Going ballistic: Graphene hot electron transistors
Introduction
The experimental realization of graphene [1] and other two-dimensional (2D)-materials [2] has opened up new opportunities for pushing the limits of the state-of-the-art in electronics [3], [4] and photonics [5], [6]. This has been motivated by graphene׳s excellent material properties, which surpass those of conventional materials in many aspects. Nevertheless, in spite of its high charge carrier mobility [7] and saturation velocity [8], graphene field effect transistors (GFETs) struggle to match or surpass the performance of conventional silicon FETs. Fundamental challenges originate in the electronic band structure of graphene. The absence of a band gap leads to high off-state currents and low on/off current ratios, which prohibit GFET applications as logic gates [9], [10]. Another consequence of the zero band gap is band to band tunneling, which reduces the output current saturation and the voltage gain, limiting the RF performance potential of GFETs [11], [12]. Recently, vertical electronic device concepts have been proposed to overcome this intrinsic limitation [13], [14], [15], [16], [17], [18], [19]. One of these novel device concepts, introduced by Mehr et al. in 2012, is vertical graphene base transistor (GBT) [13]. The concept of the GBT is based on the metal-base hot-electron transistors (HETs) introduced originally by Mead in 1961 [20]. HETs utilize high energy tunneling injected electrons (hot electrons) to reach high performance [21]. The first HETs were composed of metal emitters, metal bases, and metal collectors, which were isolated from each other by thin oxide layers. One of the main challenges for the HETs as well as heterojunction bipolar transistors (HBTs) is that the cutoff frequency is limited by base transit time. While thinning down the base mitigates this issue, it dramatically increases the base resistance, resulting in high RC delay and self-bias crowding. The graphene base transistor, in contrast, exploits the high conductivity and the single atomic-thinness of graphene as the base material in HETs to minimize the base transit time and achieve high cutoff frequencies. This concept is distinctly different from vertical graphene field effect tunneling transistors as introduced by Britnell et al. [16]. While the latter functions due to the limited density of states in single layer graphene and electrostatic gate control of the carrier transport between two isolated single layer graphene (SLG) sheets, the GBT operates through emitter–base barrier modulation analogous to the bipolar technology,. This article reviews the experimental and theoretical progress on the GBTs and related devices.
Section snippets
Working principles of the GBT
The difference between the GBT and the GFET is shown schematically in Fig. 1. In the GFET, carrier transport happens in the graphene plane between the source and the drain with a Vds bias, while the gate electrostatically controls the conductivity of the graphene channel (Fig. 1a). In the GBT, in contrast, carriers move perpendicular through the graphene plane. The graphene base is isolated from a metal or doped semiconductor emitter and collector by an emitter–base insulator (EBI) and a
Device modeling/simulation and performance projection
A zero-order estimation of the GBT performance has been performed based on quantum-mechanical simulations in [13]. For that purpose, the Schrödinger equation with open-boundary conditions was solved numerically for one-band effective potential rounded up by image force at interfaces with emitter and collector. This early model, with no scattering effects included, predicted that for a terahertz operation at emitter–base voltages around 1 V, an EBI with a barrier of 0.4 eV or smaller and a
Proof-of-concept and experimental realization of the GBT
In 2013, Vaziri et al. demonstrated the first experimental proof of concept GBT on a chip/die scale [14]. The device comprised of an n-doped silicon emitter, a 5 nm-thick thermal silicon dioxide (SiO2) EBI tunneling barrier, a graphene base, a 15–25 nm-thick atomic layer deposited (ALD) aluminum oxide (Al2O3) BCI, and a metal (titanium/gold) collector. The fabrication was done on 8-inch wafers and the GBTs were isolated by 400 nm SiO2 shallow trench isolation (STI) making the fabrication scheme
Emitter–base–nsulator (EBI)
High frequency performance of the GBT requires high on-state collector currents IC. In an ideal device with α=1, IC is equal to the emitter current IE, which consists of injected hot electrons. Hence, the EBI barrier should be low and thin enough to provide a high current of hot electrons, yet block cold electron emission. To this end, a number of potential materials have been investigated. For example, replacing a 5 nm SiO2 EBI with 6 nm thick ALD HfO2 results in an improved threshold voltage
Wafer-scale integration of vertical graphene base transistors
In parallel to the optimization of GBTs on a chip / die scale, experiments targeting their implementation in a 200 mm wafer Si pilot line were initiated. A wet graphene transfer method was adopted to cover areas of up to 20×20 mm2 on pre-patterned 200 mm wafers. Subsequently, high-k dielectrics (e.g. HfO2) were deposited directly onto the CVD graphene using atomic vapor deposition. These process modules were then combined with a standard Al-based back end of the line metallization to provide a
Theoretical limits and potential
The Graphene-Base Heterojunction Transistor (GBHT) proposed by Di Lecce et al. [19] is a promising adaptation of the GBT concept. In this device, graphene is sandwiched between an n+-semiconductor layer (emitter) and an n-semiconductor layer (collector). In this structure the carrier transport mechanism is dominated by thermionic emission over the emitter–base Schottky barrier. As a result, the GBHT is envisioned to overcome part of the GBT׳s engineering issues, eliminating the need for
Conclusions
This paper reviews the experimental and theoretical state-of-the-art in vertical hot electron transistors with graphene base contacts, GBTs and GBHTs. Simulations predict the performance of these devices surpassing 1 THz in fT and fmax. In parallel, early experimental demonstrators show the general feasibility and functionality, without reaching such impressive numbers. Nevertheless, the experimental data has been used to update and improve the models, and even as more and more realistic
Acknowledgements
The authors wish to dedicate this paper to the memory of Wolfgang Mehr, the inventor of the GBT. The authors thank the IHP cleanroom staff and the technology team for their excellent support and discussions. Support from the European Commission through a European FP7 Project (GRADE, 317839), an ERC Grant (InteGraDe, No. 307311) as well as the German Research Foundation (DFG, LE 2440/1–1) is gratefully acknowledged.
References (49)
- et al.
Solid State Commun.
(2008) - et al.
Microelectron. Eng.
(2013) - et al.
Solid-State Electron.
(2013) - et al.
Carbon
(2014) - et al.
Science
(2004) - et al.
Proc. Natl. Acad. Sci. USA
(2005) - et al.
Nat. Nanotechnol.
(2014) - et al.
MRS Bull.
(2014) - et al.
Nat. Photonics
(2014) - et al.
Nature
(2012)
Appl. Phys. Lett.
IEEE Electron Device Lett.
Nat. Nanotechnol.
IEEE Trans. Nanotechnol.
IEEE Electron Device Lett.
Nano Lett.
Nano Lett.
Science
Science
Phys. Rev. Appl.
IEEE Trans. Electron Devices
J. Appl. Phys.
Cited by (0)
- 1
Deceased.