UC Riverside
UC Riverside Previously Published Works
Title
Electronic doping and scattering by transition metals on graphene
Permalink
https://escholarship.org/uc/item/2602b784
Journal
Physical Review B - Condensed Matter and Materials Physics, 80(7)
ISSN
1098-0121
Authors
Pi, K
McCreary, KM
Bao, W
et al.
Publication Date
2009-09-21
DOI
10.1103/PhysRevB.80.075406
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Electronic doping and scattering by transition metals on graphene
K. Pi, K. M. McCreary, W. Bao, Wei Han, Y. F. Chiang, Yan Li, S.-W. Tsai, C. N. Lau, and R. K. Kawakami
*
Department of Physics and Astronomy, University of California, Riverside, California 92521, USA
共Received 10 June 2009; published 5 August 2009
兲
We investigate the effects of transition metals 共TM兲 on the electronic doping and scattering in graphene
using molecular-beam epitaxy combined with in situ transport measurements. The room-temperature deposi-
tion of TM onto graphene produces clusters that dope n type for all TM investigated 共Ti, Fe, and Pt兲. We also
find that the scattering by TM clusters exhibits different behavior compared to 1/ r Coulomb scattering. At high
coverage, Pt films are able to produce doping that is either n type or weakly p type, which provides experi-
mental evidence for a strong interfacial dipole favoring n-type doping as predicted theoretically.
DOI: 10.1103/PhysRevB.80.075406 PACS number共s兲: 73.63.⫺b, 72.10.Fk, 73.23.⫺b, 73.40.Ns
I. INTRODUCTION
Transition-metal 共TM兲 adatoms and clusters on graphene
have recently been a topic of great interest: at low density,
they are expected to induce doping, scattering,
1
and novel
magnetic
2–4
and superconducting
5
behavior; at high density
共up to continuous coverage兲, they may locally dope or
modify the band structure of graphene.
6–8
Because of their
importance for graphene-based electronics and the investiga-
tion of novel phenomena,
1–14
there have been extensive the-
oretical studies.
1–8,13,14
In contrast, the experimental explora-
tion of TM/graphene systems is much more limited.
A key issue to investigate is the charge transfer between
the TM and graphene because it is responsible for both the
local doping and the charged impurity scattering. Generally,
the relative work function 共WF兲 between the TM and the
graphene is believed to be an important factor for determin-
ing the charge transfer,
11
i.e., graphene will be p doped
共n doped兲 if the TM’s WF is larger 共smaller兲 than graphene.
Recently, density-functional calculations predict the presence
of a strong interfacial dipole that promotes the n-type doping
of graphene.
8
However, experimental studies of the local
doping by TM contacts have yet to find evidence for this
strong interfacial dipole layer.
9–12
In this work, we report in situ transport measurements of
TM/graphene systems as a function of TM coverage for sev-
eral different metals using a molecular-beam epitaxy 共MBE兲
deposition system with built-in electrical probes. Similar
techniques have been applied to study gases,
15
molecules,
16
and alkali-metal
17
adatoms on graphene. The metals used in
the study are Ti, Fe, and Pt, with WF of 4.3, 4.7, and 5.9
eV,
18
respectively 关the WF of graphene is 4.5 eV 共Refs. 19
and 20兲兴. Surprisingly, at low coverage, the TM clusters dope
graphene n type, regardless of its WF relative to that of
graphene. For the scattering at low coverage, we find that the
scattering by TM clusters exhibits different behavior com-
pared to 1/ r Coulomb scattering. Extending to high cover-
age, we make the important observation that Pt films are able
to produce doping that is either n type or weakly p type.
Because WF considerations alone would predict strong
p-type doping, this result provides experimental evidence for
the strong interfacial dipole favoring n-type doping as calcu-
lated theoretically.
8
II. EXPERIMENTAL PROCEDURES
Samples are prepared by mechanical exfoliation of Kish
graphite onto a SiO
2
/ Si substrate 共300 nm thickness of
SiO
2
兲. Single-layer graphene flakes are identified by optical
microscopy and Raman spectroscopy.
