May 2025 CMU Update
This page contains information about the state of the ALD project as of May 2025. This is the most recent update for the project.
Last updated
This page contains information about the state of the ALD project as of May 2025. This is the most recent update for the project.
Last updated
The chamber itself is designed around a modular vacuum chamber cube from Ideal Vacuum. Internally, the chamber contains a heater for the wafers being coated, and is designed to be compatible with our precursors (TMIn, TDMASn). Gas I/O is done through a ¼” Swagelok fitting (input) and a KF25 fitting (output). The chamber is equipped with a pressure gauge and an electrical feedthrough for the substrate heater. It is also designed to allow a QCM system created specifically for ALD.
Specifications
Specification of Interest
Target Value
Max. Wafer Diameter
4 in.
Max. Substrate Heater Temperature
450 °C
Base Chamber Pressure
10 mTorr
Process Chamber Pressure
200 mTorr
Substrate Heater Accuracy
± 2.2 °C
Substrate Heater Precision
± 1.0 °C
The base of the chamber is an Ideal Vacuum 9x9x9 modular chamber cube. The face breakdown is as follows:
Side(s)
Face Type
Back, Right
Blank
Left
4x KF16 Flange
Top, Bottom
1x KF25 Flange
As purchased, the faces are sealed using Viton O-rings, however Viton is incompatible with the precursor gases for ITO deposition. As such, it is crucial to replace the Viton with a sealing material that will be compatible with the process gases. Aflas was chosen for its relatively low cost (for materials in its class) and as it is one of the more available of the materials that would be suitable.
Our O-Rings are supplied by AllORings.com, but any supplier would be acceptable provided that the sizing is correct. The specific O-Ring dimensions are as follows:
Purpose
Qty.
I.D.
Thickness
AS568 Size
KF25 Centering Rings
5
1.100”
0.210”
*-320
KF16 Centering Rings
4
0.725”
0.210”
*-314
Chamber Faces
6
8.484”
0.139”
*-268
Throttle Valve
1
0.609”
0.139”
*-208
Below is a complete list of all hookups used in our ALD chamber:
Feedthrough
Face
KF25 to ¼” Swagelok
Top
KF16 Power and K-Type
Right
KF16 Pressure Gauge
Right
KF16 Blank (x2)
Right
KF25 90° Elbow
Bottom
The KF25 elbow (gas output) leads to the chamber via the following assembly:
All fittings are KF25. The assembly, top to bottom is as follows:
KF25 90° Elbow
KF25 Manual Butterfly Valve
KF25 Bellows
Dry Scroll Vacuum Pump
There are KF25 centering rings at each junction, and aside from the connection to the chamber itself, all junctions use KF25 hinge clamps. Though we have not gone far into the process development phase, it is worth noting that we found the butterfly valve to be unnecessary for regulating pressure, given our pump and mass flow controller (MFC). Instead, our process pressure is reached directly using the MFC while constantly running the pump with the throttle valve in the completely open position.
The heater mount is quite simple, consisting of a bent sheet of aluminum and ball bearings for alignment. The purpose of the aluminum is relatively self explanatory, it provides a sturdy mounting point to connect the heater module to the chamber body. The purpose of the ball bearings requires a bit of explanation.
The primary purpose of the heating module is to heat only the chips, and not the rest of the chamber, which is rated for a maximum operating temperature of 150°C. Because of this, we require a method of keeping the heater thermally insulated from the walls of the chamber while also maintaining rigid mounting. The mount uses concepts of minimum constraint design to allow the heater to mount in a repeatable and accurate way, while also ignoring the need for any screws or other mounting hardware.
The lower structural plate of the heater module has 3 slots, each of which provides 2 points of contact to the heater. 6 points of contact constrains all 6 degrees of freedom and keeps the heater in place. A diagram is below.
Lower structural plates with slots
Mounting bracket with ball bearings
The ball bearings themselves are made of silicon nitride. This material was chosen for its low thermal conductivity (~10 W/mK) and compatibility with our precursors, as well as the ability to source it from McMaster-Carr.
The heating module consists of 3 major components: structural plates, machined from sheet aluminum, electrical insulation plates made from Aluminum Nitride, and a heating element made from nichrome wire.
The upper and lower plates of the heater module are mainly for structural support and heat transfer. The upper plate provides a strong and flat resting place for chips during coating and the lower plate reinforces the module and allows it to interface with the mount. The current design revision has two 4.5 inch diameter by 0.1 inch thickness plates with the upper plate having countersunk holes for the mounting bolts (to allow the wafer to sit flat on the surface) and the lower plate having alignment features for the mount.
Upper Structural Plate
Lower Structural Plate
The heating element is a single line of 20 AWG nichrome wire, laid out in a serpentine pattern, as above. The length needed is approximately 20 inches. Nichrome wire was chosen as an easy resistive heating material which is also compatible with our precursors.
