The tech-
nique is analogous to focused ion beam processing [6–
11], but avoids damage, staining and redeposition ar-
tifacts caused by ion bombardment. EBIE is realized
using gaseous precursors such as H2O, O2, NH3, XeF2,
Cl2 and SF6, which have been used to volatilize a wide
range of materials, including graphene , single 
and multi-walled  C nanotubes, amorphous carbon
[15–18], single crystal [19–22] and nano-crystalline 
diamond, Si, SiO2, Si3N4, Cr, Ti, TaN and photoresist
[24–38]. Historical overviews and reviews of the EBIE
technique and the underlying chemical pathways can
be found in references [1–5]. The present article is fo-
cused on recent advances in EBIE modeling methods,
and emerging applications of EBIE and related electron
beam material restructuring, fabrication and analysis
EBIE precursor gases are injected into an electron mi-
croscope specimen chamber using one of two methods.
Either a capillary is positioned near the electron beam
impact point at the substrate surface, and used to in-
ject the gas into a chamber that is pumped continu-
ously using a high vacuum pumping system [39,40].
Alternatively, the entire vacuum chamber, or a sub-
chamber  is filled with a precursor gas, as is done
in environmental electron microscopy [41–44]. Ideally,
the precursor gas does not etch the substrate sponta-
neously1. Instead, the chemical reactions that give rise
to etching are driven by interactions between the inci-
1 This is, however, not true in some cases, such as whenXeF2 is used for etching of Si [45,46,31], TaN , or TaBN
2 Milos Toth
dent and emitted electrons, and surface-adsorbed pre-
cursor molecules (Fig. 1(a)). When an electron beam ir-
radiates a substrate, precursor molecule adsorbates are
consumed in the etch reaction, and the local surface
concentration decreases to a steady state value, typi-
cally within ∼ 1 millisecond . The time-evolution of
adsorbate concentration at each point on the surface is
given by a competition between adsorbate consumption
in etching, desorption, and adsorbate replenishment.
The latter proceeds through adsorption from the gas
phase and diffusion along the substrate surface (Fig.
1(b)). The vertical etch rate (∂z/∂t) generally scales
with the product of the electron flux, f(x, y), and the
concentration of surface-adsorbed precursor molecules,
Na(x, y), at each point (x, y) on the substrate surface
(Fig. 1(c)). It can be calculated as a function of elec-
tron beam irradiation time (t) by solving differential
equations of the form [1,2]:
= Λ− kNa −∂Nα∂t
= V σfNa, (2)
where a and α signify surface-adsorbed precursor molecules
and dissociation products, respectively; Λ, kNa and∂Nα
∂t are the adsorption, desorption and electron in-
duced dissociation rates (per unit area), and D is the
precursor adsorbate diffusion coefficient. The constant
V is the volume of a single molecule removed from the
substrate in the etch reaction, and σ is an electron scat-
tering cross-section for the activated process that leads
to volatilization2. Adsorption is usually assumed to pro-
ceed through a single physisorbed state (Fig. 2(a)), and
surface coverage is normally limited to 1 ML by the
Λ = sF (1−Θ), (3)
where s is the sticking coefficient, F (x, y) is the gas
molecule flux at the substrate surface (given by the gas
pressure and temperature), and Θ(x, y, ) is the precur-
sor adsorbate coverage.
, where delocalized etching occurs spontaneously and theelectron beam is used to accelerate the local etch rate.2 It is usually assumed that electrons dissociate precursor
molecule adsorbates, thereby generating reactive fragmentswhich react with and volatilize the substrate . Hence, σ isan ‘effective’  cross-section for fragment generation. How-ever, in some cases, such as XeF2 EBIE of SiO2  andXeF2 EBIE of Si3N4 , etching has been argued to pro-ceed through a cyclic process of electron induced removal ofO (or N) from the substrate surface, and spontaneous etch-ing of excess Si by XeF2. Such processes can be modeled bythe above equations provided that σ is taken to represent across-section for the electron induced restructuring step thatleads to the removal of O (or N) from the surface.
