Вы находитесь на странице: 1из 4

Epi Process Margin Improvement Using Co-implantation to Control Phosphorus Diffusion in a DRAM Manufacturing

Michael Hsiao1, Steve Ji1, Yiliang Lin1, Jay Huang1*, Chien-Hua Chu1, Pin-Yuan Yu1, Wei-Ming Wang1, Mei-Ju Chen1, Li-Yuan Cheng1, Chi-Ren Hung1 and Yu-Shan Chen1 B.N. Guo2**, W. Zou2, Lester Chiou3, Scott Wei3, H.L. Sun3, Alex Hsu3, T. Toh2, K.H. Shim2, B. Colombeau2, T. Henry2
Rexchip Electronics Corp., No. 429-1, Sanfong Road, Houli Township, Taichung County, Central Taiwan Science Park, Taiwan, R.O.C. *TEL: +886-4-25218061, E-mail: jayhuang@rexchip.com 2 Varian Semiconductor Equipment Associates, 35 Dory Road, Gloucester, MA 01930, USA **TEL : +1-978-2822469, E-mail: baonian.guo@vsea.com 3 VTL, Hsinchu, Taiwan, R.O.C. Parameter (KIP). The wafer-to-wafer variations of selective Epi thickness process were mainly from the deviation and characteristics of selective Epi process shown in the Fig. 2. To control the condition of Epi chamber wall, the selective Epi process was used by dry clean per 5 wafers. These 5 repeated Epi steps resulted in the larger variation of Epi thickness and reduced the process margin of selective Epi final thickness. Thicker Epi thickness will cause Epi-Epi short and EFT repair numbers increase (Fig. 3). Thinner Epi thickness usually results in Gate Induced Drain Leakage (GIDL) degradation due to the transient enhanced diffusion (TED) of Phosphorus (Fig. 4). Both cases will eventually cause device yield loss and device degradation. How to control Phosphorus TED and widen the process window of selective Epi thickness are critical for the yield improvement of DRAM products. Carbon and Fluorine co-implants have been shown to be effective in reducing Boron TED [1-3]. The Phosphorus diffusion mechanism is also dominated by TED [4-5]. However, the effect of Carbon and Fluorine co-implants on Phosphorus diffusion in selective Epi have been rarely investigated and documented. In this paper, the effect of Carbon and Fluorine co-implant on Phosphorus diffusion for Epi contact implant was investigated. These implanted bare wafers went through spike thermal process. Secondary Ion Mass Spectroscopy (SIMS) was used for profiling measurement to evaluate the effectiveness of co-implant conditions. Based on the Phosphorus profiles, the co-implant TED reduction approach was then applied to DRAM device wafer splits to verify its effects on the process window of the selective Epi process.
1

Abstract
Co-implantation has been proven to be an effective method to reduce Transient Enhanced Diffusion (TED). In this paper, the effect of Carbon co-implant energy, dose, and combined with Fluorine implants were investigated to control TED for a contact Phosphorus. With optimized co-implant conditions, Dynamic Random Access Memory (DRAM) device wafers were used to verify that the Epi process window can be enlarged due to better control of Phosphorus TED. The study revealed that Carbon suppresses the Phosphorus diffusion tailing and reduces Gate Induced Drain Leakage (GIDL) without degrading Vt and contact resistance performances. With the reduction of Phosphorus diffusion and GIDL, thinner selective Epi layer can be tolerated, resulting in widened process window of Epi final thickness and increased selective Epi process margin. Key Words: Co-Implantation, Transient Enhanced Diffusion (TED), Gate Induced Drain Leakage (GIDL), Carbon, Phosphorus Diffusion, Selective Epi Process Margin

1. Introduction
Selective Epi deposition for array contact followed with Phosphorus contact implant was used in a Dynamic Random Access Memory (DRAM) manufacturing. The cross-section of the device architecture was illustrated in Fig. 1. Selective Epi thickness is considered as one the Key Inline

Reference process Cell contact Photo Process a-C Dry Etch Oxide Dry Etch Resist Strip SiN deposition SiN Etch Back Resist Strip Co-Implant Split P-Implant (Contact-imp) Cell Contact Poly Depo. Contact Phos-Imp.

2. Experimental
To prove the concept of co-implant TED reduction for Phosphorus contact implant, 300mm p-type <100> Si wafers were used. All co-implant and Phosphorus implants were conducted on a Varian VHCS high current implanter. These implanted bare wafers went through the spike thermal process. SIMS was utilized to characterize the phosphorus, carbon and fluorine profiles. For device wafers co-implant experiments, selective Epi thickness splits were incorporated to simulate the real variations of Epi thickness. Germanium Pre-Amorphization Implant (Ge-PAI) was also used in device wafer splits to help to understand the fundamentals of Phosphorus TED. Key device parameters, such as Vt, contact resistance, and GIDL were measured upon process completion.