Since the heating element is nichrome, mounting the wire directly to the aluminum structural plate would cause a short circuit. Traditional mounting methods use standoffs and rely primarily on radiative and convective heating, however since the element will be working under vacuum, we are unable to rely on convective heating. For this reason, we desire a secondary plate which is electrically insulating but thermally conductive.
Our initial efforts used Boron Nitride plates, though these were found to be unsuitable due to relatively low strength and high coefficient of thermal expansion causing fractures as the wire is heated. Aluminum Nitride (AlN) is used as an alternative material in our case. A comparison of properties is below:
Aluminum (Reference)
Value
Thermal Conductivity
237 W/mK
Boron Nitride
Value
Thermal Conductivity
30 W/mK
Aluminum Nitride
Value
Thermal Conductivity
120 W/mK
In order to understand the thermal and electrical characteristics of the substrate heater, a combination of experimental and computational/theoretical techniques was employed. As previously mentioned, the initial design used Boron Nitride (BN) as the insulating agent. Upon the first experimental testing of the device, the BN disks cracked. This motivated work to understand why this cracking occurred.
Prior to getting into results, it is worth discussing the failure modes that we are concerned with. The hypothesized mode of failure for the BN disks was thermal expansion based stress formation. These stresses can form in one of two ways, “steady-state” stress formation, and “transient” stress formation.
Steady state stress formation occurs due to differences in CTEs between components, as uniform expansion across the entire part causes stress formation when one material “desires” to expand more than it is allowed to expand.
The radial direction is simpler as it is unconstrained, so we can begin with it. The critical dimensions are the distance between the screws and the diameter of the ceramic plate. In the AlN system, so long as the distance between the screws remains larger than the diameter of the disk, there will not be stress concentration, as the disk is simply held captive by its position between four screws, rather than being rigidly coupled to them.
The screws are effectively moved by the aluminum structural plates, so we can assume that the relevant quantities here are the CTE of the AlN and Al. The CTE of Al (23.5 × 10-6K-1) is greater than the CTE of AlN (5.6 × 10−6K−1). This implies that we will not have stress buildup in the radial direction for the AlN system.
Unlike the new heater, the Boron Nitride disks were rigidly coupled to the screws via mounting holes. This indicates that stress concentrations can be tensile or compressive, again depending on the relevant CTEs. Again, the CTE of Al (23.5 × 10−6 K−1) is greater than the CTE of BN (6.0 × 10−6 K−1), so the rigid coupling implies tensile stress formation.
The strain on the BN disk is effectively dictated by the Al disk because of the larger magnitude CTE, and the rigid coupling implies that the strain on the Al will be equal to the strain on the BN. Therefore:
Thus, for some change in temperature T, the stress induced in the BN will be:
Which gives a critical temperature deviation of:
We can then substitute the actual values:
Al
23.5 × 10-6 K-1
BN
6.0 × 10-6 K-1
EBN
19.5 × 109 Pa
max,BN
83.3 × 106 Pa
And calculate a critical temperature deviation of:
We can assume that the actual expansion of each component will be based off of its CTE, with all geometry scaling in length by an “effective temperature”, and the remaining “desired” expansion going into stress formation. This gives:
Solving for Delta T_eff as a function of T gives:
We can then use the effective temperature to determine the stress formation in the Aluminum Nitride:
Which can then be solved for the critical temperature deviation Delta T_crit:
Substituting in the actual material properties:
Al
23.5 × 10-6 K-1
AlN
5.6 × 10-6 K-1
EAlN
348 × 109 Pa
max,AlN
1.97 × 109 Pa
wire
14.0 × 10-6 K-1
screw
18.4 × 10-6 K-1
The steady state results explain not only why the BN exhibited cracking (tensile stress formation in the radial direction), but also give a good upper bound for the new, AlN disks.
Transient Stress Formation
Unlike steady state stress formation, which results from uniform expansion across the entire part, transient stress formation results from non-uniform temperature distributions that arise during heating. Simulations in ANSYS were used to determine if this was a major contributing factor to the BN cracking, simulating the heater using the Transient Thermal Analysis and Transient Structural Analysis modes in ANSYS Mechanical. It was found that there was minimal contribution of this mode to the cracking behavior of the disks.
In order to accommodate a Quartz Crystal Microbalance (QCM) for in-situ thickness measurements, and to reduce the cost to build, a second version of the substrate heater has been designed and is currently in the fabrication process. The only modifications are to the geometry of the structural and insulation plates, controls, heating element geometry/parameters, and the heater mount are all identical. The second iteration of both plates are shown below:
Upper Structural Plate
Lower Structural Plate
The primary modifications made are:
Change of the base shape from circular to rectangular. This accommodates the new AlN plates, which are square rather than circular.
Introduction of a large notch on one side of the heater. This is the mounting location for the QCM. A custom bracket would need to be designed and machined to thermally couple the QCM to the heater.
Longer slot for thermocouple on the lower plate
Registration marks to show users where chips should be placed on the surface.
