Six Sigma: Systematic Approach In Probe Damage...

6
36 th International Electronic Manufacturing Technology Conference, 2014 Six Sigma: Systematic Approach In Probe Damage Reduction Glenn T. Placido, Carl Olalia, Rhal Alolod ON Semiconductor Philippines ON semiconductor, Golden Mile Business Park, SEZ, Governor’s Drive Bo. Maduya, Carmona, Cavite Philippines 4116 [email protected], +632-8594000 loc .2472 Abstract This paper describes how probe damage, the major defect at sort process, was toppled down nearly to ground level using the systematic approach of Six Sigma. A thorough study on the characteristic of the probe damage was made. Using root cause analysis tools, such as process mapping, cause and effect matrix, significance testing, and other useful quality tools, the sources of probe damage were identified , validated and solutions were developed . The PPM level of probe damage was significantly reduced resulting to less wafer scrappage, high yield and improvement on sort process efficiency. 1. Introduction Probe damage is the number one visual defect contributor at wafer sort process having a direct impact on yield and cycle time. This situation led the team to launch an initiative to reduce, if not eliminate probe damage. This initiative is aimed ultimately in improving wafer quality, increasing productivity and as part of ON Semiconductor’s “Road to Zero Defect” quality program. Table 1 describes the probe damage that was studied. Table 1: Probe damage N o Defect visual aid Remarks 1 Probe marks touching the seal ring of the bond pad 2 Expose oxide on the bonding pad 2. Section 2 Header Before proceeding to root cause analysis, measurement system analysis (MSA) was performed on outgoing optical inspection system to ensure that results of inspection are effective and consistent. A detailed process mapping was done to identify all possible root causes of the probe damage. This was supplemented by fishbone diagram (refer to Figure 1) and SIPOC diagram. Using the cause and effect matrix (refer to Figure 2), a total of 144 key process input variables (KPIV) were identified [1]. A prioritization based on impact of key process input variables to key process output variables (KPOV), 78/144 were short listed. Fifteen validation plans were developed after combining similar KPIVs (see Table 2). Figure 1: Fishbone diagram Figure 2: Cause and Effect Matrix

Transcript of Six Sigma: Systematic Approach In Probe Damage...

Page 1: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

Six Sigma: Systematic Approach In Probe Damage Reduction

Glenn T. Placido, Carl Olalia, Rhal Alolod

ON Semiconductor Philippines

ON semiconductor, Golden Mile Business Park, SEZ,

Governor’s Drive Bo. Maduya,

Carmona, Cavite Philippines 4116

[email protected], +632-8594000 loc .2472

Abstract

This paper describes how probe damage, the major

defect at sort process, was toppled down nearly to ground

level using the systematic approach of Six Sigma.

A thorough study on the characteristic of the probe

damage was made. Using root cause analysis tools, such as

process mapping, cause and effect matrix, significance

testing, and other useful quality tools, the sources of probe

damage were identified , validated and solutions were

developed .

The PPM level of probe damage was significantly

reduced resulting to less wafer scrappage, high yield and

improvement on sort process efficiency.

1. Introduction

Probe damage is the number one visual defect

contributor at wafer sort process having a direct impact on

yield and cycle time. This situation led the team to launch an

initiative to reduce, if not eliminate probe damage. This

initiative is aimed ultimately in improving wafer quality,

increasing productivity and as part of ON Semiconductor’s

“Road to Zero Defect” quality program.

Table 1 describes the probe damage that was studied.

Table 1: Probe damage

N

o

Defect visual aid Remarks

1

Probe marks

touching the seal

ring of the bond

pad

2

Expose oxide

on the bonding

pad

2. Section 2 Header

Before proceeding to root cause analysis, measurement

system analysis (MSA) was performed on outgoing optical

inspection system to ensure that results of inspection are

effective and consistent.

A detailed process mapping was done to identify all

possible root causes of the probe damage. This was

supplemented by fishbone diagram (refer to Figure 1) and

SIPOC diagram. Using the cause and effect matrix (refer to

Figure 2), a total of 144 key process input variables (KPIV)

were identified [1]. A prioritization based on impact of key

process input variables to key process output variables

(KPOV), 78/144 were short listed. Fifteen validation plans

were developed after combining similar KPIVs (see Table

2).

