10 1016j apsusc 2017 07 301

Selective AuCl3 Doping of Graphene for Reducing Contact Resistance of Graphene Devices Dong-Chul Choi a,b, Minwoo Kimc, Young Jae Song c,d* , Sajjad Hussain a,b, Woo-Seok Songe, Ki-Seok

Accepted Manuscript

Title: Selective AuCl3 Doping of Graphene for Reducing

Contact Resistance of Graphene Devices

Authors: Dong-Chul Choi, Minwoo Kim, Young Jae Song,

Sajjad Hussain, Woo-Seok Song, Ki-Seok An, Jongwan Jung

This is a PDF file of an unedited manuscript that has been accepted for publication

As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain

Selective AuCl3 Doping of Graphene for Reducing Contact Resistance of Graphene Devices

Dong-Chul Choi a,b, Minwoo Kimc, Young Jae Song c,d* , Sajjad Hussain a,b, Woo-Seok Songe, Ki-Seok Ane, Jongwan Junga,b *

a Graphene Research Institute, Sejong University, Seoul 143-747, Korea

b Faculty of Nanotechnology & Advanced Materials Engineering, Sejong University, Seoul

143-747, Korea

cSKKU Advanced Institute of Nano Technology (SAINT), Sungkyunkwan University

(SKKU), Suwon 16419, Korea

d Department of Physics, Sungkyunkwan University (SKKU), Suwon 16419, Korea

e Thin Film Materials Research Center, Korea Research Institute of Chemical Technology,

Daejon 305-600, Korea

*To whom correspondence should be addressed E-mail: (Y.J.S) yjsong@skku.edu, E-mail:

(J.J) jwjung@sejong.ac.kr

Graphical abstract

AuCl3doping +heating

Highlights

 Graphene was doped selectively using AuCl3 solution only to metal-graphene contact area

 With 10 mM-AuCl3 at 80 oC, a low contact resistivity was obtained

 The stability of the contact resistivity in atmospheric environment was evaluated

 The increase of the contact resistivity due to de-doping was much lower than sheet resistance due to covering metal

Abstract

Low contact resistance between metal-graphene contacts remains a well-known challenge for building high-performance two dimensional materials devices In this study, CVD-grown graphene film was doped via AuCl3 solution selectively only to metal (Ti/Au) contact area to reduce the contact resistances without compromising the channel properties of graphene With 10 mM-AuCl3 doping, doped graphene exhibited low contact resistivity of ~897 Ω mm, which is lower than that (~1774 Ω mm ) of the raw graphene devices The stability of the contact resistivity in atmospheric environment was evaluated The contact resistivity increased by 13 % after 60 days in an air environment, while the sheet resistance of doped graphene increased by 50 % after 30 days The improved stability of the contact resistivity of AuCl3-doped graphene could be attributed to the fact that the surface of doped-graphene is covered by Ti/Au electrode and the metal prevents the diffusion of AuCl3

Keywords: Graphene; Contact resistivity; AuCl3; Stability

Introduction

Graphene, an atomically thin semi-metal with sp2–bonded carbon atoms arranged in a honeycomb lattice, is at center stage of intense research in the field of fundamental physics as well as in low-cost flexible transparent electronics, photovoltaics or microelectronics devices[1-3] For electronic devices application, the metal to graphene contact resistance is the big obstacle hindering the further progress of high-performance graphene-based devices with ultrafast carrier transport properties Contact resistance is a major limiting factor for the

on state current of nanoscale graphene field effect transistor To date, several studies for reducing contact resistances of graphene to metal have been reported Post annealing [4-6] and cleaning of exposed graphene surface are common approaches for reducing contact resistance, since the graphene surface is most likely contaminated during the fabrication process Ultraviolet/ozone treatment [7, 8], plasma treatment [4, 9-11], and atomic force microscopy cleaning for source/drain contact regions prior to metallization [12, 13] have been reported It was observed that contact resistance of two‐dimensional MoS2 thin-film was significantly reduced after laser annealing and the field‐effect mobility was increased from 24.84 cm2V‐1s‐1 to 44.84 cm2V‐1s‐1[14] These approaches however could damage the graphene Thermal annealing also affects the graphene, and heat can damage to the components or sensitive parts where thermal effects should be avoided Chan et al also found that annealing did not significantly affect the contact resistance of their device [15] It was

observed in some cases that the annealing process increases the contact resistance instead, as

a result of the structural disorders along the plasma-etched graphene edges Another effective approach is to dope graphene Most of the doping experiments have been performed using electrochemical or electrostatic techniques because these methods provide an easy way to control the Fermi energy of graphene Chemical doping of graphene using AuCl3 has been used for p-type doping [16-21] A work function of graphene could be controlled by dipping graphene films into AuCl3 solution The AuCl3 doping method has been applied to increase the conductivity of transparent graphene films [16-21] Contrary to those several reports in which graphene doping using AuCl3 to increase the conductivity of films, the effect of organic doping such as AuCl3 to graphene-metal contact resistivity has not been explored yet Here in this work, AuCl3 doping was applied selectively to source/drain contact area of chemical vapor deposition (CVD)-grown graphene to reduce the contact resistances without compromising the channel properties of graphene It was observed that AuCl3 doping between graphene and metal pad significantly improved contact resistance We also found that the stability of contact resistance is much higher than that of the sheet resistance of AuCl3-doped graphene in the atmospheric environment because the contact area is covered with metal pads while the graphene channel is directly exposed to air