To reduce costs, the 4” ⌀ x 0.1” thick circular plates which were used in version 1 are replaced with 114 mm x 114 mm x 1 mm sheets. The original disks were sourced from Heeger materials for $220 each, while the new sheets were sourced from Amazon for $81 each. This yields an effective reduction in cost of $278 with the new design. While the heater module is still by no means cheap, reducing system costs has obvious advantages for reproducibility.
Our substrate heater is run using a 10A, constant current power supply which is switched using a relay. When the circuit is closed, the system draws approximately 100W. Below is a plot of temperature vs time when running at 100W power input, noting time to reach temperature for various times of interest.
The maximum heating rate that can be achieved using these input parameters is 19.2 °C/min, with a maximum temperature of approximately 480 °C due to radiative losses. It is worth noting that using the current chamber design, it is unlikely to be safe to run the heater above 300 °C for extended periods of time. The low thermal conductivity of the viewing window causes a large temperature increase which quickly approaches the maximum operating temperature of the chamber. During characterization tests, the window temperature was measured externally at 90 °C after only a few minutes as 480 °C, suggesting not only a higher internal temperature but that we would expect failure of the seals under prolonged operation. Various control schemes were tried to determine the best achievable precision of the device. Though rigorous tuning was not done, results of the most promising parameters are presented below.
Control Scheme
Parameters
Precision
Bang-Bang
N/A
-0.32 °C/ +0.48 °C
Proportional Control
Kp = 250 °C-1 (maps from 0-255)
-0.75 °C/ +0.95 °C
More testing would be needed to ensure long term stability (multiple hours) of both control methods, but I would recommend the use of bang-bang control, given its low deviation from the target temperature.
In order to keep the precursor delivery system and the vacuum chamber positioned the same way relative to one another at all times, a simple aluminum extrusion frame is being built which is sized to fit both systems and allow for easy alignment. The primary considerations for this design are:
Keep the precursor delivery cabinet and the chamber rigidly mounted on some structure
Lift the chamber from the ground to allow the gas outlet line to pass
Construct the stand from proper materials
Aluminum T-Slot extrusion is a very common method to construct such frames, being modular and easy to source. We chose 1010 imperial extrusion (dimensions 1” x 1”) to keep consistency with the mounting hardware built into the vacuum chamber, though a similar design could easily be made from other sizes of extrusion. All corners are connected using corner blocks or brackets (external or internal corners respectively) and bolted to the extrusion using either tapped holes in either end of each section or T-Nuts.
Below is a list of relevant parts for each subsystem up to this point. Since this project is still a work in progress, it is somewhat subject to change.
Item
Qty.
Supplier
Part No.
Vacuum Chamber
Blank Flange, KF16
3
Ideal Vacuum
Ideal Vacuum Chamber Door Hinge
1
Ideal Vacuum
Ideal Vacuum Cube, 9x9 Vacuum Chamber Plate, 1x KF25
2
Ideal Vacuum
Ideal Vacuum Cube 9x9 Vacuum Chamber Plate, Blank
2
Ideal Vacuum
Ideal Vacuum Cube, 9x9 Vacuum Chamber Plate, 4x KF16
1
Ideal Vacuum
Ideal Vacuum Cube 9x9 Vacuum Chamber Plate, 3/8" Tempered Glass
1
Ideal Vacuum
Ideal Vacuum Cube Hardware Kit
1
Ideal Vacuum
Ideal Vacuum Cube 9x9x9 Aluminum Vacuum Chamber Frame
1
Ideal Vacuum
Bulkhead Clamp, KF16
4
Ideal Vacuum
Bulkhead Clamp, KF25
2
Ideal Vacuum
Ideal Vacuum Chamber Door Bolts
2
Ideal Vacuum
Centering Ring, KF16
4
Ideal Vacuum
Centering Ring, KF25
6
Ideal Vacuum
Hinge Clamp, KF25
2
Ideal Vacuum
90 Degree Elbow, KF25
1
Ideal Vacuum
24 in. Vacuum Bellows, KF25
1
Kurt J Lesker
Thermocouple and Power Feedthough, KF16
1
Ideal Vacuum
Dry Scroll Vacuum Pump
1
AJVS
Pressure Gauge
1
Fil-Tech
Power Supply for Pressure Gauge
1
HighVac Depot
Throttle Valve, KF25
1
Ideal Vacuum
Substrate Heater
Nichrome Wire (20 AWG)
1
Amazon
Crimp Wire Terminals
1
Amazon
Bare Copper Wire
1
Amazon
Inline Barrel Connector
1
Kurt J Lesker
Thermocouples
1
Amazon
Aluminum Nitride Disks
2
Heeger Materials
HM2084 (Need Quote)
Multipurpose 6061 Aluminum Sheet
1
McMaster-Carr
Silicon Nitride Balls
3
McMaster-Carr
4" Standoff (#8-32)
