Sie sind auf Seite 1von 7

Graphene for Electronic Applications – Transistors and More

F. Schwierz

Technical University Ilmenau, PF 100565, 98684 Ilmenau, Germany

Abstract — During the last five years, the new use graphene MOSFETs in real applications. We
material graphene has gained increasing attention in the discuss this issue subsequently for MOSFETs to be
device community. The progress in the development of used in digital logic and for RF MOSFETs.
graphene transistors is breathtaking and graphene-
based devices are now considered as an option for a
Modern logic is based on Si CMOS
post-Si electronics. However, to realistically assess the (complementary MOS) comprising both n-channel
potential of graphene, the existing problems with and p-channel MOSFETs. Complex logic ICs
graphene and the options to solve them have to be (integrated circuit) consist of many cascaded
analyzed carefully. The present paper provides an individual logic gates (e.g., NAND, NOR), where the
overview of the current status of graphene transistor output signal of one gate serves as the input for the
development and discusses the prospects and problems
of these devices.
ensuing gates. The cascaded gate arrangement
requires the restoration of the signal levels to ensure
that a logic gate can reliably identify if it receives a
Index Terms — Graphene, bandgap, carrier
transport, mobility, MOSFET, RF transistor, transistor. logic 1 or 0 at its input.
As shown in Fig. 1, CMOS logic gates consist of
an n-channel branch connected to ground and a p-
I. INTRODUCTION channel branch connected to the supply voltage VDD.
Only once in a blue moon, a single paper ignites a The beauty of CMOS is that it provides ideal signal
revolution in solid-state physics. Precisely this came restoration combined with the property that in steady-
about, however, when in October 2004 researchers of state, regardless of the input signals, always one of
the two branches is off and no current – except a
the University of Manchester reported on the
small off-current – can flow from VDD to ground.
preparation of a material called graphene and on the
Thus, Si CMOS offers ultimately low static power
observation of the field-effect in their samples [1].
dissipation and noticeable power is only dissipated
Since than, the interest in graphene literally exploded,
during switching.
and physicists and device engineers became intrigued
by this material [2-3]. Graphene is indeed fascinating,
yet due to the predominantly euphoric appraisal of the
potential of this material, its prospects in electronics
are frequently assessed overly optimistic.
Since so far most work on graphene devices was
related to MOSFETs (metal-semiconductor-oxide
field-effect transistor), in the present paper we focus
on the properties of graphene relevant to MOSFETs
and on the current state-of-the-art of graphene
MOSFETs. However, other options to use graphene
in electronics will also be touched.

II. IS MOSFET SWITCH-OFF INDISPENSABLE?


Conventional FETs show a reasonably good
switch-off behavior and an exponential dependence
of the drain current on the gate voltage in the
subthreshold region. This results in on-currents Ion
several orders of magnitude larger compared to the Fig. 1. Basic principle of CMOS on the example of a
currents in the off-state Ioff, and thus in large on-off NAND gate. If, for instance, the signal at input 1 (at the
ratios Ion/Ioff. Key to this behavior is a semiconducting gate terminals of transistors T1 and T3) is a logic 1 (Vin1 =
VDD) and that at input 2 is a logic 0 (Vin2 = 0), the
FET channel having a sizeable bandgap. Large-area
MOSFETs T1 and T4 are on while T2 and T3 are off. Since
graphene, however, is a zero-gap material. This leads T2 is off, the n-channel branch is off and no current can
to the situation that MOSFETs with large-area flow from VDD to ground. For any possible combination of
graphene channels cannot be switched off and the input signals one branch of the NAND gate is off.
question arises whether or not switch-off is needed to
978-1-4244-8579-6/10/$26.00 ©2010 IEEE 173
This issue is extremely important since power shows very high mobilities. In exfoliated graphene,
dissipation and the resulting self-heating is a major mobilities in excess of 10,000 cm2/Vs are routinely
problem of complex logic ICs, such as high- measured and for ideal graphene mobilities of
performance microprocessors. Si CMOS has 200,000 cm2/Vs are predicted [9].
superseded all other competing logic technologies
4
due to its low static power dissipation. Thus, a FET- 10
3p+1
based successor for Si CMOS will need transistors 3p
able to switch off. This can only be achieved if the 3
3p+2
10
FET channel is semiconducting having a reasonably

Bandgap (meV)
wide bandgap, preferably 0.4 eV or more. In other
words, gapless large-area graphene channels are not 10
2

suitable for logic FETs and ways to open a gap in Experiment


Han
graphene have to be found (and have been found as Li
Theory
we will show later). 10
1 Yang - Ideal ac GNRs
Yang - Extrapolation ideal 3p+1 ac GNRs
The situation is different in RF (radio frequency) Yang - Ideal zz GNRs

applications. Here, switch-off is not required per se.


