This is my first note on stuff related with my research on molecular beam epitaxy or MBE for short, a technique I will be using for next 3-4 years ahead to make epitaxial grown gallium nitride nanowires on silicon.
Ok, where should I begin? To describe it in simple way, MBE takes place in really huge machine (compared to other techniques like metal organic chemical vapor deposition). One advantage of using this technique is the higher purity of the grown material compared to other technique or in fact, the highest purity quality that we can get. This fact is supported by the condition that need to be maintained inside the MBE machine itself, so-called high vacuum to ultra high vacuum pressure. As an illustration, we live in the atmospheric pressure, where this pressure is equal to 1.000-1.000.000.000.000 lower than atmospheric pressure. It implies the number of contaminants is getting less and less as the pressure goes to ultra high vacuum. Actually, ultra high vacuum provides “the straight pathway”, where intended gas can travel long enough reaching surface of the material without any collision with other gases.
Growing epitaxial layer on top of the substrate can be depicted easily with lego blocks. Imagine you have green big plate and you want to make something on top of it, say a house (Figure 1). The lego house is made of many block components and colors. It can be a 2D house or detailed house, depending the creativity and what passion one has. The same thing goes with MBE.
Now think like this:
Green big plate = substrate
Block components = gas
How the machine looks like, I will write another post for it. This method of growth, where you start from the scratch of substrate where there is nothing on top of it and putting one by one intended component on the substrate, we call it as bottom-up method. Another method is called top-down method, which is similar with carving a statue from single block of stone. Bottom-up method requires more time than top-down method, but guarantee better structural quality.
Back to the title. Honestly, this is my first time for me to be trusted with one component of MBE. Since my research is gallium nitride, it means that I need nitrogen to be introduced in the process later on, and the MBE in my professor’s lab is introducing nitrogen by using a plasma.
This end of June was the second opening of the MBE made by Veeco, namely GEN930. Opening in here was done because the some sources (for gases) were nearly finished and repairing electronics of the substrate holder. After opening was finished, then we baked (put lot of panel on it, basically cover the whole surface of the MBE system) it for nearly one week. The baking was done to reduce as much as possible unwanted particles (because of the opening: exposing some part to the atmospheric pressure) and make sure the pressure can go down low enough to ultra high vacuum.
At first week of July, we turned off the baking mode and removed the panel. After connecting all necessary cables, the source outgassing part was done. Once again, the outgassing is required to make sure that the contaminants are minimized to the lowest point. Afterward, dedicated baking for the nitrogen line source was carried out, but no big panel was required. We simply cover it with conductive heating element. After another two days, the baking was finished and it was removed. We need two weeks to do all of these work. Finally, the last step must be done before getting the MBE ready to work: source calibration.
Before doing calibration, my professor would like to test the plasma source, since there was a change in aperture plate (nozzle). The reason to replace with the new one was the attempt to get high brightness mode of nitrogen plasma. This mode is required to grow gallium nitride efficiently. Honestly, I have not tried the growth with low brightness mode of nitrogen plasma source. Probably it would not be appearing as I wanted.
However, there were two main problems which we encountered. First, as we turned on the chiller (for cooling the plasma source), my friend noticed a leakage on input and output of the hose, from the autotuner to the plasma source. Not only leakage in that part, but there was noticed as well on the OD stainless pipe in the plasma source. We discovered the caused which was deformed O-rings, which we forgot to take them out before baking of nitrogen line source. After I got the required O-rings from workshop near my home (great exercise though ^^), we managed to install them and run the chiller with the cooling system not being leaked.
Now, the second problem. We got the plasma at the first time (mediated with argon gas owing to its lower ionization energy than nitrogen gas) and after decreasing the power, suddenly the plasma was off and no plasma was observed since that time. It took about 5 minutes for us to recognize a burnt smell on the autotuner unit and the worse thing was the accumulated heat on the surface near the RF shield. My friend was afraid that he broke the autotuner, but discovered afterward that the smell actually came from the remnant of the leakage liquid that had happened before. The heat was found to be high inside the RF shield itself.
The day ended as my friend sent an e-mail to Veeco engineer, describing the situation. Next day, my friend took a day off and I was the only one left in the lab. The Veeco engineer replied the message with several questions, meaning that I was the responsible person to answer those questions. I was not really familiar with the things like required flow rate, cooling flow direction and RF reflected power, so I needed a bit of time to adapt with it. It took me almost 2 hours to understand the questions, compose the answer and send it back to him.
After I had replied the e-mail, the lab engineer came to the lab and I forwarded the message from my friend. He suggested to clean the copper electrode inside the RF shield, since it was oxidized during nitrogen line baking. As a note oxidized copper is insulating. He used sand paper to remove the oxide from copper electrode. He also explained another electrode which connects to the copper is made from silver and oxidized silver is conductive, so he need to clean the copper electrode only. After putting everything back, we tried to ignite the plasma, but we failed. Still, there was abnormal heat in the same place.
Returning back to my office, I received an e-mail from the Veeco engineer. We talked a bit over the phone and he asked me to send a graph of the forward power, tuning cap and load cap over time.
The day after, I did as he requested. Based on the user manual, I tried to re-ignite the plasma and oberved the important parameter of the graph. Afterward, I sent the exported data of the information, for the two days measurement duration. It took no more than 1 hour for him to analyze the graph! That was such fast response, especially during lunch time. He called me once more and suggested to me that the chiller temperature need to be lowered. He also sent me the procedure to do regarding the difficulty in igniting the plasma.
Well, I was bit skeptic with the new procedural document. In that time, my friend who took day off the day before, coming to my office and encouraging me back to do the new step. I was surprised as the first time and so did my friend, that the required flow rate of the nitrogen was really high, 4 times higher than the highest flow rate we used in the previous experiments. We were afraid of having much high pressure as a result of massive nitrogen flow rate in the plasma source chamber. However, the document said the limit that should be kept an eye for, so we did that and it was fine actually, the pressure still in the safe zone.
After about one minute introducing such high flow rate, I stood close to the autotuner and heard a motor moved. I did not know what it mean though, but I saw a slight RF reflected power changed. I went to the computer and observed the graph. There was a slight change in tuning and load cap over time. Without any further second thought, I peeked the viewport and voila! The nitrogen plasma source was successfully ignited! You can see how the nitrogen plasma looks like in Figure 2.
I can’t describe how happy I was that time. Probably it can be described like an adventurer who finds a treasure after several days wandering in the desert. I let the nitrogen plasma on for one hour, to make it stable. At the beginning, I feared of the unusual accumulated heat in the RF shield. I monitored the heat in every 5 minutes. Surprisingly, the RF shield went warm at stable temperature. A second happiness 🙂
I stayed in the lab for another two hours, just to make sure that I can re-ignite the nitrogen plasma again and again and again and again… Well, several times. The nitrogen plasma was actually reproducible and I can play a bit with the forward current and flow rate from the set point, running from high brightness mode to low brightness mode.
That Friday evening was one of the best Friday I have ever had in my life :))