Hi Jacob,

In the last years I gained extensive experience with ALD reactors from different manufacturers. In my opinion the reactors are often too large for the purposes of researchers. If you have typical substrate sizes of a few cm, you don't need a table of 20cm diameter. Of course, sometimes you need that area, but in most cases a smaller reactor size would be sufficient. A smaller reactor would also lead to less precursor and time consumption which would increase the throughput in small-scale research.

I think the necessary key features for an inexpensive ALD are: 5-10cm sample table, 20-300°C heating, at least 3 precursors, optional plasma source. Fast processes with low precursor consumption.

We wasted many months, percursor and patience with process optimization and we made the experience that the recipes that are pre-installed or supplied by the manufacturer are often far away from optimum. Maybe it is possible to bypass or simplify this optimization process. It would definitely improve the user experience and opinion of the system!

Thank you very much Bjorn. It sounds like I had similar experiences as a grad student. Have you used a QCM in conjunction with any of the ALD systems you worked with?

Unfortunately we had no QCM in our systems. So every sample had to be analyzed by ellipsometry. We did not only fight with the deposition rate, but especially with the homogeneity of the layers.

It sounds like the flow within the reactors was not optimal and the recipes were very far off.

Having a QCM would be very helpful for fundamental research of materials.

Hi,

I think that it would be interesting the possibility of having various substrates at the same time and/or to hold them at different temperatures. In this way, one would have exactly the same ALD but on different substrates and one would be able to determine the influence of the substrate on the growth of the thin film. This is actually my main topic...

I think the research type reactor outlined by Björn Hoffmann above would be very useful in many situations. Several colleagues of mine have built small ALD reactors, and since they have experience in doing it, they can put therm together very cheaply. Usually they use a QCM as a diagnostic. I have been tempted to put one together, but there is a learning curve to get students and myself up to speed. Having a very small turnkey system would be very appealing, especially if it was validated with a simple high quality Al2O3 for a 2 cm x 2cm size sample for example. I presently use a larger system in a larger facility, but feel having a smaller system might let me dial in the parameters more carefully. Switching speeds were not mentioned above, but I also think a smaller reactor would be advantageous. I would also be interested in looking at a variety of complex materials to deposit ALD films on, i.e. felxible materials, or soft materials

So in summary, a very simple turnkey manifold with good software for switching and control. Some attention paid to safety and exhaust. One could potentially have different options for the actual reactor chamber.

The tubing carrying the precursors to the chamber must have some way to be kept clean - maybe extra heating to keep them volatilized?

We just built a prototype of a mini-ALD with an ultrasmall reactor (~1cm³) in cooperation with two small companies. We are going to test it next week with a solid copper precursor. It is not faster than typical large ALDs but consumes only a very little amound of precursor. Furthermore the reactor vessel has a glass cover and fits in our Raman microscope! In this way we can measure the Raman signal of a thin layer during growth and we can use the laser for enhanced site-selective deposition.

This may be a very specific research type of reactor, but maybe someone here is interested in our development.

Hi Björn,

the ultrasmall reactor sounds interesting! I could be interested in your development, especially the laser enhanced deposition. And I'm not so far from you (Cottbus). Why don't we get in touch?

by the way,

in our home-made reactor we sometimes have a problem with the ALD valves. They get dirty and don't work properly after some time. It happens that they don't open or become leaky. This for some precursors, only (TDMAHf for example). I think the problem is some particle deposition on the valve. Do you know how to clean ALD valves? (Swagelok ALD-valves)

Björn Hoffmann, do you have a paper describing the construction. I would be interested in knowing more about the very small reactor

I'm sorry, but we have no paper so far. It is a very recent development. We just got the prototype a few weeks ago and next week will be the first big test. Preliminary results are looking very promising! I'll keep you informed.

Massimo:

do not open the Swagelok valves. You will never be able to seal them again. A way would be to dip the valve (only the channel part) into a proper solvent and to connect the pneumatics for opening and closing. So you may dissolve the particles. Afterwards you need to purge the valve for a long time with pulsing in order to get all the liquid out.

Btw: the grey valves from Swagelok work pretty well as well and the cost is by far lower than the blue ones.....

Björn:

congratulations to your reactor. This is always a great feeling when one gets an own reactor running. Concerning your initial response to Jacob concerning the reactor size is a critical point. It is reasonable to argue that a smaller reactor size will decrease the precursor waste and the purging time. On the other hand you need to see the relation of the chamber size and the sample size. A "semi-open" chamber can sometimes be very useful for a good thermal stability of the sample itself. I do prefer a chamber size largely exceeding the typical sample size we use, since in my experience we never obtained a good homogeneity with too small chambers if we deviate from the TMA process.

Well, I write not to answer to the questions, but rather to invite you to the workshop "Synchrotron Radiation to study Atomic Layer Deposition". We will also talk about the way to construct ALD reactors!

Bests,

Massimo

http://indico.cells.es/indico/event/63/

Thanks to Patrick Gonzalez to highlighting this question. Björn Hoffmann mentioned a small ALD precursor-screening-system. I think the comments stated by Björn, got mixed up a little bit. The mentioned system was originally developed by modularflow and is now commercialized. The ALD growth can be monitored e.g. via a QCM monitor, or ellipsometry, Raman spectroscopy etc. via a viewport. A CE-certified system are also available by the joint project myplas driven by modularflow and PLASMA ELECTRONIC. Currently, we are developing (12") and built up a 4" plasma enhanced / radical enhanced ALD system. Another project is a ALD3D (ICP assisted ALD) system which is capable of handling complex 3d structured substrates. There are several systems on the market, e.g. you can find such systems at the University of Stuttgart, University of Erlangen, Furtwangen University, University of Oslo, and specialized companies. If you are interested you can contact me for additional information.

Mato Knez :

I've (somewhat) recently inherited the ALD chambers from a set of previous PhD students. Lately, we've had blockage issues with our Swagelok ALD valves with particular precursors. Notably, TiO2 and ZnO precursors. I'm looking into methods of cleaning these valves to avoid having to replace them. Do you have any recommendations for solvents or cleaning methods that are harmless to the pneumatic valve's membrane? Any suggestions would be much appreciated.