Minimization of Well-Proximity Effect by Means of 2D and 3D
Monte Carlo Simulation of Retrograde Well Implantation

 

Introduction

The formation of deep p- and n- wells using high-energy implantation has become an integral part of CMOS technology process flow. The high energy and high dose implantation into the cleared area of a thick photoresist mask generates retrograde profiles. These profiles have a relatively high peak concentration usually at the depth of approximately 1 micron and a very low surface concentration. From the first glance this process achieves its primary goal to isolate NFETs from PFETs without affecting surface areas where the transistors are formed. Unfortunately for both technology and circuit designers, this relatively simple process step brings about an unwanted Well Proximity Effect (WPE) [1] exhibited by a strong dependence of threshold voltage Vt on transistor location and even orientation within the well.

 

Origin of WPE

The WPE is caused by an extra non-uniform doping at the surface of the well area by ions scattered within photoresist and emerged from the mask edge at different angles. Figure 1 clearly illustrates this phenomenon. It shows that many of the 2000 boron ion trajectories, simulated by MC Implant Module of the process simulator ATHENA [2], terminate not inside photoresist but at different locations within PWELL. This extra doping near the surface may shift Vt by as much as 100 mV. However, the worst consequence of the WPE is the fact that the value of the shift Vt depends strongly on transistor gate position inside the well. Moreover, it is not easy to quantify and control the Vt pattern within layout. Therefore, even cumbersome compact transistor models attempting to account for the WPE (see, e.g. [3]) appeared to be very impractical.

Figure 1 Boron ion trajectories started at a single impact point 0.1μ of the vertical mask edge.

 

How to Minimize WPE

Polishchuk, et.al, [4] experimentally showed that it is possible to reduce the WPE by altering mask thickness and shape as well as by introducing an extra screen layer under the mask. The goal of this work is to use Monte Carlo implant simulation for optimizing masking conditions in order to greatly reduce or even eliminate consequences of the WPE in most cases. Figure 2 illustrates the main idea of such optimization. Majority of trajectories of high-energy ions impacted in a single point are confined within a distinctive cone which height and slope are determined by ion type and energy as well as density and composition of the amorphous photoresist material. If the mask is slightly thicker than the height of the imaginary cone and the mask edge slope is slightly larger than the slope of the cone then most of the ions will stop within the mask. The same is valid for ions impacting the mask slope. In the case shown in Figure 2, only 4 out of 2000 simulated trajectories have emerged from the sloped mask edge, which suggests that the WPE could be considerably suppressed.

Figure 2 Boron ion trajectories started at a single impact point exactly at the upper corner of the sloped mask edge.

 

 

2D Simulation Results and Analysis

To confirm that using the sloped mask edge could indeed minimize the WPE, full 2D Monte Carlo simulations were performed for both vertical (Figure 3) and sloped (Figure 4) mask edges. To obtain a reasonable accuracy of profiles near the PWELL surface, 15 million 300 keV boron trajectories were simulated using parallel version of the MC Implant [2]. It took approximately 10 hours on a multiple CPUs Linux workstation. Figure 5 compares lateral boron profiles right under the surface. It clearly shows that using mask edge with optimized slope decreases boron surface concentration by at least one order of magnitude. This means that the Vt shift measured near the sloped mask edge will be as small as the Vt shift measured as far as 1 μm from a vertical edge. Figure 6 compares the vertical profiles for unmasked implant with profiles obtained for vertical and sloped masks. It clearly shows that in the case of vertical mask edge the surface concentration of ions scattered out of the mask could be as high as that of Vt adjust implant. The sloped mask decreases this extra concentration by the order of magnitude. Figure 7 and Figure 8 illustrate effect of a screen layer. Figure 7 shows that if a 0.1 μ screen photoresist layer is used in attempt to further reduce the WPE then the 400 keV boron implantation should be used to maintain the same depth and retrograde shape of the profile obtained for 300 keV implant. Figure 8 shows that the screen layer by itself could not sufficiently reduce the WPE, while combination with the sloped mask edge can practically eliminate the WPE.

Figure 3. 2D boron PWELL implant distribution in case of vertical mask edge.

 

Figure 4. 2D boron PWELL implant distribution in case of sloped mask edge.

 

Figure 5. Effect of mask edge shape on lateral boron profile along the surface of PWELL and STI regions. “Vertical Mask Edge” profile is extracted from Figure 3, while “Sloped Mask Edge” profile is extracted from Figure 4

 

Figure 6. Effect of mask edge shape on vertical boron profile at 0.3μm to the left from STI/PWELL boundary. “Vertical Mask Edge” profile is extracted from Figure 3, while “Sloped Mask Edge” profile is extracted from Figure 4

 

Figure 7. Effect of 1D PWELL implantation through 0.1μm layer of the screen photoresist.

 

Figure 8. Effect of 0.1μ screen photoresist layer on lateral boron profile along the surface of PWELL and STI regions. Upper profile corresponds to 300 keV boron implant (see Figure 3). The middle profile corresponds to 400 keV boron implant simulated in case of mask with the vertical edge as in Figure3 but with extra 0.1μ screen photoresist layer over PWELL and STI surface. The lower profile corresponds to 400 keV boron implant simulated in case of mask with the sloped edge as in Figure4 but with extra 0.1μ screen photoresist layer over PWELL and STI surface.

 

 

3D Simulation Results and Analysis

To further analyze 3D corner effects we performed a full 3D Monte Carlo implant simulations in a L-shaped mask with the same mask edge settings as in Figures 3,4. Figure 9 and 10 show shallow surface boron distributions obtained by simulation of 100 million 300 keV boron trajectories. The simulated distributions demonstrate pronounced corner effects in a L-shaped implant window. These two surface contour plots also unequivocally show that implantation into the mask window with optimally sloped mask edges provides a considerable gain of area with low near surface concentration. This way, the corresponding area inside the well without critical Vt shift greatly expands toward mask edges.

 

Figure 9. Contour density plot of shallow surface concentration of 300 keV boron implant into L-shaped mask. For this simulation the implant window opening had vertical walls of the photoresist mask as in Figure 1.

 

Figure 10. This contour plot is simulated with the same implant as in Figure 9 but the photoresist was exposed and developed in such a way as to have an optimal sloped mask edge (see Figure 2).

 

Conclusion

Comprehensive 2D and 3D simulations of high energy boron ion implantation are used to investigate the origins of the Well Proximity Effect as well as possible methods to avoid negative consequences of this effect It is shown that WPE could be greatly decreased and practically eliminated if the right combination of process parameters and shape of the resist mask is selected by careful analysis of ion trajectories and results of full 2D/3D implant simulations.

References

  1. T.B. Cook, et.al, “Lateral ion implant straggle and mask proximity effect”, IEEE Trans. Electron. Devices, Vol. 50, pp. 1946-1951, 2003.
  2. ATHENA Users Manual, Silvaco Data Systems, Santa Clara, CA, USA, www.silvaco.com, 2007.
  3. Yi-Ming Sheu, et.al, “Modeling the Well Edge Proximity Effect in Highly Scaled MOSFETs”, IEEE Trans. Electron Devices, Vol. 53, pp. 2792-2798, 2006.
  4. I. Polishchuk, et. al, “CMOS Vt-Control Improvement through Implant Lateral Scatter Elimination”, IEEE International Symposium on Semiconductor Manufacturing, ISBN: 0-7803-9143-8, pp. 193-196, 2005.

Download PDF Version of this article