Picture and Location
Fiji1 and Fiji2 are physically contained in a single frame, although each tool is completely separate and autonomous from the other. They are located at the end of the first aisle on your left as you enter from the gowning room.
Atomic layer deposition (ALD) is a type of chemical vapor deposition that utilizes a cycling of different precursor gases to produce monolayers of a particular solid film. In conventional LPCVD as performed in the SNF furnaces the precursor gases are mixed continuously and are so reactive that the solid films are formed everywhere including occasionally in above the substrate surface. Conversely, in ALD first one precursor is introduced which will ideally completely and with self-limitation coat the substrate with a single molecular layer. The second precursor again in a self-limiting process reacts with the first molecular layer at the surface to produce a monolayer of the desire material. The key is finding a chemistry for the ALD precursors that can react at under the appropriate pressures and thermal energy the system can deliver, but will not desorb from the surface in the reactor conditions. This "energy window" for proper ALD growth is shown schematically in below (fromAtomic Layer Deposition for Nanotechnology by Arthur Sherman):
Because of the self-limiting nature of the sequential percursors, ALD systems can achieve near monolayer-at-a-time growth conditions with extreme uniformity over large areas. Additionally, the thin films have tremendously conformal coatings even under extreme topology. The films in general are dense and provide excellent interface preservation. However, the deposition rate is exceedingly slow (in general on the order of an Angstrom/minute) and this technique is not ideal for films thicker than a few tens of nanometers.
With the addition of plasma, the ALD window is no longer only dependent on temperature to supply the appropriate energy for a reaction to take place. Because of this many recipes are possible at temperatures not possible with thermal processing only. Additionally, some ALD processes can be achieved only by plasma assistance.
A more detailed introduction to ALD can be found in this presentation (given by Dr. J Provine 11/1/12 at Stanford University): ALD Introductory Tutorial 2012-11-01
A summary of more in-depth ALD processing information can be found in this presentation (given by Dr. J Provine 11/1/12 at Stanford University): ALD In Depth Tutorial 2012-11-01
The Fiji1 is classified as semi-clean. The temperature range of the chamber is determined largely by the materials to be used. The minimum temperature is set by the requirement that a positive thermal gradient exist from the precursor to the substrate (this prevents undesired condensation and potential valve clogging). Thus the precursor input lines/manifold must be hotter than the precursor and the chamber must be hotter than the manifold/input lines. The upper temperature limit is hard limited by the software at 350C for the chuck and 200C for the input line/manifold. In an effort to maintain cleanliness in the system the following materials are not allowed:
1. Polymers not specifically approved by the superuser.
2. Samples from gold-contaminated equipment
3. Wet samples4. Glass
5. Plastic, including Teflon
6. Non-encapsulated particles
7. Chips that are so small they will come off the substrate and fall into the pump line.
A Note on III-V Materials
How to make a Carrier Wafer for the Fiji
2. Pattern SiO2 with resist. The pattern should be a larger area than your piece as this pattern defines your pocket.
3. Dry etch SiO2. Note: wet etch does not work because you need to keep the backside oxide.
5a. Wet etch Si wafer with SiO2 as a mask in TMAH heated to about 90C for a few hours. This makes a recess with depth dependent on your initial mask because TMAH stops on the Si 111 planes. The chips can sit on the crystal smooth angled sidewalls of the trenches, which are very flat allowing good thermal conduction from the substrate, and hence the substrate heater, to your chip.
Performance of the Tool
What the Tool CAN do
- Deposit high quality metal oxides
- Deposition of oxide free metal nitrides and elemental metal films with low contamination
- Tightly controlled thickness through monolayer at a time growth.
- Thin film deposition with angstrom precision
- Ultra-conformal deposition thermal processing
- Depositions assisted by H2, N2, or O2 plasmas.
- Uniform coverage on aspect ratios of roughly 20:1 for plasma deposited films
What the Tool CANNOT do
- Deposit thick films. This is because of both the slow deposition rates and the expense of material precursors.
Films thicker than 50nm require prior approval from the quality circle.
- Selective ALD is a difficult issue and current research topic...it should not be expected that materials will deposit on only certain materials.
- ALD of certain films has not been demonstrated (Au for instance). If you want to try something not currently available, please consult the quality circle or the ever-expanding ALD literature.