Let us consider, for example, a small-signal low-noise 10
0

0.1 1 10 100
amplifier. In such a circuit, the FET is permanently
GNR width (nm)
operated in the on-state and the small RF signals to be
amplified are superimposed onto the dc gate voltage.
To summarize we state that (i) for graphene Fig. 2. Calculated and measured GNR bandgap as a
transistors to be used in a CMOS-like logic, switch- function of ribbon width, after [3]. Stars: experimental data
off is indispensable and a semiconducting channel is [4-5], triangles: simulation [7].
needed, and (ii) for RF graphene MOSFETs the
ability to switch off would certainly be useful but is On the other hand, theoretical investigations
not mandatory. suggest that the gap opening in narrow GNRs will be
accompanied by a dramatic mobility reduction. This
is not a big surprise since a decreasing mobility for
III. PROPERTIES OF GRAPHENE
increasing bandgap is a well known trend for
Single-layer graphene is a purely two-dimensional semiconductors. This trend is indicated in Fig. 3
material with a lattice consisting of regular hexagons showing the electron mobility in several conventional
with one carbon atom at each corner. By nature, semiconductors, Si nanowires, carbon nanotubes,
large-area graphene is gapless, but several approaches large-area graphene, and GNRs.
to create a gap in graphene are explored. Most
popular is constraining one dimension of large-area
graphene thus forming narrow GNRs (graphene 10
6 Large-area graphene
GNRs
nanoribbon) [4-5], while a second option is biasing CNTs (sim, exp)

bilayer graphene [6]. The two ideal GNR types are


Electron mobility (cm /Vs)

Ge (bulk
5
10 Si (bulk)
armchair (ac) and zigzag (zz) nanoribbons. It has
2

Si MOS
been predicted that both these GNR configurations Si nanowires
III-Vs (bulk)
show a bandgap EG that depends on the GNR width w 10
4 SiC (bulk)

roughly according to EG ∝ 1/w [7].


The gap opening has been verified experimentally
for GNRs with widths down to about 1 nm [4-5]. It 10
3

should be noted, however, that experimental GNRs


do not have ideal armchair or zigzag edges but 2
10
possess rough non-regular edges instead. Already 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
modest edge disorder obliterates any difference in the Bandgap (eV)
gap between GNRs of different edge geometry [8].
Figure 2 shows a compilation of reported theoretical Fig. 3. Electron mobility vs. bandgap. Black circles: III-
and experimental data for the bandgap of GNRs as a V compounds, from left to right InSb, InAs, In0.53Ga0.47As,
function of GNR width. It can be seen that, to get a InP, GaAs, In0.49Ga0.51P, GaN. Blue squares and blue lines:
carbon nanotubes, simulated [10-11]. Open blue circle:
gap of about 0.4 eV, very narrow GNRs with a width carbon nanotube, experiment [12]. Red diamonds: Si
around/below 10 nm are needed. nanowires, simulated [13]. Magenta stars and line: large-
We now move to the discussion of carrier transport area graphene [9, 14-16]. Light blue area: GNRs, simulated
and start with the mobility. Most frequently the high [17-18]. Green triangle (up), red full and open triangles
carrier mobility is stated to be the major advantage of (down): bulk Ge, bulk Si and Si MOS channels.
graphene. Indeed, large-area (but gapless) graphene
174
Figure 3 indicates that for a given bandgap, with graphene channels produced by one of the
graphene does not offer a distinct advantage in terms methods mentioned above:
of electron mobility compared to the more - Back-gated graphene MOSFETs with either
conventional semiconductors. Less frequently the gapless large-area graphene channels, e.g., [1,
high-field transport in graphene is discussed. 26] or narrow semiconducting GNR channels
Simulations have shown that graphene shows much [7, 27]. Back-gate MOSFETs are easy to
higher electron velocities at high fields (around 4 × fabricate and useful for proof-of-concept
107 cm/s [19-20]) than the conventional materials. purposes but are not suitable for practical
Note that at high fields, the electron saturation applications.
velocity in Si and in most of the III-V compounds is - Top-gated MOSFETs with either gapless large-
only around 107 cm/s. area graphene channels [24, 28-34] or narrow
Graphene offers another feature that is highly GNR channels [35].
desirable for MOSFETs. Aggressively scaled Si We start the discussion with MOSFETs having
MOSFETs with nm gates are known to suffer from gapless large-area graphene channels. So far, these
short-channel effects, such as threshold voltage roll- transistors constitute the majority of experimental
off, drain-induced barrier lowering, and impaired graphene devices. As mentioned above, the gapless
saturation of drain currents [21-22]. The scale length channels of these transistors cannot be switched off.
theory predicts that a MOSFET having a thin gate Figure 4 shows the transfer characteristics of a top-
dielectric and a thin gate-controlled channel region gate graphene MOSFET with a large-area epitaxial
will be robust against short-channel effects [23]. graphene channel on SiC. It can clearly be seen that
Since graphene offers the thinnest possible MOSFET this transistor does not switch off. For negative gate
channels (only one atomic layer thick), nm-gate voltages, the transistor has a p-type channel, at the
graphene MOSFET are expected to be very robust Dirac point (the point of minimum drain current) the
against short-channel effects and probably will be conductivity type of the channel switches from p- to
scalable beyond the limits of Si MOSFETs. It is n-type, and for positive gate voltages the transistor
important to note, however, that the scale length behaves as an n-channel FET.
concept is valid only for MOSFETs having a
semiconducting channel. Thus, our statement
regarding the scalability does not apply to graphene
MOSFETs with gapless channels.