Selective Epi

Fig. 1. The process flow and cross-section of a DRAM device was illustrated. Array Epi and Phosphorus contact implant location and direction were marked.
50 Selective Epi Thickness (nm)

48

46

44

42 0 5 10 15 20 Wafer slots by 5X clean 25

3. Results
Bare wafers were used to determine the best co-implant condition to suppress TED for contact implant (Phosphorus 25keV, 1e13/cm2). Post-annealed Phosphorus SIMS profiles under various Carbon co-implant energy conditions were shown in Fig. 5. Carbon energies were varied from 2.5 to 15keV with 1e15/cm2 dose. The reference wafer (Phosphorus only) displayed the typical TED signature. With the introduction of Carbon, the Phosphorus profile tails diffused less and junction depth was decreased significantly. The profile improvement benefits from increasing Carbon energy and the improvement was saturated around 10keV. The predictive profile peak for 10keV Carbon is ~35nm which is comparable to the peak for 25keV Phosphorus. Overlap Carbon SIMS profile to Phosphorus profile appears to be the most effective in controlling Phosphorus diffusion. Fig. 6 depicted the comparison of phosphorus profiles as a function of Carbon doses (1e14 to 1e15/cm2 with energy at 5keV). The 1e15/cm2 condition demonstrated slightly better Phosphorus SIMS profile and was selected for further device wafer experiments. The combination of Carbon and Fluorine co-implants were also evaluated. The comparison of post annealing Phosphorus SIMS profiles with Carbon only and Carbon plus Fluorine co-implant conditions were presented in Fig. 7. The Carbon co-implant was 10keV, 1e14/cm2. The Fluorine co-implant was 3.5keV, 1e14/cm2. For the comparison purpose, the Phosphorus SIMS profile from Carbon 10keV,

Fig. 2. The selective Epi thickness by wafer slots shows the high variation of selective Epi thickness.
90

EFT Repair Num bers

85

80

75

70 40 42 44 46 48 50 52 Selective Epi Thickness (nm) 54

Fig. 3. The repair numbers of EFT yield will increase by thicker selective Epi thickness.
90

GIDL R epair N bers um

85

80

75

70 40 42 44 46 48 50 52 54 Selective Epi Thickness (nm)

Fig. 4. The repair numbers of GIDL will increase by thinner selective Epi thickness.

1e15/cm2 was also shown in Fig. 7. There is no noticeable improvement by adding additional Fluorine into the co-implant process. The combination of Carbon and Fluorine produce similar Phosphorus SIMS profile as Carbon only. Based on the bare wafer results, DRAM device wafers were used to evaluate the co-implant effects on device performance. Carbon 8 or 10keV at 1e15/cm2 were selected for device splits. Selective Epi thickness splits were incorporated to simulate the thickness variations of selective Epi chambers. The Epi split thickness ranged from 34.5 to 46.5nm. A Germanium 20keV 3e14/cm2 was used on two device wafers for comparison. The comparisons of Vt and contact resistance under various co-implant conditions were shown in Figs. 8 and 9, respectively. The compatible results make co-implant more adoptable and reliable than the baseline process flow.
1.E+19 P only C 10KeV 1e15+P C 5KeV 1e15+P C 2.5KeV 1e15+P C 15KeV 1e15+P

1.E+19

P only C 10KeV 5E14 + F 3.5KeV+P C 10KeV 1E15+P

Concentration (atom/cm 3)

1.E+18

1.E+17

1.E+16

1.E+15 0 100 200 300 400 500

Depth(nm)

Fig. 7. Comparison of post annealing Phosphorus SIMS profiles under Carbon and Carbon plus Fluorine co-implant conditions.
W/o Co-IMP
0.60 0.55 Vt(V) 0.50 0.45 0.40 0.453 0.459 0.444 0.461 0.447 0.454 0.470 0.457 0.447 0.453 0.35 0.461
.5 34 .5 40 ) ) ) ) ) ) ) ) .5 8K (8K (8K 10K 10K 10K 20K 20K 46 ( ( ( ( ( C( C C 5- 0.5- 6.5- .5-C .5-C .5- C -Ge -Ge . 34 4 4 34 40 46 40.5 46.5
VTCRB C-8K

C-10K

Ge-20KeV

Concentration (atom/cm 3)

1.E+18

1.E+17

1.E+16

Epi Thickness(nm)

1.E+15
0 100 200 300 400 500

Depth(nm)

Fig. 8. Vt comparison of DRAM device under various co-implant conditions.