21
Figure 1共a兲 shows a
scanning electron microscope 共SEM兲 image of a typical
graphene device with Au/Ti electrodes defined by e-beam
lithography. The device is annealed under Ar/ H
2
environ-
ment at 200 °C for1htoremove resist residue
22,23
and
degassed in ultrahigh vacuum at 90 °C for 1 h. The room-
temperature MBE deposition of TM atoms 共growth
No Pt (0 ML)
0.025 ML
0.071 ML
0.4
0.6
0.2
0.0
-50 0 50
(e) Pt-1
0.127 ML
-100
No Fe (0 ML)
0.041 ML
Gate Voltage (V)
(d) Fe-1
Conductivity (mS)
2.0
No Ti (0 ML)
0.0038 ML
0.0077 ML
0.015 ML
(c) Ti-1
2.0
0.5
1.5
1.0
0.0
0.123 ML
0.205 ML
1.0
1.5
0.5
0.0
0nm
10 n
m
2 μm
(a)
(b)
FIG. 1. 共Color兲共a兲 SEM image of a graphene device with
Au共100 nm兲/Ti共10 nm兲 electrodes. 共b兲 AFM image of 0.01 ML of Pt
deposited on single-layer graphene. 关共c兲–共e兲兴 The gate-dependent
conductivity at selected TM coverage for Ti, Fe, and Pt,
respectively.
PHYSICAL REVIEW B 80, 075406 共2009兲
1098-0121/2009/80共7兲/075406共5兲 ©2009 The American Physical Society075406-1
pressure⬍7⫻ 10
−10
torr兲 is calibrated by a quartz deposition
monitor. The coverage is converted from atoms/ cm
2
to
“monolayers” 共ML兲, where 1 ML is defined as 1.908
⫻10
15
atoms/ cm
2
, the areal density of primitive unit cells in
graphene. For low coverage, the room-temperature deposi-
tion of TM leads to clustering as shown in the atomic force
microscope 共AFM兲 image of 0.01 ML Pt on graphene 关Fig.
1共b兲兴. The presence of isolated adatoms cannot be ruled out
by the AFM but are unfavorable theoretically.
6
In situ trans-
port measurements are performed using standard lock-in de-
tection 共1
A excitation兲.
III. RESULTS AND DISCUSSION
The fine control of TM deposition provides the ability to
probe the effect of small amounts of material on the transport
properties of graphene. Figure 1共c兲 shows representative
gate-dependent conductivity scans for various thicknesses of
Ti in the low-coverage regime. The minimum in the gate-
dependent conductivity identifies the position of the Dirac
point 共V
D
兲 while the slope corresponds to the mobility of
charge carriers in the graphene. With increasing coverage,
two characteristic behaviors are observed. First, the introduc-
tion of Ti on the graphene surface results in shifting the
Dirac point toward more negative gate voltages, indicating
that the Ti is a donor, producing n-type doping in the
graphene. Second, the slope of the conductance curves away
from the Dirac point decreases, indicating that the Ti intro-
duces additional scattering to lower the mobility. Both of
these characteristics are also observed for Fe doping 关Fig.
1共d兲兴 and Pt doping 关Fig. 1共e兲兴.
Figure 2 highlights the relation between the Dirac point
shift 共 V
D,shift
=V
D
−V
D,initial
兲 and TM coverage for a collec-
tion of Ti, Fe, and Pt samples in the low-coverage regime.
Despite the sample-to-sample variations which may be due
to differences in the graphene surface purity, growth rate
uncertainties, and the possible dependence of graphene WF
on flake size or edge roughness,
24
several important features
are discovered. First, all samples, including the Pt samples
with WF greater than graphene, result in n-type doping. Sec-
ond, the three different TM result in three different ranges for
slopes, with the Ti samples exhibiting the most negative ini-
tial slopes 共−2169 to −4602 V/ ML兲. From this value the
doping efficiency or number of electrons transferred per Ti
atom to graphene is determined by knowing the carrier con-
centration associated with the given change in gate voltage
共⌬n =
␣
⌬V
g
, where
␣
=7.2⫻ 10
10
V
−1
cm
−2
based on calcu-
lated capacitance values兲. The doping efficiency is in the
range of 0.082–0.174 electrons per Ti atom. The Fe shows
the next strongest efficiency 共0.017 to 0.046兲 while the Pt is
the weakest electron donor with the efficiency of 0.014 to
0.021 electrons transferred for each Pt atom. Upon recalling
the bulk WFs of Ti 共4.3 eV兲,Fe共4.7 eV兲, and Pt 共5.9 eV兲,it
is apparent that the WF of the TM is related to the doping
efficiency, with electrons being more easily transferred from
the lowest WF material, Ti, compared to the highest WF
material, Pt. However, the magnitudes of the doping effi-
ciency do not vary linearly with the WF of the TM. There-
fore, in addition to the work function, other effects such as
wave-function hybridization or structural modifications may
contribute to the electronic doping of graphene.
