Protocols for deposition of metal on wafers for heaters, electrodes, etc.
CPA metal deposition -- recipes & tips
Ti (adhesion layer) 20 mtorr Ar 1 kW power 40cm/min track speed 1 pass ~ 25 nm (etch with 20:1:1 vol H2O/49% HF/30% H2O2)
Ni 10 mtorr Ar 2 kW power 15 cm/min 4 passes ~ 1 um (etch with Ni etchant in old lab)
Al/2%Si 3 kW 9 mtorr/140 sccm Ar 25 cm/min -> deposition rate: 175 nm/pass 20 cm/min -> deposition rate: 215 nm/pass
Al will often give cloudy films in CPA, especially for thicker films. A representative of CPA says that for > 1 um thick films, cloudiness is unavoidable. Here are ways you can minimize cloudiness:
1. pump down to less than 2.5e-7 torr
2. pre-sputter w/ Ti to react all the water in the chamber: 3 min. / 1.5 kW /whatever pressure you normally use for Al sputtering
3. pre-sputter Al for 5 min.
4. wait for 5 minutes between each pass. 10 minutes if you are patient.
5. For best results, turn off target during every waiting period. For the less patient, do every other pass.
6. Don't use more than 4 kW power. 3 kW recommended, although 4 kW won't hurt too much if you need a fast deposition rate.
7. Don't use less than 20 cm/min track speed.
1. Anneal your films after you pattern them! This will reduce the resistance substantially and improve the electrical stability of your contacts/heaters/wires over time. I use 250C overnight, but for at least an hour. Serious people probably use higher temperatures and in reducing environments (forming gas). The reduction in resistance with annealing may be due to reduction in the strain of the films and improvement in the crystallinity. Improvements in stability -- well, if anything like oxidation is going to take place, you might as well get it over with before your real experiments.
2. Make sure the Nitrogen knob (on the left side of CPA) is OFF, unless you're interested in AlN or TiN.
3. It's a real pain to try to keep track of your wafer inside the chamber for the purpose of speeding up/slowing down the chain speed. I timed a dummy pallet once, and so now i just use those timings instead of sitting around watching my pallets. That way, i can go off and do other stuff as long as i have my stopwatch with me. Even if I change the chain speed, i can just scale the timings. Here are my timings:
Direction speed time (m:ss) ---left--- 80 cm/min 1:52 ---left--- 20 cm/min 5:43 wait/cool 5:00 --right--- 20 cm/min 5:43 --right--- 80 cm/min 1:52 repeat ad nauseum
4. I use CPA to deposit films of Al for heaters AFTER I have bonded my glass wafers. This saves the agony of making your heaters if your wafers don't bond. It also beats Randex. Obviously, your films are going to be much rouger due to the imprinting from bonding. You should use the graphite blocks in the vacuum oven if you can, but it still works with the Macor. You should use ~1 um of Al to compensate for this roughess, and don't use linewidths less than 200 um. I piranha clean my wafers first, which means that there is some piranha in the channels after they come out of the sink. I use a syringe attached to tubing with an upchurch fitting and o-ring out of the end so that I can suck out the piranha. I then flush the channels with water several times to get rid of the piranha. Then I bake in VWR for 5 min. to dry it out. The other important point is that the Al doesn't get into the channels as long as the chip is sitting via side down flush on the pallet.
Fun Facts (for qualification, etc)
1. Why is there 2% Si in the Aluminum?
Device people often anneal their films/devices to get better contact between the Al and Si (see above note about annealing) Unfortunately, the phase diagram of Al and Si means that there is a eutectic which causes Al to diffuse into the Si and vice versa, resulting in shorting. In order to prevent this, the Al is saturated with 2% Si so that the Si can't diffuse into the Al, leaving no holes for the Al to diffuse into the Si.
2. Can I get pure Al in CPA? Does the 2% Si do anything bad?
Yup. Just request the pure Al target on the mailing list. It's a pretty rare occurrence. The 2% Si doesn't make any difference, except maybe a slight increase in resistance. For heaters, etc, it's fine.
3. Which pallet do I use if I want to deposit two metals, like Ti first, then Ni?
Always use the pallet of the LAST metal you will deposit. So for Ti/Ni, you would use the Ni pallet.
4. The power supply won't turn on.
You forgot to enable CPA.
