One of the most overlooked outgas items are the sample heaters: the outgas stage in the preparation chamber and the manipulator (sample heater) in the growth chamber. Let’s review: You have just opened the entire system to atmosphere, replenished the source material and fixed the many little problems, pumped the system down and baked it and finally outgassed all the sources (see Post bake tasks: Cell outgas). After that you put in your sample, heat it up to oxide remove and grow, right?
Wrong! The heaters need just as thorough an outgas as you would give the cells. Let’s face it they are much closer to the substrate. Failure to outgas the heaters will delay your return to optimum material quality by a week or more. A thorough outgas will mean your first sample is already 90% of the way there. The second sample should be 100% of the quality of your best before you came down for maintenance. So how do you save a week or two at the start of your growth campaign?
First, load a sample plate (platen) with either a 2” Ta disc or a 2” Si substrate in place of your usual III-V substrate. These dummy samples are very useful as the temperature limit is >2000°C for Ta (>1200°C for Si), and therefore way beyond the range of your III-V heater. The procedure to outgas the heaters is nigh on identical to that established for outgassing new, cleaned platens (see Essential Maintenance: Sample holders). In brief all you need to do is load the dummy platen into the heater stage and approach the maximum temperatures, gradually, and monitor the background pressure. Keep the pressure in the low 10-7 mBar range. If the pressure rises too swiftly back off the temperature and hold it until the pressure recovers. Aim to hold each heater stage at maximum temperature for an hour. The process may take a whole day, though you can run it in parallel with your cell outgassing so no real time is lost…and viola! An instant return to fully quality. This procedure was used to produce both excellent mobility in InSb on GaAs and excellent optical quality from InGaAs SQW in a GaAs/AlGaAs DBR.
Suddenly the dreaded “MBE downtime” seems a lot more manageable.