Figures 3共a兲–3共c兲 show the conductivity as a function of
carrier concentration 关n=−
␣
共V
g
–V
D
兲兴. The electron and
hole mobilities are determined by taking the slope of the
conductivity away from the Dirac point
共
e,h
=兩⌬
/ e⌬n兩兲.
15,17
Figures 3共d兲–3共f兲 illustrate the de-
tailed dependence of mobility on the TM coverage for Ti, Fe,
and Pt samples in the low-coverage regime. Comparing the
different samples at equivalent coverages, the Ti exhibits the
strongest scattering and Pt has the weakest scattering. Noting
that the trend in the scattering 共Ti⬎Pt兲 matches that of the
doping efficiency, we investigate this relationship by plotting
the normalized mobility
25
against the Dirac point shift 关Fig.
3共g兲兴. The average mobility,
=共
e
+
h
兲/ 2, is plotted for Ti
and Pt. The Fe samples typically exhibit a reduction in hole
mobility which is most pronounced in sample Fe-2, so
e
and
h
are plotted separately. Comparing the different mate-
rials shows that the mobility reduction in Ti, Pt, and Fe 共elec-
trons兲 is much more strongly related to the Dirac point shift
than the TM coverage 关Fig. 3共g兲兴. Because the Dirac point
shift not only measures the doping level in the graphene but
also the average charge density of the TM, the data shows
that the scattering is related to the average charge density of
the clusters—a characteristic, that is, plausible for Coulomb
scattering. However, we point out that this behavior is actu-
ally different from what is calculated for Coulomb scattering
by pointlike scatterers with 1/ r potential.
26
Specifically, in
Ref. 26, the scattering per impurity does not scale linearly
with the impurity charge 共
␣
兲 and instead has a strong qua-
dratic component, resulting in scattering that scales as
␣
2
n
imp
=
␣
共
␣
n
imp
兲⬃
␣
共V
D,shift
兲. Due to the presence of the
material-dependent
␣
factor 共i.e., doping efficiency兲, the
mobility vs Dirac point shift curves should be significantly
different for different materials. Therefore, the observed scat-
tering by TM clusters exhibits behavior that differs from 1/ r
Coulomb scattering by isolated impurities.
1
Additionally, we analyze the power-law relationship be-
tween the scattering and doping effects. The total scattering
rate is ⌫=⌫
0
+⌫
TM
, where ⌫
0
is the scattering rate of the
Pt-1
Pt-2
Fe-2
Fe-1
Fe-3
Ti-1
Ti-2
Ti-3
TM Covera
g
e
(
ML
)
-40
0
-80
-120
0.000 0.025 0.050 0.075
0.1
0.2
0.3
F
erm
iL
eve
l Shif
t
(
e
V)
0
Pt-3
V
D,shift
(
V
)
FIG. 2. 共Color兲 Dirac point shift vs coverage for nine separate
samples. The dashed lines indicate the linear fit used to define the
doping efficiencies, which are: 0.174, 0.092, and 0.082 electrons/
atom for Ti-1, Ti-2, and Ti-3, respectively; 0.017, 0.040, and 0.046
electrons/atom for Fe-1, Fe-2, and Fe-3, respectively; 0.014, 0.021,
and 0.019 electrons/atom for Pt-1, Pt-2, and Pt-3, respectively.