The Fiji system consists of two identical chambers (Fiji1 and Fiji2), which are completely separate and autonomous. The differences between the systems are their cleanliness level (Fiji1 is semi-clean) and the presence of BOOST capability on Fiji2 for low vapor pressure precursors. THE OZONE GENERATOR FOR FIJI1 HAS BEEN REMOVED. They each have the following components (diagram courtesy of Cambridge Nanotech):
- A vacuum loadlock chamber
- A pump to maintain low pressures in the chamber and remove unreacted precursor
- A electronic control box (called the "e-box") that controls all of the heaters, valves, etc and communicates with the control computer via USB
- A control computer (a non-networked PC running Windows 7)
- H2, N2, and O2 gas lines for plasma
The films available on the Fiji1 is determined by the precursors installed in the system. In general, anything possible on the Savannah or the Fiji2 is possible on the Fiji1. The Fiji1 tool is currently maintained without a DI water source so primarily plasma assisted recipes are recommended. If you need to run a thermal film on the tool, please discuss with Michelle Rincon (email@example.com) or J Provine (firstname.lastname@example.org) to discuss. The list of available precursors is maintained on this page and also on a file on the desktop of control computer. The file on the computer is the final word on what precursors are installed on the system currently. In the event that this website and the file on the computer disagree - follow the document on the control computer.
Current films available (in stock precursors):
*Please note film deposition rates listed are specified to one significant digit for estimation purposes. For applications with tight thickness specifications it is recommended the user perform some process characterization to ensure precise depositions.
For the list of currently installed precursors consult the Fiji1 control program and badger for the most recent changes.
If the precursor or film you want is not shown here, please contact the quality circle. New materials and special runs can be possible as needed and with appropriate planning.
Contact List and How to Become a User
The following people make up the Tool Quality Circle:
- Process Staff: Michelle Rincon (mmrincon at snf dot stanford dot edu)
- Maintenance: Michelle Rincon (mmrincon at snf dot stanford dot edu)
- Super-Users: J Provine (jprovine at stanford dot edu)
Training to Become a Tool User
Fiji1 and Fiji2 have a the same training protocol. If you are qualified for one, you will be simultaneously be qualified for the other.
Becoming a Fiji1/2 operator is a three stage process.
- Shadowing. Contact a current user of the Fiji (either 1 or 2) and request that they introduce you to the system and demonstrate the use of the system for you. If you do not know a Fiji user, contact the user list (fiji1 at snf dot stanford dot edu or fiji2 at snf dot stanford dot edu) or check badger for upcoming user reservation. It is up to the user if they are willing to have you shadow them. If you have trouble finding someone to shadow, please feel free to contact the quality circle for help. (Helpful hint: Be familiar with the information here on the wiki before you shadow and be sure to ask lots of questions during the shadowing.)
- Written Quiz. After shadowing a user, contact Michelle Rincon or J Provine to take the short written quiz for the Fiji1/2. The quiz is closed book and all the information should be known to a someone who has studied the online documentation and has gone through a shadowing.
- Oral Qualification Exam. After the written quiz has been passed, you may schedule a final oral qualification exam with a superuser. At this exam you are expected to demonstrate your ability to run the tool safely and appropriately without any input. Once this final stage is satisfactorily passed, you will be given full access and utilization for the Fiji1 and Fiji2 as an operator.
- Provide a project proposal outlining the needs of the engineer status and the parameters under which you will use the system.
- Read all documentation relevant to the Fiji system.
- Pass a closed book exam about the concerns both safety- and process-wise related to changes in the Fiji process conditions and set-points.
Basic Operation Instructions
- Make reservation in badger.
- At least 24 hours prior to reservation time, make a comment in the maintenance section to request the specific precursors you would like to use for your run. The precursors are swapped frequently so do not assume that a precursor that is on the tool when you make your reservation will still be there when you are going to be running. If your reservation is for the weekend, please make the comment by the end of the day Thursdays.
- Enable the system on badger. Check the maintenance comment labeled "precursor status" to verify your precursor is available. Do not operate the system without being enabled on badger. Wafers found in the system without badger enabled will be removed at the discretion of the enabling user.