Experimental details

First, highly crystalline graphene was grown on a Cu foil by rapid thermal CVD, and the CVD-grown graphene was transferred to 300 nm-SiO2/Si substrate FeCl3 etchant was used for etching of the Cu foil and poly (methyl methacrylate) (PMMA) was used as a supporting film for the transfer process The active areas were opened by photolithography, and a thin

Cu film was evaporated using an evaporator, in which the Cu film (~20 nm) was used for a shadow mask for etching graphene After lift-off, the exposed graphene area was etched out

by O2 plasma using a reactive ion etcher (RIE) The Cu film was then etched out by FeCl3

etchant Source and drain pad areas for graphene transistors were opened by photolithography again AuCl3 solution was dropped and spin-coated with 2000 rpm for 1 min on the open contact area, while the channel region of graphene was protected by photoresist Even though nitromethane is the most common solvent for AuCl3, water was used instead for solvent since

we noticed that nitromethane remains after lift-off process The samples were heated to

50 °C-140 °C on a hot plate for dopant activation Then, Ti/Au (8/50 nm) was evaporated using a thermal evaporator for source and drain metal and lifted off The overall fabrication scheme for device is illustrated in Fig 1

Results and Discussion

The Raman spectra of the graphene before and after doping using AuCl3 solution are shown

in Fig 2 First, the CVD-grown graphene film was identified as a single layer graphene based

on I2D/IG (>2) and transmittance data (~97.3% @ 550 nm) on a glass The G and 2D peaks of the pristine graphene (non-treated graphene with AuCl3) are observed at 1583.4 cm−1 and 2677.9 cm−1, respectively, as shown in Fig 2a and Fig 2b On the other hand, the G peaks of the doped graphene using different AuCl3 concentration are found to be at 1586.7, 1589.5, 1590.8 and 1593.1 cm−1, respectively for AuCl3 concentration of 2.5, 5, 10, and 15 mM, respectively While 2D peak shifted upward to 2683.5, 2683.9, 2688.7, and 2691.5 cm−1, respectively for 2.5, 5, 10, and 15 mM of AuCl3, respectively The upward shift of G and 2D peak positions is a clear indication of p-type doping of graphene [16, 22, 23] The ratio of

ID/IG was changed to 0.28 after AuCl3 doping, respectively, as compared to the 0.02 ratio for the pristine CVD-grown graphene, most likely because the AuCl3 doping increased the disorder of the graphene basal plane [24] On the contrary, Raman spectra were not

changed noticeably on a graphene channel (Fig 2c) since the graphene channel was protected by photoresist during AuCl3 doping process as shown in Fig 1

X-ray photon spectroscopy (XPS) was analysed for non-treated and AuCl3-doped graphene The C 1s peak, corresponding to sp2 C – sp2 C bond at ~284.9 eV was not changed with heat treatment up to 140 °C In Fig 3b, the Au4f core level was resolved to AuCl3 and Au0, respectively These results demonstrate a successful doping of Au or Cl on the graphene layer Fig 3c and Fig 3d show that Au0 4f 5/2 peak position shifted monotonically downward with increasing annealing temperature (87.3 eV for 50 °C to 86.9 eV at 140 °C), while Cl 2p 3/2

peak position was not changed, indicating an increase in the size of Au0 clusters [17, 25] This could be attributed to the fact that Au 3+ ions were reduced and agglomerated into Au0cluster Fig 3e shows that atomic composition of Cl and Au3+ ions presents a slight peak at

80 °C and decreases slightly beyond 80 °C The decrease of Au3+ ions above 80 °C indicates that Au3+ ions were reduced to Au0 due to transformation of the unstable Au3+ in AuCl4- to a stable Au0 cluster by Cl desorption during heat treatment [17, 25]

Scanning Kelvin probe microscopy (SKPM) was employed to analyze the surface potential of graphene sheets A contact potential difference (CPD) was measured with modulations of amplitude of ~0.3-0.5 V and a frequency of 17 kHz in a Pt- coated AFM probe Fig 4a and Fig 4b show topography and surface potential mapping image of non-treated and AuCl3-doped graphene as functions of molar concentration (annealing temperature was set to 50 °C)

As mole concentration of AuCl3 increases, the Au-cluster size increases Work functions of graphene sheets increased from ~4.65 eV to ~4.79 eV with increasing the AuCl3

concentration from 0 mM to 10 mM as shown in Fig 4c And the work function saturates in AuCl3 doping above 10 mM In this work, work function of raw graphene was 4.65 eV which was a quite higher than the reported value of 4.5 eV for ideal graphene[24, 25] It is believed

that organic residues still remain on the film or substrate during transfer process The residual contaminants and oxygen trapped at the interface of graphene and substrate make graphene p-doped Fig 4d shows the effects of annealing temperatures on work function of graphene With increasing the temperature, work function of doped graphene also increases However, when the annealing temperature exceeds 80° C, the work function of the doped graphene becomes saturated and decreases again It is believed that beyond 80° C, Au3+ ions were transformed to Au0, consistent with the previous XPS results The topography and surface potential mapping image of as-doped graphene and as-annealed graphene at different temperatures are shown in Fig S2

