Questions for Faebian?
If you have a specific question related to MBE and you either cannot find the answer in my existing blog posts or require further clarification, please ask them below. In order to avoid clutter on this page I will periodically remove old questions and either update my blog posts or add them to the FAQs page.
Hello, Dr. Bastiman
We are running all solid source MBE with GaP decomposition K-cell and As-cracker cell for group-V sources.
I have a question about As-P change step. How to make clean interface of GaInP-GaAs or GaAs-GaInP ?
Hello Seokjin Kang,
Yes that is a common problem with switch both the group III and V at an interface. It reminds me on an InAs/GaSb problem. We forced both interfaces to be InSb, and used the compressive strain of the InSb to strain compensate the InAs and GaSb super lattice.
You can perform a P “purge” after the GaInP layer: i.e. all shutters and valves closed to allow the P to be purged from the surface for a number of seconds (my estimate 10-20s). You will need to watch it on RHEED for the group V terminated reconstruction changing to a group III terminated reconstruction. Then you can supply the As. Allow the surface to recover then start growing the GaAs. After the GaAs layer, purge the As, apply P, then start growing the GaInP.
What you will create is a single ML of GaInAs and a single ML of GaP at each interface. What this means of course is that you are creating a strained interface. One compressive, the other tensile. That is the most abrupt interface you can achieve in a 4 atomic layer system.
In order to keep the interface abrupt you will need to anneal the surface before each purge. By which I mean stop the group III flux after the desired layer thickness and apply a lower group V flux at the growth temperature until the RHEED recovers into a pattern with very short streaks. The surface will flatten out and the multiple-ML steps will planarize. That way your interface will not be split over several MLs. Of course if you are growing a super lattice, that will significantly lengthen your growth times.
Hope this helps,
Thank you for your kind answer. I have one more question. The growth temperature of GaAs and GaInP is different. When we change from GaInP to GaAs, the growth temperature should be increased. What kind of problems can be caused by this and how to solve it?
Can you tell me your typical growth temperatures for InGaP and GaAs?
490-500C for GaInP and 570-580C for GaAs (pyrometer measured),
The solution will depend on what you are growing.
If you are growing thick layers, then you can simply purge the P from the InGaP layer and replace it with As at the InGaP growth temperature (500*C). Then grow 2.5nm (10ML) of GaAs at this temperature (500*C). Then you can ramp up to 580*C and grow the remainder of the GaAs layer. When you want to grow InGaP again, you can purge the As and replace it with P during the cool down step. The single ML of GaP will be stable from 580*C down to 500*C. Once the GaP has formed and you have reached 500*C you can start growing your InGaP layer.
On the other hand if you are growing thin layers (10-15nm) I would say that 500*C is not a terrible growth temperature for GaAs. You can simply grow the entire layer structure at 500*C and compare it to one with temperature ramps like the thicker layer described above. It is then simply a question of layer quality vs growth time.
Thank you for your answer, it will be very helpful to optimize the group-V change steps.
Hello, Dr. Bastiman,
I read your blog and I find it very interesting and helpful. Recently we bought a LAB-10 MBE system from Omicron coupled with a VT-SPM. For now I deposited Au on Si (111) 7×7. To determine the deposition rate and film thickness the system is fitted with Beam Flux Monitor only (no quartz microbalance). For an effusion cell temperature of 1240 degrees Celsius the flux on BFM is 5.1e-8 mbar. What can I do to know which is the deposition rate for this flux? It may sound strange question but I can not get past this obstacle. Thank you.
The honest answer is that there is no simple way to do this. The “Thin film processes” book by John Vossen and Werner Kern had a nice equation:
atoms/nm2/s = 3.51E22 * PA / ( (d^2) * (MT)^0.5 )
Where P (in Torr) is the pressure inside the cell at temperature T (in K), A (in cm2) is the cell orifice area which is ~2-3 cm2, d (in cm) is the distance of the substrate from the cell and M is the atomic mass (169 for gold). Unfortuantely I could never get it to give me a sensible answer without assuming the pressure inside the cell is 500x larger than the pressure measured by the beam flux monitor.
If we say, for Ga, that P =500 x 1.76E-7mBar = 500 x 1.32E-7 Torr, A is π cm2, d is 20cm, M = 69 (Ga), and T is 951 °C = 1224K. Then the flux is 0.62E13 atoms/cm2/s which is 0.1ML/s on GaAs(100) or 0.28 nm/s.
Then we put in Au (196) at 1240°C and 5.1e-8 mBar and we get 9.68E-12 atoms/cm2/s, which is 0.02 ML/s on GaAs(100) or 0.0043 nm/s.