The above modeling approach is also applicable to
the related technique of electron beam induced depo-
sition (EBID) [1–5,47–49], and is often used in studies
of deposition and etch kinetics. Reaction rate kinetics
are of interest because they affect spatial resolution,
proximity effects, fabrication rates, composition and the
topography of nanostructures fabricated by EBIE and
EBID [50–53,49,54,23,55,16]. In recent years, the basic
model defined by Eqns. 1–2 has been used (or modified)
to account for the following phenomena:
– The gas pressure distribution inside the electron mi-
croscope specimen chamber [39,40,56,57], which gov-
erns the gas molecule flux F (x, y) across the sub-
strate surface. The precursor pressure can vary sub-
stantially across the surface region irradiated by the
electron beam, particularly when the precursor gas
is delivered into the vacuum chamber using a capil-
lary located near the beam impact point at the sur-
face. Pressure distributions are significant because
they causes the EBIE rate to vary across the sub-
strate surface through Eqn. 3.
– Precursor transport into high aspect ratio pits .
The gas flow conductance of a pit decreases with
increasing aspect ratio, and hence alters the replen-
ishment rate of precursor molecules consumed in
EBIE. The replenishment rate affects Na which de-
termines the etch rate through Eqn. 2. Etch pit con-
ductance can be the dominant, etch-rate-limiting
process when fabricating high aspect ratio pits. It
causes the etch rate to decrease as the etch pit grows
during EBIE, giving rise to a characteristic, sub-
linear dependence of etch pit depth on electron beam
processing time .
– The behavior of etch reaction products (α) at the
substrate surface . The diffusion, desorption, and
electron induced re-dissociation of etch product molecules
can be modeled by setting up a differential equation
analogous to Eqn. 1 for each molecular species at the
substrate surface. Etch reaction product kinetics are
relevant if the molecules have a significant residence
time at the surface, or if multiple reaction steps are
needed to produce a volatile species that ultimately
desorbs from the surface. Electron induced dissoci-
ation of reaction products acts to reverse EBIE. It
can limit the etch rate, and can give rise to complex
dependencies of etch rate on electron flux .
– EBIE performed using a precursor gas mixture com-
prised of an etch precursor and a deposition pre-
cursor [16,54]. Mixtures are implemented by setting
up a differential equation for each molecular species
making up the gas, and can be used to simulate the
resulting dependencies of the etch (and deposition)
rate on electron flux. Mixtures play a role in EBIE
Advances in gas-mediated electron beam induced etching and related material processing techniques 3
when residual contaminants such as hydrocarbons
are present in the vacuum chamber and give rise to
unintended EBID that competes with etching [18,
29,16]. Mixtures can also be used to intentionally
modify and control reaction kinetics, and hence con-
trol the morphology  or composition [59–61,53,
62–66,49,67] of nanostructures fabricated by elec-
tron beam fabrication techniques.
– The potential well and energy barrier associated with
activated chemisorption , a type of adsorption
in which a gas molecule overcomes an energy bar-
rier and forms a chemical bond with a surface (Fig.
2(b)). Activated chemisorption alters the tempera-
ture dependence of the adsorbate concentration Na,
and enables electron beam chemical processing at el-
evated temperatures where the surface coverage of
physisorbed precursor molecules is negligible. This
has a number of benefits such as accelerated desorp-
tion of unwanted adsorbates (e.g. C-containing con-
taminants), and enables control over the species of
surface-adsorbed precursor molecules through par-
tial, thermal decomposition of the adsorbates.
– Spontaneous decomposition of precursor molecules
at the substrate surface, which can occur in paral-
lel with electron induced dissociation. This effect
has been modeled for the case of XeF2  which
can fragment through a dissociative chemisorption
pathway, leading to fluorination of many surfaces
[45,46,70,71] at room temperature. The model used
to simulate the spontaneous and electron induced
dissociation of XeF2  is a variant of a model
of activated chemisorption  that had been de-
veloped to describe the temperature-dependence of
EBID preformed using tetraethoxysilane (TEOS) as
the precursor gas.