1
McMaster-Carr
#6-32 Screws
1
McMaster-Carr
Substrate Heater Mount Bracket
1
Sendcutsend
#6-32 Hex Nuts
1
McMaster-Carr
#6 Split Washer
1
McMaster-Carr
#8-32 Screws
1
McMaster-Carr
1/4"-20 Screws
1
McMaster-Carr
#6 Flat Washer
1
McMaster-Carr
Chamber Stand
1010 x 10.1” Al Extrusion
5
8020 Inc.
1010 x 4” Al Extrusion
5
8020 Inc.
1010 x 27.6” Al Extrusion
3
8020 Inc.
1010 x 19” Al Extrusion
1
8020 Inc.
1010 Inside Corner Bracket
4
8020 Inc.
1010 3-Way Square Corner Connector
8
8020 Inc.
1010 2-Way Square Corner Connector
1
8020 Inc.
1/4"-20 0.75” Button Head Cap Screw
1
McMaster-Carr
10-32 ⅜” Button Head Cap Screw
1
McMaster-Carr
#10-32 T-Nuts
3
McMaster-Carr
The Precursor Delivery System is responsible for housing and delivering precursors (TMIn, TDMASn, and H2O) with N2 carrier gas to the ALD chamber for the deposition of Indium Tin Oxide (ITO) thin films. ITO offers excellent electrical conductivity and optical transparency, making it a promising candidate for advanced applications, particularly as a channel material in thin-film transistors (TFTs).
This project is part of the larger ALD project to unlock Hacker Fab’s capability to deposit high quality thin films with nanometer control over thickness and potential for a wide range of materials. ALD is also isotropic in nature, which allows the team to build new structures.
Our aim is to create a system with similar capabilities as the ALD machine in the CMU Nanofab at a fraction of the cost. We also want our ALD system to be brought to the Nanofab in order to lower the barrier of entry to thin-film research at CMU.
The overall ALD Delivery System has two main elements:
The Delivery System consists of precursor ampules, ALD valves, manifold, heating elements, mass flow controller, and piping that are responsible for pulsing and purging of precursor chemicals into the ALD chamber
The Delivery Storage insulates the chemicals from the lab via an enclosure connected to the exhaust, and provides structural support for the Delivery System.
Metric
Specifications
Operating Vacuum Pressure
100 mTorr
Pyrophoric chemical compatibility
Stainless steel piping and fittings
Cylinder pressure to MFC stepdown
Two stage pressure regulator from ~2000 psi → <70 psi
Temperature
60 C maintained for entire precursor flow path with thermocouple monitoring
Pulse control
ALD valves pulse for 3-0.625 s
MFC flow rate
10-40 sccm
Precursor ampules
Swagelok cylinder assembly holding TMIn, TDMASn, and H2O
Carrier gas
N2 cylinder
Vacuum rating:
We chose VCR connections for the gas lines in our vacuum system to ensure a reliable, leak-tight seal, essential for maintaining system integrity. These metal-to-metal sealing connections are ideal for ultra-high vacuum (UHV) and high-purity gas applications, reducing the risk of leaks, outgassing, and contamination.
Their robust design and reusability also made them a practical choice for our setup, where frequent assembly and disassembly are required.
Carrier gas delivery:
We use a two-stage regulator at the carrier gas cylinder as it ensures consistent and precise pressure control, reducing the high-pressure gas from the source to a stable, manageable level for downstream components. This stability is essential for maintaining uniform gas flow in the system.
The gas at the correct pressure is then delivered through a mass flow controller that regulates the flow rate of gases entering the system, providing precise control to meet process requirements. It ensures accurate delivery of gases to the precursor manifold.
We are receiving a precursor manifold having three Swagelok ALD3 valves donated by the Claire & John Bertucci Nanotechnology Laboratory at CMU. The precursor manifold serves as a distribution hub, directing gases to valves.
Due to our system’s reduced complexity compared to the Nanofab, we opted to remove a section of the manifold to fit our intended gas flow path. We also removed one of the ALD valves to account for only using three precursors. This led to footprint and weight reduction for the assembly, as well as leaving us with only ¼” VCR inlet and outlet for the carrier gas.
The ALD valves in the manifold are high-speed, precise valves that control the pulsed delivery of precursors into the vacuum chamber. These valves are critical for achieving the sequential gas flows required in ALD processes.
We are also receiving precursor ampules pre-filled with TMIn and TDMASn from the Claire & John Bertucci Nanotechnology Laboratory at CMU. In an effort to reduce cost, we quoted a custom welded empty cylinder assembly from Swagelok directly as opposed to STREM, which allowed us to almost half the price from ~$800 to ~$480. The empty ampule will then be filled with DI water for our third precursor source.
The precursors and the carrier gas mix must be heated for certain temperatures before entering the chamber both to sublimate from their solid states in the ampules as well as to keep them from condensing in the gas line. This is incredibly important due to the pyrophoric properties of the precursor chemicals. Insufficient temperature will leave precursor residue in the gas line, which may lead to contamination in the deposition process or combustion.
