• 沒有找到結果。

Evaluation of advanced chemical mechanical planarization techniques for copper damascene interconnect

N/A
N/A
Protected

Academic year: 2021

Share "Evaluation of advanced chemical mechanical planarization techniques for copper damascene interconnect"

Copied!
6
0
0

加載中.... (立即查看全文)

全文

(1)

0040-6090/04/$ - see front matter䊚 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2003.07.015

Department of Applied Physics, National Chiayi University, Hsin-Chu, Taiwan, ROC

Department of Material Science and Engineering, National Cheng Kung University, Tainan, Taiwan, ROC e

Abstract

According to rapid development of CMP technology, the difference of polishing methods would be applied for copper damascene interconnect. These methods include conventional rotary, linear, oscillation platform. The advantages of these platform would highlighted in uniformity control, stability of removal rate, planarization efficiency, throughput promotion, lower cost of ownership, even dishing and erosion effect integrated with slurry. This paper presented our experience to compare the uniformity, removal rate and planarization efficiency of Cu-CMP between various polishing platforms. In convection, the rotary or oscillation platform would be usually applied in oxide and tungsten CMP, due to the robust air-back carrier and rigid platen to polish the wafer, which performs the good reliability on wafer-to-wafer thickness and endpoint detection. However, the linear polisher is made of the air-bearing moving belt and self-rotated carrier and would provide the wider uniformity and planarization control window than others. In addition, the simulation model would explain the difference from mechanic design and wafer moving paths between various polishing techniques. Hence, the trade-off advantage and application between various platforms would be compensated and integrated with slurry, conditioners and coming thickness profile from copper plating. It can achieve the optimization of Cu-CMP process.

䊚 2003 Elsevier B.V. All rights reserved.

Keywords: Copper; Chemical mechanical planarization(CMP); Computer simulation

1. Introduction

Copper chemical mechanical polishing(Cu-CMP) has

been an important process in copper interconnect tech-nology. A considerable amount of work has been done in achieving the best planarization, dishing and erosion minimization and polishing control in CMP. There stud-ies mostly addressed on slurry chemical effects w1x, pad deformation w2x and the optimization between pattern density and overpolish on copper CMP planarization

w3,4x. Although these approaches tended to increase the

polishing window, there is still a hampered work and discussion to integrate the various CMP process methods under different electroplating conditions and control the *Corresponding author. Tel.: 6-5052000x8069; fax: q886-6-5052081.

E-mail address: ylwang@tsmc.com.tw(Y.L. Wang).

consistent process way in the uniformity and removal rate. For example, in convectional copper electroplating process, there is a phenomenon of faster copper depo-sition rate on the edge of wafer. It is caused from the non-uniformity of the electric field distribution between multiple electrodes cycled and contacted around wafer edge. The copper plating thickness profile is shown on Fig. 1. Based on the plating result, the CMP process is always requested to increase the edge-polishing rate. However, the requirement with the faster edge-polishing rate would be correlative to the polishing platforms designs, slurry delivery characteristics, planarization efficiency and overpolish windows. To integrate the relationship and optimize the process windows, this paper will investigate the process comparison and capa-bilities on various polishing methods and figure out the correlation between polishing methods and process par-ameters; then in advance simulate these polishing

(2)

para-Fig. 1. Copper thickness profile across the edge of the wafer after electroplating process.(Z level means the distance between the

cath-ode and anodic electrcath-odes.)

Fig. 2. Platen and wafer configuration for conventional rotary polisher.

(Platen and wafer rotate at the same direction.)

Table 1

Comparison of process performance in various polishing platform

Platform Removal rate Within-wafer Endpoint u-scratch Remarks

(polisher) deviation(3s) uniformity detection defect

deviation(1s)

Rotary 10–15% 4–8% Stable More Rigid platen to

or orbital contact the wafer

Linear 20–30% 2–5% Less stable Less More tracking in

air-bearing belt meters to understand the performance of uniformity and

removal rate on various polishing platforms.

2. Description-copper polisher techniques

There are more than two types of commercial CMP platform applied in copper polishing process. By the pad moving direction, they can often split into two major types: one is rotary or oscillation polisher and the other is linear.