PI et al. PHYSICAL REVIEW B 80, 075406 共2009兲
075406-2
undoped sample and ⌫
TM
is the scattering rate induced by the
TM. Because mobility is inversely proportional to scattering,
the quantity 1/
−1/
0
is proportional to ⌫
TM
. The relation-
ship between the Dirac point shift and ⌫
TM
is investigated by
plotting −⌬V
D,shift
vs 1/
−1/
0
on a log-log scale 关Fig.
3共h兲兴. The dashed lines are power-law fits, −⌬V
D,shift
⬃共⌫
TM
兲
b
, with values of b ranging from 0.64–1.01 as indi-
cated in the figure caption. Compared to the results of Chen
et al.
17
which find values of b =1.2–1.3 for scattering by
isolated potassium impurities, our results with bⱕ 1 indicate
a different behavior for scattering by TM clusters.
A surprising result from the studies at low coverage 共Figs.
1–3兲 is the n-type doping of graphene by Pt. If the WF is the
only factor affecting the transfer of electrons between mate-
rials, Pt is expected to dope graphene strongly p type, since
the WF of Pt 共5.9 eV兲 is significantly larger than that of
graphene 共4.5 eV兲. Density-functional calculations of bulk
TM on graphene
8
present a possible explanation for this ob-
served behavior by predicting the formation of an interfacial
dipole layer resulting in a potential step to favor n-type dop-
ing 共⌬V=0.9 eV兲. So far, however, there has been no ex-
perimental evidence for such a strong dipole layer forming at
the interface between a bulk TM and graphene.
9–12
To inves-
tigate the theoretical prediction of a strong interfacial dipole
layer between the graphene and bulk TM, we extend the
Pt-doping study to higher coverage to study the charge trans-
fer from Pt films. Figure 4共a兲 displays V
D
as a function of
coverage for several Pt-doped samples. An initial rapid shift
toward negative voltages is observed in all samples. As more
Pt is deposited, bringing the sample into the medium-
coverage regime, the rate of shift in V
D
slows and reaches a
minimum value before gradually increasing toward more
positive voltages. At high coverage, the Dirac point stabilizes
and shows very little variation with additional deposition.
The sample morphology is measured by ex situ AFM. The
AFM image for 0.62 ML of Pt shows that the Pt is still in the
form of isolated clusters 关Fig. 4共b兲兴. At the higher coverage
of 3.19 ML, the Pt forms a connected film with some uncov-
ered regions of graphene 关Fig. 4共c兲兴. The connected film pro-
vides a parallel conduction pathway that contributes to the
measured conductivity value but should not be gate depen-
dent. The gate dependence of the conductivity is primarily
due to the chemical-potential shift in the graphene that is not
covered by the metal. For graphene in direct contact with the
metal, the local chemical potential is pinned, exhibiting no
gate dependence. However, the gate dependence of the un-
covered graphene regions and the voltage of the conductance
minima 共V
D
兲 still provide a reliable measure of the electronic
-V
D,shift
(V)
1.0
0.5
0.0
0
-20 -40 -60 -8
0
Normalized mobility, μ/μ
0
Dirac Point Shift (V)
0.008 ML
0.029 ML
Pt-1
Pt-2
Pt-3
Ti-1
Ti-2
Fe-2 electron
Fe-2 hole
10
-4
10
-3
1/μ-1/μ
0
(Vs/cm
2
)
No Ti
0.0038 ML
0.0077 ML
0.015 ML
No Fe
0.003 ML
0.012 ML
0.018 ML
No Pt
0.0051 ML
0.010 ML
0.018 ML
Electron
Hole
-3 0 3
2.0
1.0
0.0
0 0.01 0.02
3
2
1
2.0
1.0
0.0
3
2
1
3
2
1
0.5
0.0
1.5
1.0
Electron
Hole
Electron
Hole
n (10
12
cm
-2
)
Conductivity (mS)
Coverage (ML)
Mobility, μ (10
3
cm
2
/Vs)
Pt-1
Pt-2
Pt-3
Ti-1
Ti-2
Fe-2 electron
Fe-2 hole
(a)
(b)
(c)
(d)
(e)
(f)
(g)
Fe-2
Ti-1
Pt-2
(h)
0.102 ML
FIG. 3. 共Color兲关共a兲–共c兲兴 The conductivity vs carrier concentra-
tion for Ti, Fe, and Pt, respectively, for four different TM coverages.