I would honestly be amazed if this was accurate. The only good way to check is to cool the Si(111) to 0°C and then open the Au shutter for an hour. You can then measure the Au film thickness with SEM, TEM or XRR and work out your deposition rate in nm/h, and then scale that back to atoms/cm2/s. In so doing you can check to see how accurate the crude estimate was from the above equation (if at all!)
Hello, Dr. Bastiman
I am always getting many help from your posts, thank you.
I am wondering what is the suitable amount of source when we are doing source charging. (in case of SSMBE)
In addition, could you explain about the relation between recommanded amount of source and volume of PBN crucible?
For example, we are using 25 cc crucible for Si and Be and 40 cc crucible for In, Al, Ga. (K-cell)
I will expect your kind answer. Thank you
The 25cc and 40cc tells you the exact volume of the crucible in cubic centimeters. This is simply stated by the manufacturer as a means of determining the crucible size. With regard to how much material you should put into a crucible, that is not straight forward.
On the one hand you should put in as “much as possible” since this will give a larger flux at the respective temperature, however on the other hand “over filling” can cause serious problems in the case of In/Ga spitting and Al creeping. So for Al/Ga/In I would suggest you only fill the crucible to about 66% capacity maximum.
Si is a very high temperature source and a very low cost element (one typically uses smashed up high resistivity Si substrates that cost less that $10). One also does not (usually) melt the Si at 1414*C, so for Si you can really fill it up to about 90% of capacity if you wish and run it at a slightly lower temperature. However since doping with Si consumes very little Si, you can simply put half a Si wafer inside at it will last you years.
Be on the other hand is a very expensive element, something like $500/g! So I would recommend you only put a little inside the crucible: 2 cc (~2 g) is all that is really needed. You will need to run the cell a little hotter, but you can still dope e19 cm-3 with the cell below 950*C.
Thank you for your kind answer. I will try to put more sources next maintanance.
I have one more question about substrate holder. (VG-V80H, As and P based)
Could you recommand indium free substrate holder which does not need indium soldering? I am using Mo block which needs to indium solder to hold the sample (a piece of wafer, 1-2.5×1-2.5 cm2). I think it would be better if I can handle indium free block because I don’t need to indium removal process when I am making a device using general lithography process. Actually, I found one from Veeco but I am not sure about that and the price was about $3000 for each.
Contact Ross as SEP: http://www.semiconductorengineeringproducts.com and inquire about V80 platens. He typically provides parts for the production sized V90/100/150 systems, but he may be willing to provide a few V80 sample holders. The V90+ sized reactors were mostly vertical, which allowed a simple sample plate and backing ring design. The V80 (being horizontal) needs extra fitting: 3 Ta washers and 3 pieces of Ta wire. These fiddly bits add to the machining costs, on the other hand V80 platens are only 4″ across so they use less Mo. In the end it is difficult to estimate how much they should be, but the V90 platens are ~$500 each.
The platen I described above is for a 2″ wafer. The first thing you lose when you switch from “In solder” to “In free” mounting is flexibility: all your substrates must be exactly the same size. In order to accommodate these smaller pieces you can create an “insert”. The insert is a 2″ circular piece of Mo that perfectly fills the 2″ hole the substrate would usually occupy. You can then order a number of these “inserts” with different holes cut into them (e.g. 10 x 10 mm2 and 25 x 25 mm2) with substrate holding shelves like the original 2″ circular hole. The inserts will be around half the price of the sample plate, since they will also need fiddly bits machined into them to hold the sample down. There may already be designs for V80 sample holders that accommodate smaller pieces, at present I am only aware of the 1/4 of 2″ type. It should not be too difficult to work with a Mo machinist to create the sample holder you desire.
I just came across your blog and I find it exceptionally interesting; I can’t stop reading it.
My main interest is oxide MBE.
What companies do you recommend looking into for such systems?
In addition, do you have ballpark estimate for prices of such systems (i.e. DCA M600 etc). Can you think of ways of obtaining prices other than asking the companies directly? (the ‘quantum measurement problem’)
Assuming you are setting out to do research… the Riber Compact 21 is a really nice small research reactor and can be adapted for all materials. A ball mark price for a new MBE system from ANY supplier is €500-900k depending on the transfer (automatic/manual) and cell configuration. The only other way to obtain prices is to find out who recently bought a system and ask politely how much they paid…
I was indeed very interesting to stumble upon your blogs.I was very impressed on the type of MBE related work carried on by you.
I work on an EPI930 MBE system.Our substrate rotation/Flux monitor positioner went bad. We were looking for getting the equipment or its parts if available at a cheaper price.Please let me know if you have any idea where it can be found at a cheaper rates. Thanks.Really appreciate your time.
Certainly third party suppliers are typically cheaper than OEM for spare parts. Without knowing specifically what happened it is difficult to advise. However if it is the bearings of the substrate rotation or a mechanical probelm, I would suggest you contact Ross Davies email@example.com. He focuses mainly on old VG production system, but may be able to help you with the EPI930.