In order to increase replicability of our system, we opted to use heat tapes from BriskHeat for heating our ampules and manifold as opposed to custom-machined heating blocks previously used in the Nanofab. This change drastically reduced the weight and footprint of the delivery system, as well as reducing assembly complexity.
Heating will be implemented with closed loop feedback control using thermocouples to measure the pipe/ampule surface temperature as was done in the studies we referenced for this project.
In order for the ALD valves to operate properly, it needs N2 gas at sufficient pressure, which has led to modification of the N2 gas supply line in our lab. This resulted in splitting the N2 supply between our plasma etcher and ALD and required purchasing of a number of NPT and push to connect fittings, as well as a pressure regulator and valves to control pressure.
Part
Part Number
Quantity
Date
Supplier
Unit Cost
Price
Link
100 SCCM Flow Controller
1
10-21-24
Alicat
$1,480.00
$1,480.00
50ml Swagelok cylinder assembly
CSDP-2025-1025
1
3-11-25
Swagelok
$482.53
$482.53
Trimethylindium, elec. gr. (99.999%-In) PURATREM
98-4056
1
STREM
$4,777.00
$4,777.00
Tetrakis(dimethylamino)tin(IV), 99% (99.99%-Sn) TDMASn PURATREM
98-4050
1
STREM
$1,959.00
$1,959.00
Stainless Steel Tubing
2
10-31-24
McMaster-Carr
$20.22
$40.44
Stainless Steel Tubing
1
10-31-24
McMaster-Carr
$9.26
$9.26
Female VCR to tube fitting
SS-4-WVCR-6-400
2
2-19-24
Swagelok
$101.20
$202.40
1/4" VCR face seal metal gaskets
SS-4-VCR-2
40
2-19-24
Swagelok
$1.90
$76.00
1/4" tube fitting to VCR connector body (Male)
SS-4-VCR-6-400
1
10-31-24
Swagelok
$22.40
$22.40
KF25 to 1/4" tube fitting
QF25X4SWG
1
10-31-24
Kurt J Lesker
$66.35
$66.35
1/8" Male NPT connection tube fitting
2
10-31-24
Swagelok
$18.20
$36.40
ALD Valves
# 34C-ABA-GDFC-1KT
3
Provoast
$41.20
$123.60
3120A Series Dual-Stage High-Purity Regulator (0-50 psig, 580 CGA)
SEQ3121A580
1
1-30-24
Matheson
$854.40
$854.40
1/2" VCR face seal metal gaskets
SS-8-VCR-2
4
2-3-25
Swagelok
$2.20
$8.80
1/2" Plug VCR Face Seal Fitting
SS-8-VCR-P
1
2-3-25
Swagelok
$15.80
$15.80
1/2" Cap VCR Face Seal Fitting
SS-8-VCR-CP
1
2-3-25
Swagelok
$33.60
$33.60
1/4" Cap VCR Face Seal Fitting
SS-4-VCR-CP
3
2-19-24
Swagelok
$47.70
$143.10
Compressed Air Regulator
8812K52
1
2-26-25
McMaster-Carr
$30.96
$30.96
High-Pressure Brass T fitting
50785K222
1
2-26-25
McMaster-Carr
$5.71
$5.71
High-Pressure Brass Pipe Nipple
50785K152
1
2-26-25
McMaster-Carr
$2.30
$2.30
1/4" NPT to 1/8" Push-to-Connect
7880T114
1
2-26-25
McMaster-Carr
$2.43
$2.43
1/8" OD Firm Polyurethane Tubing - 25 ft
5648K22
1
2-26-25
McMaster-Carr
$7.00
$7.00
Brass Compact Threaded On/Off Valve
4912K2
2
2-26-25
McMaster-Carr
$6.87
$13.74
High-Pressure Brass 90 Degree Elbow
50785K36
1
3-3-25
McMaster-Carr
$4.71
$4.71
Standard-Wall Brass Pipe Nipple
4568K135
1
3-3-25
McMaster-Carr
$4.05
$4.05
Push to connect reducer 5/32" x 1/8"
5779K351
1
3-14-25
McMaster-Carr
$6.66
$6.66
Standard Insulated Heating Tapes (B00) - 120V, 0.5" width, 4 feet length
BIH051040L
2
BriskHeat
$66.25
$132.50
Standard Insulated Heating Tapes (B00) - 120V, 0.5" width, 2 feet length
BIH051020L
2
BriskHeat
$64.75
$129.5
Thermocouples
-
1
4-14-25
Amazon
$44.99
$44.99
KF25 Clamps
QF25-100-SSC
3
4-20-25
Kurt J Lesker
$17.55
$52.65
KF40 Clamps
QF40-150-SSC
2
4-20-25
Kurt J Lesker
$19.25
$38.50
KF25 Centering Rings
QF25-100-SRV
5
4-20-25
Kurt J Lesker
$11.25
$56.25
KF40 Centering Rings
QF40-150-SRV
3
4-20-25
Kurt J Lesker
$13.50
$40.50
PVC hose - 1"ID, 50 ft
-
2
4-20-25
Amazon
$75.99
$151.98
PVC hose - 1.5"ID, 50 ft
-
1
4-20-25
Amazon
$139.99
$139.99
KF25 to hose
QF25-100-SH
6
4-20-25
Kurt J Lesker
$27.60
$165.60
KF40 to hose
QF40-150-SH
3
4-20-25
Kurt J Lesker
$33.20
$99.60
Hose clamps
5415K17
1
4-20-25
McMaster-Carr
$12.27
$12.27
Total
$12132.48
Metric
Specifications
Storage capacity
L 430mm x W 340mm x H 375mm
Material compatibility
Steel and aluminum structural parts
Integration with ALD system
Pass through holes for all inlet and outlets
Ergonomics
Door with hinge and latch
Ampules accessible for replacement
Manufacturing
All components accessible during assembly
Built in adjustability for manifold position
Sheet metal enclosure
1.6mm thick 5052 Al bent sheet metal
Safety
Exhaust connection on top of enclosure
Difficulty with rivets
Time and effort to manufacture
Components accuracy
Improving replicability
Structural rigidity
All solutions implemented to address these challenges came with a trade off for price, which was necessary to meet the targets set for this semester.