- Check that the system is in working condition: the loadlock is pumped down, all the heaters enabled and within range, the precursor lines have the materials you need, and the reaction chamber is pumping to below 100mTorr with a carrierflow set to 0 sccm for both the Ar lines. If any of these features are not found to be in range, report the situation on badger.
- Load your desired recipe: right click on the recipe in the control software and select from the Windows file list.
- Edit your recipe as needed: see the More on Recipes subsection for more information about writing and editing recipes.
Loading the substrate
- Transfer your wafers into the load lock (Be Careful It can be HOT depending on when last in the chamber). Please wear clean, new gloves when you are in the load lock. Note if there are any particles present and wipe them up with clean-tech wipes if needed.
- 2.1 raise the arm to almost max height
2.2 insert arm
2.3 should go in until 1cm or so of maximum arm depth
2.4 push in with some force until the arm is fully inserted (you will feel a click)
3.0 without turning, pull out the substrate holder at the same height. Check through the window in the loadlock that the sample carrier and your sample were retrieved.
- Close the load-lock lid and pump the chamber by pushing the "TRANSFER SAMPLE" button in the VACUUM section of the control software
- Introduce the sample by using the magnetic transfer arm. NEVER push forcefully in any direction on the transfer arm. Keep the arm elevated and push it all the way into the system.
- Lower the arm 1/8th of a turn below the neutral point and fully retract the arm, leaving the sample behind in the chamber.
- Click "OK" on the software to close the gate valve.
- Check to make sure the chamber reaches base pressure (this should only take a couple of minutes).
- When the system has reached base pressure (check that it is similar to what base pressure you found the system in - different by not more than a few mTorr), run your recipe by pushing the "START" button
- Check to make sure the pressure pulses of your recipe are in the correct range by monitoring the pressure readout in the control software. Typical pressure pulses are one hundred mTorr or more above the base pressure.
- If you have a plasma component to your recipe, make sure the plasma strikes and the reflected power quickly drops to within a few W of zero.
- When the process finishes, check that the base pressure has returned to the initial value. This checks that the system is not leaking and has returned to its original state.
- Push the "Transfer Sample" button in the VACUUM section of the control software.
- When instructed, insert the transfer arm at the neutral height all the way into the chamber to the stop point. NEVER force the arm or push with unnecessary force. You may need to "wiggle" the height a little to ensure the transfer arm correctly mated with the substrate holder.
- Lift the substrate holder up approximately a quarter turn, and gently remove the transfer arm fully. You should feel little to no friction or resistance to this action. Check through the window in the loadlock that the sample carrier and your sample were retrieved.
- Click "OK" on the control software.
- Vent the Loadlock by pushing the "Vent LL" button in the VACUUM section of the control software.
- Retrieve your samples from the substrate carrier using proper tweezers and gloves (it will be hot, be careful you do not melt anything onto the carrier).
- Shut the lid to the load lock and transfer the empty substrate holder back into the chamber following steps 2-6 of the Loading the substrate section.
- Run the standby process (00-STANDBY on each system). By having the substrate holder in the chamber and running the standby program we improve chamber cleanliness by reducing particles from temperature cycling and keeping a small background flow of Ar in the chamber. Also the standby program provides quick insurance that all process gases are off except for the carrier.
- If the next user asked you to adjust the chamber temperatures for them, please do so.
- Disable the system on badger.
If you need to stop your deposition at any time during the process, press the "STOP" button which will leave the heaters and the pump on but stop the recipe. It is not necessary to close the control program overall. If you do accidentally log off the computer or close the control program, open the program up again, run it, and load & run the STANDBY program.
A Note on Heating and Cooling
The recipes are all maintained in the folder C://Recipes/. Standard recipes listed in this document are kept in that folder.