Thank you Dr.Faebian. I will definitely contact him.
Hi, Dr. Bastiman
I asked about Indium free Mo holder a few month ago and now I made some Indium free Mo holders through Mo machining company (for 16×16 mm2 size sample). However, it contains two main problems. First, the temperature measured by pyrometer (modline 5G) is 50~100C higher with thin Mo diffuser or no Mo diffuser than what I got before. I think some radiation from heater goes in to the pyrometer when the diffuser is thin or none, so I got 50~100C higher temperature. Second, when Mo diffuser is thick, it blocks radiation from heater I think but the temperature difference between GaAs wafer and Mo surface is 30~40C at 600C. It makes the side of the sample very bad because it is contacted by Mo holder. Now I figured out that using Indium is a good solution for these problem. How do you solve these problems using Indium free Mo holder? If you have any comment about this, please let me know.
I will be glad to assist you with this. Please send an email to the address on my homepage together with a picture of the In-free holder. Can you also confirm this is the same V80 system from earlier and that the substrate is GaAs?
Hi Dr. Bastiman
It’s Rahul from Indian Institute of Technology, Kharagpur, India. I read your blog and look forward to read the new contents. Thank you for posting such informative articles on your blog. It’s always nice to read serious topics in an easy way.
We are facing a problem with the GaAs chamber in our four cluster tool based MBE system. GaAs chamber need to be opened to change the filament of the flux gauge as somehow it got dis-functional. We are worried about the security measures before opening up the GaAs chamber.
Can you please tell me the required security measures before opening up the arsenide chamber
Yes I fully understand your concerns regarding arsenic! You can gain some good general advice by reading the following report:
Click to access ashpaper.pdf
Some other points:
Some of the As in the system combines with H to create AsH (Arsine), this is a very dangerous gas, but because of the continuous pumping the amount of this gas in your system is negligible. In order to be safe you can buy an cheap hand held mini Arsine detector and hold it near the system when you first break vacuum. From experience this never, ever goes off. The above report makes the same conclusion. You can check the composition of your vacuum before you open up using a RGA and my blog post:
The main issue with As (also highlighted in the report) is the As dust. Almost all of the As freezes onto the chamber side walls though a very small amount of As vapour exists in the vacuum. As dust is heavy and not readily air born, but it will cover all surfaces around the MBE system in a very thin layer during maintenance. Because of this you should wear a face mask during maintenance and wipe down all surfaces with an isopropanol (IPA) cloth after maintenance. In my experience the As dust is also very, very low.
Obviously you should also wear other safety clothes, like gloves and a white lab coat or other suitable clothing.
In order to get the system ready to vent follow the steps in my other post:
Hi Dr Bastiman,
I have noticed that some MBE growers give the material flux in terms of “nm/min”, instead of sccm.
Since the flux is, by definition “amount of material/area”, I’m a bit confused with the unit of “nm/min”.
Is there any specific reason for using the unit “nm/min” in explaining the MBE growth?
Can you please explain this to me?
MBE growers have fallen in the nasty habit of using non-SI units when stating their fluxes. There are two common non-SI units. The first is to quote the flux in terms of beam equivalent pressure (BEP) in mBar/Torr. This is perhaps the worst unit, since not only does it vary significantly from system to system, it actually varies significantly on a single system either over time or by changing the monitoring ion gauge head or cell position. The second non-SI unit is to quote the flux in terms of a “flux induced growth rate”. The latter is the one you mention in your question in terms of “nm/min”. The MBE grower is saying the cell flux induced a growth rate of X nm/min (on a certain substrate). Each grower chooses a growth rate that is most suitable to their applicaiton, so you will see the “flux” stated in several units: ML/s, µm/h, Å/s, nm/min, etc. The problem being the growth rate varies between substrates of different types for a given cell flux. For example the growth rate on GaAs is different to that on InP for the same cell flux.
Since in MBE we deal with particle fluxes, the SI unit would not be SCCM but rather would be [number of particles]/m^2/s. The unit would then be atoms/m^2/s. However since the atomic flux is so small it is more convenient to use: atoms/nm^2/s. You can find some more information in my blog post:
Hi Dr Bastiman,
Thank you so much for the explanation.
Since I’m an MOCVD grower it was strange to me; we frequently use material flow in “mol/min” rather than flux.
Your post in the given link was really helpful to understand the idea.
Hello Dr Bastiman,
I am Braj Bhusan singh from National University of Singapore, started to upgrade an old EIKO MBE system. There was a old PBN crucible, I want to reuse the crucible. Can you please suggest me the procedure for cleaning it? Thanks.