Manufacturing time and effort is greatly reduced via outsourcing sheet metal manufacturing to SendCutSend, a quick turnaround (~2 weeks) company with reasonable pricing ($307.6). This also led to improvements in rigidity by changing the previous thickness of 0.050” (1.27mm) to 0.063” (1.6mm). This was justified given CMU’s limited water jetting availability and ALD team’s aggressive timeline. This method also gives us better confidence in dimensional accuracy of sheet metal features.
Another way that rigidity was improved was through the placement of flanges. By ensuring each side has at least two flanges parallel to each other, and at least one other free side bolted to a flange of an adjacent sheet metal part, the bending stiffness for each face of the enclosure is greatly improved by increasing the mass moment of inertia.
To address difficulty in working with rivets, all fasteners were switched to bolted connections for ease of installation. All connections between sheet metal parts are standardized to M3 bolts and nuts.
A few options for supporting the manifold were considered. The first idea was to support the manifold from the bottom with sheet metal components, but this was not considered due to the need to easily remove ampules and interference with heating tape wrapped around them. The general approach of supporting from the top was chosen and had the following requirements:
Fully constrain the manifold in all degrees of freedom
Allow adequate access to ALD valves, inlet and outlet fittings, and ampules
Off the shelf components to reduce manufacturing time
The best approach we arrived at was using pipe U-bolts and retaining plates to secure the manifold by collars coming out of the ALD valves. This solution would prevent putting stress on the main precursor and carrier gas line and leave space for installing all fittings , ampules, and heating tape. U-bolts were found readily available on McMaster Carr with inner diameters that closely matches the valve collars, and retaining plates can be waterjetted easily on campus.
The U-bolts would mount to the enclosure ceiling and the mounting holes are slotted to allow for adjustment of the manifold position relative to the chamber in order to account for any manufacturing tolerances, especially with the outlet tube.
The exhaust was placed at the top of the enclosure to direct airflow upwards, and small vent holes placed at the bottom of side walls to allow fresh air to be sucked in. The exhaust has a KF40 to 1.5” PVC hose adapter mounted to allow for sufficient cross sectional area for air removal from the enclosure.
In addition to adding the exhaust, we also had to plan out how this integration would be implemented for the rest of the lab, with the plasma etcher and the ALD vacuum pump also needing consideration. The decided routing makes sure that the three lines are separate to prevent unwanted chemical reactions in the lab. The planning is illustrated below:
Additionally, in order to increase negative pressure within the storage cabinet, grommets and covers were added to the various ports. The non-circular ports for the inlet and outlet are designed as 3D prints with ABS due to proximity to the heating tapes. The rest of the circular openings use cut-to-size grommets.
Holes are placed in the sheet metal enclosure to allow electrical and gas connections be to made with the manifold assembly as shown below:
In order to further ensure smooth integration, we have decided to share a common stand with the ALD chamber constructed out of aluminum extrusions. This would allow further control over the relative positions between the Delivery System and Chamber to ensure the tube connecting both will connect successfully. 10-32 holes on the floor of the enclosure will be used for mounting to the common stand.