– Chemically active (and inactive) surface sites that
enable (or inhibit) EBIE, and dynamic activation
of surface sites through electron beam restructuring
of the substrate . Surface site activation can en-
able EBIE of materials that can not be etched in
their virgin, unmodified state. It can also alter etch
kinetics, and give rise to a characteristic super-linear
dependence of etch pit depth on electron beam pro-
The above continuum models of EBIE [16,23,54,
55,58,69] and analogous (continuum and Monte Carlo)
models of EBID have been used to simulate the time-
evolution of the geometries (and in some cases the com-
position ) of structures fabricated by these tech-
niques. The key limitations of existing models are that:
(i) the model input parameters are often not known for
precursor-substrate combinations of interest, (ii) changes
in sample geometry caused by EBIE (or EBID) and
their consequences for the electron flux profile, f(x, y),
have not been modeled realistically over large areas
and for long processing times, and (iii) all of the in-
dividual effects listed above have thus far been stud-
ied in isolation, and are yet to be consolidated into
a generic, predictive model of etching and deposition.
Furthermore, a number of physical and chemical pro-
cesses have, to date, not been incorporated explicitly in
published EBIE models. These include electron stim-
ulated desorption [72–74], knock-on damage and sput-
tering caused by high energy (& 100 keV) electrons and
other mechanisms through which an electron beam can
alter the substrate composition and nanostructure [75–
82], the use of actual (rather than effective ) cross-
sections for EBIE, realistic multi-step reaction path-
ways, surface roughening caused by EBIE (Fig. 3(a-d)),
electron beam induced heating , and the effects of
charging  caused by electron injection into insula-
tors or electrically isolated substrate regions.
Traditionally, EBIE has been used as a direct-write sub-
croanal. 7(5), 397 (2001)57. G. Danilatos, J. Microsc.-Oxford 234, 26 (2009)58. M.G. Lassiter, P.D. Rack, Nanotechnology 19(45),
455306 (2008)59. A. Perentes, P. Hoffmann, J. Vac. Sci. Technol. B 25(6),
2233 (2007)60. A. Perentes, P. Hoffmann, Chem. Vapor Depos. 13(4),
176 (2007)61. M. Shimojo, M. Takeguchi, K. Furuya, Nanotechnology
17(15), 3637 (2006)
6 Milos Toth
62. A. Folch, J. Tejada, C.H. Peters, M.S. Wrighton, Appl.Phys. Lett. 66(16), 2080 (1995)
63. A. Folch, J. Servat, J. Esteve, J. Tejada, M. Seco, J. Vac.Sci. Technol. B 14(4), 2609 (1996)
64. K. Molhave, D.N. Madsen, A.M. Rasmussen, A. Carlsson,C.C. Appel, M. Brorson, C.J.H. Jacobsen, P. Boggild,Nano Lett. 3(11), 1499 (2003)
65. K. Molhave, D.N. Madsen, S. Dohn, P. Boggild, Nan-otechnology 15(8), 1047 (2004)
66. S. Wang, Y.M. Sun, Q. Wang, J.M. White, J. Vac. Sci.Technol. B 22(4), 1803 (2004)
67. F. Porrati, B. Kampken, A. Terfort, M. Huth, J. Appl.Phys. 113(5), 053707 (2013)
68. J. Bishop, C.J. Lobo, A. Martin, M. Ford, M.R. Phillips,M. Toth, Phys. Rev. Lett. 109, 146103 (2012)
69. S.J. Randolph, A. Botman, M. Toth, Particle 30(8), 672(2013)
70. M. Loudiana, A. Schmid, J. Dickinson, E. Ashley, Surf.Sci. 141, 409 (1984)
71. M. Hills, G. Arnold, Appl. Surf. Sci. 47(1), 77 (1991)72. R.D. Ramsier, J.T. Yates, Surf. Sci. Rep. 12(6-8), 243
(1991)73. W. Li, D.C. Joy, J. Vac. Sci. Technol. A 24(3), 431 (2006)74. W.F. van Dorp, T.W. Hansen, J.B. Wagner, J.T.M.
De Hosson, Beilstein J. Nanotechnol. 4, 474 (2013)75. F. Banhart, Rep. Prog. Phys. 62(8), 1181 (1999)76. R.F. Egerton, P. Li, M. Malac, Micron 35(6), 399 (2004)77. R.F. Egerton, R. McLeod, F. Wang, M. Malac, Ultrami-
croscopy 110(8), 991 (2010)78. A.V. Krasheninnikov, K. Nordlund, J. Appl. Phys.
