掩膜版（mask）制造技术
Photomask Making
Roger Robbins 3/6/2007
The University of Texas at Dallas
Erik Jonsson School of Engineering
Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Photomask Making
Roger Robbins 3/6/2007
Table of Contents
Photomask Making (2)
Table of Contents (2)
Photomask Making (3)
Purpose (3)
Introduction (3)
Mask “Color” (3)
Process Description (4)
Step 1: Plasma Cleaning (4)
Step 2: HMDS Application (5)
Step 3: Resist Application (6)
Step 5: Lithographic Patterning (8)
Step 6: Post Bake (10)
Step 7: Develop (10)
Step 8: Plasma De-scum (11)
Step 9: Cr Etch (12)
Step 10: Resist Removal and Mask Clean (12)
Step 11: Final Inspection (14)
Conclusion (14)
Appendix A (15)
Appendix B (18)
Title: Photomask Making Page 2 of 24 Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Photomask Making
Roger Robbins 3/6/2007
Purpose
This paper describes a starting process for making photomasks with the tools in the UTD Cleanroom. Special needs may dictate variations in this process flow.
Introduction
Photomasks are generally Chrome coated glass lithographic templates designed to optically transfer patterns to wafers or other substrates in order to fabricate planar type devices of all types. Basically the pattern information is created in a drawing package and stored in a database, reformatted and transferred to a lithography tool –laser writer or e-beam writer in our case – then printed in a layer of photoresist coated onto the photomask plate. The imaged pattern is next developed to form a template over the opaque Chrome and then the Chrome is etched away where the resist is clear. After the etch process is complete, the remaining photoresist is removed, the plate cleaned, and then stored for later use in an optical printer.
This paper will describe in detail all the fundamental process steps required to fabricate a photomask in the UTD Cleanroom Labs.
Figure 1. Example of standard photomasks: “Clear field” on left, “dark field” on right
Mask “Color”
First, there are some key details to discuss that determine what “color”photomask you will need to make, (Figure 1). Normally, we use “positive” photoresist to make masks. This means that wherever light exposes the photoresist, the developer will
Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
which light will pass to expose the substrate during the pattern transfer process – (Dark Field mask color).
Following that logic, if the substrate also has positive photoresist, the mask will allow exposure to the substrate in the same pattern as the clear regions on the mask. The development of the substrate resist will produce exposed areas in the same location as the clear areas on the mask. This will allow an etch process to transfer the clear area pattern to an underlying film by etching away the underlying film in the clear area.
This may seem simple to the casual observer, but there is a complication when you consider the common “lift-off” process which will change the “color” of the pattern on the substrate. In the lift-off proc
ess, the patterned substrate is first developed and then a film is deposited on top of the patterned resist. The next step strips the resist under the deposited film. This will leave a pattern of deposited film in the areas where the positive photoresist was exposed and leave bare the areas under the unexposed resist that were washed away, taking the deposited film away with the wash.
Then to further complicate a complicated logic, there is negative photoresist which stays put when exposed and washes away in unexposed areas. This negative process does the opposite thing to the above positive resist. The resists can be used either on the mask or the substrate in any order. Thus you must keep exact logical understanding and control of your process in order to make what you want.
Process Description
The following process description will assume that we start with a bare, blank photomask and describe all the steps to fabricate a complete mask. There may be sections that are optional or unnecessary due to the particular situation you have, such as a pre-coated mask that does not need to be coated with photoresist.
Step 1: Plasma Cleaning
Both new and recycled bare photomasks may have a thin invisible layer of organic contamination on the Cr surface. This organic layer will sometimes cause adhesion problems between the Cr and the photoresist later in the process. It also may interfere with the Cr etch process after photomask imaging. However it is easy to remove the contamination with a short, but aggressive Oxygen plasma treatment before starting the photomask process. The current process designed and tested for this purpose is shown in Table 1. It uses the March Asher1 tool, (Figure 2), but fortunately it can be used in any other month of the year as well. ?
Table I
March Asher Cleaning*
Step Parameter Value
1 Vacuum Pressure 230 mTorr
2 Gas O2
3 Gas Flow 31 sccm
4 Time 600 sec
*Set the photomask Cr side up on the powered
electrode plate for RIE conditions.
Figure 2. March Plasma Asher tool: RF power supply at left, Control section in center,
and access door to RF plasma chamber on right.
Step 2: HMDS Application
In order to enhance the adhesion of the photoresist to the Cr, we bake the mask in a special oven that applies a molecular monolayer of Hexamethyldisilizane (HMDS), to the surface2. This molecule chemically bonds to the Cr and then bonds to the photoresist so that the photoresist will not allow developers or etch solutions to lift the edges of resist patterns away from the mask. This step is pretty much mandatory for coating a mask prior to exposure.
This is a simple process. Basically you open the oven door, insert your mask (riding on a quartz boat), close the door and push the start button. During the oven process, the mask is brought to temperature (120 C), and pressure is automatically cycled between 10 mTorr and 1200 mTorr several times to drive off moisture from the surface of the substrate. The oven fills with HMDS vapor for 5 minutes. This is th
e application step in which HMDS chemically attaches one end of its molecule to the substrate. The atmosphere in the oven is then cleared of vapor by introducing N2 into the oven and pumping it out again over several cycles. After about 27 minutes, the oven
special ? in thick Cr plated Stainless steel table normally used to level SU8 photoresist. It will take about 3 – 5 minutes to cool to room temperature.
2 Daggett, Joe, Villareal, Sam, and Robbins, Roger, “UT Dallas IC Fabrication Laboratory HMDS Process
Setup,”www.doczj/doc/feb9526d3369a45177232f60ddccda38366be17c.html
/research/cleanroom/documents/HMDS_Process.pdf,
(2/21/2003).