Part
Part Number
Quantity
Date
Supplier
Unit Cost
Price
Link
Easy-to-Weld 5052 Aluminum Sheet, 0.050" Thick x 12" Wide x 48" Long
88895K235
1
11-26-24
McMaster-Carr
$70.04
$70.04
Easy-Squeeze Manual Tool for Blind Rivets
97543A120
1
11-26-24
McMaster-Carr
$113.51
$113.51
Draw Latch (Screw on, Nickel-Plated Steel, 2-3/4" Long x 1-7/16" Wide)
1766A2
1
11-26-24
McMaster-Carr
$15.20
$15.20
Adhesive-Back Bumpers (Polyurethane Rubber, 1" OD, 1/4" High, Durometer 65A, Black)
95495K891
1
11-26-24
McMaster-Carr
$12.42
$12.42
Aluminum Blind Rivets with Steel Mandrel (Domed Head, 3/16" Diameter, for 0.020"-0.125" Material Thickness)
97517A045
1
11-26-24
McMaster-Carr
$12.55
$12.55
Sheet Metal Enclosure
1
2-20-25
SendCutSend
$320.56
$320.56
U bolts, 1-1/8" ID
8880T955
3
2-26-25
McMaster-Carr
$1.88
$5.64
Surface-Mount Hinge with Holes
1586A34
1
2-26-25
McMaster-Carr
$7.15
$7.15
M3 Screws
91290A113
1
2-26-25
McMaster-Carr
$12.52
$12.52
M3 nylon lock nuts
90576A102
1
2-26-25
McMaster-Carr
$4.82
$4.82
#6 screws, countersunk
91263A514
1
2-26-25
McMaster-Carr
$8.26
$8.26
#6 screws, flat head
91251A146
1
2-26-25
McMaster-Carr
$11.90
$11.90
#6 nylon lock nuts
90631A007
1
2-26-25
McMaster-Carr
$3.93
$3.93
3/8"-16 Lock Nuts
90630A121
1
2-26-25
McMaster-Carr
$4.50
$4.50
Addressable LED Strip Light
1
2-26-25
Amazon
$15.99
$15.99
KF40 Bulkhead Clamp
QF40-150-BC
1
2-26-25
Kurt J Lester
$36.10
$36.10
10-32 lock nuts
90631A411
1
3-14-25
McMaster-Carr
$4.83
$4.83
10-32 bolts 3/4" long
90128A945
1
3-17-25
McMaster-Carr
$9.21
$9.21
3/8" washers
92141A031
1
3-14-25
McMaster-Carr
$6.15
$6.15
1.5" grommet
2633N3
1
4-20-25
McMaster-Carr
$6.72
$6.72
2.25" grommet
2633N33
1
4-20-25
McMaster-Carr
$7.34
$7.34
0.5" grommet
2633N11
1
4-20-25
McMaster-Carr
$8.80
$8.80
1.0" grommet
2633N26
1
4-20-25
McMaster-Carr
$6.36
$6.36
Total
$704.50
The assembly process starts with the modification of the manifold, where the top ½” section was removed and replaced with VCR caps.
Then the installation of all structural components, which includes the assembly of the cabinet as well as mounting the manifold. The cabinet was also mounted to the stand loosely for adjustment later.
Note:
The retainer plates should not be tightened down too tightly, otherwise they will fail in bending. They only need enough preload to prevent the manifold from wobbling.
Having the M3 socket head screws for the cabinet sheet metal face inwards made tool access easier.
There are three pairs of nuts for the U-bolts: 1 for retainer plates, 1 for clamping to the inner ceiling, and 1 lock nuts for clamping to the outer ceiling.
The space allowed at the top is limited, so make sure when tightening the U-bolts that the top of the VCR caps are not pressing against the ceiling.
The three holes on the floor were cut with hole saws after ordering due to the unforeseen change in length of the Swagelok cylinder assembly.
The next step is gas line integration, where all the mass flow controller, two stage regulator, and fittings were connected together with bent tubes.
Due to the number of moving parts, we decided to start at the N2 cylinder, and adjusted tube bending points, stand position, and cabinet positions as needed during the installation.
Note:
Numerous tutorials were used during the tube bending and VCR installation process, for which we used tools borrowed from the CMU Nanofab:
Due to the difficulty in lining up all the tubing perfectly, the cabinet had to be shifted closer to the chamber, which meant that the left most two stand mounting points were not used.
After the line was installed, we ran a vacuum test on the entire line. After some time with the pump running, the line was able to hold 20 mTorr of pressure with roughly 0.6 mTorr/min of leak rate. This was acceptable to us, thus we were able to validate the gas line.
The next step was to hook up the lab N2 gas supply to the ALD valves. This led to the removal of the existing connection to the plasma etcher, and adding in fittings to split the line, along with an additional pressure regulator for ALD. The final connection to the ALD system was made using a push to connect fitting.
We were able to validate this also by checking for major leaks and verifying that the pressure regulator can hold to the desired 50 psi. The valves were also confirmed to be able to actuate using the gas pressure.
The heating tape was then installed to the gas line. The 4 ft heating tape was able to cover the line from near the cabinet inlet to the chamber inlet, allowing us to heat the line throughout the mixing area. I was also able to verify that the 2 ft heating tape was able to wrap around one of our ampules.
Two thermocouples were also placed on the line, with one under the heating tape, and another on a bare section between the tape to verify thermal uniformity during heating.