107(7), 071301 (2010)79. A. Botman, C.W. Hagen, J. Li, B.L. Thiel, K.A. Dunn,
J.J.L. Mulders, S. Randolph, M. Toth, J. Vac. Sci. Tech-nol. B 27(6), 2759 (2009)
80. J. Li, M. Toth, V. Tileli, K.A. Dunn, C.J. Lobo, B.L.Thiel, Appl. Phys. Lett. 93(2), 023130 (2008)
81. J.T. Li, M. Toth, K.A. Dunn, B.L. Thiel, J. Appl. Phys.107(10), 103540 (2010)
82. F. Porrati, R. Sachser, C.H. Schwalb, A.S. Frangakis,M. Huth, J. Appl. Phys. 109(6) (2011)
84. J. Cazaux, J. Appl. Phys. 95(2), 731 (2004)85. M. Toth, C.J. Lobo, W.R. Knowles, M.R. Phillips, M.T.
Postek, A.E. Vladar, Nano Lett. 7(2), 525 (2007)86. J. Bishop, M. Toth, M. Phillips, C. Lobo, Appl. Phys.
Lett. 101(21), 211605 (2012)87. N. Vanhove, P. Lievens, W. Vandervorst, Appl. Surf. Sci.
255(4), 1360 (2008)88. N. Vanhove, P. Lievens, W. Vandervorst, Surf. Interface
Anal. 43(1-2), 159 (2010)89. T. Shanley, A.A. Martin, I. Aharonovich, M. Toth, Appl.
Phys. Lett. in press (2014)90. A. Botman, A. Bahm, S. Randolph, M. Straw, M. Toth,
Phys. Rev. Lett. 111(13), 135503 (2013)91. T. Lukasczyk, M. Schirmer, H.P. Steinruck, H. Marbach,
Langmuir 25(19), 11930 (2009)92. M. Walz, M. Schirmer, F. Vollnhals, T. Lukasczyk,
H.P. Steinruck, H. Marbach, Angewandte Chemie-international Edition 49, 4669 (2010)
93. F. Vollnhals, P. Wintrich, M.M. Walz, H.P. Steinruck,H. Marbach, Langmuir 29(39), 12290 (2013)
94. F. Vollnhals, T. Woolcot, M.M. Walz, S. Seiler, H.P.Steinruck, G. Thornton, H. Marbach, The Journal ofPhysical Chemistry C 117(34), 17674 (2013)
95. M.M. Walz, F. Vollnhals, F. Rietzler, M. Schirmer, H.P.Steinruck, H. Marbach, Appl. Phys. Lett. 100(5), 053118(2012)
96. K. Muthukumar, H.O. Jeschke, R. Valentı, E. Begun,J. Schwenk, F. Porrati, M. Huth, Beilstein J. Nanotech-nol. 3, 546 (2012)
97. M.N. Khan, V. Tjong, A. Chilkoti, M. Zharnikov, AngewChem Int Edit 51(41), 10303 (2012)
98. H. Ebinger, J. Yates, Phys. Rev. B 57(3), 1976 (1998)99. J.B. Lassiter, M.W. Knight, N.A. Mirin, N.J. Halas, Nano
Lett. 9(12), 4326 (2009)
Advances in gas-mediated electron beam induced etching and related material processing techniques 7
Distance above the substrate surface
Distance above the substrate surface
gas phasephysisorbed state
Fig. 2 Potential energy diagrams for physisorption (a) andactivated chemisorption (b) used in models of gas-mediatedelectron beam induced processing [68,69].
Nanotechnology 23 (2012) 145301 N A Roberts et al
Figure 1. SEM images of the process steps of the standard conductive AFM tip. (a) As-received standard conductive AFM tip. (b) 30 nm Ptsputtered onto AFM tip. (c) 50 nm SiNx PECVD. (d) Electron beam induced etch with XeF2. (e) Electrochemical deposition of Au.(f) Schematic of the setup for electroplating of the isolated AFM tip. (g) Current measurement during electrochemical deposition. Thedeposition begins at the location labeled start and ends at the location labeled end.