In conventional rotary or oscillation CMP platform design, the wafer is located under the carrier film, diaphragm or air-back polishing head; then closely contact to the polishing pad, which is attached above the rigid platen. This design is useful to control consis-tent removal rate for wafer-to-wafer polishing, due to exist stable contact and provide less variation of slurry film between the wafer and the pad. However, there is a limit capability in non-uniformity and polishing profile control w5x. It is lack of the flexible space to compensate the pad deformation and down-force distribution on the wafer. Hence, many developments focus on the novel air-back carrier w6x, multiple-zone backside pressure w7x or in-situ dynamic full-wafer endpoint detection w8x. All wish to provide the tunable polishing profile and improve the control of non-uniformity.

However, linear polisher would relatively provide more flexible uniformity and profile control with the air-bearing platen design w9x. It is due that the design provides the adjustable belt above air-bearing platen to

fit the full-wafer surface. However, the floating air-bearing belt also performed less stable removal rate control for wafer to wafer than rotary one. Hence, we try to summarize the process comparison of these platforms in Table 1 and explain the advantage and disadvantage of polishing methods for the endpoint system, uniformity and removal rate control. It will be useful to understand the difference for various polishing methods and optimize the process between plating and polishing.

3. Simulation for various CMP technique

Although the various CMP platform design applied and required in the process optimization, some common parameters would be adjusted in each approach. These parameters mainly include with down-force, average relative velocity and polishing sweeping trajectory. By the adjustment of these parameters, it is possible to benchmark with different polisher and perform the similar results in the uniformity and removal rate. Hence, the following simulation will calculate the correlation of these parameters and understand the process control in advance.

(3)

Fig. 3. Belt and wafer configuration for linear polisher.(Wafer rotated

on the center of the belt.)

where V and V are the x and y components of thex y

velocity, V is the angular velocity of the platen, f is an angle, which is illustrated in Fig. 2.

The relative velocity vector of a point on the pad with respect to wafer is the pad velocity vector in the reference coordinate system, which is fixed on the rotating wafer. Furthermore, the relative velocity vector can be discomposed into two components, one parallel to and the other perpendicular to the radius of the concerned point on the wafer. The parallel component is

vparasV cos vt qV sin vtx

Ž

.

y

Ž

.

syVRsinfcos vt qVRcosfsin vt

Ž

.

Ž

.

(2) sVRsin vtyf

Ž

.

sVrsin v

Ž

t

.

where ris the distance between the center of the platen and the center of the wafer, vis the angular velocity of the wafer ration, and t is time. Note that R in Eq. (2)

is a variable and its value depends on the rotation angle of the wafer. The perpendicular component of the relative velocity vector is

vperpsyV sin vt qV cos vt yvrx

Ž

.

y

Ž

.

sVRsinfsin vt qVRcosfcos vt yvr

Ž

.

Ž

.

(3) sVRcos vtyf yvr

Ž

.

s

Ž

Vyv

.

rqVrcos vt

Ž

.

The name of modulus of the relative velocity vector will be abbreviated to relative velocity in all following paragraphs for the reason of simplicity. The relative

velocity is equal to 2 2

y

vs vparaqvperp (4) 2 2 2 2

y

s V r q Vyv

Ž

.

r q2 Vyv Vrrcos vt

Ž

.

Ž

.

The average of relative velocity over a wafer rotation cycle is given by v 2pyv ¯v r sŽ .

|

v dt 2p 0 2p S B E B E2 2 B E W T 1 v r v r T U C F C F C F X sVr 1q 1y q2 1y cosu du T

|

y

T D G D G D G 2p V r V r V 0 Y (5) 3.2. Linear polisher

The linear polisher, such as LAM Teres CMP system

w9x, has a belt, which moves linearly. The illustration

and notation are shown on Fig. 3. Hence, the belt velocity vector is

V s0x

(6) V sVy

whereV is the linear motion speed of the belt.

The parallel component of the relative velocity vector is

v sV sin(vt) (7)

para

wherev is the angular velocity of wafer rotation. The perpendicular component of the relative velocity vector is

(4)

Fig. 4. Comparison of planarization efficiency between rotary and linear polishers.(Step height change measured on the 100=100 mm copper

pad.)

The relative velocity is equal to

2 2

y

vs vparaqvperp (9) 2 2 2

y

s V qv r y2Vvrcos vt

Ž

.