IV. GRAPHENE TRANSISTORS


The currently most popular approaches to prepare
graphene for electronic applications are (i)
mechanical exfoliation of pyrolytic graphite [1], (ii)
epitaxial graphene growth on metals and subsequent
transfer of the grown graphene to insulating
substrates [24], and (iii) thermal decomposition of
SiC [25]. It should be noted that, although method (ii)
involves epitaxy while method (iii) is not a true
epitaxial process (since no material is deposited but
rather Si is removed from the surface of SiC wafers), Fig. 4. Transfer characteristics of a graphene MOSFET with
graphene produced by method (iii) is commonly large-area epitaxial graphene channel [32]. The point of
called epitaxial graphene. Method (i), i.e., exfoliation, minimum drain current is called the Dirac point. © IEEE
is still popular for lab purposes but not suited to the 2008.
semiconductor industry since it is extremely labor- While due to the missing switch-off MOSFETs
intensive and provides graphene flakes of only with gapless large-area graphene channels are not
limited size (several µm × several µm). The other two suitable for logic applications, they may be useful for
methods, however, have the potential for the RF applications, and indeed significant efforts have
production of wafer-scale graphene. Once the been spent on investigating (and improving) the RF
graphene layer is produced, well-established performance of these transistors.
semiconductor processing technology (lithography, As an example, Fig. 5 shows information on a top-
etching, layer deposition, etc.) can be applied to gate graphene MOSFET reported by an IBM group
fabricate graphene transistors. [34]. The channel consists of exfoliated graphene
A growing number of groups are now successfully deposited on a SiO2-covered (300 nm) Si wafer
fabricating different versions of graphene MOSFETs serving as a back-gate. The top-gate dielectric is a 12-
nm Al2O3 layer. As shown in Fig. 5c, the output
175
characteristics of this transistor do not show a pHEMTs. Compared to the many reported fT data of
saturation of the drain current. In [3] it has been graphene MOSFETs, much less experimental data on
discussed in detail that a FET should be operated in the maximum frequency of oscillation fmax has been
drain current saturation to exploit the full high-speed published so far. Current the record fmax for a
potential of the channel material. In spite of the graphene MOSFET is 14 GHz [25]. So far, no data on
missing current saturation, the transistor from Fig. 5 the RF noise performance of graphene transistors is
shows a remarkably high cutoff frequency fT of 50 available.
GHz. Due to the fast improvement of the high-speed
Figure 6 compares the cutoff frequency of performance of graphene MOSFETs during the last
experimental graphene MOSFETs (all with gapless two years (note that the first graphene MOSFET with
large-area graphene channels) with the best fT values GHz operation has been reported in late 2008 [30]),
reported for other types of RF FETs: InP HEMTs we expect significant further enhancement of the RF
(high electron mobility transistor), GaAs mHEMTs performance of graphene MOSFETs in the nearest
(metamorphic HEMT), Si MOSFETs, GaAs pHEMTs future.
(pseudomorphic HEMT). So far, the fastest (in terms
of fT) graphene transistors are a 240-nm gate 1000
graphene MOSFET showing an fT of 100 GHz and a
transistor with an fT of 53 GHz and a 550-nm gate
[25].

