Field effect of in-plane gates with different gap sizes on the Fermi level tuning of
graphene channels
Meng-Yu Lin, Yen-Hao Chen, Cheng-Hung Wang, Chen-Fung Su, Shu-Wei Chang, Si-Chen Lee, and Shih-Yen Lin
Citation: Applied Physics Letters 104, 183503 (2014); doi: 10.1063/1.4875583 View online: http://dx.doi.org/10.1063/1.4875583
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/104/18?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
Alleviation of fermi-level pinning effect at metal/germanium interface by the insertion of graphene layers Appl. Phys. Lett. 105, 073508 (2014); 10.1063/1.4893668
Fermi-level shifts in graphene transistors with dual-cut channels scraped by atomic force microscope tips Appl. Phys. Lett. 104, 023511 (2014); 10.1063/1.4862275
Spin filter and molecular switch based on bowtie-shaped graphene nanoflake J. Appl. Phys. 112, 104328 (2012); 10.1063/1.4766914
Quantum Hall-like effect in gated four-terminal graphene devices without magnetic field Appl. Phys. Lett. 99, 222101 (2011); 10.1063/1.3663625
Device model for graphene bilayer field-effect transistor J. Appl. Phys. 105, 104510 (2009); 10.1063/1.3131686
Field effect of in-plane gates with different gap sizes on the Fermi level
tuning of graphene channels
Meng-Yu Lin,1,2Yen-Hao Chen,3Cheng-Hung Wang,4Chen-Fung Su,5 Shu-Wei Chang,2,3,a)Si-Chen Lee,1and Shih-Yen Lin1,2,3,a)
1
Graduate Institute of Electronics Engineering, National Taiwan University, Taipei, Taiwan
2
Research Center for Applied Sciences, Academia Sinica, Nankang, Taipei, Taiwan
3
Department of Photonics, National Chiao-Tung University, Hsinchu, Taiwan
4
Institute of Display, National Chiao-Tung University, Hsinchu, Taiwan
5
College of Photonics, National Chiao-Tung University, Tainan, Taiwan
(Received 3 March 2014; accepted 27 April 2014; published online 5 May 2014)
Tuning of the Fermi level is investigated in graphene channels using two in-plane gates with significantly different-sized isolating gaps. While the n-type tuning was achievable in both schemes, the wide-gap device had an enhanced minimum drain current and less prominent current modulation than the narrow-gap device. In addition, further p-type tuning was not observed in the wide-gap device at negative gate biases. These phenomena indicated that both devices had distinct field-strength dependences and Fermi level tuning effects, which may be critical for the practical design of devices.
VC 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4875583]
Owing to its unique electrical, chemical, and thermal properties, graphene, a two-dimensional material formed by carbon atoms in a honeycomb lattice, has attracted a lot of attention.1 To prepare graphene films, many methods, including mechanical exfoliation using scotch tape, sublima-tion of silicon carbide in a high-vacuum chamber, and chem-ical vapor deposition (CVD) on metal templates,1–7 have been performed. To achieve large-area and high-quality gra-phene films, the CVD growth of gragra-phene on copper (Cu) foil has been regularly adopted. However, chemical contami-nation or water adsorption during the sample transferring process causes most graphene films to appear heavily p-type doped.8Often, positive biases that result in minimum drain currents (crossing between the Fermi level and the Dirac point) are observed in graphene field-effect transistors.8,9To operate around the Dirac point, effort has been devoted to n-type doping graphene films. Possible approaches to achieve this goal include: applying an ammonia gas flow during the graphene growth process or immersing the gra-phene films in a chemical solution.9Although these methods are effective in obtaining n-type doped graphene films, they do not provide precise control over the carrier density. Conversely, the control over carrier densities can be achieved electrically by employing the architecture of a dual-gate device.10,11 Through tuning either the top or the bottom gate voltage, the Fermi level of a graphene film can be manipulated. In doing this, the original positive bottom/ top gate voltage at the minimum drain current can be shifted to a zero or negative bias. However, the main disadvantage of this approach is that the top dielectric layers can possibly influence the graphene channels, leading to a more compli-cated fabrication procedure. To solve these problems, an in-plane-gate device architecture is an alternative choice for graphene transistors.12,13This in-plane-gate device architec-ture has been widely applied for one-dimensional transistors and quantum transport devices.14,15
