of 10
Supporting Information for
Selective-area, Water-Free Atomic Layer Deposition
of Metal Oxides on Graphene Defects
Michael F. Mazza,† Miguel Cabán-Acevedo†, Harold J. Fu†, Madeline C. Meier†, Annelise C.
Thompson†, Zachary P. Ifkovitz†, Azhar I. Carim†, and Nathan S. Lewis*†,
127-72, 210 Noyes Laboratory, Division of Chemistry and Chemical Engineering, California
Institute of Technology, Pasadena, California 91125, United States
Beckman Institute, California Institute of Technology, Pasadena, California 91125, United
States;
*Corresponding Author:
nslewis@caltech.edu
Materials and Methods
Monolayer Graphene
CVD-grown graphene on Si was purchased from Advanced Chemicals Supplier Material
(ACS Material) and was used as received. The Si(100) substrate was doped p-type with a
resistivity < 0.01
Ω
cm. Monolayer graphene was grown on Cu foil and transferred with a
PMMA (poly methylmethacrylate) protection layer
via
a wet transfer process. The grain size of
the as-deposited monolayer CVD-graphene film was ~ 50 μm prior to transfer.
Atomic Layer Deposition
Titanium dioxide (TiO
2
), aluminum oxide (Al
2
O
3
) and titanium aluminum oxide
(Ti
x
Al
y
O
z
) were deposited using a Cambridge Nanotech S200 ALD system. All depositions were
performed with 20 sccm N
2
carrier gas at a base pressure of ~ 0.6 torr. Titanium (IV) iso-
propoxide (TTIP, 98% STREM) was heated to 75
̊C
and used as the titanium source. Trimethyl
aluminum (TMA, 98% STREM) at room temperature was used as the aluminum source.
The ALD recipe for the water-free deposition was:
TTIP(0.013s) – 40s – TMA(0.011s) – 40s
(1)
Scanning Electron Microscopy
Scanning electron microscopy (SEM) was performed using a FEI Nova NanoSE 450 at
accelerating voltages between 3 and 15 keV with a 5 mm working distance.
X-Ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) was performed using a Surface Science M-
probe ESCA/XPS. The system had a chamber base pressure of <2 x 10
-8
. The sample was
illuminated by a 1486.7 eV Al K
monochromatic x-ray source at
35 ̊
relative to the surface.
Photoelectrons were collected by a hemispherical energy analyzer with a 100 mm radius and a
pass energy of 25 eV. The analyzer had a resolution of ~ 0.8 eV. Data were acquired using the
ESCA25 Capture (Service Physics, Bend OR; V5.01.04) software and were evaluated using
CasaXPS software. The high-resolution peaks were referenced to the graphene carbon 1s peak at
284 eV.
Kelvin Probe Force Microscopy
Atomic-force microscopy (AFM) was performed using a Bruker Dimension Icon.
Samples were affixed to a steel disc using Ag paint. The scan was performed with the PeakForce
setpoint at 5.0 nN, the scan rate at 0.501 Hz, and a PeakForce amplitude at 150 nm using a SCM-
PtSi (Bruker) probe with a nominal tip radius of 15 nm. PeakForce Kelvin Probe Force
Microscopy (PF-KPFM) was performed during the scan, in which height, adhesion, and potential
maps were collected concurrently. Adhesion maps provided qualitative data comparing relative
adhesion of the heterogeneous surface scanned and were not calibrated to a reference sample.
Raman Spectroscopy
Raman spectra were collected with a Renishaw inVia Raman microprobe equipped with a
Leica DM 2500 M microscope, a Leica N Plan 50x objective (numerical aperture = 0.75), an
1800 lines mm
-1
grating, and a CCD detector configured in a 180° backscatter geometry. A
λ
= 532 nm diode-pumped solid-state (DPSS) laser (Renishaw RL532C50) was used as the
excitation source and a ~100 μW radiant flux was incident on the surface of the sample. A
λ/4
plate was used to circularly
polarize
the incident excitation. No polarizing collection optic was
used.