Figure 1: Fishbone diagram

Figure 2: Cause and Effect Matrix

Page 2: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

Table 2: Validation Plans

No. Hypothesis

1 Prober XY shift will increase if prober was

relocated

2 Using defective prober footings will cause

planarity problem and can induce probe

damage

3 Card relative Z height and overdrive does

affect probe mark size.

4 “Model not found" and "auto focus error"

will induce probe damage if not properly

corrected.

5 /6 Incorrect soak time will have effect on

probe mark size

7 Procedures is not yet robust or is not yet

define

8 Using out of spec screws in PC stiffener

(ELTESO PC) can induce probe damage.

9 Loading an wrong/not updated BOP disk

can cause smash the probe card and eventually

causes probe damage

10 Use of Sodium Hydroxide can cause

weakening of probe needle which can lead to

needle misalignment and eventually probe

damage.

11/12 Replacing the cleaning material incorrectly

and worn out cleaning materials will eventually

mis-align the needle during needle cleaning.

13 Accumulation of debris on the probe needle

can cause probe damage.

14 Leaning on the test head and removal of

cable on cable setup will induce probe damage.

15 Probe card with planarity issue can cause

probe damage.

3. Section 3 Header

The following are the results of validation experiments.

Corrective actions were defined for valid KPIV.

#1 Prober XY shift will increase if prober was

relocated

Result: During the first relocation, there was no

significant difference between XY measurements. However,

during the second relocation, X measurement showed

significant change. The same is true with the third

replication. Y measurement showed that there was no

significant difference during the three trials (see Figure 3).

Therefore the identified KPIV is VALID. Use of brackets

and quantitative measurement were the corrective actions

identified (refer to Table 3).

Figure 3: KPIV # 1 Hypothesis Testing

Table 3: Implement bracket and quantitative

measurement

# 2 Using defective prober footings will cause planarity

problem and can induce probe damage

Result: There is a significant difference between the XY

shift using good footings versus defective footings.

Therefore the identified KPIV is VALID. Figure 4 shows

the ANOVA for defective footings. As a corrective action,

checking of prober footings was implemented (Refer to

Table 4).

Figure 4: ANOVA for defective footings

Page 3: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

Table 4: Include footing check

# 3 Card relative Z height and overdrive does affect

probe mark size.

Result: Setting of card relative is critical as this will

dictate the needle cleaning height. As Z height increases

needle cleaning height also increases causing probe card to

smash and eventually probe damage will occur. Therefore

the identified KPIV is VALID. Table 5 lists the corrective

actions for this issue.

Table 5: Action implemented for KPIV # 3,4, 7

# 4 “Model not found" and "auto focus error" will

induce probe damage if not properly corrected.

Although “Model not found” error was not duplicated,

the team still consider this as valid since “Model not found”

error is due to probe card versus prober theta alignment and

corrective action to address this has to be defined. Therefore

the identified KPIV is VALID. Please refer to Table 5 for

the corrective action.

# 5 /# 6 Incorrect soak time will have an effect on probe

mark size

(Prober soak – 2hrs; PC soak- 10min)

Result: The results show that prober soak and PC soaking

are required as these affect the probe mark size and position

which can lead to probe damage. Probe mark size will

increase as PC soak time increases whereas probe mark

position shifts as prober soak time increases (see Figure 5).

Therefore the identified KPIV is VALID. The corrective

action was to implement automate soaking (See Table 5).

Figure 5: XY Shift

Table 5: Automate soaking

# 7 Procedure is not yet robust or is not yet defined

Result: There are no procedures defined on some of the

critical activities as listed on Table 6. As for the procedures

that exist, there is a need to enhance them to increase

detection of the problem. Each operator/technician has

his/her own procedure which may result to poor detection of

probe damage. Therefore the identified KPIV is VALID.

Process documentation for the identified processes were now

completed.

