Chapter 10.1

Chemical-Mechanical Polisher

(cmp)

1.0        Title

Chemical-Mechanical Polisher

2.0        Purpose

The Strausbaugh CMP tool is a chemical-mechanical polisher primarily used for polishing polysilicon and silicon dioxide films on 4" and 6” wafers. The system processes wafers serially with manual loading and unloading of each wafer. The chemical-mechanical polishing process uses the combination of chemical and mechanical polishing to planarize wafer surfaces. Materials that can be polished right now are: various oxides, nitride, polysilicon, and polysilicon-germanium.  No metal is allowed in CMP as of now.

3.0        Scope

The tool consists of a polishing table, polishing overarm, load/unload   station, and pad conditioning device. The load/unload station centers the wafer and holds the wafer with vacuum before the polishing overarm picks up the wafer.  The load/unload station also has a sprayer for cleaning the wafer after polishing.

The polishing over arm holds the wafer in the wafer carrier, which includes a pad underneath the wafer.  The quill can spray air and water through the pad, or hold the wafer with vacuum.  Backpressure is applied during polishing through the holes in the wafer pad.

The polishing over arm can exert down force up to 15 psi, and backpressure up to 1 psi less than the down force (larger back-pressures might blow the wafer off the chuck during polishing, damaging both the machine and the wafer).

The polishing overarm is connected to the overarm spindle and the eccentric wheel, which controls the lateral travel of the arm across the pad during polishing.  The travel is set by adjusting a bolt attached to the overarm spindle – this adjustment should only be made by Microlab staff

The polishing pad is glued onto the polishing table.  The pad, (we are currently using a IC1000/SUBA IV composite pad) is made of a felt-like material adhered to a stiff back structure.   If the pad ever dries out, the residual slurry will stick in the pad fibers and destroy the pad.  It is critical that the wafer carrier pad and the polishing pad are never dry.

The conditioning arm is used to condition  (brush) the table during  or  in  preparation  for  wafer  polishing.   This arm automatically retires to the rinse bowl located behind the machine after polishing is completed.

The slurry consists of DI water, KOH, ~ 200 nm diameter silicon dioxide particles.   When the slurry dries it is difficult or impossible to remove, so it is important that all wafer surfaces stay wet until they are properly cleaned.

Wafers are cleaned in CMPWC, a dedicated machine for post-cmp clean.  Please refer to its manual (Chapter 10.2) for the operation procedure.  In the case CMPWC is down or unavailable, there is a blue box labeled CMP transfer. This box can be used to transfer wafers underwater to the Microlab for cleaning after polishing.  Cleaning can be done in spindryer 3, recipe 2, or sink 8 piranha and rinse.

4.0        Applicable Documents

Revision History

CMP Process Characterization Report 1999

5.0        Definitions & Process Terminology

N/A

6.0        Safety

N/A

7.0        Statistical/Process Notes

N/A

8.0        Available Process, Gases, Process Notes

N/A

9.0        Equipment Operation

9.1        Check the machine setup. The label on the machine indicates whether it is set up for either 4” or 6” operation. If change is desired, e-mail cmp@silicon2 and request a tooling change. Be sure to include 4” or 6” in your request. Cmp users will be notified when this change can be made.

9.2        Enable the CMP on the wand.

9.3        The machine should be in the Wet Idle process.   Press Stop Wet Idle, then press Exit Wet Idle to exit this mode.

9.4        Log out idle user and log in:

          Choose the Login from the menu.

          Press Log Off to log out the user idle.

          Login with your own username and password.

          Make sure the Chuck vacuum meter reads more than 500, otherwise report the fault and refrain from using the tool.

9.5        Prepare the system for polishing:

          Turn the valves on the DI water and slurry lines to the slurry configuration as shown on the diagram by the slurry tank.

9.6        Prime slurry lines:

          Load recipe: slurry.prime -- this recipe is used to prime the lines with slurry.

          Go to Auto mode -- download recipe.

          Make sure table is clear.

          Press start conditioning button: conditioning arm moves out and onto the table, slurry line begins to dispense. Slurry should start dripping onto the table after ~ 2 minutes. The conditioning ends after three minutes.

          No dummy wafer needed for slurry line prime as the polish overarm should stay over the table throughout the process.

9.7        Process your wafers:

          You are now ready to process. Go back to recipe  page and load oxide.std.00 (for standard thermal oxide, LTO, PSG polishing on 4”) or 6ox6.00 (for standard thermal oxide, LTO, PSG polishing on 6”).

          Go to Auto mode -- DO NOT SAVE YOUR RECIPE MODIFICATIONS -- download recipe.

Note:    If you need to change any parameter other than the time, you will need prior approval from the staff (see the Appendix).

          Put in your device wafers and press start polish.