Using Eq. (8), the average relative velocity over a wafer rotation cycle is given by

v 2pyv ¯vs

|

v dt 2p 0 (10) 2p S B E2 W T 1 vr vr T U C F X sV 1q y2 cosu du T

|

y

T D G 2p V V V 0 Y

Assumed the polishing processes obey the Preston’s law, the correlations between polishing profile and the average velocity (the removal rate, too) are calculated

by the Eqs. (5) and (10), responded to the rotary and

linear polisher, respectively. The correlation can figure out the difference of the uniformity between various polishers. Besides, in the rotary polisher, the distance from the center of platen to the center of wafer would be applied with different carrier sweeping trajectory, such as usual simple harmonic motion (SHM) or sinu-soid oscillation. Thus, it can change the uniformity and polishing profile from the parameters tuning to achieve the optimization from polishing profile to plating profile on the copper films.

4. Experiments

All experiments are carrier out on the 8 inch wafers. For the blanket wafer with copper film, they are polished in the example of rotary polisher, AMAT Mirra-Mesa CMP, and in the linear polisher, LAM Teres CMP. These polishers would be applied with the same slurry and conditioners. In addition, a four-point probe tool, Tencor

Rs-75 can be used to measure the removal rate and uniformity. The step height changes during polishing are measured on the 100=100 mm copper pad by the alpha-step profiler. The planarization efficiency is calculated by the difference between the initial step height and the minimum of step height over the polishing time.

5. Results

Beyond the comparison between various polisher, shown on Table 1, the planarization efficiency has been done in the Fig. 4. The data show that the planarization rate (namely, the ratio of the pattern step height reduc-tion to polishing time) on linear polisher is slightly

faster than on the rotary one before the lowest step height reached. Besides, the planarization rate is posi-tively dependent on the local removal rate w10x

(there-fore, the local relative velocity too, if polishing obey Preston’s law.) From the Eqs. (4) and (8), the ratio of

the relative velocity of linear to rotary polishers would be greater than 1 on the assumption ofV being equal to Vrand V close tov,(i.e. assumed linear belt speeds

platen’s rotary speed and platen’s angular velocity s wafer’s angular velocity). That is, the belt of linear

polisher seems to perform more rigid pad deformation and slightly faster relative velocity across the local area of the wafer than the platen of rotary polisher. Therefore, the planarization efficiency of linear polisher is better than ones of rotary polisher w10x.

In order to realize the difference between various platforms, we take the advantage of the Eqs. (5) and

(10) to simulate the profiles of the average velocity

across the radius of wafer for rotary and linear polisher, shown in Fig. 5. (It is same meaning for the profile of removal rate if the process also obeys the Preston’s law.) From the simulation results, it shows the velocity

profile changing with various polishing sweep trajecto-ries in the rotary polisher. The sweeping trajectotrajecto-ries are

(5)

Fig. 5. The relationship between normalized average relative velocity profile and the distance from the center of wafer for various polisher and sweeping trajectory in rotary polishers.(Therefore, removal rate too if the polishing obey the Preston’s law.)

Fig. 6. The various sweeping trajectories of polishing carrier in the rotary polishers.(Some of the data in Fig. 5 is calculated from these trajectories.)

just usual examples applied in the copper CMP system. All are illustrated in Fig. 6. That replies the uniformity of rotary polishers would be worse than the linear ones. Of course, the uniformity control is also changed by the various polishing sweeping trajectories. For example, the thickness profile of the electroplating, like the result of Fig. 1, would be suit to polish with the parabolic sweeping trajectory. It can achieve the faster removal rate in the wafer edge and perform the better uniformity and process optimizations between plating and polishing. Besides, if the plating profile of Fig. 1 is polished in the linear polisher, it is easy to suffer the copper remaining on the edge of the wafer, due to the more flat profile in the linear polisher, shown in the Fig. 5. Hence, the development of the extra air-zone force would be applied and designed under the edge of the belt w9x.

Sometimes the simulation equations can explain and calculate the correlations of mechanic parameters between rotary and linear polishers. For instance of 8 inch polishers, if the value of linear belt speed (V) is setting three times than the value of angular velocity

(V) of rotary platen, there is the same removal rate

performed in both platforms. Besides, the carrier rotation speed is less impact on the uniformity, while the belt speed is greather than the carrier rotation speed (i.e. V4vr) in the linear polisher or while the angular

velocity (V) of rotary platen is close to the angular

velocity of the carrier. All of these correlations would clearly figure out the characteristics of various polishers in polishing profile and uniformity control.