First of all it is good to know what state the crucible is in, and what material (if any) is still in there. Whilst it is possible to reuse most PBN crucibles, I have found crucibles that have been used with Aluminium are particularly difficult to clean.
If there are metallic residues in the crucible you can start by simply up-ending the crucible and trying to shake them out. Most metals to not wet PBN and shrink on freezing, so you may be able to remove them like this. Next you can try etching. Always use chemical in a flow box and wear protective clothing.
Attempt #1: Try filling the crucible with HCl and leaving it to stand for 24 hours. Rinse with DI water and finally IPA.
Attempt #2: If the residues persist try using Aqua Regia (a mixture of HNO3 and HCl in a 1:3 ratio). Again filling the crucible and leaving it for 24 hours. If the metal is etching slowly, consider leaving it another few days. Afterwards neutralize the Aqua Regia with sodium bicarbonate and flush with plenty of DI water, followed by IPA again.
Attempt #3: If the Aqua Regia does not work the only other option is Bromine Methanol, but that stuff is very dangerous especially via inhalation. Unless you have a specific safety person already qualified to use it, I would suggest you simply buy a new crucible. You can order new ones for around $750. Contact SciTech Solutions Ltd for a quote: http://www.scitechsolutionsltd.com/
If it does clean, after a final IPA rinse, it is a good idea to bake it to 100*C in an oven before you load it into the cell. It is also a good idea to outgas the newly cleaned crucible in an outgas rig (a special vacuum chamber just for outgassing cells) before you load it onto your growth chamber.
Hello Dr . Faebian
Thank you very much. I dipped PBN crucible in HNO3 for 24 hrs, and it get cleaned. Thanks.
Hello Dr. Bastiman
I asked about the amount of source that is charged in PBN crucible and you recommanded 66% of crucible volume.
We have charged about 50~55 gram for Ga and 40~45 gram for In and the crucible size is 40 cc.
So, I am planning to charge 40% of crucible volume because bottom four ports of our VG-V80H is horizontaly tilted. (Ga1,Ga2,In1,In2)
Then the charging volume will be 16 cc and it equals 97.52 gram for Ga and 112.32 gram for In. I am surprised that I have to put sources 2-3 times more. I wondered we used too small sources and it will be ok to proceed my plan (40% of crucible volume)
Please give your idea about this.
Hello Seokjin Kang,
I am afraid I do not fully follow the question. I believe you are asking if 100g of Ga/In is sufficient for refilling effusion cells on a VG-V80. If so, yes, I believe 2 x Ga cells and 2 x In cells with 100g of Ga and In is enough to grow 1000 microns of material before you need to refill. If you need more information, please send an email to firstname.lastname@example.org and I will be glad to discuss it with you.
Hello Dr. Bastiman,
I have a question regarding growth of InAs nanowires on silicon substrate. I have grown InAs on Silicon and I have seen that there are large amount of island growths.I did a time series and the NW length increases linearly with time while the diameter increases linearly for sometimes and then saturates.The starting diameter was 32 nm (I use nucleation centers to initiate the growth) and the saturated diameter is 75 nm. The island growth also increases with time and during diameter saturation, 50 % of the silicon surface is covered with islands.To check diameter saturation I changed the starting diameter to 75 nm and did the time series again.The length increased linearly but the diameter did not change from 75 nm. The surface was 50 % covered at zero time for 75 nm starting diameter as compared to 32 nm starting diameter which had 12 % surface coverage .
My assumptions are
1. As the wires are not tapered and have not shown super linear growth ,i assume that the length of NW is always smaller than the side wall diffusion length.So all the adatoms impinging on the sidewalls is utilized in axial growth.Radial growth should be zero.
2.No catalyst (Indium nor gold).axial growth from direct impingement is negligible.No radial growth here as well
3.Island growth is increasing with time so amount of ad atoms reaching the NW is decreasing as exposed silicon surface is decreasing with time.For 75 nm starting diameter probably very less adatoms reached the base of the NW
4. Birth and spread growth of sidewalls .The base of the NW and silicon surface provides low nucleation barrier.Radial growth takes place from here.
I have knowledge upto this much and these assumptions are not helping me to understand why the diameter saturated.???
Is there any good explanation for such a behaviour.
I think this diameter normalisation is covered in the GaN NW literature. Take a look here:
Hi Dr. Faebian,
I have question regarding the substrate temperature calibration. Since in MBE, we used load lock system, so radiative heating is obvious. The T/C will be far from substrate. How would we know the the actual substrate temperature?. Some people have used eutectic alloys for substrate temperature calibration. I want to grow Bi2Se3 on GaAs substrate. I am not very familiar about these alloys, please suggest me in this regard. Thanks.