9.8        Purge slurry lines:

          Turn the valves on the water and slurry  lines  back to  the  original  configuration  as  shown  on  the diagram by the slurry tank.

          Go to recipe page, and load rinse.purge recipe.

          Go to Auto mode, press start conditioning.

9.9        Clean off all surfaces -- dried slurry is difficult or impossible to clean!

9.10     Thoroughly rinse the load station, the spindle, and the table. MAKE SURE the WHOLE table is covered with water.

9.11     Log out.

9.12     Log in user: idle, password: (blank).

9.13     Go to wet idle page.

9.14     MAKE SURE TO CHECK Enable Quill  Flush,  Enable  Chuck Speed, Enable Table Rinse.

9.15     Press Start Wet Idle.

9.16     Disable CMP on the wand.

9.17    See Chapter 10.2 on how to use CMPWC in conjunction with CMP to clean your wafers.

10.0    Troubleshooting

10.1     I forgot to put a wafer on the chuck  before  starting  a recipe,  now it is just sitting there looking for a wafer, what do I do?

Press the Stop polishing button on the screen.   Go  to the  Manual  page  on  the  computer, and press the arm position button to move the polish  arm  back  above  the table.   Now  you  can open the front door of the chamber and place a wafer on the wafer  load  chuck,  then  press Start Polishing again.

10.2     There is a wafer on the load chuck, but the machine keeps turning  the  vacuum  on and off, as if it cannot find the wafer.

The vacuum is probably not strong  enough.   Wait  a  few seconds and the pump should start to pump down the vacuum chamber.

10.3     I forgot to run the wet idle and then left for the  weekend.  Is this a problem?

YES -- if the pad ever dries out it must  be  replaced. This costs ~ $300 for the pad, plus about 3 hours of technician time.  DO NOT FORGET TO RUN THE WET IDLE!

10.4  Helpful Suggestions (by Emmanuel Quévy, 7/04)

10.4.1          Local Non-uniformity

Many users complain about non-uniformity, which is really Within Die Non Uniformity (WIDNU), not within wafer non-uniformity.

Local non-uniformity occurs because of the uneven distribution of patterns to be planarized (up areas) and non-planarized patterns (down areas) within the die. This non-uniformity is reproduced everywhere on the wafer, and is really a matter of design. What I did to trick this was that I included dummy structures on my design to roughly get a uniform distribution of Up and Down areas (~ 50%). I got way better results than what I did with previous designs.

But it is still not enough, since I do not have a hard etch stop. So, at this point, other parameters to play with are the speed of the pad, and the down force applied. Basically, each design requires a little trimming of the recipe to minimize WIDNU. I can improve that by modifying the recipe a little.

10.4.2          Within Wafer Non-Uniformity

Down force is a very efficient parameter when within wafer uniformity is critical. Less down force, however, produces a lower removal rate but does provide better uniformity.

Example: My run with CMP went absolutely fine. No scratches whatsoever, and the wafers never went off the pad. I planarized 2 µm of SiGe with a removal rate of 0.2 µm/min roughly. Cross wafer and cross load Uniformity is really good (~1%) provided that you rotate your wafer. I did 6 times 2 min, each time rotated the wafer on the load station by 60 degrees.

10.4.3          Monitoring

For any monitoring purpose, I would advise to just run blanket removals on oxide and poly-Si, and measure uniformity to assess the machine is operating as expected.

11.0          Figures & Schematics

12.0          Appendix

Standard Recipes

oxide.std.00

Step

1

2

3

4

5

Time (sec)

15

5

5

60

15

Down Force (psi)

0

2

6

6

2

Table RPM

100

100

100

100

100

Chuck RPM

10

10

10

10

10

Back Pressure (psi)

-2

-2

-2

1

-2

Table Temp (ºC)

30

30

30

30

30

Slurry 1 (ml/min)

50

50

50

50

0

Rinse (on/off)

off

off

off

off

on

6ox6.00

Step

1

2

3

4

5

Time (sec)

0

15

5

60

4

Down Force (psi)

0

0

3

6

0

Table RPM

33

33

33

33

15

Chuck RPM

15

15

15

15

15

Back Pressure (psi)

-2

-2

-2

2

-2

Table Temp (ºC)

30

30

30

30

30

Slurry 1 (ml/min)

50

100

100

125

0

Rinse (on/off)

off

Off

off

off

on

Poly.polish

Step

1

2

3

4

Time (sec)

15

2

20

5

Down Force (psi)

0

5

8

0

Table RPM

24

24

24

24

Chuck RPM

6

6

6

6

Back Pressure (psi)

-2

-1

1

-2

Table Temp (ºC)

30

30

30

30

Slurry 1 (ml/min)

100

100

100

100

Rinse (on/off)

off

off

off

on

rinse.purge

Step