1994 BipoladBiCMOS Circuits zyxw 8, Technology Meeting 12.1 z The Importance of Including Lattice Self-Heating and Hot-Carrier Transport in BJT Simulation Minchang Liang and Mark zyxwv E. Law 335A Larsen Hall, University of Florida Gainesville, FLorida 32611-6200 ABSTRACT As the size of VLSl devices shrink, lattice self-heating and hot-carrier transport effects become significant in their operation. As a result, the drift-diffusion model which ignores local heating and models the hot-carrier transport effects with simple local field-dependent rela- tionships begins to fail. Simulations were performed to demonstrate how these effects affect device operation, and the importance of simultaneously including both effects in device simulation is discussed. 1. INTRODUCTION As device technologies improve, the size of the semi- conductor devices has been shrunk dramatically. Since the applied bias has not be reduced proportionally, the electric field and field gradient become large enough that the self-heating and hot-carrier transport effects become non-negligible. As a result, the traditional drift-diffusion model which ignores self-heating and models the hot- carrier transport with a simple field-dependent relation begins to fail. For correct modeling of state-of-the-art semiconductor devices, thermal effects have to be included in the device simulations.[l][2] In this paper, a more advanced carrier transport system which includes the carrier energy transport and the lattice self-heating effect as part of the solutions will be briefly presented. A new general-purpose device simulator, FLOODS ( a o r i d a Qbject-Qriented Qevice Simulator) [3], was developed. The simulator was used been made such that the underlying device physics is more complete. The system describes the conservation of carriers and carrier energy at any location in the devices during any period of time, while the transport of the carrier and carrier energy inside devices are derived from the Boltzmann Transport Equations under the relax- ation time approximation. The resultant semiconductor equations are: -Ve(EV@) = p = q(p-n+C) (1) at = -(Vdn+Rn) zyxwv 3 2 Wn = E, + -ne KBTn awl - = -VeSn - R,,,, zyxwv at (3) (4) WL = CL. TL (6) awL at - = -Ve$, - R,, where @, zyxwvu n, zyxwv p, T,, Tp and TL are respectively the electro- static potential, electron and hole concentration, electron and hole temperature, and the lattice temperature. Rn and Rp are respectively the electron and hole recombi- nation rate; if transient trapping is ignored, Rn+Rp = 0. R,, , Rwp and R, model the energy transfer among the electrons, holes and crystal lattice, and the physical mechanisms include energy transfer through scattering to study the influence and significance of the lattice self- heating and carrier energy transport on advanced device performance. In this paper, we will focus on discussing the simulation results of a BJT structure. Both the DC and AC characteristics will be studied, and the results show that, for modern BJT structures, including both lattice and carrier thermal effects is required for accurate prediction of device performance. and recombination[5]. Sn , S ’ , 3, , 3, and 3, are the particle and energy transport equations, and they can be written as: jn = - (npnVEc) -DnVn - ( 1 -a) npnV( -) (7) (8) KBTn 4 ?p = PCL,VE,-Q Va- ( 1 - a) PCL~V( KBTp zyx 7) Jn=( (l ) B n) 4 GTn E,+ --a K T Sn- (5/2-a)nyn-VT, (9) 2. THE THERMODYNAMIC SYSTEM The thermodynamic system is based on the system described in [4], although some modifications [5] have 187 0-7803-21 17-0/94/$4.00 0 1994 IEEE