In this paper, tuning the Fermi level of graphene chan-nels is achieved with in-plane-gates, and examined using bottom gates. Two devices with significantly different-sized isolating gaps (10 lm versus 100 nm or less) between their in-plane gates and channels are investigated. The channels of both devices were p-type doped in nature and could be electrically converted to n-type with comparable positive biases applied to the in-plane gates. Upon being tuned into n-type, the 10-lm-gap device had an enhanced minimum drain current and a less prominent current modu-lation as its bottom-gate voltage was swept around its bias point, corresponding to the Dirac point. In addition, with a negative in-plane gate voltage, only the 100-nm-gap device could be tuned further p-type effectively. These phenomena may originate from the field effect, owing to the distinct sizes of the insulating gaps and the non-uniform carrier density profiles in the graphene channels as a result of the in-plane-gate geometry.16
The graphene films were prepared with a low-pressure CVD system. Copper foils about 25 lm in size were annealed at 1030C for 30 min in a quartz tube fur-nace that was filled with 320 mTorr hydrogen gas (H2)
to remove the metal oxide. After annealing, the gas mix-ture was composed of methane and H2 with flow rates 7
and 15 sccm, respectively, which was sent into the tube to grow the graphene. During the 10-min growth period, the pressure was maintained at 650 mTorr. The samples were then removed from the tube and underwent a stand-ard graphene transfer procedure.10 After these steps, the graphene films were attached to 600-nm silicon-dioxide/ silicon (SiO2/Si) substrates, which were pre-patterned
with titanium/gold (Ti/Au) electrodes. The same film preparation and transferring procedure has been discussed elsewhere.17
For the fabrication of the 10-lm-gap device, a pattern was formed using standard photolithography and oxygen (O2) plasma etching was performed to define the transistor
channel on the film. A scanning electron microscope (SEM)
a)
Electronic addresses: swchang@sinica.edu.tw and shihyen@gate.sinica.edu.tw
0003-6951/2014/104(18)/183503/4/$30.00 104, 183503-1 VC2014 AIP Publishing LLC
image of the device is shown in Fig.1(a). The in-plane gate, drain, and source electrodes were made of Au. The channel itself and the separations between the channel and in-plane gates were all 10 lm. The drain currents (ID) are plotted as a
function of the bottom gate voltage (VGS,b) at different
in-plane gate biases (VGS,in¼ 100, 0, and 100 V) in Fig.
1(b). As a positiveVGS,inwas applied, the bottom-gate
volt-age (VGS,b,min) at the minimum drain current (ID,min) shifted
from 70 to nearly 0 V. This result suggests that the in-plane gate voltage shifted the Fermi level in the graphene channel and shows the potential for Fermi level tuning. The mini-mum drain current (ID,min) nearly doubled from 20 to 40 lA.
The corresponding modulation of ID with the bottom-gate
voltage,VGS,baroundVGS,b,min, was not as sharp as its
coun-terpart at VGS,in¼ 0 V. Conversely, the VGS,b-ID curve at a
negative in-plane gate voltage ofVGS,in¼ 100 V remained
similar to that atVGS,in¼ 0 V. This observation indicated that
further inducing holes through electrical control of VGS,in
was ineffective. The distinct trends in the current modula-tions at opposite VGS,inpolarities could be demonstrated by
plottingVGS,b,minandID,minas a function ofVGS,in, as shown
in Fig. 1(c). The effective tuning with an increasing VGS,in
was accompanied by a negative shift in theVGS,b,minand an
increase in ID,min. However, variations in these two
quanti-ties were much less significant (also non-monotonic for ID,min) as VGS,in became negative, implying that electrical
p-type tuning did not function properly.
The device with 100-nm-wide isolating gaps between the channel and the in-plane gates exhibited different behav-ior. An image of the device layout is shown in Fig.2(a). In this device, a crisscross graphene sheet was scraped with atomic force microscopy (AFM) tips at the joints between two of the graphene arms, resulting in two trenches (isolating gaps). The AFM scraping over the graphene surfaces proved to be effective at electrically isolating the graphene sheets at the two sides of the trench.18 The trenches were about
FIG. 1. (a) SEM image of an in-plane gate graphene transistor with 10-lm isolating gaps. (b) The drain current,IDof this device as a function of the
VGS,batVGS,in¼ 100, 0, and 100 V. (c) The minimum drain current, ID,min
and corresponding bottom-gate bias,VGS,b,minversusVGS,in.
FIG. 2. (a) An image of a crisscross graphene film after AFM scraping (100-nm trenches). (b) TheVGS,b-IDcurves of the narrow-gap device under
VGS,in¼ 100, 0, and 100 V.
