Modeling of Diffusion and Activation of Low Energy Arsenic Implants in Silicon

  • PDF / 243,033 Bytes
  • 6 Pages / 612 x 792 pts (letter) Page_size
  • 83 Downloads / 290 Views

DOWNLOAD

REPORT


Modeling of Diffusion and Activation of Low Energy Arsenic Implants in Silicon Srinivasan Chakravarthi, Chidambaram P.R., Charles Machala, Amitabh Jain, and Xin Zhang Silicon Technology Development, Texas Instruments, Dallas, TX 75243. Ultra-shallow n source/drain extensions require low energy arsenic implants combined with rapid thermal anneals in order to minimize diffusion. However, annealing of low energy arsenic implants exhibits anomalous activation/diffusion. Accurate models are needed to enable TCAD driven transistor design/optimization. The actual extent of arsenic extension diffusion/activation is critical to modeling nMOS transistors. In this work, we develop a physics-based model for arsenic that successfully models both diffusion and activation/deactivation during both front-end and back-end processing. An As4 V cluster coupled with a arsenic precipitation model was used to model anomalous behavior of low energy arsenic.

INTRODUCTION Transient enhanced diffusion (TED) has been the dominant effect in determining junction depths for the past decade and will continue to be important. However, the use of ultra-low energy implants and short, high temperature Rapid Thermal Processing (RTP) annealing has greatly diminished the importance of TED in ultra-shallow arsenic junctions. Reducing the implant energy is particularly effective as it puts the damage closer to the surface where it can be more readily annihilated, thus reducing the time period over which TED is present. Recent work shows that TED can be nearly eliminated for low implant energies.1 However, ultra-shallow arsenic data suggests other effects controlling arsenic diffusivity/activation.2 This paper focuses on understanding and modeling these effects.

EXPERIMENTS Arsenic was implanted at energies of 500 to 5000 eV and doses from 5 10 14 to 2 1015 cm 2 in 200mm p-type wafers. All implants were carried out on an Applied Materials LEAP-II. Implants into bare silicon were capped with an oxide before annealing. Anneals were carried out on a Centura XE+ (also Applied Materials). The RTP anneals were carried out in pure nitrogen (5 ppm O2 ). Secondary Ion Mass Spectroscopy (SIMS) analysis was carried out by Evans Texas using a 1 keV Cs beam. Electrical sheet resistance measurements were performed on a Prometrix Rs-75 instrument using a type-H four-point in-line probe. 





MODELS Initial Damage Each implanted ion creates a defect cascade, producing a large number of interstitials and vacancies. Fig. 1(a) shows a typical set of defect and dopant profiles from UT-MARLOWE, a Monte Carlo ion implantation simulator. Although the large initial interstitial and vacancy profiles

C3.7.1

1.4

22

10

Net I Net V Tot V Tot I As

10

1.2

Plus ā€™nā€™ Factor

Concentration (cm )

āˆ’3

21

20

10

1

0.8

19

10

0.6

18

10

0.4 0

50

100 o Depth (A)

150

200

0

2

4 6 Energy (keV)

8

10

Fig. 1: (a) Simulations from UT-Marlowe for a 2keV 1 1015 arsenic implant showing total interstitial and vacancy profiles. The net profiles are also plotted. (b) Calculate