Field-Effect Transistors

Controlled Ambipolar-to-Unipolar Conversion in Graphene Field-Effect Transistors Through Surface Coating with Poly(ethylene imine)/Poly(ethylene glycol) Films

Zheng Yan, Jun Yao, Zhengzong Sun, Yu Zhu, and James M. Tour*

Since its first isolation as a single-layered sheet of graphite, graphene has attracted enormous interest due to its extraordinary electronic properties.[1–4] In particular, the extremely high mobility in graphene has made it a promising candidate for transistors.[5,6] However, the zero bandgap elec-tronic structure of graphene results in ambipolar transistor behaviors that are different from the unipolar behaviors in conventional transistors. In this regard, many efforts have been made to modify the electronic structure in graphene to achieve electronic transport dominated by the preferred type of carrier.[7–14] As conventional doping usually only moves the Fermi level relative to the Dirac point, many n- or p-type doping methods largely only shift the neutrality point while maintaining the ambipolar traits.[7–10] Chemical doping can introduce defects to the lattice, potentially reducing the car-rier mobility.[11,12]

Much research has been done in the ambipolar-to- unipolar conversion of carbon-nanotube-based field-effect transistors (FETs).[15–17] Amine groups in poly(ethylene imine) (PEI) are effective electron donors since they each bear a lone pair of electrons, hence PEI has been gener-ally used to make n-type carbon nanotube transistors.[17–21] Avouris et al. also have used PEI to make ambipolar n-type graphene transistors.[10] However, limited research has been conducted to produce unipolar n-type graphene transistors.[22] In this report, we disclose an ambipolar-to-unipolar (n-type) conversion in graphene field-effect transistors (FETs) along

© 2012 Wiley-VCH Verlag Gmbsmall 2012, 8, No. 1, 59–62

DOI: 10.1002/smll.201101528

Z. Yan, Z. Sun, Dr. Y. Zhu, Prof. J. M. TourDepartment of ChemistryRice University6100 Main Street, Houston, TX, 77005, USAE-mail: tour@rice.eduZ. Yan, Z. Sun, Dr. Y. Zhu, Prof. J. M. Tour Richard E. Smalley Institute for Nanoscale Science and TechnologyRice University6100 Main Street, Houston, TX, 77005, USAJ. Yao, Prof. J. M. Tour Applied Physics Program through the Department of BioengineeringRice University6100 Main Street, Houston, TX, 77005, USA

Prof. J. M. Tour Department of Mechanical Engineering and Materials ScienceRice University6100 Main Street, Houston, TX, 77005, USA

with an apparent increase in the electron mobility. By coating the graphene layer with a film of PEI in poly(ethylene glycol) (PEG), we observe a suppression in the hole conduction and an increase in the electron conduction with an increase in the transconductance, demonstrating an unipolar n-type (electron-doping) behavior. The recorded unipolar n-type behavior is similar to that observed in unipolar carbon nanotube-based devices.[17,21] Moreover, the electron mobility in the PEI-coated graphene FETs also increased by several times. These traits are desirable improvements in making graphene FETs of high quality. Furthermore, this method is controllable. Both unipolar and ambipolar n-type graphene FETs can be easily obtained by adjusting the thickness of the PEI film atop the graphene channel. We propose that the amine groups in the PEI film provide doping/dedoping to the electron/hole carriers in the graphene layer modulated by an external gate bias voltage. If PEI alone is used, delamina-tion of the polyamine from the graphene occurs upon drying. However, with the addition of PEG, the dried film remains physisorbed to the graphene. This type of enhanced phys-isorption using PEG could also prove to be useful in other film-forming formulations with graphene.