W/o Co-IMP
5000 4500 4000
RCSNSC1_CELL C-8K

Fig. 5. Comparison of post annealing Phosphorus SIMS profiles under various Carbon co-implant energies.
1.E+19 P only C 5KeV 1E15+P

C-10K

Ge-20KeV

Rs(ohm) Data

3500 3000 2500 2000 1500

Concentration (atom/cm 3)

C 5KeV 5E14+P 1.E+18

1.E+17

.5 34

.5 40

.5 K) 8K) 8K) 0K) 0K) 0K) 0K) 0K) 46 (8 ( ( 1 1 1 2 2 -C 5- C 5-C -C( -C( - C( Ge( Ge( .5 . . 5 5 5 46 34. 40. 46. 0.5- 6.534 40 4 4

1.E+16

Epi Thickness(nm)

Fig. 9. Contact Resistance (Rs) comparison of DRAM device under various co-implant conditions.
0 100 200 300 400 500

1.E+15

Depth(nm)

Fig. 6. Comparison of post annealing Phosphorus SIMS profiles under various Carbon co-implant doses.

Fig. 10 depicts the GIDL performance under various implant conditions. Both Carbon co-implant energy conditions show the improvements in GIDL performance, even more significant in thinner Epi cases. The results indicate that the Carbon co-implant can reduce Phosphorus lateral diffusion into the extension region and achieve lower GIDL. The GIDL

reduction makes the thinner selective Epi acceptable and allows the process integration to extend the lower specification limit (LSL). Therefore the selective Epi process margin is greatly improved. The wafers with Ge-PAI prior to Phosphorus implant show GIDL performance degradation which might have caused by the End of Range (EOR) defects generated by the Ge-PAI implants [6]. Fig. 11 shows the similar trend of STH (Static Hold) performance. Carbon co-implant can suppress the Phosphorous TED effectively that is not sensitive to the Epi thickness and therefore improves the process margin of selective Epi process caused by high variations of Epi thickness (especially thinner Epi thickness).
100 W/O Co-IMP C-8K C-10K Ge-20K

support that overlap of Carbon and Phosphorus SIMS profiles reduces the Phosphorus TED. Interestingly, matching profile peaks for Carbon and Phosphorus shows the best Phosphorus TED control. DRAM device studies have verified the co-implant benefits. The compatible Vt and contact resistance, improved device GIDL performance, enlarged selective Epi thickness process window and process margin, make co-implant more adoptable and reliable than the baseline process flow. In the future, this study should be helpful for the yield improvement and advanced technology nodes.

Acknowledgments
Authors would like to thank VSEA Application Lab. for their technical suggestions and SIMS analysis supports.

80

60 G L ID

40

References
[1] G. Impellizzeri, J.H.R. Santos, S. Mirabella, and F. Priolo, Suppression of Boron Diffusion by Fluorine Implantation in Preamorphized Silicon, Mat. Res. Soc. Symp. Proc. Vol. 810, C5.9.1-6, (2004). [2] H. Li, P. Kohli, S. Ganguly, T.A. Kirichenko, P. Zeitzoff , K. Torres, and S. Banerjee, Boron Diffusion and Activation in the Presence of Other species, 0-7803-6441-4/00 2000 IEEE. [3] B. Colombeau, S.H. Yeong, D.X.M. Tan, A.J. Smith, R.M. Gwilliam, C.M. Ng, K.R.C. Mok, F. Benistant, and L. Chan, Ultra-Shallow Junction Formation Physics and Advanced Technology, Int. Conf. Ion Implantation Technology Proc., 11-18 (2008). [4] Y. Momiyama, K. Okabe, H. Nakao, M. Kojima, M. Kase, and T. Sugii , Extension Engineering using Carbon co-Implantation Technology for Low Power CMOS Design with Phosphorus- and BoronExtension, IWJT 63-64, (2007) . [5] A. Vanderpool, Importance of the Carbon Kick-out Mechanism in Reducing Transient Enhanced Diffusion, Int. Conf. Ion Implantation Technology Proc., 213-216, (2008). [6] C.O. Chui, L. Kulig, J. Moran, and W. Tsai, and K.C. Saraswat, Germanium n-type shallow junction activation dependences, App. Phys. Lett., 87, 091909, (2005).

20

0 30 35 40 Epi Thickness(nm) 45 50

Fig. 10. GIDL comparison versus Epi thickness under various co-implant conditions.
W/o Co-IMP 180 150 120 90 60 30 0 30 35 40
Epi thickness (nm)

C-8K C-10K Ge-20K

STH R epaired num bers

45

50

Fig. 11. STatic Hold (STH) comparison versus Epi thickness under various co-implant conditions.

4. Conclusions
In this paper, we have presented experimental results and analyses using co-implants on TED control for DRAM contact Phosphorus implant. The results

Вам также может понравиться