Can you tell me what growth temperature you are aiming for? I have no experience with using eutectic alloys for temperature determination, since I tend to use the calibration points available from As-based transitions on GaAs(100):
I assume the idea is to detect the “solid state” of the eutectic alloy with RHEED (or light scattering or another method) and then locate the freeze-melt point. Upon melting the RHEED pattern would lose any features that correspond to facets of the solid alloy. A friend of mine did a similar temperature calibration for ~157°C with pure Indium on GaAs. I had thought to do the same with Bismuth for ~272°C. You would begin by putting down material at a high enough temperature for liquid droplets to form, then stop depositing material and cool the substrate down until facets appear on the RHEED.
If you are looking for temperatures >400°C I would recommend a pyrometer (the actual lowest detectable temperature depends on the working distance and background IR sources. Whilst the manufacturer may quote 280°C, in an MBE system the lowest temperature is about 400°C) . Pyrometers make not be accurate (due to them relying on emissivity) but they are repeatable. Most people use a pyrometer nowadays. I recommend the Photrix Series with an 880 spectral range:
If not need to cover the range from >250°C then you would need some bandedge thermography, like the Bandit system:
But those tend to be fairly expensive!
Hello, Dr. Bastiman
I wonder about shutter speed.
My linear motion shutter has some problem about opening and closing time.
It is getting longer and sounds something like scratch.
Sometimes it is stuck especially for In and Al.
I failed to grow samples several times because these shutters were not opened.
Could you give me some advises for this?
It sounds like you have III-V material on the linear bearing. Can you disconnect the linear actuator (the motor or pneumatic shutter driver on the outside of the machine) and slide the shutter open and closed by hand? Do it get stuck any where in particular? You can compare several and see if there is a difference to their motion, particularly the In and Al ones.
One short term fix is to simply increase the pneumatic gas pressure (if they are pneumatic) to provide more force to open and close and counter act the high friction bearing. Ultimately you would need to take each mechanism off and sent it away to be serviced.
The alternatives are either a badly coated shutter blade that is sticking in the cell divider housing or simply a dry magnetic coupling.
If it is the shutter blade the shutter motion will only be stuck at the point it is fully closed. If it is a dry mechanism you can simply try greasing the external mechanism, that is greasing the body of the short metal rod that the magnetic coupler slides along to open and close the shutter.
Hello, Dr. Bastiman
Based on your advice, I solved the shutter problem by cleaning the actuator but I think the real problem is inside the chamber. Sometimes it is still stuck and the experiment failed. So I am going to clean the shutter blade, separator and the rod in coming maintenance.
I wonder about source charging, especially As. I am using 500cc valved cracker cell made by applied epi. Should I pumped out all remaining As before source charging? What do you think about adding the source on the remaining source like III-materials?
Hello Seokjin Kang,
Yes you will need to clean the entire internal shutter mechanism. You will probably find some III-metal material on rod.
You definitely need to fully deplete the As charge before you replace it. This is done by running the As bulk (reservoir) at 450°C until the flux from the As source drops to background levels. Since it is typical to operate the cell at ~380°C this can take several weeks to fully deplete the source. You can go up in 10°C steps and still use the As for some growth until it finally runs out.
It is a good idea to fully deplete the Al cell (and Bi cell if you have one) before refilling, then you get away from problem with crucible damage on cooling. The others are ok to “top up” the existing material and not fully deplete. Remember to keep your Ga cell at 30-40°C while you top it up, otherwise the Ga can freeze and shatter the crucible.
You can find more info here:
Thank you for kind answer.
Could you explain the reason for depleting the As?
The reason for depleting the As is to do with the internal design of the reservoir. The As coats the entire inside of the reservoir and if there is any left at all you will find that when you remove the reservoir flange the inside is inaccessible because of the As build up. Worse, you cannot usually get a good seal when you replace the flange seal. The result is you need to send the cracker away to get refurbished, which means replacing the entire inside (because As coats the inside of the cracker tube too). In this case it is around 60%-70 of the cracker total new value to fix.
I found that you manualy clean the chamber inside on your post. I always wonder how to clean the chamber. However, because of Arsenic, i think it can be very dangerous even if i wear the mask. So, could you give me some advice for cleaning chamber inside?
In addition, I wonder my PBN crucible design is good. As i found in “www.scitechsolutionsltd.com” VG semicon 40cc crucible is ID:38 mm, IH:88.1 mm, Lip OD:50.8 mm, but our one is ID:32.5 mm, IH:110.8 mm, Lip OD:51 mm. And there is another one that has long IH of 149.8 mm. I wonder why crucible shape is not cylinder. If you have some idea for this could your please answer to me or post it?
