Frank Schwierz
Institut für Mikro- und Nanoelektronik, Technische Universität Ilmenau, PF 100565, 98684 Ilmenau, Germany
Email: frank.schwierz@tu-ilmenau.de
Abstract
During the last few years, graphene has gained
remarkable attention in the device community. Graphene
transistors are evolving at a rapid pace and
graphene-based devices are considered as an option for a
post-Si electronics. To assess whether graphene can meet
the high expectations or not, the properties and specifics
of this new material have to be analyzed carefully. The
present paper provides an overview of the current status
of graphene transistor development and reviews the
prospects and problems of these devices.
1. Introduction
In October 2004 researchers of the University of
Figure 1. Transfer characteristics of an experimental graphene
Manchester reported that they had prepared and analyzed
MOSFET with large-area gapless channel [7]. Shown is also
single sheets of carbon atoms arranged in a regular the band diagram indicating the shift of the Fermi level and the
honeycomb lattice [1]. Since than, this new material changeover from p- to n-type conduction at the Dirac point.
called graphene has attracted the attention of engineers
working in various fields [2-4]. Targeted applications of On the other hand, large-area graphene shows very high
graphene range from serving as an additive for aviation carrier mobilities. In exfoliated graphene on SiO2-
fuel and car tires to transparent electrodes for solar cells covered Si wafers, mobilities of 10,000-15,000 cm2/Vs
and to high-performance transistors.
are measured routinely [1, 8], an upper mobility limit of
Graphene is indeed a fascinating material, yet due to the
70,000 cm2/Vs has been suggested [9], and in the
predominantly euphoric appraisal of its potential, the
absence of charged impurities and ripples, mobilities of
prospects of graphene in electronics are sometimes
200,000 cm2/Vs have been predicted [10]. These high
assessed overly optimistic.
mobilities make graphene a promising material for fast
In the following, I will discuss the properties of graphene
RF (radio frequency) transistors, in particular since
relevant for transistors, review the state-of-the-art of
switch-off is not required per se for transistors in RF
graphene transistors, and discuss the prospects of
graphene from a device engineer's point of view. Since circuits.
so far most work on graphene devices was related to The above statement that large-area graphene is gapless
MOSFETs, the main emphasis in the present paper is put does not mean, however, that a gap in graphene cannot
on graphene MOSFETs. be opened. Instead, several approaches to create a gap
are explored. Most popular is constraining one
2. Properties of graphene dimension of large-area graphene thus forming narrow
Currently the most popular approaches to prepare GNRs (graphene nanoribbon) [11-12], while a second
graphene for electronic applications are (i) mechanical option is biasing bilayer graphene [13]. The two ideal
exfoliation of pyrolytic graphite and transfer to oxidized GNR types are armchair (ac) and zigzag (zz)
Si wafers [1], (ii) epitaxial graphene growth on metals nanoribbons. It has been predicted that these two GNR
and subsequent graphene transfer to insulating substrates configurations show a bandgap EG that depends on the
[5], and (iii) thermal decomposition of SiC [6]. GNR width w roughly according to EG v 1/w [14]. The
Large-area graphene is a semi-metal having a zero gap opening has been verified experimentally for ribbons
bandgap and cone-shaped conduction and valence bands with widths down to about 1 nm [11-12]. Figure 2 shows
meeting each other at the K point of the brillouin zone. a compilation of theoretical and experimental data for
Due to the missing bandgap, MOSFETs with large-area the bandgap of GNRs as a function of GNR width. It can
graphene channels cannot be switched off (see Fig. 1) be seen that, to get a gap of about 0.4 eV, very narrow
and are not suitable for CMOS-like logic applications. GNRs with widths around/below 10 nm are needed.