关共d兲–共f兲兴 The electron and hole mobilities for Ti, Fe, and Pt, respec-
tively, as a function of TM coverage. 共g兲 The normalized mobility
共
/
0
兲 vs Dirac point shift. The data points corresponding to 0.102
ML Pt, 0.008 ML Ti, and 0.029 ML Fe on samples Pt-1, Ti-1, and
Fe-2 are circled. 共h兲 −V
D,shift
is plotted vs 1/
–1/
0
. The dashed
lines are power-law fits to the equation, −⌬V
D,shift
⬃共⌫
TM
兲
b
, where
b is 0.64, 1.01, 0.85, 0.83, 0.86, and 0.95 for Ti-1, Ti-2, Pt-1, Pt-2,
Pt-3, and Fe-2 electrons, respectively.
0nm
10 nm
3.19 ML
0.62 ML
02468
0 0.87 1.75 2.62 3.50
0
-40
-80
-120
-160
Pt Coverage (ML)
Pt Coverage (
Å
)
Dirac Point
(
V
)
AFM 1
AFM 2
AFM 1
AFM 2
(b) (c)
(a)
Pt-3
Pt-4
Pt-5
Pt-6
FIG. 4. 共Color兲共a兲 The Dirac point as a function of Pt coverage
up to high coverage. 共b兲 AFM image of 0.62 ML Pt exhibits isolated
clusters. 共c兲 AFM image of 3.19 ML Pt indicates a connected film
with some areas of bare graphene.
ELECTRONIC DOPING AND SCATTERING BY… PHYSICAL REVIEW B 80, 075406 共2009兲
075406-3
doping by the TM due to the continuity of the chemical
potential. Thus, the final values of V
D
in the high-coverage
regime clearly show that Pt films can produce either n-type
or weak p-type doping of the graphene. This sample-to-
sample variation is most likely due to differences in the ini-
tial surface purity among samples. Although hydrogen clean-
ing is performed on all samples, trace amounts of resist
residue could remain, directly affecting the TM-graphene
spacing. Due to the highly spacing-dependent interfacial di-
pole strength,
8
any variation in the spacing will directly af-
fect the type and amount of doping. The fact that n-type
doping is observed provides experimental evidence for the
presence of a strong interfacial dipole layer favoring n-type
doping as predicted theoretically
8
because the expected dop-
ing based only on WF considerations would lead to strong
p-type doping.
An interfacial dipole whose strength decreases with in-
creasing equilibrium spacing 共d
eq
兲共Ref. 8兲 provides a pos-
sible explanation for the nonmonotonic behavior of the Dirac
point shift in Pt samples. Based on theoretical calculations,
the d
eq
between TM adatoms and graphene is less than 3 Å
共Ref. 6兲 while for bulk TM the distance increases to
⬃3.3 Å.
8
The n-type doping observed in samples at low
coverage is an indication of a strong interfacial dipole favor-
ing n-type doping, as expected for low coverages exhibiting
a small d
eq
. As the bulklike regime is approached, the in-
creasing d
eq
decreases the dipole strength and hence reduces
the n-type doping efficiency as observed by the shift in the
Dirac point toward positive voltages. We emphasize that the
interfacial dipole provides just one possible scenario to ex-
plain the nonmonotonic evolution of the Dirac point shift. A
quantitative understanding is complicated by the fact that the
WF can differ from bulk values for small clusters 共⬍4nm
lateral size兲共Ref. 27兲 and the corresponding quantity for ada-
toms 共should they be present兲 is the first ionization energy.
Therefore, further theoretical calculations are needed to fully
understand the doping effect of clusters. Regardless of the
exact mechanism for doping by clusters, an interfacial dipole
is still necessary to explain the n-type or weak p-type doping
measured in the bulklike regime.
IV. CONCLUSION
In conclusion, the exploration of TM/graphene systems
leads to several important observations. At low coverage, the
doping efficiency is found to be related to the TM WFs but
Ti, Fe, and Pt all exhibit n-type doping even for materials
with higher WF than graphene 共i.e., Fe and Pt兲. Extending
the Pt-doping study to higher thickness, the doping can either
be n type or weakly p type. Because WF considerations
alone would generate strong p-type doping, this result pro-
vides experimental evidence for the strong interfacial dipole
favoring n-type doping as predicted by theory.
