Specific Process Knowledge/Thin film deposition/Temescal

Feedback to this page: click here

E-Beam Evaporator (Temescal)

The Temescal E-beam evaporator in cleanroom A-5

The Temescal is a system for depositing metals by electron-beam evaporation. In e-beam evaporation, the deposition is line-of-sight directed from the source, which means it will coat only the surface of the sample facing directly towards the source. This makes it very useful for example for metals for lift-off. The system also has an ion source for in-situ Argon sputtering that can be used either for cleaning samples prior to deposition or to modify the film during deposition.

Wafers are loaded into the top of the chamber, which acts as a loadlock as it can be separated from the rest of the chamber by a large gate valve. Deposition will happen on all samples that are loaded together. You can load up to four 6" wafers or three 8" wafers for deposition on surfaces facing the evaporation source, or on up to one 6" wafer for tilted deposition (smaller samples may be tilted more). By using sample holder inserts, you can deposit metals on samples of different sizes and shapes. Only one metal can be deposited at a time, but you can deposit many layers of different metals one after the other. The system contains 6 metals at a time and the metals are exchanged based on user requests, so please request the metals you wish well in advance.

The user manual, user APV, and contact information can be found in LabManager:

E Beam Evaporator (Temescal) in LabManager

Training videos may be found here:

Training videos on Youtube

Deposition rate and thickness measurement accuracy

The machine measures the thickness of the growing film with a Quartz Crystal Microbalance or QCM. The machine calls it the Xtal (crystal). This is a very thin piece of quartz that resonates at about 5-6 MHz when a voltage is applied across it. The resonance frequency varies with the mass of the crystal, and when material is deposited on one side of it, the frequency changes. This is measured by the crystal monitor, which can then calculate the deposited thickness.

Quartz crystal microbalance with gold electrode. Image from RLC on EC21.com

The thickness calculation depends on the material density as well as other physical factors. A tooling factor is used to adjust the calculation based on the geometry of the setup, since the crystal is not in the same place as the wafers so the thickness deposited on the crystal is lower than on the samples.

The tooling factor is calibrated for a particular deposition rate. It may not be perfectly accurate for other deposition rates, but should easily be within the 10 % accuracy that we test for in our quality control measurements. For very thin films the thickness measurement will be less accurate than for thicker films.

The machine gives a rough number for the crystal lifetime simply based on how thick a layer it calculates has been deposited on it. If many layers have been deposited and there is stress in the layers (e.g., in Cr or Ni), there may be partial delamination, which can make the thickness measurement inaccurate. In this case the lifetime estimate given by the machine will be inaccurate. If you think the crystal is not measuring correctly, please let us know. We exchange the crystals usually around 20 % lifetime use.

Heating during the deposition

If the material you are depositing requires a lot of heat to evaporate, the substrates may get warm during the deposition.

A temperature test of a 100 nm Al deposition at 10 Å/s showed that the back of the wafer stays below 37 ^oC. The same is true for 10 nm Ti plus 90 nm Au at 10 Å/s. In contrast, deposition of 100 nm Nb at 2 Å/s heated the substrate to above 104 ^oC (but the back of the wafer stayed at less than 110 ^oC). Deposition of 60 nm W at about 1 Å/s heated the substrate to more than 123 ^oC.

You may get some idea of how much the substrate will be heated by looking at the temperature required to give a reasonable vapor pressure for the evaporation. You can find a collection of Honig's vapor pressure curves at the bottom of this page (external link).

Please contact us if you would like to test the heating of the substrates (write to thinfilm@danchip.dtu.dk).

The e-beam and the Cu hearth

Empty 6-pocket copper hearth (from Fil-Tech)

E-beam impinging on the target from the filament and a 6-pocket Temescal copper hearth and filament assembly enclosed by shields (from Scotech)

At the start of the deposition and for every 100 nm, please check that the e-beam hits the target material rather than the hearth next to the pocket or the bottom of the pocket.

1) Check that the e-beam sweep has not shifted away from the material onto the hearth. This could happen if the filament assembly has been distorted by heat for example.

2) Check that there is enough material in the pocket, so that the beam does not hit the bottom of the pocket. If you burn a hole in the bottom of the pocket, cooling water can leak into the chamber and it may flood!

Particulates on the films

1. Testing November/December 2018 by Rebecca Ettlinger and Evgeniy Shkondin

The number of particles that end up on the film depends on the material being deposited and the deposition parameters as well as the cleanliness of the wafer and the chamber. We have done some tests to compare particulates on the films in the Temescal and the Wordentec for different materials and process conditions.