Cutoff frequency (GHz)


100

Record Graphene FETs


240-nm gate, 100 GHz fT
550-nm gate, 53 GHz fT

10

InP HEMT, GaAs mHEMT


Si MOSFET
GaAs pHEMT
Graphene FET

1
0.01 0.1 1 2
Gate length (µm)

Fig. 6. Cutoff frequency vs. gate length for different RF


FET types. The symbols indicate experimental data
compiled from the literature [36] and the lines are a guide
for the eye. Red open circles: GaAs mHEMTs and InP
HEMTs, blue full circles: Si MOSFETs, green triangles:
GaAs pHEMTs, stars: graphene MOSFETs.

Back-gate MOSFETs with narrow GNR channels


(GNR width down to less than 5 nm) have been
reported [7, 27]. These transistors showed excellent
switch-off (on-off ratios up to 106) and provided
evidence for the suitability of GNR MOSFETs for
Fig. 5. Details of a 350-nm gate graphene MOSFET
reported by an IBM group [34]. (a) cross-section, (b) logic applications. Recently also the first GNR
layout, (c) output characteristics, (d) small-signal current MOSFET with top-gate has been demonstrated [35].
gain vs. frequency and cutoff frequency. © IEEE 2010. The on-off ratio of this transistor is around 70 which
significantly exceeds the on-off ratios of MOSFETs
with large-area graphene channels (typically 2…20)
The reported record fT data for the competing RF but is still too low for logic application where on-off
FET types are as follows: 660 GHz for a 20-nm gate ratios of 104 to 107 are required [42]. The GNR
GaAs mHEMT [37], 628 GHz for a 30-nm gate InP channel of this transistor was about 10-20 nm wide
HEMT [38], 485 GHz for a 29-nm gate Si MOSFET and apparently did not offer a bandgap wide enough
[39], and 152 GHz for a 100-nm gate GaAs pHEMT to sufficiently suppress the off-current. However, a by
[40]. For comparison, the fastest carbon nanotube further reduction of the ribbon width we can expect
MOSFET reported so far has a 300-nm gate and much better on-off ratios of top-gated GNR
shows an fT of 80 GHz [41]. MOSFETs in the near future. So far, experimental
As can be seen from Fig. 6, in terms of cutoff data on the dynamic behavior (e.g., on the cutoff
frequency the best graphene MOSFETs already frequency or the switching delay) of GNR MOSFETs
outperform the fastest Si MOSFETs with comparable is not available.
gate length and aggressively attack the GaAs
176
The graphene transistor discussed so far are based support current densities in excess of 108 A/cm2 [52].
on the conventional MOSFET principle, i.e., Note that this is two orders of magnitude more than
controlling the carrier concentration and thus the the current density supported by copper and
channel conductivity by the applied gate voltage. In comparable with that supported by carbon nanotubes.
spite of the impressing performance obtained with Graphene devices for sensing applications look
experimental devices we have seen that these also promising. Graphene virtually consists only of
"conventional" graphene MOSFETs suffer from surface without a bulk and therefore should be very
several fundamental problems. Some of these are: (i) sensitive to its environment. Graphene sensors able to
the loss of mobility in narrow GNRs, (ii) the need of detect single gas molecules have been reported [53].
extremely narrow GNRs (only a very few nm wide) Another possible class of sensor devices is graphene-
to open a gap useful for switch-off, (iii) the based NEMS (nano electromechanical system) that
unsatisfying drain current saturation of graphene benefit from the superior mechanical properties and
MOSFETs with large-area graphene channels. This the extremely small mass of graphene [54].
has motivated research on novel graphene-based FET
concepts. Two examples, that have already gathered VII. CONCLUSION
considerable attention in the device community, are
the graphene tunnel FET [43] and the graphene During the past 30 months, we have witnessed
BiSFET (Bilayer PseudoSpin FET) [44]. Although huge progress in the development of graphene
these two transistor concepts are still at an embryonic transistors. Most impressing was the demonstration of
stage, they already found their way into the Emerging a 100 GHz fT graphene MOSFETs and of the
Research Devices section of the most recent issue of excellent switching behavior of MOSFETs with
the ITRS [42]. narrow GNR channels. The progress has been