The latter necessitates a reduced interaction region of the apexof the tip, and ideally a higher aspect ratio of the conicalpart of the probe . One method for the size reduction orincreased aspect ratio is to attach a high aspect ratio material,such as carbon nanotube (CNT) or nanowire, at the probeapex . In this paper, we demonstrate a synthesis route foran electrically shielded carbon nanotube scanning probe tipoffering high resolution and electrical isolation.
2. Standard insulated scanning probes
Illustrated in figure 1 is a sequence of scanning electronmicroscopy (SEM) images demonstrating the steps involvedin synthesizing an insulated (covered by insulator) orshielded (having additional external shield) scanning probetip. Figure 1(a) shows the as-received silicon scanning probeand 1(b) shows the tip after a 30 nm Pt conductive layerand etch stop layer was deposited (albeit thinner at the tip
due to geometric considerations of the nearly line-of-sitedeposition). The platinum sputtering conditions were 10 Wdc power (150 mm diameter target), 3 mTorr pressure andan approximate source to substrate distance of 7.5 cm.Figure 1(c) shows the tip after a 50 nm thick silicon nitridefilm was deposited (process conditions given below) andfigure 1(d) shows the tip after a selective focused electronbeam induced etching at the tip apex. For more informationon electron beam induced processing (EBIP) see [15–17].In addition to the listed reviews of EBIP, recent work byLobo et al  demonstrated sub-1 nm length scales with acombined electron beam induced etch (EBIE) and deposition(EBID) process via simulations.
The high uniformity of dielectric coating can bedemonstrated by the electroplating experiment illustratedin figure 1(e). Here, the shielded SPM tip is mountedin an Asylum Research MFP-3D AFM platform and isthen submerged in a droplet of gold electroplating solution
it is possible to perform a direct etching of graphene byinjecting oxygen gas in the SEM chamber during imaging atlow voltage (5 kV in this case) and exploiting the electronbeam in order to locally create ozone and oxygen radicals.36
We verified that the EBIE is particularly effective on SLGand we were able to monitor the etching process in real timeby observing how the contrast changed during a line scan onthe graphene flake. Figure 5(a) shows the SEM image of thegraphene flakes together with the vertical scan line (dashedred line) where the single-line lithographic process wasapplied. Figure 5(b) reports the SEM image of the SGL afterthe spatially resolved etching. Two of the three larger blistersalready shown in Fig. 4(a) were strongly modified by theetching and the graphene membrane relaxed onto the sur-face. One of the blisters was pierced by direct EBIE [Fig.5(c), red arrow]. In this case the result was an almost com-plete flattening of the suspended SLG membrane [note thechange in the dark image contrast between Fig. 5(a), 5(b),
and 5(c)]. Strain relaxation probably originates from dam-ages to the crystalline structure of graphene induced by oxy-gen. This interpretation was strengthened by the emergenceof a strong D peak in the Raman spectra [inset in Fig. 5(c)]performed after the spatially resolved etching, in the vicinityof the exposed areas.
Blisters in graphene produced by standard microme-chanical exfoliation (at much lower density compared to thepresent case) were observed and “bubbles” were alsoinduced in exfoliated graphene by chemical methods andproton irradiation, as described in Ref. 29. Similar results ongraphene or other single-layer materials obtained by PDMS-based transfer printing were reported: in those cases the pres-ence of blistering was not highlighted by the authors but isevident from the published AFM images.37 Recently, anothergroup demonstrated that gold and silver nanoparticles depos-ited on top of SiO2 substrates can create suspended graphenemembranes,38 as we observed in correspondence of the nano-scale debris present on our substrates.
We believe that a more precise control of the pressureapplied in the PDMS transfer-printing technique will favorthe production of graphene layers having a surface topogra-phy ranging continuously from “flat” to “blistered” shapes.