The graphene film was grown using a solid carbon source, poly(methyl methacrylate), as described previously.[12] Devices were made from monolayer graphene, as determined by Raman spectroscopy (see Figure 1a). The two most pro-nounced peaks in this spectrum are the G peak at 1580 cm−1 and the 2D peak at 2690 cm−1. The 2D/G intensity ratio is about 4 and the full width at half maximum height (FWHMH) of the 2D peak is about 30 cm−1, indicating that the graphene is monolayer.[12] The graphene was transferred to a doped Si substrate (ρ = 0.005 Ohm cm) capped with a 100-nm-thick SiO2 layer. Standard electron-beam lithography and lift-off processes were used to define the source and drain electrodes (30-nm-thick Pt) in the graphene devices, with the doped Si substrate serving as the back-gate electrode. Graphene stripes (5 μm wide) were further defined by oxygen-plasma etching. The scanning electron microscopy (SEM) image of the as-made device is shown in the inset of Figure 1b. Electrical characterizations were performed in a probe station (Desert Cryogenic TT-probe 6 system) under vacuum (10−6 Torr). The current–voltage (I–V) data were collected using an Agilent 4155C semiconductor parameter analyzer at room tempera-ture. Figure 1b shows the transport properties of the as-made graphene FET in vacuum (10−6 Torr) after pumping for

59wileyonlinelibrary.comH & Co. KGaA, Weinheim

Z. Yan et al.

60

communications

Figure 1. Characterisitics of pristine graphene. a) Raman spectrum (514 nm excitation) of a monolayer graphene film transferred on a SiO2/Si substrate. b) The FET I–V curve of pristine graphene on a SiO2/Si substrate (drain–source voltage, Vds = 1 V). The inset is the SEM image of a graphene-based FET device (5 μm × 5 μm (length, L × width, W)) atop a 100 nm SiO2/Si wafer with 30 nm Pt as the source and drain electrodes.

∼3 days to exclude effects from air adsorption.[12] It shows a typical ambipolar transport behavior with the neutrality point near 0 V.

The thin layer of PEI film in PEG, with a thickness of ∼5 μm, was formed atop the graphene channel by a spin-coating the mixture solution of PEI (average molecular weight 800) and PEG (average molecular weight 400) in methanol (see Supporting Information for the spin-coating processes). Figure 2a shows a schematic of the PEI-coated device. After coating the device with a layer of PEI/in PEG, several altered electrical characteristics were observed in the graphene FET (the measurement was made under vacuum, see curve 2 in Figure 2b). First, the hole conduction was suppressed with respect to the gate voltage in the negative bias region, whereas the electron conductance in the positive biased region was increased. These two changes result in a transi-tion from ambipolar behavior to unipolar n-type behavior. Note that the conduction from the PEI film is negligible (see Supporting Information, Figure S1) and the effect of PEG alone on the transport behavior of graphene FET is limited (see Supporting Information, Figure S2). Hence the observed ambipolar-to-unipolar conversion comes from modification of the graphene channel induced by PEI in the film as amines (pKb = 3.3) are far more electron donating than ethers

www.small-journal.com © 2012 Wiley-VCH Verlag GmbH & Co. KGaA,

Figure 2. a) Schematic diagram of the graphene FET coated with a PEI/PEG layer. b) Electrical transport curves from a typical graphene FET before (curve 1) and after (curve 2) coating with the PEI/PEG (1:2 w/w) layer with thickness of ∼5 μm) (Vds = 1 V). The curved arrows show the increase in the electron conduction and the suppression in the hole conduction. Ids, Vg and Vds represents the drain–source current, back-gate voltage, and drain–source voltage.

(pKb = 17.5). Meanwhile, we also observed an apparent change in the graphene field-effect mobilities. The field-effect mobilities were extracted from the drain–source cur-rent (Ids)/back-gate potential (Vg) curves using the following formula: μ = [(ΔIds/Vds)(L/W)]/CoxΔVg.