Always thanks for your kind answer and posting
Cleaning an MBE chamber that has been used for As can become problematic once the As layer is above a certain thickness. If it is thin enough (a few millimeters) you can clean it pretty well with a special purpose toxic vacuum cleaner. However to do this you need to wear full breathing apparatus, which also means you need to attend a breathing apparatus course. I also recommend wearing a one time use disposable white full body suit and shoes and gloves. My chamber was relatively small, so I could lift out the LN2 cooling shroud and place the entire shroud in a fume cupboard. There is only negligible chance of As inhalation even without breathing apparatus, but the As dust can form a thin coat on any nearby surface. Everything needs to be thoroughly wiped down afterwards and all As coated material must be disposed of safely. The alternative is to send the entire chamber off to be cleaned once every 5 years for around $20-30k. Larger chambers (like a V80) are not so easy to clean yourself.
With regard to crucible sizes. It is fairly common in MBE to have non-standard sizes of cells and crucibles. Companies tend to list only the most common sizes. I would assume your crucible dimensions are correct and merely uncommon. You can either measure the crucible dimensions or send a crucible to be copied. Sending a crucible off to be copied is the easiest way.
One of the reasons crucibles are not cylindrical is that the sidewalls interfere with the evaporated flux beam shape. When a cylindrical crucible is near empty it gives a very different beam shape to a nearly full crucible. To this end the crucible shape is very carefully chosen based on the diameter and length to give a uniform beam pattern over a wide range of material fill percentages.
I stumbled across your blog while looking for some info in MBE growth….so far, I have still not found it.
Short of calling up some of the commercial MBE growth companies, I am looking for some very rough pricing info.
Specifically, what does it cost (rough ball park number will do) to grow Type II superlattice infrared detectors (some typical “recipe”) on 3 for 4″ GaSb wafers?
A ballpark cost for a commercial grown type II SL is around €2k per 4″ wafer. Perhaps you can get some discount if they have a system capable of growing 3 x 4″ simultaneously.
Hello Dr. Bastiman,
Thank you for writing this blog, it is very helpful and informative.
Our group is growing III-V semiconductor lasers, but we often deal with many crystal defects. We see a variety of defects, not just the ubiquitous oval-defects.
Do you have a recommendation for a book, website, or other source that gives detailed information about III-V crystal defects? It would be especially helpful if we could see pictures of defects and compare to what we see in our lab.
Unfortunately I cannot suggest any reference material for you. However assuming you are growing DBRs or QCLs from standard III-V alloys, then I recommend you email a TEM expert familiar with those material/devices and pose your questions directly. I would suggest Richard Beanland (Warwick, UK) as a start.
We are two students currently trying to grow GaN in a home-made MBE system at our university.
We are currently considering the introduction of a very simple Al effusion cell, and thus have a few questions.
You say: “Finally Al has a very different thermal expansion rate to PBN. The wet PBN cannot contract as fast as its Al film, hence cooling Al too rapidly from standby (850°C) to solid (660°C) can readily shatter the crucible.”
We were wondering, what you think if one were to use an alumina crucible instead?
How slowly would you cool it? And how many cycles of heating, followed by cooling to room temperature do you think it could sustain, if any?
We are most likely using a small alumina crucible, so we would have to do refills rather often. The crucible will simply be held by some wolfram wire.
Do you have any thoughts about this simple design? E.g. about length vs diameter of the crucible, heating, how much should we fill it?
The Al will be used to grow some AlN, which only needs a thickness of 20-50 nm, so having a large flux is not crucial.
We are going to do some simple testing of the Al effusion cell in a separate chamber before introducing it into the MBE system.
Certainly effusion cells can be very simple. I do not have any direct experience with Alumina crucibles, so I cannot comment on its suitability. I would suggest you contact a crucible supplier directly for information. However for the cell operation, I would not recommend to more that 25% fill by volume your crucible. You could consider using two crucibles (one inside the other) with the filament around between to protect the filament in the case of Al creep and overflow? You would need to create two small holes for the filament feedthrough. Certainly experiment on a separate chamber first. You could use a standard crucible. You could use a 10cc crucible, though obviously a 40cc would last longer. Are you going to use a thermocouple or just drive it directly with known power? Do not wind the filament all the way to the tip (or decrease the density of coils) to create a cold lip source. My initial thoughts is that due to the high temperature the filament alone would not be strong enough to hold the crucible, you probably need an additional support. Also, without cooling, the heat load will cause your MBE system to locally outgas.
Experiment and see…
I am a PhD student in Taiwan. I think I can give you some experience about Al crucible. It should be choose double-wall PBN crucible to avoid damaging Al cell when the Al crucible crecking. Also, as you know, we need to cool down very solwly. According to my experience, the cooling down rate between 860oC-450oC is 1 oC/min and 450oC-RT is 4oC/min. I bought the double-wall PBN crucible from CVT. Hope these can help you.