100 nm in width. The graphene sheet between the two trenches was 10 lm wide and played the role of the channel, while the two disjointed graphene arms functioned as in-plane gates. The VGS,b–ID curves for this device, measured
with VGS,in¼ 100, 0, and 100 V, are shown in Fig. 2(b).
The comparison between the two curves at VGS,in¼ 0 V in
Figs.1(b)and2(b)show that the device with AFM-scraped gaps had a smallerVGS,b,min¼ 40 V, indicating that the
corre-sponding channel was less p-type doped in nature. This
char-acteristic may originate from the change in the
graphene-SiO2interface, induced by the AFM scraping.
17
At VGS,in¼ 100 V, the bottom-gate voltage, VGS,b,minat the
min-imum drain current was lowered to 0 V. This bias shift indi-cated that the n-type tuning was effective. In contrast to the doubling of theID,minin the 10-lm-gap device, the minimum
drain current did not alter much. Meanwhile, the current modulation around VGS,b,min remained as sharp as that at
VGS,in¼ 0 V. For a negative bias of VGS,in¼ 100 V, the
voltage VGS,b,min increased to around 70 V, indicating that
further p-type tuning through electrical control of the in-plane gates was possible. In addition, the sharp current modulation remained under these circumstances.
The differences between these two devices may have been caused by the non-uniformity of the carrier profiles that were induced by distinct field effects.16The drain currentID
was proportional to the channel conductivity rch, which is
related to the average local surface carrier density along the channel cross section:
rch¼ ql ðdx
WnðjEF EDiracðxÞjÞ; (1) whereq and l are the charge and mobility of the carriers, respectively, of which the product ql is roughly identical and has the same sign for both electrons and holes;W is the width of the channel; andn(|EF–EDirac(x)|) is the surface
car-rier density (regardless of carcar-rier types) at positionx, which only depends on the difference in the magnitude between the Fermi level,EF, and local Dirac-point energy,EDirac(x). The
surface density (n) nearly vanished at EF¼ EDirac, except for
some minor contributions such as charged impurities.19 However, the densityn always increased, regardless of whether EF was above or belowEDirac. From this viewpoint and Eq.
(1), switching off the channel conductivity rchwith a constant
EFbecame difficult whenEDirac(x) had a large spatial
fluctua-tion or simply deviated from a constant in the channel because EF–EDirac(x) cannot simultaneously vanish everywhere. This
issue is not common in semiconductors with considerable bandgaps. As long as the Femi level is deep inside the bandgap, the local carrier density is always small and the spa-tial variations of band edges play a minor role in this aspect.
Figure3shows a schematic band diagram of the 10-lm-gap device under a positive VGS,in. Only half of the channel is
depicted owing to its symmetric layout. Because the widths of the isolating gap and channel were comparable, a consid-erable voltage drop was present across the channel region shown in Fig.3. This voltage drop lifted the Fermi levelEF
(VGS,b¼ 0) at a zero bottom-gate voltage close to/above the
Dirac-point energy,EDirac(x) (tuned into n-type electrically).
Although the wide gap led to a mild electric field (F1) near
the graphene edge, this field persisted in the channel and induced a wide-range of variations in the Dirac-point energy EDirac(x). In this case, when the bottom-gate voltage,
VGS,b,min,1that minimizes the drain current was applied, the
corresponding Fermi level EF (VGS,b¼ VGS,b,min,1) did not
conform well withEDirac(x), and the minimum drain current
was enhanced. It was also expected that a change in EF
within twice the standard deviation, DE1of the energy
differ-ence EF(VGS,b¼ VGS,b,min,1) EDirac(x) would not suddenly
change the ID, which partially explains the smooth
modula-tion of IDwith VGS,b. However, the mild modulation could
have also been caused by the non-uniform in-plane field dis-tributed along the channel (from the drain to the source).16
The band diagram for the 100-nm-gap device at a posi-tiveVGS,inis shown in Fig.4. With an intense external field
F2 F1,EDirac(x) varied sharply around the graphene edge.
Because of the lack of a bandgap, the spatial charge could easily be induced, even when the graphene sheet had a low carrier density. With this, the electrons induced byF2
accu-mulated around the edge of the film, screening the intense field and confining the sharp variation of EDirac(x), only at
the edge. The accumulated electrons experienced more disor-der and extrinsic scattering than those in the graphene sheet, causing them to become less mobile. As a result, the modula-tion of ID with VGS,b around the minimum-current voltage
VGS,b,min,2 was mainly influenced by the flat profile of
EDirac(x) in the channel. Because the deviation (DE2) of
FIG. 3. The weak field (F1) induced a wide-range variation in theEDirac(x),
which smoothed the current modulation of the wide-gap device.