Main conclusions:

• A lower deposition rate gave fewer particles for both TiAu films and Al films (comparison of 10 Å/s and 2 Å/s).
• Optimizing the deposition parameters (the soak and rise times and perhaps power level) can reduce the number of particles. If you would like to optimize the process for a material that you are working with, please contact the process responsible staff.
• For TiAu layers, there are relatively many particles compared to, e.g., Al or Ni. Specifically for Au, this is apparently due to carbon contamination of the target material, which is reduced by long soaking times before the deposition starts.
• There will be more particles on the film if the loading of the wafer does not go smoothly, so it is worthwhile to be careful and use the vacuum tweezers if possible.
• There are fewer particles from just loading/pumping/venting/unloading wafers with no processing in the Wordentec than the Temescal, at least when tested in Nov/Dec 2018.

The full results of the testing can be found here: File:particles-pinholes test Temescal.pptx.

2. Testing July/August 2019 by Evgeniy Shkondin and Patama Pholprasit

Deposition of Ti/Au revealed a big number of particles on the wafers. The amount has been heavily reduced by optimizing deposition conditions and choosing the right crucible. The particles are in fact gold droplets of various size ejected from the melt.

Main guidance and conclusions:

• Ensure the chamber, shutters etc. are properly cleaned.
• Use as pure Au target as possible.
• Do not use ceramic crucible, put Au pellets directly into the dedicated Cu heart pocket. W-crucible can be considered (but so far we did not tested it).
• Sweep pattern needs to be optimized so it coveres the bigger area of the target but avoid getting the beam to close to the edges.
• Optimize soak/rise powers and times.
• Use deposition rate of 2 Å/s. Ensure that the rise2 power is adequate for that rate.
• Ensure appropriate PID values, since the deposition rate has to be stable during the evaporation.

The full results of the testing can be found here: Media:Au issues with Temescal.pptx.

Process information

• Acceptance Test. Describes thickness uniformity tests, side wall deposition tests, sheet resistance tests and tests of the ion source for substrate cleaning.

Materials for e-beam evaporation

• Aluminium (Al)
• Chromium (Cr)
• Copper (Cu)
• Germanium (Ge)
• Gold (Au)
• Molybdenum (Mo)
• Nickel (Ni)
• Niobium (Nb)
• Palladium (Pd)
• Platinum (Pt)
• Silver (Ag)
• Tantalum (Ta)
• Tin (Sn)
• Titanium (Ti)
• Tungsten (W) - thinner layers

Note that to date (Dec 2018) we have processes available for deposition of Al, Cr, Au, Ge, Nb, Ni, Pd, Pt, Ag, Ti, and W as well as Ru. More will be available as they are requested.

Equipment performance and process related parameters for the Temescal E-beam evaporator

Purpose	Deposition of metals	E-beam evaporation of metals Line-of-sight deposition Possible to tilt sample Possible to ion clean samples Possible to modify deposition by Ar ion bombardment
Performance	Film thickness	10Å - 1µm* (for some materials)
	Deposition rate	0.5Å/s - 10Å/s
	Thickness uniformity	up to 3 % Wafer-in-Wafer variation, Wafer-to-Wafer and Batch-to-Batch variation **
	Thickness accuracy	May vary by up to about +/- 10 % Less accurate for films below 20 nm
Process parameter range	Process Temperature	Approximately room temperature. Higher for refractive metals that require a lot of heat to evaporate, see above.
	Process pressure	Below 1*10-6 mbar before deposition starts Below 5*10-6 mbar during deposition
	Source-substrate distance	69.85 cm
Substrates	Batch size	Up to four 6" wafers per standard run Or up to three 8" wafers Up to one 6" wafer with tilt Deposition on one side of the substrate
	Substrate material allowed	Silicon wafers Quartz wafers Pyrex wafers See also the cross-contamination sheet
	Material allowed on the substrate	Silicon oxide Silicon (oxy)nitride Photoresist PMMA Mylar Metals See also the cross-contamination sheet

* For thicknesses above 600 nm please request permission so we can ensure that enough material will be present.

** Defined as the ratio of the standard deviation to the average of the measurement made using the DektakXT. For further details see the acceptance test.

Quality control (QC) for the Temescal

We are still developing the QC procedure for the Temescal (December 2018).