accompanied, however, by the exposure of numerous
problems. Due to the missing gap, MOSFETs with
V. ARE GRAPHENE BIPOLAR TRANSISTORS AN
large-area channels cannot be switched off and
OPTION?
therefore are not suited for logic applications.
The FETs we discussed in the previous Section did Moreover, they show an unsatisfying drain current
not use intentionally doped graphene. However, saturation behavior limiting the RF performance.
graphene can be doped and both n-type and p-type Another problem is the fact that so far the extremely
graphene can be realized [44-45]. Thus, in principle narrow GNRs needed to open a useful gap face
bipolar graphene transistors are imaginable. However, serious fabrication issues.
due to the purely two-dimensional structure of Currently it seems that the biggest advantage of
graphene, these devices should involve lateral npn (or graphene MOSFETs may not be, as commonly stated,
pnp) structures. This in turn would require the base the high mobility but rather the ultimately thin
dimensions to be defined by lithography. In common channel and the resulting ultra-short scale length.
bipolar transistors, the critical device dimension in the Experiences from the past have shown that RF
direction of current flow is the base thickness, i.e., a chipmakers are more open to novel device and
vertical dimension defined by the thickness of an material concepts than chipmakers in Si mainstream
epitaxially grown layer in heterojunction bipolar VLSI electronics. Therefore, graphene might make its
transistors or the thickness of an implanted region in first appearance in RF applications rather than in
homojunction bipolar transistors. Thus, a graphene logic circuits [3].
bipolar transistor would suffer from the limitations of When assessing the prospects of graphene in
conventional lateral bipolar transistors, most notably electronics, it is important to recognize that all other
a wide base. So far, no reports on graphene bipolar options considered as possible successors of the
transistors could be found in the literature. conventional mainstream transistors also face serious
problems. Concepts that have been investigated
already for many years, such as spin transistors or
VI. FURTHER POSSIBLE APPLICATIONS OF GRAPHENE molecular devices, seem to be farther from
Since large-area graphene is gapless and behaves applications than graphene transistors, and it is not
like a semimetal, it may be used for interconnects. clear if they will ever reach the production stage. At
This would open the possibility of all-graphene ICs the moment it is impossible to assess which of the
with both the active devices and the wiring made of alternative device concepts, if any, has the potential to
graphene [47-48]. This vision has stirred more replace Si MOSFETs and III-V HEMTs. The race is
detailed investigations on the potential of graphene still open and the chances for electronic graphene
for use as interconnect. It has been shown that devices are at least as promising as those of other
graphene interconnects show comparable or even device concepts. The commercialization of graphene-
better performance compared to copper interconnects based devices and ICs in the near future, however, is
at room temperature [49-51] and that graphene can not likely.
177
ACKNOWLEDGEMENT [18] T. Fang et al., “Mobility in semiconducting
nanoribbons: Phonon, impurity, and edge roughness
This work was financially supported by the 2008- scattering”, Phys. Rev. B, vol. 78, p. 205403, 2008.
2009 Excellence Research Grant of TU Ilmenau. [19] A. Akturk and N. Goldsman, “Electron transport and
full-band electron-phonon interactions in graphene”, J.
Appl. Phys., vol. 103, p. 053702, 2008.
REFERENCES
[20] R. S. Shishir and D. K. Ferry, “Velocity saturation in
[1] K. S. Novoselov et al., “Electric field effect in intrinsic graphene”, J. Phys.: Cond. Matter, vol. 21, p.
atomically thin carbon films”, Science, vol. 306, pp. 344201 (2009).
666-669, 2004. [21] Y. Taur and T. H. Ning, Fundamentals of Modern
[2] A. K. Geim and K. S. Novoselov, “The rise of VLSI Devices, Cambridge University Press, Cambridge
graphene”, Nature Mater., vol. 6, pp. 183-191, 2007. 1998.
[3] F. Schwierz, “Graphene transistors” Nature Nanotech., [22] F. Schwierz, H. Wong, and J. J. Liou, Nanometer