Contact resistance between the doped graphene and metal (Ti/Au) was measured by using transmission line method (TLM) The TLM was designed with contact distances from 5 to 30

mm, contact width (W) of 20 mm, and contact length (d) of 100 mm Fig 5a shows a schematic image of the designed TLM devices, and typical images of the fabricated devices are shown in Fig 5b and Fig 5c Current and resistance values between the TLM pads were measured using a parameter analyzer, B1500A (Agilent Co.) Fig 6 shows the measured resistance and extracted contact resistance values for the fabricated graphene samples With respect to naming of the contact resistance, Rc refers to the contact resistance (measured in Ω), the specific contact resistivity (or specific contact resistance) is denoted as ρcspecific (in Ω mm2) and the contact resistivityρc= Rc∙𝑊 (in Ω μm) The contact resistance was calculated based

on the y-intercept of a linear data fit of the total resistance (RT) The sheet resistance of graphene channel (Rsh) was extracted from the slope of a linear data fit It can be noted that

Rsh (the slope of the curves) was not much changed for all samples because only the source and drain contact area were doped In TLM, the equation between RT and channel length L provides an estimate of ρc specific through the transfer length LT The intersection of the RT

curve for RT=0 (Lx), gives 2 LT for Rsk =Rsh , Rsh being the sheet resistance of the graphene channel and Rsk the sheet resistance of the graphene under the contact [26] For a precise extraction of LT, one should take into account the difference of the sheet resistance in the channel and underneath the metal contact [27] In our experiment, the graphene beneath the contact was doped by AuCl3, whereas, the graphene channel was not Besides, it has been reported that Rsh under the contact is strongly dependent on the deposition process [28] Comparison of transfer length LT with the actual contact length d indicates whether the

current flow is crowded to the edge of the contact or flows into the whole contact area Because the contact length d (100 mm) of the TLM device is much larger than the intersection

of the RT curve for RT=0, current crowding is so high that the contact resistivity is described effectively as the contact resistivity rc =Rc∙W (in Ω mm) The contact resistivity was ~1774 Ω

mm for the raw graphene, and it was decreased to ~978, ~897, and ~909 Ω mm, for 5 mM, 10

mM, and 15 mM AuCl3-doping, respectively (Fig 7a) The smallest contact resistivity (~897

Ω mm) value was obtained from the doped graphene with 10 mM-AuCl3 at 80 °C Please note that the contact resistivity difference at 10 mM and 15 mM is very small, and this result is consistent with the previous surface potential mapping data (Fig 4c) Fig 7b shows the contact resistivity change at different annealing temperatures The contact resistivity of doped graphene exhibited the lowest value (~897 Ω mm) at 80 °C and slight change at different annealing temperatures (Fig S3) The choice of contact metal is critical to reduce the graphene contact resistance For Ti/Au metal, the contact resistivity varies largely by many orders of magnitude from ~103 to ~106 Ω mm [29] The obtained specific contact resistivity values are positioned in the low range value compared with the previous results from Ti/Au metal/graphene contact (103 to ~106 Ω mm [29], 600-1000 Ω mm [30], 800 Ω mm [31], 568

Ω mm [32], 7500 Ω mm [33], 2000 Ω mm [34]) We also investigated the stability of contact

resistivity over time The sheet resistance of doped graphene and contact resistivity was compared Fig 8 shows the sheet resistance of doped graphene and contact resistivity of doped graphene with Ti/Au electrode over time The sheet resistance of AuCl3-doped graphene exposed in an air environment was increased by 15 % after 5 days and by 50 % after 30 days On the contrary, the contact resistivity of doped graphene increased by 13 % after 60 days The improved stability of the contact resistivity of graphene-metal could be attributed to the fact that the surface of doped graphene is covered by Ti/Au electrode, and the metal electrode obstructs the diffusion of AuCl3, while the graphene channel is directly exposed to the air These results show that chemical doping by using AuCl3 solution is an effective and easily applicable approach for reducing contact resistances of graphene

Conclusion

In summary, we report a treatment to improve the graphene-metal contacts through doping by AuCl3 and annealing at various temperatures With 10 mM-AuCl3 doping, and annealing at

80 °C, the doped graphene devices exhibit a contact resistivity of ~897 Ω mm which is half

of the raw-graphene We also noticed that the contact resistivity has much higher stability than the sheet resistance of graphene channel exposed to an air environment The improved stability of the contact resistivity of AuCl3-doped graphene could be attributed to the fact that the surface of doped graphene is covered by Ti/Au electrode and the metal prevents the diffusion of AuCl3 These results show that chemical doping by using AuCl3 solution is an effective and easily applicable approach for reducing contact resistances of graphene

Acknowledgements