In conclusion, PDMS transfer-printing represents a veryuseful approach for the production of graphene from bulkgraphite with much reduced contamination with respect toother transfer-printing methods that rely on scotch tape, resinor PMMA-based materials. We believe that it constitutes astraightforward methodology to investigate strain-inducedeffects in single-layer graphene. Moreover further improve-ments and optimization of the PDMS transfer-printingmethod reported here may allow the controlled and orderedself-assembly of blistering and rippling on single graphenelayers on various substrates (e.g., as recently studied in sus-pended graphene membranes with thermally-generatedstrain),39,40 allowing a systematic investigation of the effectof blistering on the electronic and optical properties of gra-phene and their links and interactions with different substratespecies. Finally, in order to prove the nature of the blisterednanostructures, a spatially resolved direct etching of gra-phene was successfully demonstrated employing e-beaminduced ionization of O2 in a SEM vacuum chamber. Wethink that this technique could become a valuable tool forgraphene lithography once its impact on electronic and opti-cal properties of the resulting patterned nanostructure isproperly assessed.41
We would thank Dr. Marco Cecchini for technicalassistance during the PDMS sample preparation andfunctionalization.
1K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V.Dubonos, I. V. Grigorieva, and A. A. Firsov, Science 306, 666 (2004).
2A. K. Geim and K. S. Novoselov, Nature Mater 6, 183 (2007).3Y. Zheng, Y. W. Tan, H. L. Stormer, and P. Kim, Nature 438, 201 (2005).
FIG. 5. (Color online) SEM images of the same SLG region analyzed inFigs. 3 and 4. (a) Large view of the sample: the dashed line is the one whereEBIE was performed. (b) SEM picture of the SLG after the spatially resolvedEBIE: the obtained cut has a width of 37 nm and some blisters disappeared,relaxing on the substrate. (c) The same region after EBIE pricking of the cen-tral blister (evidenced by the red arrow). Inset: Raman spectrum on the sameSLG after the EBIE process. Notice the appearance of the D peak.
064308-5 Goler et al. J. Appl. Phys. 110, 064308 (2011)
Downloaded 30 Oct 2011 to 184.108.40.206. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
a) b) c) d)
EBIO gap formation. During the EBIO process, a constantsource-drain voltage VSD of maximum 1 V was applied andthe current continuously measured in intervals of 50 ms.
To cut the carbon nanotubes, an electron-beam line scanwas executed across the nanotube while injecting the oxy-gen. During line scans, the microscope magnification wasadjusted to be either 25 kX or 50 kX, yielding a line scanwidth of 4.57 lm and 2.29 lm, respectively. The primaryelectron-beam current used was !100 pA, which yields aline dose of !21.8 lC/m and !43.7 lC/m per second,respectively. Acceleration voltage of the primary electrons(PEs) was set to 10 kV. Scale-calibrated images were used toassess the reproducibility of gap sizes.
Figs. 1(b) and 1(c) show SEM images of an individualmSWNT on SiO2/Si and on Si3N4 after EBIO-induced gapformation. Both images were recorded with an in-lens detec-tor that is expected to be sensitive to SE1 and SE2 secondaryelectrons.14 For better visibility of the mSWNT on SiO2/Si,we used voltage-contrast scanning electron microscopy(VCSEM) to suppress the SEs of the SiO2/Si substrate.15
The required backgate voltage Vg"#10 V was applied withrespect to the source-drain electrodes only for imaging. Aneffect of Vg on the EBIO process was not observed.Although Fig. 1(c) was recorded without VCSEM, the mem-brane appears dark due to its limited SE2 generation.16 TheEBIO-induced gaps shown in Figs. 1(b) and 1(c) are !20 nmwide. This is a typical result since the gap size average over62 devices on SiO2/Si substrates is (19 6 5) nm, as shown inFig. 1(d). We should note that the formation of a gap hasalso been confirmed by topography measurements with anatomic force microscope (not shown). EBIO is not limited tothe cutting of individual nanotubes but can also be extendedto form gaps in devices with multiple nanotubes. Fig. 2(c)
shows such an example. With EBIO, gaps of similar size canbe produced in all of the nanotubes at the same relative posi-tion. Such a result cannot be obtained by current-inducedoxidation.17–19
The critical line dose nC that is required for gap forma-tion has been determined from in-situ conductance measure-ments. Fig. 2(a) shows the typical conductance reduction ofan mSWNT-device on SiO2/Si and Si3N4, during the EBIOprocess. The traces of the conductance G have been normal-ized to their initial values Gi, at the beginning of the EBIOprocess at t" 0. Gi is typically on the order of (40 kX)$1 af-ter current-annealing. G decreases with n roughly exponen-tially by two orders of magnitude, before a sudden drop byadditional 3-4 orders of magnitude down to the electron-beam-induced residual conductance is recorded< (10GX)$1. This last step indicates the formation of a gap in thenanotube. There are no plateaus in G which could indicate astep-wise introduction of defects. Typically, nC is on theorder of a few mC/m. Conductance traces for several thin-film devices during the EBIO process are shown in Fig. 2(b).The critical dose is comparable to or larger than the dose forsingle-tube devices, which is likely due to the presence ofoverlapping nanotubes in some of the thin-film devices.