[23] Here, L and W are the length and the width of the graphene channel, respectively. And Cox is calculated by ε0εrA/d, where d is the thickness of SiO2, A is unit area, ε0 is permittivity of free space, and εr is 3.0.[23] The calculated elec-tron mobility for the PEI-coated graphene FET (∼4100 cm2 V−1 s−1) is ∼4 times that of the pristine FET before PEI/PEG coating (∼900 cm2 V−1 s−1). In contrast, the hole mobility decreases to ∼60 cm2 V−1 s−1

from ∼1100 cm2 V−1 s−1 after coating with the PEI/PEG film. Both the ambipolar-to-unipolar transition and the changed mobilities are highly reproducible in more than five FETs that were independently prepared (see Supporting Informa-tion, Table S1).

We propose that the observed phenomena result from the effective carrier density modulation in the graphene layer by the PEI/PEG thin film. As a strict 2D electronic system, a capacitor model has been effective and frequently adopted to describe transport in graphene FETs.[24,25] For the weak dependence of carrier mobility on carrier density in graphene,[26] the conductance change is approximately pro-portional to the change in carrier density that is linearly mod-ulated by an external gate voltage. This results in the typical linear ambipolar behaviors in our as-made pristine graphene FETs (curve 1 in Figure 2b). The coating of the PEI/PEG layer on the graphene surface breaks the linear relationship between the increase of carrier-density and the gate voltage as described in a simple capacitor model by introducing an additional carrier injecting/withdrawing source. As the amine groups in PEI are effective electron donors since they bear lone pair electrons,[17–21] electron carriers from these sites are pulled into the graphene layer by the downward electric field induced by a positive back-gate voltage (see illustration in

Figure 3a), resulting in a sudden increase in electron density and hence conductance increases (curve 2 in Figure 2b). Meanwhile, the amine groups are electronegative (via the sigma-bond framework) and can serve as effective hole traps. This type of lone-pair electron donation with concomitant electronegativity through the C-N bonds is commonly see in the reactivity of organic compounds.[27] The increased hole carriers generated by a negative back-gate voltage are pulled into the PEI layer and trapped by the amine groups (see illustration in Figure 3b), resulting in an almost flat Ids–Vg curve in the negative gate voltage region (curve 2 in Figure 2b). Since there are a finite number of amine groups near the graphene, the carrier modulation is also

Weinheim small 2012, 8, No. 1, 59–62

Ambipolar-to-Unipolar Conversion in Graphene FETs through Surface Coating with PEI/PEG Films

Figure 3. Schematic of carrier (electrons, represented as black circles, and holes represented as gray circles) migration at the PEI film/graphene interface when a a) positive or b) negative Vg is applied.

expected to be finite. This is seen in the curve 2 in Figure 2b; above +10 V the Ids–Vg curve goes back to the original trend (paralleling the curve 1 of the pristine graphene FET in Figure 2b). A similar trend was observed in the negative back-gate voltage region; once all the amine groups (hole-trapping sites) are saturated, Ids begins to increase (see Sup-porting Information, Figure S3).

We also find that the spin-coating speed has a limited effect on the transport behavior of graphene FETs. As shown in Figure 4a, similar ambipolar-to-unipolar conversions were observed when using spin-coating rates of 2000, 4000, or 8000 rpm. Under these three different coating speeds, the thickness of the PEI film was measured by SEM and was between 3 and 6 μm, suggesting that the doping comes from the local amine groups near the PEI/graphene interface. When the graphene device coated with the PEI film was put into methanol for 1 h to remove most of the PEI/PEG, the thick-ness of the PEI/PEG film decreased to ∼2 nm, as measured by elliposmetry. In this case, the ambipolar-to-unipolar con-version disappeared. Ambipolar n-type behavior is observed in the graphene FET and the Ids–Vg curve is dominated by the shift of the neutrality point to negative gate voltages (see black curve in Figure 4b). This shift is due to a doping from the PEI at the PEI/graphene interface and has been observed in graphene FETs with very thin layer PEI films atop the graphene channel.[10] A transition between ambipolar and unipolar n-type behavior was observed when the thickness of PEI/PEG film reached ∼200 nm (by SEM) (Figure S4a in the Supporting Information). In addition, when changing the weight ratio of PEI and PEG, similar unipolar n-type behav-iors were observed (Figure S4b,c).