vol. 5, DOI: 10.1038/NNANO.2010, to appear in the CMOS, Pan Stanford Publishing, Singapore 2010.
printed version in the July issue, 2010. [23] D. J. Frank, Y. Taur, and H.-S. P. Wong, “Generalized
[4] M. Han et al., “Energy band-gap engineering of scale length for two-dimensional effects in
graphene nanoribbons”, Phys. Rev. Lett., vol. 98, p. MOSFETs”, IEEE Electron Device Lett., vol. 19, pp.
206805, 2007. 385-387, 1998.
[5] X. Li et al., “Chemically derived, ultrasmooth [24] J. Kedzierski et al., “Graphene-on-insulator transistors
graphene nanoribbon semiconductors”, Science, vol. made using C on Ni chemical-vapor deposition”, IEEE
319, pp. 1229-1232, 2008. Electron Device Lett., vol. 30, pp. 745-747, 2009.
[6] E. V. Castro et al., “Biased bilayer graphene: [25] Y.-M. Lin et al., “100-GHz transistors from wafer-
Semiconductor with a gap tunable by the electric field scale epitaxial graphene”, Science, vol. 327, p. 662,
effect”, Phys. Rev. Lett., vol. 99, p. 216802, 2007. 2010.
[7] L. Yang et al., “Quasiparticle energies and band gaps [26] C. Hummel et al., “Ambient and temperature
in graphene nanoribbons”, Phys. Rev. Lett., vol. 99, p. dependent electric properties of backgate graphene
186801, 2007. transistors”, phys. stat. sol. B, vol. 247, pp. 903-906,
2010.
[8] M. Evaldsson et al., “Edge-disorder-induced Anderson
localization and conduction gap in graphene [27] X. Wang et al., “Room-temperature all-
nanoribbons”, Phys. Rev. B, vol. 78, p. 161407, 2008. semiconducting sub-10-nm graphene nanoribbon field-
effect transistors”, Phys. Rev. Lett., vol. 100, p.
[9] S. V. Morozov et al., “Giant intrinsic carrier mobilities 206803, 2008.
in graphene and its bilayer”, Phys. Rev. Lett., vol. 100,
p. 016602, 2008. [28] M. C. Lemme et al., “A graphene field-effect device”,
IEEE Electron Device Lett., vol. 28, pp. 282-284,
[10] X. Zhou et al., “Band structure, phonon scattering, and 2007.
performance limit of single-walled carbon nanotube
transistors”, Phys. Rev. Lett., vol. 95, p. 146805, 2005. [29] Y.-M. Lin et al., “Operation of graphene transistors at
gigahertz frequencies”, Nano Lett., vol. 9, pp. 422-426,
[11] V. Perebeinos, J. Tersoff, and Ph. Avouris, “Electron- 2009.
phonon interaction and transport in semiconducting
carbon nanotubes”, Phys. Rev. Lett., vol. 94, p. [30] I. Meric et al., “RF performance of top-gated, zero-
0786802, 2005. bandgap graphene field-effect transistors”, Tech. Dig.
IEDM, p. 21.2, 2008.
[12] G. Pennington and N. Goldsman, “Low-field
semiclassical transport in semiconducting carbon [31] I. Meric et al., “Current saturation in zero-bandgap,
nanotubes”, Phys. Rev. B, vol. 71, p. 205318, 2005. top-gated graphene field-effect transistors”, Nature
Nanotech., vol. 3, pp. 654-659, 2008.
[13] S. Jin, M. V. Fischetti, and T.-W. Tang, “Modeling of
electron mobility in gated silicon nanowires at room [32] J. Kedzierski et al., “Epitaxial graphene transistors on
temperature: Surface roughness scattering, dielectric SiC substrates”, IEEE Trans. Electron Devices, vol.
screening, and band nonparabolicity”, J. Appl. Phys., 55, pp. 2078-2085, 2008.
vol. 102, p. 083715, 2007. [33] J. S. Moon et al., “Epitaxial-graphene RF field-effect
[14] J.-H. Chen et al., “Intrinsic and extrinsic performance transistors on Si-face 6H-SiC substrates”, IEEE
limits of graphene devices on SiO2”, Nature Electron Device Lett., vol. 30, pp. 650-652, 2009.
Nanotechnol., vol. 3, pp. 206-209, 2008. [34] Y.-M. Lin et al., “Dual-gate graphene FETs with fT of
[15] F. Chen et al., “Dielectric screening enhanced 50 GHz”, IEEE Electron Device Lett., vol. 31, pp. 68-
performance in graphene FET”, Nano Lett., vol. 9, pp. 70, 2010.
2571-2574, 2009. [35] L. Liao et al., “Top-gated graphene nanoribbon
[16] A. Geim, “Graphene update”, http://meetings.aps.org/ transistors with ultrathin high-k dielectrics”, Nano
Lett., vol. 10, pp. 1917-1921, 2010.
link/BAPS.2010.MAR.J21.4.
[36] F. Schwierz, “Microwave Transistors: State of the art
[17] B. Obradovic et al., “Analysis of graphene in the 1980s, 1990s, and 2000s – A compilation of
nanoribbons as a channel material for field-effect 1000 top references”, TU Ilmenau, unpublished, 2010.
transistors”, Appl. Phys. Lett., vol. 88, p. 142102,
2006. [37] M. Schlechtweg, Private communication, 2010.