The conductance drop remains irreversible even atVSD % 10 V. This behavior is in marked contrast to the con-tinuous electron-beam-induced metal-to-insulator transitions,which are caused by charges trapped in the substrate sur-face.13 It has been shown that such transitions require a linedose of !200 lC/m at 10 kV PE energy. During the EBIOprocess, such a small dose causes only a small reduction inG. We assume that charging is not important here since oxy-gen ions are compensating electron-beam-induced surfacecharges.
FIG. 1. (Color online) (a) Schematic of the EBIO setup: Primary-beam electrons are repeatedly scanned across an mSWNT in the presence of injected oxygen.The mSWNTs are wired to Pd electrodes and supported by SiO2/Si substrates or Si3N4 membranes (inset). Indicated are the trajectories of secondary (SE1,SE2) and BSE. (b) SE image of an mSWNT on SiO2 after gap formation, recorded with Vg"#10 V. (c) SE image of an mSWNT on a Si3N4 membrane aftergap formation. (d) Histogram of nanogap sizes of 62 devices on SiO2/Si substrates, fitted to a normal distribution with l" 19 nm and r" 5 nm.
FIG. 2. (Color online) (a) Normalized conductance G/Gi vs. line dose n recorded during the EBIO process for a single-tube device on SiO2/Si and on a Si3N4
membrane. (b) Normalized conductance G/Gi vs. line dose n recorded during the EBIO process for several thin-film devices on SiO2/Si. The spread in criticaldose is likely due to the presence of overlapping nanotubes in some of the thin-film devices. (c) SE image of an mSWNT thin-film device after gap formation.
173105-2 Thiele et al. Appl. Phys. Lett. 99, 173105 (2011)
Downloaded 15 Mar 2012 to 220.127.116.11. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
Fig. 3 (a-d) Four frames from a movie showing the formationof a nano-gap in a carbonaceous nanowire on a bulk SiO2 sub-strate by high resolution, H2O EBIE , (e) scanning probemicroscopy tip sculpted by XeF2 EBIE , (f) etch pit inchrome fabricated by XeF2 EBIE , (g) 37 nm gap etchedin graphene by O2 EBIE , (h) gap cut into a single-walledcarbon nanotube by O2 EBIE , (i) false-color electron im-age of a Ga-filled GaF3 microcapillary grown using XeF2 asshown in Fig. 4(c) .
8 Milos Toth
a) b) c) e- e- e-
Fig. 1 General steps involved in electron beam induced etching: (a) electron beam irradiation, and emission of secondary andbackscattered electrons from the substrate, (b) consumption of adsorbates in the etch reaction, and precursor replenishmentthrough adsorption from the gas phase and diffusion along the substrate surface, (c) volatilization of the substrate in thevicinity of the electron beam.
a) b) e-
Fig. 4 Three forms of material processing driven by electron dissociation of XeF2 adsorbates: (a) fluorination of the substratesurface , (b) electron beam induced etching, and (c) fabrication of gallium-filled, GaF3 capillaries by ion bombardmentof GaN . GaF3 forms due to XeF2 decomposition by secondary electrons which are emitted from GaN and irradiate thesidewall of the growing pillar. An electron image of such a microstructure is shown in Fig. 3(i)