Supporting Data
Thermal ALD Controls
ALD was performed on graphene/Si surfaces using water as the counter-reactant to
determine how the oxidant-free deposition chemistry differed from conventional ALD. The ALD
recipes were performed at mutually the same temperature, purge times, precursor dose times, and
number of cycles. The recipes are shown below:
TMA(0.011s) – 40s – H
2
O (0.015s) – 40s
(2)
TTIP(0.013s) – 40s – H
2
O(0.015s) – 40s
(3)
Figure S1 shows SEM images of the two controls. Aluminum oxide formed a near
conformal film after 200 cycles (Figure S1a). Line features assigned to initial defect nucleation
were observed through the film. Little deposition was observed with only moderate selectivity
for line defects for the titanium oxide control (Figure S1b). This result is consistent with a
substantially lower growth rate for TTIP relative to TMA at 110
̊C.
Water-based ALD recipes
supersaturate the basal plane with adsorbed water, resulting in fractional nucleation on the basal
plane. The film therefore becomes less selective, with precursors with higher growth rates more
quickly becoming conformal. These control experiments indicate that removing water from the
recipe directed anisotropic deposition and allowed for robust anisotropic films to form on the
surface before creating a conformal film.
Figure S1.
SEM images of monolayer graphene samples after 200 ALD cycles. A) Water-based
aluminum oxide deposition and B) Water-based titanium oxide deposition.
KPFM Mapping Data
KPFM mapping data were obtained concurrently with the AFM data shown in the main
text. These maps showed clear differences in the contact potential and in the adhesion force
between the line defects selectively passivated with Al
x
Ti
y
O
z
and the graphene basal plane.
These data were not included in the main text because the surface scanned was not calibrated to a
reference sample.
Figure S2.
KPFM mapping data showing A) contact potential and B) surface adhesion force
differences between the deposited defect regions and the pristine basal plane. Scale bars are 500
nm.
Temperature Dependence of Deposition Selectivity
The combined TTIP/TMA recipe was performed at various temperatures to optimize the
anisotropic growth behavior as well as to probe the surface chemistry (Figure S3). Each sample
was deposited according to the recipe outlined in the Methods section, and SEM data were
collected after 200 cycles. The coverage was strongly dependent on the deposition temperature.
At low temperatures, selectivity was lost and the metal oxide deposition resembled a
nonconformal film (Figure S3a). The deposition was most anisotropic at 110
̊C,
with very little
deposition observed outside of the line patterns (Figure S3b). This temperature-dependent
behavior is observed across similarly patterned regions on the sheet, with a notable increase in
selectivity at 110
̊C
(Figures S4a,b) At higher temperatures, deposition occurred at random
nucleation sites as well as at large-scale line networks (Figure S3c).
These results are consistent with a narrow regime in which selective-area ALD occurred.
At very low temperatures, precursors condensed on the graphene basal plane. At higher
temperatures, the metal-oxide precursors more readily reacted with the graphene basal plane,
leading to more conformal coverage and less deposition concentrated on defects.
Figure S3.
SEM images showing the temperature-dependent selectivity of the water-free ALD
recipe. Images show the deposition behavior at A) 95
̊C,
B) 110
̊C,
and C) 125
̊C.
Figure S4.
Lower-magnification SEM images showing differences in selectivity with similar
deposition patterns at A) 95
̊C
and B) 110
̊C.
Raman Spectroscopy
Raman spectra were taken on two cut regions of the same CVD-graphene on silicon
sample. One chip was left pristine and the second chip was exposed to 200 cycles of the water-
free ALD recipe at 100
̊C
as detailed in the methods section above. The spectra show that the
ALD recipe does not damage the graphene sheet, as shown by the relative intensities of the D
and G peaks.
Figure S5.
Raman spectrum showing the quality of the graphene layer without deposition
(black) and after 200 ALD cycles (red).
Statistical Estimate for Average Film Thickness
To experimentally determine an
average
film thickness, line traces were taken across 10
spots of the AFM map shown in Figure 2. The average thickness was determined to be 8 nm with
a standard deviation of 2 nm.
Figure S6. AFM map of the 10 spots used for the statistical determination for average film
thickness. The thickness is estimated at 8 ± 2nm.