Table 6: Processes that needs documentation

No Processes without documents/ Procedure

needs enhancement

1 Prober relocation

2 PC/LB installation

3 Test head docking

4 Inspection methodology to detect theta

alignment problem

5 Setting z-ht and overdrive

6 Detect prober XY shift problem

7 Correct prober XY shift problem

8 Methodology after equipment repair

9 Probemark inspection criteria

10 Different criteria PCB vs prodn

11 Inspection of probecard needle prior use

– Has procedure

Page 4: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

As corrective action we have created procedure on all the

processes identified.

# 8 Using out of specifications screws in probe card

(PC) stiffener (ELTESO “Elevated temp sort” - PC) can

induce probe damage

Result: Special screws provide leeway for the board to

expand during high temp probing. On the other hand,

standard screws suppress the thermal expansion resulting to

probe damage (refer to Figure 6). Therefore the identified

KPIV is VALID. Table 7 list the corrective actions made.

Figure 6: Result of correct screw and not correct screw

Table 7: Use of specified/correct screws

#9 Loading a wrong/not updated BOP disk can smash

the probe card and eventually cause probe damage

Figure 7: Result of BOP disk simulation no damaged

Result: Incorrectly loading a BOP disk causes smashed

probe card. However, during validation, probe damage was

not induced. This however does not mean that this type of

probe damage will not happen, as previous data suggested

otherwise. Thus the KPIV is VALID. A more robust

system replaced the BOP disk (see Table 8).

Table 8: Remove BOP disk and replaced with auto

loading.

#10 Use of Sodium Hydroxide can cause weakening of

probe needle which can lead to needle misalignment and

eventually probe damage

Result: Sodium Hydroxide is the chemical used during

needle etching. Application of this chemical is subjective

and operator dependent while use of Duralco epoxy does not

effectively hold the needle of the probecard. This eventually

weakens the needle and will cause premature needle

misalignment (see Figure 8). The KPIV is therefore VALID.

The corrective action was to stop the use of sodium

hydroxide.

Figure 8: Needle misalignment

# 11 / # 12 Replacing the cleaning material incorrectly

and worn out cleaning materials will eventually mis-align

the needle during needle cleaning

Correct

screw NOT Correct

screw

Page 5: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

Table 9: Procedure on cleaning material

Table 10: Online needle cleaning

# 13 Accumulation of debris on the probe needle can

cause probe damage

Result shows that probe damage will occur if there is an

accumulated dirt/debris on the probe card needle. The size

and location of the probe damage coincide with that of the

debris on the probe card needle (see Figure 10). The KPIV

therefore is VALID. Tables 9 and 10 define the corrective

actions made.

Figure 10: Need tip dirt accumulation

# 14 Leaning on the test head and removal of cable on

cable setup will induce probe damage

Result: Result shows that after reseating the cable, probe

damage occurred (see Figure 11). Therefore the identified

KPIV is VALID. Table 11 shows the corrective actions.

Figure 11: Leaning on test head result

Table 11: Set-up on cable and hardware set-up

# 15 Probecard with planarity issue can cause probe

damage

Result: The result in the summary table (see Table 12) is

so convincing that even without the Chi-squared test one can

confidently state that non-planar probe card is conclusively

inducing probe damage. The KPIV therefore is VALID. As

a corrective action, ensure that probe card was planarized

using automated equipment that checks planarity before

running production.

Page 6: Six Sigma: Systematic Approach In Probe Damage Reductionewh.ieee.org/r10/malaysia/cpmt/IEMT2014/Technical... · 36th International Electronic Manufacturing Technology Conference,

36th

International Electronic Manufacturing Technology Conference, 2014

Table 12: Non planer PC result and Chi-square test

4. Conclusions

The team confirmed that the probe damage can be

reduced by thoroughly looking on all possible sources using

the right tools and prioritizing it using the cause effect

matrix. Below is the before and after performance of probe

damage.

Acknowledgments

The author would like to thank OSPI management for the

support that they provided over the course of this endeavor.

Peter Awayan for six sigma mentoring. The team for the

time spent during weekly meeting and technical expertise.

References

1. Pyzdek Thomas, The Six Sigma Handbook.