5. When I turn the gas flow knob, the gas flow doesn't increase.
Check to see if the valve on the Ar cylinder is open. If it is, check to see if there's any Ar left. If there's not, report it on faults.
6. Can we get other metals like Pt in CPA?
We asked once upon a time, and the Microlab said: "Sure, if you buy the target. A Pt target runs $10,000." So that would be a no.
(Notes prepared by Emory Chan, 1/17/2006.)
Electroplating -- Ni & Au
Please refer to lab manual for sinkplate protocol.
The sinkplate is a dedicated sink utilized for electrochemical processes such as electroplating (Ni, Au). The sink is located in the old lab in 432D. All other items such as personal glassware, tweezers, or plating setup items should be stored in your personal storage areas.
Equipment Operation for Au plating General Preparation
1. Reservations Plating can be accomplished with the use of small or large beakers that can be obtained in the glass storage are in 432ºC or checked out from the Microlab office. Otherwise the plating sink contains everything necessary to carry out electroplating.
A light switch is located on the lower left of the sink and controls the lights within the sink. Please turn off the light when you are finished.
Au plating is performed at 55°C. Turn the plate to 100°C for ~30min to get the solution temperature up. Stir at 250 rpm to maintain solution homogeneity during plating.
3. Power Supplies and Multimeters
There are currently two power supplies available at the station. One supply (Small HP unit) is for lower current requirements (up to 120 mA), while the second supply (Soresnson model) can be used for processes that require high currents (up to 12 amps). Multimeters to monitor the current flow or voltage in electroplating setups are also available in this cabinet.
Au plating can be performed at ≤3 asf. Attached is an excel conversion chart. Insert the area to be plated in cm2. Change the current value so that current density is ≤3 asf. I plate a 0.45 cm2 area at 0.0014A (2.89 asf).
Next determine the time to plate based on the desired gold thickness. For the conditions above, 6 min of plating gives 1.18 um of gold. Similarly, 40 min gives 7.9 um of gold.
Cables with banana connectors and alligator clips can be found on the wall to the right of the sink. Please replace cables when you are through using them. Do not leave them dangling in the sink.
Below is a schematic of the plating setup.
Store gold plating solution in plating cabinet.
Please leave the sink as you would like to find it (dry and clean).
Remove glassware and other plating setup items once you are done.
(Notes prepared by Nick Toriello, 2/6/06)
Randex sputtering system
Please refer to the lab manual at http://microlab.berkeley.edu/labmanual/chap6/6.03.html for randex protocol.
The Randex system sputters metal and dielectric materials. We have used it for depositing, Ti, Pt, Al and SiOx. This is not a piece of equipment you can learn by reading a manual. You must observe a qualified user to learn the nuances.
The Randex system consists of three parts:
- The deposition chamber and control panel
- The vacuum system
- The ENI OEM 12A RF power supply
Inside the deposition chamber are three targets, a substrate table, and an etch station circulating system. The control panel selects the target and substrate position and controls the sputtering power. The vacuum system contains a mechanical pump, a cryopump, throttle valve, automatic valve controls and chamber. An ENI OEM 12A RF power supply provides the RF power needed to generate plasma in the deposition chamber.
Sputtering power should never exceed 1000 Watts. Reflected power should never exceed 100 Watts or 10% of the forward power.
Note: A large inventory of targets exists for the Randex. Should you require a target changed, please observe the following target change policy.
Target changes are made by staff and labmembers who have been trained for target changes. Labmembers qualified for the randex are not allowed to do target changes.
Current targets in the randex are listed on the WAND in equipment comments.
To request a target change, send email to firstname.lastname@example.org, stating what target(s) you would like and when you need them. Staff will add your requested targets to the target change queue in the order they were received and do whatever is possible to meet your schedule. Staff will enter the target change schedule into equipment comments and send a copy to labmembers qualified to use randex. The latest schedule can be accessed on the wand at any time.
Announced target changes will not take place if the labmember(s) who requested the target(s) has no reservations for the period which the target(s) is scheduled to be installed. Anyone wanting to purchase a new target should contact Bob Hamilton (bob@silicon).
Loading a Sample – Pumpdown
1. Users of the randex must supply their own aluminum or copper disc to cover the substrate table, or wrap the substrate with oil free aluminum foil available in the dispenser on the orange clean workstation. The disc should be 6 inches in diameter, and ~1/8 inch thick.