FIG. 4. The induced electrons screened the intense fieldF2, maintained the
flatness of theEDirac(x), and kept the current modulation of the narrow-gap
device sharp.
EF(VGS,b¼ VGS,b,min,2) EDirac(x) was narrow in this case,
the current modulation remained sharp. For the negative in-plane gate voltage, the incapability of further inducing holes in the 10-lm-gap device may be attributed to the al-ready strong screening from holes, which further prevented the weak external field from penetrating the graphene chan-nel. Conversely, the electrical p-type tuning might still be effective in the 100-nm-gap device because of the intense external field and the weaker screening (less natural p-type doping because of AFM scraping).
In conclusion, tuning of the Fermi level in graphene channels was achieved with two in-plane gates of different sizes. The device architecture simplified the fabrication pro-cedure. Modulation of the drain current through the in-plane gate was influenced by the external field strength, carrier screening near the graphene edge, and the uniformity of the Dirac-point energy inside the channel. These factors affected the capability of Fermi level tuning and the sharpness of the current modulation, and have to be taken into account in the practical design of devices.
This work was supported in part by the National Science Council Projects NSC 102-2221-E-001-032-MY3 and NSC 102-2622-E-002-014.
1K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V.
Dubonos, I. V. Grigorieva, and A. A. Firsov,Science306, 666 (2004).
2C. Riedl, A. A. Zakharov, and U. Starke,Appl. Phys. Lett.
93, 033106 (2008).
3
N. Camara, G. Rius, J.-R. Huntzinger, A. Tiberj, L. Magaud, N. Mestres, P. Godignon, and J. Camassel, Appl. Phys. Lett. 93, 263102 (2008).
4
N. Camara, G. Rius, J.-R. Huntzinger, A. Tiberj, N. Mestres, P. Godignon, and J. Camassel,Appl. Phys. Lett.93, 123503 (2008).
5K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, K. S. Kim, J. H. Ahn,
P. Kim, J. Y. Choi, and B. H. Homg,Nature457, 706 (2009).
6
X. Li, W. Cai, L. Colombo, and R. S. Ruoff,NanoLett.9, 4268 (2009).
7
G. A. Lopez and E. J. Mittemeijer,Scr. Mater.51, 1 (2004).
8T. Lohmann, K. V. Klitzing, and J. H. Smet,NanoLett.9, 1973 (2009). 9X. Wang, X. Li, L. Zhang, Y. Yoon, P. K. Weber, H. Wang, J. Guo, and
H. Dai,Science324, 768 (2009).
10
X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, A. Velamakanni, I. Jung, E. Tutuc, S. K. Banerjee, L. Colombo, and R. S. Ruoff,Science 324, 1312 (2009).
11
S. Kim, J. Nah, I. Jo, D. Shahrjerdi, L. Colombo, Z. Yao, E. Tutuc, and S. K. Banerjee,Appl. Phys. Lett.94, 062107 (2009).
12A. D. Wieck and K. Ploog,Appl. Phys. Lett.56, 928 (1990). 13A. D. Wieck and K. Ploog,Surf. Sci.
229, 252–255 (1990).
14
J. Nieder, A. D. Wieck, P. Grambow, H. Lage, D. Heitmann, K. v. Klitzing, and K. Ploog,Appl. Phys. Lett.57, 2695 (1990).
15H. O. Li, T. Tu, G. Cao, L. J. Wang, G. C. Guo, and G. P. Guo,Quantum
Transport in Graphene Quantum Dots, New Progress on Graphene Research, edited by J.-R. Gong (InTech, 2013), Chap. 6.
16
F. Molitor, J. G€uttinger, C. Stampfer, D. Graf, T. Ihn, and K. Ensslin, Phys. Rev. B76, 245426 (2007).
17M. Y. Lin, Y. H. Chen, C. F. Su, S. W. Chang, S. C. Lee, and S. Y. Lin,
Appl. Phys. Lett.104, 023511 (2014).
18
L. Weng, L. Zhang, T. P. Chen, and L. P. Rokhinson,Appl. Phys. Lett.93, 093107(2008).
19K. Nagashio, T. Nishimura, and A. Toriumi, Appl. Phys. Lett.
102, 173507 (2013).