The parameters you can control in a recipe are listed below with comments about each:
|Heater||Each heater has an unique item number and a value given in degrees Celsius. Do not overheat precursors (all precursors should be below 115C and some should not be heated at all - consult with quality circle if unsure). The reaction chamber should always be hotter than the manifold which should always be hotter than the precursor lines. The upper limits on temperature are protected by setpoints internal to the system.|
|Flow||This controls the flow of through a number of MFC. The plasma process gases are controlled by these MFCs as are the carrier gases. It is defined in sccm. It is not recommended to adjust the standard flow values in recipes. The MFCs for H2, N2, and O2 additionally have a hard off state and should be set to zero and then closed. This is already established in all recipes on the tool.|
|Pulse||This command is for pulsing a precursor line. It requires an ALD valve number and the amount of time you want the valve open in seconds. The fastest these valves can fire is roughly .015 seconds, so note that even if you define a shorter time that is likely the valve open time you will get. (NOTE: when writing a recipe the time in seconds needs a digit to the left of the decimal place; thus you should use "0.015" instead of ".015" for the minimum duration pulse.)|
|goto||Used to define loops in the recipes. This command takes as an input the step to which the recipe should return. The value for this command defines how many times the loop will run.|
|stabilize||This command is used to hold a recipe until a heater has reached the desired value. It takes as input a heater ID number and will wait until that heater demonstrates the set temperature with a degree C over a few seconds.|
|wait||This command takes as input a value in seconds that you would like the system to wait before proceeding to the next command. (NOTE: when writing a recipe the time in seconds needs a digit to the left of the decimal place; thus you should use "0.015" instead of ".015" for the minimum duration pulse.)|
|plasma||This command indicates the power that should be generated by the RF plasma system in Watts.|
|stopvalve||This command will close or open the output valve for the reaction chamber depending on a Boolean input. This command is currently not used in any of the standard recipes, but development is beginning for recipes using this feature.|
|line ac out||Users should not use this command. It changes the heater voltage on precursor heater wraps.|
Example 1: Thermal Alumina Recipe
The table on the left is how the recipe looks on the tool. The description on the right describes what each command does.
Example 2: Plasma Alumina Recipe
Tool Qualification Run
Qualification runs are the responsibility of the SUMO team. They will be completed with monthly frequency and also whenever additional data is requested to investigate reported or potential changes in the tool’s performance (i.e. when a part is replaced). The tool will be reserved for 2 hours on Badger, as the run normally takes just over an hour. A comment will be made requesting precursor Hf for the qual date.
Two freshly RCA cleaned 4" L test 100 Si wafers will be run in the plasma HfO2 recipe, which may need to be modified in order to utilize the Hf precursor for 100 cycles (always check pulse number). Wafers are to be placed on the left and right sides of the chamber with flats facing the center, as shown below.
A 9-point spectroscopic ellipsometry measurement will be made on the Woollam for each wafer. The analysis strategy will be “HfO2 w floating n,” located within the ALD folder.
The report will be made as a comment on Badger, as an addition to the tool’s Google Spreadsheet located in the SUMO folder, and as an update to the tool trend chart on the Fiji1/2/3 website. Both reports will include a) the recipe, b) the Woollam analysis strategy, and for each wafer: the thickness—c) average, d) min, and e) max— f) the index of refraction n at 633 nm, g) the standard deviation, and h) the non-uniformity (standard dev./avg. thickness). Machine status will be adjusted according to the description below.
Machine Status States
User reports and tools qualification runs are used to determine the appropriate machine status state on Badger. Acceptable ranges for the parameters listed above are as follows:
c-e) 100 - 140 A (not nm)
f) 1.9 - 2.32
g) 0 - 5 A
h) 0 - 5%
Red: The tool is not capable of depositing any films and needs maintenance or verification of proper working conditions.
Yellow: The tool is capable of depositing some films or films of non-standard quality. Reasons could include lack of necessary precursor, issues with the machine (pump, heaters, etc), or contamination that needs to be cleaned. The cause will be explored by the quality circle.
Green: The tool is in working condition for all available films, and those films were found to be within specifications.
Process Monitoring Results & General Data
Thermal HfO2 Deposition rate of 1.08A/cycle for standard recipe at 200C
Plasma HfO2 Deposition rate of 1.01A/cycle for standard recipe at 200C
The breakdown voltage was found to be .4V/nm for plasma Hafnia films (no anneal) of thickness ~10nm.Many thanks to Tim Holme and Cheng-Chieh Chao and the team at QuantumScape for the breakdown data.
Plasma TiN Deposition rate of 0.4A/cycle for standard recipe
Thermal ZrOxide data (7/12/12)
Thanks to Elizabeth Friend from Rambus for the data!