8
Analysis of
the scattering at low coverage indicates that the scattering by
TM clusters exhibits different behavior compared to 1/ r
Coulomb scattering.
ACKNOWLEDGMENTS
We acknowledge helpful discussions with F. Guinea and
M. Fuhrer. We acknowledge the support of ONR 共Grant No.
N00014-09-1-0117兲, NSF 共Grants No. CAREER DMR-
0450037, No. CAREER DMR-0748910, and No. MRSEC
DMR-0820414兲, and CNID 共Grant No. ONR/DMEA-
H94003-07-2-0703兲.
*
roland.kawakami@ucr.edu
1
M. I. Katsnelson, F. Guinea, and A. K. Geim, Phys. Rev. B 79,
195426 共2009兲.
2
B. Uchoa, V. N. Kotov, N. M. R. Peres, and A. H. Castro Neto,
Phys. Rev. Lett. 101, 026805 共2008兲.
3
B. Uchoa, C.-Y. Lin, and A. H. Castro Neto, Phys. Rev. B 77,
035420 共2008兲.
4
A. V. Krasheninnikov, P. O. Lehtinen, A. S. Foster, P. Pyykko,
and R. M. Nieminen, Phys. Rev. Lett. 102, 126807 共2009兲.
5
B. Uchoa and A. H. Castro Neto, Phys. Rev. Lett. 98, 146801
共2007兲.
6
K. T. Chan, J. B. Neaton, and M. L. Cohen, Phys. Rev. B 77,
235430 共2008兲.
7
Y. Mao, J. Yuan, and J. Zhong, J. Phys.: Condens. Matter 20,
115209 共2008兲.
8
G. Giovannetti, P. A. Khomyakov, G. Brocks, V. M. Karpan, J.
van den Brink, and P. J. Kelly, Phys. Rev. Lett. 101, 026803
共2008兲.
9
E. J. H. Lee, K. Balasubramanian, R. T. Weitz, M. Burghard, and
K. Kern, Nat. Nanotechnol. 3, 486 共2008兲.
10
F. Xia, T. Mueller, R. Golizadeh-Mojarad, M. Freitag, Y.-M. Lin,
J. C. Tsang, V. Perebeinos, and P. Avouris, Nano Lett. 9, 1039
共2009兲.
11
B. Huard, N. Stander, J. A. Sulpizio, and D. Goldhaber-Gordon,
Phys. Rev. B 78, 121402 共2008兲.
12
J. Park, Y. H. Ahn, and C. Ruiz-Vargas, Nano Lett. 9, 1742
共2009兲.
13
H. Sevincli, M. Topsakal, E. Durgun, and S. Ciraci, Phys. Rev. B
77, 195434 共2008兲.
14
M. I. Rojas and E. P. M. Leiva, Phys. Rev. B 76, 155415 共2007兲.
15
F. Schedin, A. K. Geim, S. V. Morozov, E. W. Hill, P. Blake, M.
I. Katsnelson, and K. S. Novoselov, Nature Mater. 6, 652
共2007兲.
16
D. B. Farmer, R. Golizadeh-Mojarad, V. Perebeinos, Y.-M. Lin,
G. S. Tulevski, J. C. Tsang, and P. Avouris, Nano Lett. 9, 388
共2009兲.
17
J.-H. Chen, C. Jang, S. Adam, M. S. Fuhrer, E. D. Williams, and
M. Ishigami, Nat. Phys. 4, 377 共2008兲.
18
CRC Handbook of Chemistry and Physics 共CRC Press, Boca
Raton, FL, 2008兲.
19
S. J. Sque, R. Jones, and P. R. Briddon, Phys. Status Solidi A
204, 3078 共2007兲.
20
N. Ooi, A. Rairkar, and J. B. Adams, Carbon 44, 231 共2006兲.
21
A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri,
F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth, and
A. K. Geim, Phys. Rev. Lett. 97, 187401 共2006兲.
PI et al. PHYSICAL REVIEW B 80, 075406 共2009兲
075406-4