Our PBN crucible containing Ga cracked, so the liquid Ga ran into the effusion source. We tried to remove it by heating the source and turning it upside down for a long time. Indeed, some Ga was extracted from the source in this manner. We also tried to heat the source to 1100 degree celsius and more in vacuum in order to (hopefully) evaporate remaining Ga.
We then got a new crucible, and it seems we still have some Ga in the source. It could seem that this excess Ga is sitting (possibly incorporated?) on the filaments of the source, especially the top most part of the filaments, and thus reduces the resistance on this part of the filament.
We cannot heat the tip of the source well enough compared to the main part of it. For instance we have the main part at 1300 degree celsius (way above our normal operating temperature) now, while the tip cannot go higher than 1050 degrees.
See e.g. this picture of the source in action right now:
where it is only a few windings that are heating properly.
Do you have any advice on how to fix this?
It seems that our Ga flux has severely dropped, because the tip of the source is not properly heated, i.e. when we deposit with the main temperature at 1100 degrees, the tip is maybe 1000-1050 degrees. Is this “small” temperature difference enough to reduce the Ga flux drastically?
We are growing GaN, and had a drop in growth rate from 200 nm/h to 50-60 nm/h due to this.
The cracking of a crucible is somewhat inevitable. You should read my post
for some tips on how to avoid this in the future. As for dealing with your current problem: do you know the manufacturer of the cell? You could contact them and ask for some advice directly. I have never performed the cleaning myself, but I know it involves some selective etching and costs around €2000. If not you could contact MBE Komponenten
they repair third party cells.
Thank you for the fast reply!
Yes I have read your post 🙂 we are now always keeping the Ga liquid at 50 degrees celsius and we are very careful about how much Ga we fill into the crucible. We have been in contact with the manufacturer and got some advice. It is those that we are currently following.
It seems we cannot “fix” it by just heating it up. We are currently doing our masters thesis, so the time is limited, and we hoped to avoid having to send the source in for some cleaning. We will consider some wet etching procedure, I think.
Have a nice day.
Good luck. In answer to your other question, yes the hot lip temperature can have a significant effect on the flux.
Thank you very much. I enjoy your blog, you give much useful information on MBE, which cannot be found in most text books. It has been very useful for someone like me, who was new to MBE half a year ago.
If it is no bother, I wanted to ask another question regarding this hot lip on the Ga source, in case you know the answer.
What is the normal temperature relationship between the lip and the main part of the effusion source?
I am wondering if you should keep the temperature equal all over the source, or if you actually want the lip to be hotter than the main part?
What we have done, was to keep the temperature equal for both parts.
For Ga and In you want the lip hotter than the melt. The term is a hot lip source. The idea is to suppress the formation of small droplets which cause the infamous Oval Defects. In practice you cannot fully suppress them, but the hotter the lip the better. Most people use a delta T of 50 – 100 degrees.
Conversely for Al you want to keep the lip colder to suppress Al creep. Al actually climbs up the side walls of the crucible and flows out. The cold lip increase the viscosity and slows the creep. This is termed a cold lip source.
Thank you very much for the useful answer, it is very helpful. Hopefully we can make the Ga source work better again, and then we should try to make the lip hotter than the melt.
Dear Dr. Bastiman,
I am a PhD student in Taiwan. Our group have a problem on the cleaning the SUMO Mn cell. Our SUMO cell was coating the Mn because the crucible cracking…. Could you give me some suggention on how to clean Mn which is coating on the cell?
Or do you know any company can help this?
Another question is about N2 plasma gun, I have a veeco N2 plasma gun and its cooling water pipe is leaking. Is pissible to repiar it by ourself? Do you have any suggentions?
Thanks a lot,
You are not the first person to have a leaky N2 plasma cell, I would recommend it returns to supplier for repair.
Regarding Mn contamination of a cell, I would suggest you contact MBE Komponenten in Germany. They will repair any cell from any manufacturer and give advice. They also speak English. https://www.mbe-komponenten.de/
Hello Dr Faebian,
I am using an Riber 32P system equipped with In, Ga, Al, Si, Be, a valved As cracker cell and a valved P cracker cell. I am thinking of adding a Bismuth cell into the system using a single filament effusion cell. Could Bi contaminate other existing cells and cause major problem in the MBE system where you have problem growing the normal GaAs or InP materials? Could it increase the background/unintentional doping level of und-GaAs?
Loke Wan Khai
We never experienced any negative side effects of using Bi in our GaAs or InGaAsP systems. It actually rather difficult to get it to incorporate at all. It tensd to ride on the surface as a surfactant than incorporate as background doping or indeed intended doping.
Hi, I am a PhD student working on Arsenide MBE. Is it a good thing to evaporate AlGaAs composition to clean the chamber and reduce the background pressure of the chamber.
You can use a Ti sublimation pump to mop up general background contaminants.
