Stacked Metal Layers as Gates for MOSFET Threshold Voltage Control

  • PDF / 91,369 Bytes
  • 6 Pages / 612 x 792 pts (letter) Page_size
  • 17 Downloads / 294 Views

DOWNLOAD

REPORT


D1.4.1

Stacked Metal Layers as Gates for MOSFET Threshold Voltage Control Wei Gao, John F. Conley and Yoshi Ono Sharp Labs of America, 5750 N.W. Pacific Rim Blvd., Camas, WA 98607 U.S.A ABSTRACT Two layer metal gate stacks allow the effective work function to be tuned by varying the thickness of the first metal layer. Metal-oxide-semiconductor (MOS) capacitors were fabricated by using two metals of very different work functions on thermal oxide gate dielectric where the bottom layer thickness is varied over a range from 0 to 50nm. Electrical and thermal stability measurements were performed on the Al on TaN metal gate stack. The effective workfunction is seen to shift from the value of one metal to the other rapidly as the thickness of the first metal layer is varied from 0 to approximately 10nm. The flat band voltage (Vfb) transition matches the workfunction difference of the two metals in the stack. The advantage of this approach when applied to metal-oxide-semiconductor-field-effect-transistors (MOSFETs) is that it allows the effective workfunction of the metal stack, and the threshold voltage (Vth) of the device to be fine tuned. It also allows for eventual dual gate complementary MOS (CMOS) device fabrication where two different work function metal stacks are necessary, without processing directly on the gate dielectric. A model is proposed to elucidate the workfunction tuning mechanism. INTRODUCTION As the gate length of silicon CMOS devices is scaled below 100nm, new high-k materials are expected to replace thermal silicon oxide, and metal gates to replace polycrystalline silicon. Metal gate not only solves the polysilicon gate depletion problem but could also enable threshold voltage adjustments without altering channel doping. However, it is a challenge to introduce metal gates to CMOS process without significantly complicating the process. The following issues must be considered: 1. Since two different metals with proper workfunction will be required for CMOS devices, removing one metal over the gate dielectric to deposit another is detrimental to the gate dielectric integrity. 2. The metal workfunction is closely correlated with its chemical activity and etching properties. NMOS metals are highly reactive and normally unstable in contact with the gate dielectric, while PMOS metals are more stable but harder to process. 3. A stable metal in contact with the gate dielectric may still exhibit poor adhesion or diffusion into the channel. Many efforts have addressed these issues, still a metal gate metal system that solves all of these problems remains a challenge. Polishchuk, et al. used a diffusion method to mix Ti and Ni to achieve NMOS and PMOS workfunction by varying the Ti thickness. The workfunction is successfully adjusted, however, the stability issue is not addressed [1]. Ranade et al. used N implantation to tune the workfunction of Mo to achieve as low as 4.5eV. For NMOS, the channel doping will be compromised in order to achieve the targeted threshold voltage [2]. Samavedam, et al. did dual metal