Title: Photomask Making Page 5 of 24 Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Figure 3. HMDS Oven showing logbook, mask, quartz holder with extractable handle and open door. T
he start button is the black button in the upper center of the control box. The red button is the reset and buzzer silencer.
Step 3: Resist Application
After the blank photomask is prepared, the next step is to apply the photoresist. This is done on the CEE spinner3 manually, (Figure 4). We use two photoresists for photomasks in our lab: S1813, and AZ1518. (See Appendix A and B for resist spec sheets). Normally, the AZ1518 is pre-applied by the photomask vendor, and we don’t have to worry about applying that resist. But if we are coating a blank one, we normally use the S1813 or a new, faster resist AZ TFP650.
3 Robbins, Roger “CEE Spin Coater/Hotplate Operation,”
Figure 4. Manual application of S1813 photoresist to a blank
photomask in the CEE spinner.
You must be trained on this tool before using. The design of the CEE spinner requires that you learn how to program the spin cycle. This is described in detail in reference #3, but verbal instruction with demonstration is required. The parameters of a nominal coating program are listed in Table 2. These p
arameters are regularly changed by users, so you must first check the installed values and put in your own values before coating.
Table 2
CEE Spinner Parameters for Mask Photoresist Coating
Step Parameter Value
1 Dispense 0
2 Spin Speed #1 500 rpm
3 Acceleration 500 rpm/sec
4 Spin Time Duration #1 2 sec
5 Spin Speed #2 4000 rpm
6 Acceleration 4000 rpm/sec
7 Spin Time Duration #2 60 sec
lid and starting the CEE spin cycle. For a 5 inch square photomask, the amount of photoresist is about 2 ml. The first spin cycle quickly spreads the resist without throwing it off the mask, and the second spin cycle stretches the resist film into a highly uniform thin film at the desired thickness based on a spin-speed curve from the manufacturer. For a photomask, you would typically want about 6,000 –10,000?of thickness.
Title: Photomask Making Page 7 of 24 Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Step 4: Pre-Bake
After the resist is coated it still contains a considerable amount of solvent that needs to be driven out via a moderately high temperature bake. This is usually done on the integral CEE spinner hotplate shown in Figure 5. This step is called “pre-bake”because it happens before exposure.
Figure 5. CEE spinner hotplate with mask baking under exhaust lid.
The CEE spinner has a special programmable hotplate with an integral hinged lid
with fume exhaust capabilities. This lid captures the solvent vapors emanating from the
hot photoresist, and ports them away from your nose so you w on’t come down with
central nervous system problems. Table 3 lists the baking parameters for S1813.
Temperature and time are important here and vary with the photoresist type.
Table 3
Bake Parameters for S1813 Photoresist
Step Parameter Value
1 Temperature 115 C
2 Bake Time 90 sec
Step 5: Lithographic Patterning
Lithography is a complex process and will not be described here in detail. In the UTD clean room there is one method for making photomasks – the Heidelberg
Title: Photomask Making Page 8 of 24 Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Instruments “D WL-66” laser mask writer. The principle steps involved in making a mask are listed below.
Create a device design.
Lay out the mask geometries using a drawing package such as AutoCad.
Convert the drawing data into a form familiar to the DWL-66 computer
(.cif), using a special conversion software and special separate computer.
Note: You should write a label on each mask you make to avoid using the
wrong mask during your device fabrication.
Send the converted data to the DWL-66 computer.
Reserve time on the DWL-66 laser writer
Sign the Logbook
Load the resist coated photomask blank onto the laser writer stage.
Set up the layout on the DWL-66 computer. Setup the job on the
DWL-66 computer.
Focus the DWL-66.
Find the center of the photomask blank.
Start the Job.
Wait a long, long time before the mask is complete (many hours).
Remove the mask from the DWL-66.
Basically, the exposure step modifies the chemistry of the photoresist so that the exposed portion either dissolves in the developer and washes away - (Positive resist), or solidifies and remains on the mask while all the unexposed resist washes away in the developer - (Negative resist). The two photo r
esists listed in this document are positive resists which wash away after exposure.
There are a lot of pitfalls involved with the stream of actions listed above – you must follow them exactly – deviations may cause computer confusion crashes. The lithography step will require extensive training and practice along with recurrent help –ask if you are uncertain. Use the available instruction sheets for detailed guidance so you won’t forget anything. Let us assume here that the lithography exposure goes well. Figure 6 shows the illustrious DWL-66 mask writer.
Figure 6. DWL-66 Laser Mask Writer
Title: Photomask Making Page 9 of 24 Author: Roger Robbins 3/6/2007 Document Number: SP-07-001 The University of Texas at Dallas
Step 6: Post Bake
Some resists require a short bake after exposure to “activate” the exposure. The two photomask resists, S1813 and AZ1518 do not, so we just skip right by this step in this document.
xposedStep 7: Develop
After exposure, the resist needs development to show the pattern. Each photoresist has its own developer and we have automatic programs set up in the CPK spin developer tools to develop your mask or wafer, (Figure 7). For optical resists, the developer chemical is basically a solution of Tetra Methyl Ammonium Hydroxide (TMAH). The rinse cycle utilizes de-ionized water.
Figure 7. Photo of the CPK Spin Develop/Etch tool. The process fluids are forced
through the spray nozzles via pressurized source tanks (black tanks with green bolt
handles in lower drawers). Note the “POLOS” controller panel at the top left of the overhead panel – this is where the tool is programmed and where process progress is displayed. Also note that if a light goes on above one of the switches at the top right, the