Metal-InGaAs contact resistance calculations from first principles Troels Markussen and Kurt Stokbro QuantumWise A/S Fruebjergvej 3, Postbox 4 DK-2100 Copenhagen, Denmark Email: troels.markussen@quantumwise.com Abstract—The metal-semiconductor contact resistance is an important factor in the performance of MOSFETs and a detailed understanding of the contact resistance is necessary in order to simulate and eventually optimize the devices. In this work we cal- culate metal-InGaAs contact resistances using Density Functional Theory (DFT) combined with non-equilibrium Greens function (NEGF) methods as implemented in Atomistix ToolKit (ATK). We have calculated specific contact resistivities, ρc, of metal-InGaAs and metal-InAs interfaces, where the metals studied are Ti, W, and Mo. We find that ρc is approximately inversly proportional to the doping density with values that are in close agreement with experimental results from the literature. I. I NTRODUCTION Interfaces between metals and semiconductors occur in a wide range of technologies such as electronic and optoelec- tronic devices, thermoelectric modules, as well as quantum computing devices[1]. The metal-semiconductor contact re- sistance is an important factor in the performance of both MOSFETs[2] and thermoelectrics[3] and a detailed under- standing of the contact resistance is necessary in order to simulate and eventually optimize the devices. In this work we report on theoretically calculated contact resistances. The calculations are performed with Density Func- tional Theory (DFT) combined with non-equilibrium Green’s function (NEGF) methods as implemented in ATK[4]. This approach allows for first-principles studies of arbitrary material combinations and does not require fitted parameters as is the case for e.g. tight-binding (TB) methods. The DFT-based approach is thus particularly well suited for interface studies as TB parameters between metals and semiconductors are generally not available in the literature. We have calculated specific contact resistivities of metal- In 0.5 Ga 0.5 As and metal-InAs interfaces, where the metals studied are Ti, W, and Mo, which have been studied ex- perimentally [5]. We further study both In- and As termi- nated semiconductor surfaces. For both metal-InAs and metal- InGaAs interfaces the calculated specific contact resistivities are in close agreement with experimental values. II. METHOD A. Metal-InGaAs interface atomic structure When studying an interface the first step is to setup an appropriate atomic configuration. In all calculations presented below, the In 0.5 Ga 0.5 As is oriented such that the [001] di- rection is normal to the interface plane. We do not simulate In 0.5 Ga 0.5 As as a random alloy, but always represent it with periodically repeated alternating planes of In-As-Ga-As-In- etc. The orientation of the metal has been chosen as a compro- mise between minimizing the strain, which is always applied to the metal, and at the same time keeping the system size computationally feasible. The average strain applied to the metal is never larger than 5%. Since we are not aware of any experimental results for the crystal orientation of the metal, we have generally taken the liberty to choose a metal orientation that minimizes the strain with the smallest possible super-cell area. Having defined an initial interface structure, we first form a slab geometry periodic in x and y but with finite sizes of Ti and InGaAs in the z-direction and with a large vacuum region. We then allow for a structural relaxation of the atoms closest to the interface. The remaining atoms are allowed to move as a rigid body, i.e. their internal coordinates are fixed and only the centre of mass is moved [6]. The structural relaxation is performed until the maximum force on all the freely moving atoms (close to the interface) is below 0.05 eV/ ˚ A. The structural relaxation is performed with the Perdew-Burke- Ernzerhof (PBE) [7] generalized gradient approximation for the exchange-correlation potential with a transverse k-point density of typically 6 ˚ A (this corresponds to having one k- point in a direction with real space unit cell length of 2π × 6 ˚ A). The right end of the In 0.5 Ga 0.5 As which is facing vacuum is passivated using hydrogen atoms in order to avoid metallic surface states. Having relaxed the interface, we construct a device config- uration by extending both the metal and In 0.5 Ga 0.5 As using the fixed structures at the ends. The device configuration consists of a central region shown in Figure 1 connected to two semi-infinite electrodes. The atomic structure as well as the potentials are fully periodic in the electrodes, whereas in the central region there can be a non-periodic potential profile as will be shown below. B. Device calculations For the device calculations we have used the Meta-GGA exchange-correlation functional by Tran and Blaha [8] with 978-1-5090-0818-6/16/$31.00 c 2016 IEEE Simulation of Semiconductor Processes and Devices 2016 Edited by E. Bär, J. Lorenz, and P. Pichler 373