6. Conclusions

In this study, rotary and linear polishers are used as the main polishing platforms in copper CMP process. On the difference of mechanical designs in both polish-ers, there is more stable removal rate in the rotary polisher, but better uniformity control in the linear polishers. It is due to the rigid contact and support between wafer and the pad for the rotary polisher and flexible motion of the linear belt and air-bearing platen across the wafer for the linear polisher. Besides, from the simulation Eqs. (5) and (10), the local relative

(6)

rotary polisher. Hence, in the planarization efficiency, the linear polisher is better than the rotary one.

In addition, simulation equation of linear and rotary polishers can explain the difference of profile and uniformity control. By various sweeping trajectories of polishing carrier on the rotary polisher, the optimized polishing profile can totally resolve the thicker copper thickness issue on the wafer edge from the non-unifor-mity of the electric field distribution on the electroplat-ing process.

Finally, the mechanic parameters in the simulation equations are not only the factors on the uniformity and polishing profile, but other factors, such as slurry effects, pad deformation and conditioner wearing, also affect the removal rate and uniformity performance. Hence, under the understanding of these simulation parameters on various polishing platforms, it is useful to distinguish the difference between mechanical designs of polishers and the effects of consumable parts and judge the evaluation of the polishers in the optimization of copper process.

References

w1x J. Reid, S. Mayer, E. Broadbent, Solid State Technol. 43(7) (July 2000) 86.

w2x J. Kelly, C. Tian, A. West, J. Electrochem. Soc. 146(1999)

2540.

w3x T.C. Tsai, M. Naujok, L. Clevenger, H.C. Chen, P. Leung, K.C.

Lin et al., Proceedings of the 7th CMP-MIC Conference Proceedings, March, Santa Clara, USA, 2002, p. 19.

w4x M. Fayolle, F. Romagna, Microelectron. Eng. 37y38 (1997)

135.

w5x K. Sakai, T.K. Doy, M.M. Touzov, M. Satoh, T. Kasai,

Semiconductor Manufacturing Conference Proceedings, 1999 IEEE International Symposium on, Santa Clara, USA, 1999, p. 183.

w6x T.K. Doy, K. Sakai, H. Jeong, H. Ohmori, T. Kasai,

Semi-conductor Manufacturing Conference Proceedings, 1999 IEEE International Symposium on, Santa Clara, USA, 1999, p. 225.

w7x M.M. Touzov, T. Fujita, T.K. Doy, Semiconductor

Manufactur-ing Symposium, 2001 IEEE International, San Jose, USA, 2001, p. 337.

w8x B.W. Adams, B. Swedek, R. Bajaj, F. Redeker, M. Birang, G.

Amicoet, Semicond. Fabtech. 12(2000) 283.

w9x A. Jensen, Solid State Technol. 44(6) (June 2001) 101. w10x H. Treichel, R. Frausto, S. Srivatsan, B. Whithers, T. Meyer,

R. Morishige, J. Vac. Sci. Technol. A 17 (4) (July–August

數據

Fig. 2. Platen and wafer configuration for conventional rotary polisher.
Fig. 3. Belt and wafer configuration for linear polisher. (Wafer rotated
Fig. 4. Comparison of planarization efficiency between rotary and linear polishers. (Step height change measured on the 100=100 mm copper
Fig. 6. The various sweeping trajectories of polishing carrier in the rotary polishers

參考文獻

相關文件

好了既然 Z[x] 中的 ideal 不一定是 principle ideal 那麼我們就不能學 Proposition 7.2.11 的方法得到 Z[x] 中的 irreducible element 就是 prime element 了..

volume suppressed mass: (TeV) 2 /M P ∼ 10 −4 eV → mm range can be experimentally tested for any number of extra dimensions - Light U(1) gauge bosons: no derivative couplings. =>

For pedagogical purposes, let us start consideration from a simple one-dimensional (1D) system, where electrons are confined to a chain parallel to the x axis. As it is well known

The observed small neutrino masses strongly suggest the presence of super heavy Majorana neutrinos N. Out-of-thermal equilibrium processes may be easily realized around the

incapable to extract any quantities from QCD, nor to tackle the most interesting physics, namely, the spontaneously chiral symmetry breaking and the color confinement.. 

(1) Determine a hypersurface on which matching condition is given.. (2) Determine a

• Formation of massive primordial stars as origin of objects in the early universe. • Supernova explosions might be visible to the most

The difference resulted from the co- existence of two kinds of words in Buddhist scriptures a foreign words in which di- syllabic words are dominant, and most of them are the