2. Push the VACUUM button to terminate pump cycle. This will close the high vacuum valve.
3. Vent the chamber by pushing the VENT button. (The system has a 10 second time delay relay to assure the gate valve has closed.) Vent until an N breeze can be felt around the baseplate. Turn off the 2 VENT.
4. Put on clean polyethylene gloves and raise the top hoist.
5. Change gloves. Load the substrate plate and sample.
6. Before starting pumpdown, confirm that the ion gauge is off.
7. Push the VACUUM button. The roughing valve should open and the chamber should pump down. When the pres- sure falls to ~500 microns, the roughing valve should close and the high vacuum valve should open. If the crossover does not occur, push the VACUUM button to close the roughing valve, and report a problem on FAULTS. Make sure the throttle valve is not engaged during pump down. After the hi-vac valve opens, turn on ion gauge filament.
8. When the system achieves a pressure in the mid 10-7 range, turn on the cooling water (both knobs to the left). It usually takes 5-6 hrs for the chamber to pump down. You usually pump down overnight or during the day. Don’t wait for the chamber to pump down. Make a reservation for ~12 hrs/run.
9. Turn off the ion gauge filament, close the throttle valve, and set the flow rates of the gases you'll be using to the desired values. The mass flow controller is mounted on the lower right of the Randex. Push toggle up to POWER to turn on. Push toggle down to SET to set desired flow. Set to READ to read flow. Position 1 is Ar, position 2 is O : Ar [1:1], and position 3 is N.
For metals run MFC at 100sccm
Table movement and target selection.
The table with the wafers on it move, while the targets are stationary. The target location is listed on the front of the machine (3 targets at a time).
Below is a schematic of the targets locations and the table locations.
Always presputter (clean off the target with no wafer underneath) before sputtering on a wafer. For example, to presputter on from target 1, place the table in the “Target 3” position. To presputter from target 2, place the table in the “Etch” position. Which position do you use to presputter from target 3?
Operating Instructions for ENI OEM 12A RF Generator Operation
1. Increase the power setting on the ENI generator to 100 watts reflected. No forward power with flow.
2. Push IGNITE button to ignite the plasma on your target - this is essential for proper tuning. It is easier to ignite the plasma at higher pressures, i.e. at 10 - 20 mTorr, and then reduce the pressure for processing.
3. Tune target load for maximum forward power and minimum reflected power using load and tune knobs.
4. Adjust output level for desired power. Read watts on forward power meter. Reflected power should not exceed 10% of forward power.
Do not apply more than 300 watts of power to a dielectric target because the poor thermal conductivity of dielectrics heat is not effectively transferred to the coolant and the target may crack.
1. To sputter clean a target, place the system in the following position:
Introduce about 50 Watts of RF power using POWER ADJUST. Reflected power should never exceed 100 Watts. If plasma is not formed, adjust tune and load so that reflected power is at a minimum. Ignite by pushing the button on the power adjust knob.
2. After plasma is clearly visible, retune the system to minimize the reflected power. The TUNE knob has a greater effect on reflected power than the LOAD knob.
Slowly adjust the TUNE knob first, and then use the LOAD knob to minimize reflected power. Adjust the power to the desired forward rate and sputter clean for the desired time.
3. After Sputter clean, turn the table position to your target and begin deposition. Check frequently and minimize the reflected power for the first 20 minutes.
4. After desired sputtering time, turn the power to adjust to zero.
5. Cool down the system (~5 to 10 minutes depending on sputtering power used); turn power off on the Henry power supply. Turn off gas toggle valves. The mfc controller should be left on. DO NOT SHUT OFF THE GASES AT THE TANKS.
6. Place the throttling valve to the open position, and close the hi-vac valve. Wait 10 seconds. Press the vent button until the chamber is at atmosphere, unloading sample and substrate plate with fresh gloves.
7. Turn off the cooling water.
8. Place the radex in the standby mode by:
- Pump down system. Turn off AC breaker to RF supply.
- Check HV valve (ON), Roughing valve (OFF), Gas valves (OFF), Throttling valve (open), and pressure in chamber is less than 5 x 10-6. You may leave the ion gauge power and filament on.
(notes prepared by Nick Toriello, 2/6/06)
- Eric Chu 17:15, 15 July 2009 (PDT):
or instead, discuss this protocol.