178
[38] D.-H. Kim and J. A. del Alamo, 30-nm InAs
pseudomorphic HEMTs on InP substrate with a
current-gain cutoff frequency of 628 GHz, IEEE
Electron Device Lett., vol. 29, pp. 830-833, 2008.
[39] S. Lee et al., “Record RF performance of 45-nm SOI
CMOS technology”, Tech. Dig. IEDM, pp. 255-258,
2007.
[40] L. D. Nguyen et al., “Characterization of ultra-high-
speed AlGaAs/InGaAs (on GaAs) MODFETs”, IEEE
Trans. Electron Devices, vol. 36, pp. 2243-2248, 1989.
[41] L. Nougaret et al., “80 GHz field-effect transistors
produced using high purity semiconducting single-
walled carbon nanotubes”, Appl. Phys. Lett., vol. 94, p.
243505, 2009.
[42] The International Technology Roadmap for
Semiconductors ITRS, 2009 edition, Semiconductor
Industry Association (2009). See also at
http://www.itrs.net/Links/2009ITRS/Home2009.htm.
[43] G. Iannoccone et al., “Perspectives of graphene
nanoelectronics: Probing technological options with
modeling”, Tech. Dig. IEDM, pp. 245-248, 2009.
[44] S. K. Banerjee, et al., “Bilayer pseudospin field-effect
transistor (BiSFET): A proposed new logic device”,
IEEE Electron Device Lett., vol. 30, pp. 158-160,
2009.
[45] S. S. Yu, W. T. Zheng, and Q. Jiang, “Electronic
properties of nitrogen-/boron-doped graphene
nanoribbons with armchair edges”, IEEE Trans.
Nanotechnol., vol. 9, pp. 78-81, 2010.
[46] Y.-C. Lin, C.-Y. Lin, and P.-W. Chiu, “Controllabe
graphene N-doping with ammonia plasma”, Appl.
Phys. Lett., vol. 96, p. 133110, 2010.
[47] C. Berger et al., “Electronic confinement and
coherence in patterned epitaxial graphene”, Science,
vol. 312, pp. 1191-1196, 2006.
[48] R. Van Noorden, “Moving towards a graphene world”,
Nature, vol. 442, pp. 228-229, 2006.
[49] A. Naeemi and J. D. Meindl, “Conductance modeling
for graphene nanoribbon (GNR) interconnects”, IEEE
Electron Device Lett., vol. 28, pp. 428-431, 2007.
[50] C. Xu and K. Banerjee, “Graphene nano-ribbon (GNR)
interconnects: A genuine contender or a delusive
dream?”, Tech. Dig. IEDM, p. 8.5, 2008.
[51] R. Murali et al., “Resistivity of graphene nanoribbon
interconnects”, IEEE Electron Device Lett., vol. 30,
pp. 611-313, 2009.
[52] J. Moser, A. Barreiro, and A. Bachtold, Current-
induced cleaning of graphene, Appl. Phys. Lett., vol.
91, p. 163513, 2007.
[53] F. Schedin et al., “Detection of individual gas
molecules adsorbed on graphene”, Nature Materials,
vol. 6, pp. 652-655, 2007.
[54] J. S. Bunch, “Electromechanical resonators from
graphene sheets”, Science, vol. 315, pp. 490-493,
2007.

179

Das könnte Ihnen auch gefallen