Methanol and chloroform were also used as solvents for PEI (no PEG) to make films. After coating with a fresh film using either of these solvents, similar ambipolar-to-unipolar transistions were observed (see Supporting Information,

© 2012 Wiley-VCH Verlag Gmbsmall 2012, 8, No. 1, 59–62

Figure 4. a) Transport curves (Ids–Vg) of a graphene FET after spin-coating the PEI/PEG films at different spin-coating rates (Vds = 1 V). b) Transport curves (Ids–Vg) of a graphene FET before (gray curve) and after (black curve) coating with an extremely thin PEI/PEG layer having a thickness of ∼2 nm (Vds = 1 V).

Figure S5). However, if the PEI film is permitted to dry in vacuum (10−6 Torr), becoming nearly methanol or chloroform- free, a simple shift in the neutrality point was observed, which could be due to poor interfacial contact between the PEI and graphene. The use of PEI/PEG mixtures can address this deficiency. After being placed in vacuum (10−6 Torr) for 24 h, five out of six devices were successfully prepared and demonstrated an obvious ambipolar-to-unipolar conversion, indicating around 80% reproducibility. In the sixth device, the PEI film delaminated and only one simple shift in the neutrality point was observed in this device (see Supporting Information, Table S1). Figure S6 demonstrates that the obtained device shows ∼15% change in air after 48 h. As an example, a simple method for making a P–N junction using the PEI/PEG film is shown in the Supporting Information, Figure S7.

In summary, we have demonstrated an ambipolar-to-unipolar conversion method in graphene FETs by coating them with PEI films. Unlike other surface engineering that produces instant doping in graphene, the PEI film here serves as both an electron and hole reservoir that can be modu-lated by the external gate voltage. This leads to an apparent unipolar n-type transport behavior and increased electron mobility. This method is controllable. By adjusting the thick-ness of PEI/PEG films, we can obtain both unipolar and ambipolar n-type graphene FETs. The study here provides a guide to engineering graphene transport properties by doping/dedoping from an external source.

Experimental Section

The detailed fabrication of devices are described in the Supporting Information. Raman spectroscopy was performed with a Renishaw Raman microscope using 514 nm laser at ambient temperature. The I–V data were collected by an Agilent 4155C semiconductor parameter analyzer. The electrical properties were measured in a probe station (Desert Cryogenic TT-probe 6 system) under vacuum (10−5–10−6 Torr) at room temperature.

Supporting Information

Supporting Information is available from the Wiley Online Library or from the author. The Supporting Information includes materials and methods, Raman spectra, and the transport curves of PEI and graphene FETs.

Acknowledgements

We thank the AFOSR (FA9550-09-1-0581), the AFOSR through a STTR with PrivaTran (FA9550-10-C-0098), the Lockheed Martin LANCER IV Program, the ONR MURI program (#00006766, N00014-09-1-1066) and the AFRL through Universal Technology Corpora-tion (FA8650-05-D-5807).

61www.small-journal.comH & Co. KGaA, Weinheim

Z. Yan et al.

62

communications

[1] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang,

S. V. Dubonos, I. V. Grigorieva, A. A. Firsov, Science 2004, 306, 666–669.

[2] A. K. Geim, K. S. Novoselov, Nat. Mater. 2007, 3, 183–191. [3] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang,

M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos, A. A. Firsov, Nature 2005, 438, 197–200.