Note:
It’s important to leave as little tape area as possible not in contact with metal in order to prevent the heating tape from overheating. The same is true for overlapping heating tape, which should also be avoided.
We have not yet been able to validate the heating tapes due to waiting for electrical parts needed for supplying the 120 VAC.
This is a part of the assembly that has not yet been completely finished due to lead times for all the necessary parts. So far the covers were printed and grommets installed with appropriate holes cut to size. The KF40 bulkhead clamp has also been installed to the top of the cabinet.
Lastly, we were able to fill the custom cylinder assembly from Swagelok with DI water and install it to the manifold.
We filled the ampule to about 2/3 of the way in order to allow space for water vaporization.
Note:
During the filling process, a pipette was used, and a repeated shoving motion was needed to get the water in due to the small opening. This contact is not ideal for the VCR fitting, but seems to be the simplest process.
The project has proven successful over the course of the semester despite some outstanding actions that still need to be taken. The previous design was improved upon, and most of the construction is complete.
Below are the overall R&D costs of the delivery system so far:
With Precursors
Precursor Delivery
$11427.98
Precursor Storage
$704.5
Total
$12132.48
Without Precursors
Precursor Delivery
$4691.98
Precursor Storage
$704.5
Total
$5396.48
Below are action that still need completion:
Verification of heating tape effectiveness via thermocouple measurements
Integration of exhaust lines into the lab building
Procurement of the remaining precursors from the CMU nanofab
Once the overall ALD system is fully operational, the team will begin testing the effect of valve pulsing on chamber pressure. Once the machine’s behavior is understood, the team will move forward with attempting growth of common thin films like aluminum and hafnium oxides before moving on to ITO deposition. During this process, the team will begin film characterization and process development to determine recipes for finally growing high quality thin films with atomic layer precision.
Below are additional links to files and documentation:
The ALD system requires controlled temperatures for the substrate and the precursors, as well as a pressure controlled vacuum chamber. There are four subsystems:
Carrier Gas Flow
Chamber Pressure
ALD Valves
Heating and Thermocouple Elements
Material
Quantity
Price
Vendor
KAMRUI GK3 Plus Mini PC, 16GB RAM 512GB M.2 SSD Mini Computers,12th Gen Alder Lake N95 (up to 3.4GHz) Micro PC
1
$170
12V 3A AC Adapter Power Supply
1
$14
Manual Butterfly Valve
1
$345
4-Channel K-Type Thermocouple Sensor MAX31855 SPI Arduino Shield (ch4-7)
1
$79.95
Arduino Uno Rev3
2
$27.60 each
Power Strip
1
$29.99
3-wire heavy duty power cord
2
$15.99 each
Fuse kit
1
$11.99
Power Relay Featherwing
5
$9.95
16 gauge wire, 100 ft
1
$14.99
[]
[]
Chamber faces should be installed following Ideal Vacuum’s chamber .
Each chamber port should be fitted with the relevant feedthrough, a properly sized centering ring, and properly sized bulkhead clamps. Follow the for installation.
[]
[]
The axial direction is marginally more complicated. The upper and lower surfaces of the plates are secured using stainless steel screws, which effectively limits the axial expansion and causes a concentration of compressive stress in the AlN. Assuming the screws have an initial length , a CTE of , and an initial stress of 0 Pa, an application of some Delta T would result in a lengthening of the screws to:
The sum of the lengths of the other components must equal , which can be stated as:
Gives a critical temperature deviation of
[]
The table below is the BOM for delivery system components. The entire BOM for the ALD Machine so far, or a more detailed look into the ALD BOM, can be found here:
Building off previous work that can be found in the Delivery Storage section at this link: , the current enclosure design aims to address the following challenges from the previous design:
The table below is the BOM for delivery system components. The entire BOM for the ALD Machine so far, or a more detailed look into the ALD BOM, can be found here:
Each subsystem will be described below in detail. LabVIEW is run on a mini PC communicating with Arduino Uno microcontrollers to manage various system values. The associated LabVIEW models and code can be found .
This Mass Flow Controller (MFC) is responsible for regulating the amount of Nitrogen gas flowing through the tubing over time. A is used to regulate N2 flow. For now, we are using manual control of the MFC.
This valve is responsible for modulating the pressure of the vacuum chamber and the evacuation rate of the pump; we require a pressure of less than 100 mTorr for our processes. This can be used to control the pressure of the chamber, or it can be done manually with this .
We are using these donated by the CMU Nanofab. These valves control the duration for which precursors are open to the carrier gas line. This uses a truth table to assess the cycle number and the specific precursor ratio to determine whether to be on or off. A relay board is used to switch the power supply based on these truth table commands.
An Arduino Uno is used with a and . The relay boards are used to drive current through heating elements in the system, and the thermocouple shield is used to measure the temperature at these points in the system, thereby allowing them to be maintained at a controlled temperature. The code for driving the boards and reading temperatures can be found in the repo .