When you oxide remove in the growth chamber you do add extra local contamination. This can indeed be mopped up by growing AlGaAs on the substrate. The oxygen binds to the AlGaAs interface, so you actually need to grow a super lattice of alternative AlGaAs and GaAs layers to trap the oxygen under GaAs and avoid it just riding up to the upper surface and contaminating your actual growth structure.
I do not have an As source in my MBE chamber and I perform CIGS growth on GaAs(100). Are there any protocol to deoxidize GaAs substrates without As without damaging the surface?
I answered your previous question. Try Ga assisted oxide removal.
Thank you Faebian.
Hello Dr Faebian,
I am using an Compact 21 system equipped with In, Ga, Al, Si, Be and valved As cracker cell. I have some problems with flux measurements. From the manual: “Depending on the gauge’s previous history, a gauge that has already been ‘arsenic-saturated’ or if Group III fluxes are measured last, the flux will almost immediately reach its equilibrium value or will slowly approach it.” Deviations in my system reach 1-2 orders of magnitude and equilibrium is reached within hours. This is normal for Compact 21? In my other MBE system, such problems exist, but they are much smaller.
All the best,
It always takes a little bit of time to “wipe” the other material from an ion gauge head and get an accurate measurement. You should minimize switch of material and get all the readings you want for one source before switching to the next. The time to reach equilibrium should be in the order of minutes in most cases, even for As. I have never used a Compact 21, however I suspect you need to replace your flux monitor ion gauge head.
Hi Dr. Faebian,
I am Yuanpeng from University of Michigan, Ann Arbor. I have a question regarding the beam equivalent pressure measurement.
If the two materials, say Ga and In, arrives the beam flux monitor (the ion gauge) with the same real beam flux (in atoms/cm^2), how much difference would the reading of the ion gauge looks like concerning that Ga and In are different materials and have different ionization energy.
I am afraid it is not so simple. T
You also need to consider the angle the cells hit the BFM. I seem to remember the In flux was 3-4x higher than Ga for the same growth rate on my system. Unfortunately you will need to calibrate the cells on your own system.
Thanks for the suggestion!
Hi Dr. Faebian,
First of all, thanks a lot for sharing this much knowledge and your experience with us.
I am a Ph.D. student. Now, after going through the Flux determination post, I can determine my III and V fluxes for any system in system independent unit. But what if one is working with TMDCs?
I think metal flux can still be determined in the same way as group III flux because transition metal flux is the one that determines the growth rate in TMD growth similar to group III flux in the III-V compound.
But I don’t have much idea about the chalcogen flux. Can we still use a reconstruction map for them and how?
I have no direct experience with TMDs, but the general principles apply. I do not know what compound you are growing, but taking WSe2 as an example… Can you grow 2D layers of this on something and observe RHEED oscillations? Is there a reconstruction map in the literature? Can you growth a well known II-VI alloy on a II-VI substrate like CaSe or MgSe? Can you deposit W and Se individually at 0°C onto your substrate for a few hours and measure the thickness with SEM (assumes unity sticking coefficient)? I hope this helps. Faebian
I would like to add some details that in our TMDs MBE system, we are also having the facility of a QCM (quartz crystal microbalance) which is placed very close to the substrate for determining the thickness of the growing film. Will this be helpful to determine system independent flux and growth rates if we don’t have a reconstruction map for any material?
Well, yes, a QCM is perfect. Since it is in situ, you can monitor deposition rates without the need for SEM. The same rules apply as for SEM layers… (1) you will be depositing amorphous layers, not crystalline structures, so you need to use the amorphous atomic density to calculate growth rate (2) you need to assume a unity stick coefficient for impinging atoms, but in reality this may now be the case, hence the flux may be under estimated.
Thanks a lot, sir for your kind suggestion!
Dear Dr. Bastiman,
When we talk about the growth mode, what makes it easier to get layer by layer growth in MBE than in the other growth techniques like CVD? Are the slow and controlled deposition rate and absence of any chemical reaction the only possible reasons behind it or some other feature of MBE?
Also, could you please comment on the factors that we should focus on during an MBE growth to achieve layer by layer growth instead of 3d island growth?
Thanks in advance for your kind suggestions and discussion.
I am no expert general 3D vs 2D growth, and different material systems all have their specific nuances. One of the advantages of MBE is the lower growth temperatures. If you tell me exactly what you are trying to grow on what, I can perhaps assist further.
I am trying to grow a few layers TMDs such as MoS2 with strict control over the layer numbers (from monolayer to bi and trilayers). To achieve it, the growth must occur in layer-by-layer mode. However, we often tend to get island growth. So the question is how an MBE grower can control it. Please assist if you have any suggestions.
TMDs are difficult to grow via MBE. As of 2020, there had been no real success. This is outside my area of expertise I am afraid
thanks for your comment Dr. Faebian.