[4] R. S. Ruoff, Nat. Nanotechnol. 2008, 3, 10–11. [5] Y. M. Lin, C. Dimitrakopoulos, K. A. Jenkins, D. B. Farmer,

H. Y. Chiu, A. Grill, Ph. Avouris, Science 2010, 327, 662. [6] X. Liang, Y. Jung, S. Wu, A. Ismach, D. Olynick, S. Cabrini, J. Bokor,

Nano Lett. 2010, 10, 2454–2460. [7] X. Wang, X. Li, L. Zhang, Y. Yoon, W. Weber, H. Wang, J. Guo,

H. Dai, Science 2009, 324, 768–771. [8] X. Li, H. Wang, J. Robinson, H. Sanchez, G. Diankov, H. J. Dai, Am.

Chem. Soc. 2009, 131, 15939–15944. [9] X. Dong, D. Fu, W. Fang, Y. Shi, P. Chen, L. Li, Small 2009, 5,

1422–1426.[10] D. B. Farmer, R. Golizadeh-Mojarad, V. Perebeinos, Y. Lin,

G. Tulevski, J. Tsang, P. Avouris, Nano Lett. 2009, 9, 388–392.[11] D. Wei, Y. Liu, Y. Wang, H. Zhang, L. Huang, G. Yu, Nano Lett. 2009,

9, 1752–1758.[12] Z. Sun, Z. Yan, J. Yao, E. Beitler, Y. Zhu, J. M. Tour, Nature 2010,

468, 549–552.[13] Z. Yan, Z. Sun, W. Lu, J. Yao, Y. Zhu, J. M. Tour, ACS Nano 2011, 5,

1535–1540.[14] B. Guo, Q. Liu, E. Chen, H. Zhu, L. Fang, R. Gong, Nano Lett. 2010,

10, 4975–4980.[15] Y. Lin, J. Appenzeller, P. Avouris, Nano Lett. 2004, 4, 947–950.

www.small-journal.com © 2012 Wiley-VCH V

[16] D. Wei, Y. Zhang, Y. Yang, D. Hasko, D. Chu, K. Teo, G. Amaratunga, W. Milne, ACS Nano. 2008, 2, 2526–2530.

[17] Y. Zhou, A. Gaur, S. Hur, C. Kocabas, M. Meitl, M. Shim, J. Rogers, Nano Lett. 2004, 4, 2031–2035.

[18] T. Ozel, A. Gaur, J. Rogers, M. Shim, Nano Lett. 2005, 5, 905–911.

[19] G. Siddons, D. Merchin, J. Back, J. Jeong, M. Shim, Nano Lett. 2004, 4, 927–931.

[20] M. Shim, T. Ozel, A. Gaur, C. Wang, J. Am. Chem. Soc. 2006, 128, 7522–7530.

[21] M. Shim, A. Javey, A. Wong, S. Kam, H. Dai, J. Am. Chem. Soc. 2001, 123, 11152–11513.

[22] H. Li, Q. Zhang, C. Liu, S. Xu, P. Gao, ACS Nano 2011, 5, 3198–3203.

[23] Y. Q. Wu, P. D. Ye, M. A. Capano, Y. Xuan, Y. Sui, M. Qi, J. A. Copper, T. Shen, D. Pandey, G. Prakash, Appl. Phys. Lett. 2008, 92, 092102.

[24] E. S. Snow, P. M. Campbell, M. G. Ancona, J. P. Novak, Appl. Phys. Lett. 2005, 86, 033105.

[25] I. Meric, M. Han, A. Young, B. Ozyilmaz, P. Kim, K. Shepard, Nat. Nanotechnol. 2008, 11, 654–659.

[26] V. Perebeinos, P. Avouris, Phys. Rev. B. 2010, 81, 195442.[27] L. G. Wade in Organic Chemistry, (Eds: N. Folcjetti, J. Challice,

R. Mullaney, C. Pritchard-Martinez, K. Schiaparelli, M. J. Richards, D. King, P. Burns, C. Long), Pearson Education, Little River, SC 2006, p. 780.

Received: July 27, 2011Revised: August 23, 2011Published online: November 10, 2011

erlag GmbH & Co. KGaA, Weinheim small 2012, 8, No. 1, 59–62