Faebian Bastiman
Keeping a journal of your weeks activities may seem like an monumental effort, but personal i found the experience very insightful. It enabled me to realise where I have been spending my time, wasting my time and what needs more attention. Let’s summarise the individual days: Monday, Tuesday, Wednesday, Thursday, Friday.
Firstly working in a clean room is thirsty work; it is in fact similar to spending that day on an airplane. You can expect your body to require an extra litre of water compared to office work. Failure to realise this and keep your fluids topped up can lead to dehydration and a string of related problems.
The next thing we notice is that the days comprise routine tasks. Every day is based around:
- Drinking water
- System checks
- Growing sample 1
- Growing sample 2
- Transferring samples
- Loading/unloading samples
- Updating paper work
- Preparing samples for characterisation
- Planning next samples
- Analysising and summerising recent samples (includes preparing results for publications)
- Software and system development
System checks are essential. The pumps, gauges, vacuum and fluxes should be checked daily. Auto logging and flux collection could be utilised to speed up the whole process. This would be particularly useful for the flux checking, which currently consumes just over an hour in the morning, and would be longer when checking more sources. Automatically running these checks at 0700, and having the system ready at 0850 to start growing the first sample would allow 3 samples a day.
This work was performed on a Oxford (VG) V80 system circa 2003. The sample transfer is manual (sadly) but not too cumbersome. Of course the sister system (the V90) possesses automatic sample transfer, pump down/vent and batching. In this particular case that would save us around an hour a day. This particular project requires constant RHEED analysis and so batching will not free up that much of our time either. In order to really save time on the sample transfer we would need: auto transfer and auto RHEED capture (jpg). In that way we could run 24/7. The current throughput peaks around 0.25 samples per hour, 2 samples a day. With automation we could get 0.5 samples per hour, 12 samples a day… however, this is an absolute maximum. Honestly after growing 12 samples in one day, one would require the next day to do the analysis.
Note that preparing to present result at a weekly meeting was perhaps the most useful exercise of the week. The detailed analysis enabled new insights (and in fact led to a publication) that would otherwise have been missed. There is a tendency to analyse productivity on throughput rather than worth. Sitting down and looking at what you have is certainly worth it. The advantage of automation is therefore to free up time for analysis. Note that the analysis was actually performed post work on a Wednesday evening in this particular week. That is a feature that is all too common. Let’s try to get ourselves down to a productive 40 hour week.
10 samples in a 40 hour week from a lone operator is about all you can expect from a manual machine. Having two growers on that manual machine could enable 20 samples in 80 hours, but then forces the growers to endure the “unsociable” hours of shift work with no compensation. If you would like more than 10 samples a week consider buying an automatic machine. Automation enables working smarter, not harder. A lone grower can easily grow 60 samples in a 40 hour week, feeding multiple projects. Or, alternatively, 5 growers could grow 12 samples a day and share the automatic system for one day a week each. This also enables the grower to do his/her own characterisation and analysis. The “samples” I mention are characterisation test structures and are typically only 1-3 hours each. Should you require 8-12 um thick pins or DBRs you can of course only grow 2 samples a day by MBE. MBE is limited to around 1-2 um/h, and strictly the best quality growth seems to be at 0.57um/h (for some reason).
Typically, a research topic will require anywhere form 5 to 30 samples per publication. That is around 1 to 3 weeks of growth, plus a whole week of analysis. Running an MBE system at a 2 sample per day pace may seem unproductive at first glance, but if you take the care and effort to ensure every sample contributes to the topic (i.e is logical, systematic and provides a valuable data point) then the result is 12 peer reviewed journal publications a year, which in is in fact very productive indeed.
Updating paperwork is also essential and should be performed meticulously. A growth sheet, run log and maintainence/problem log are mandatory. The growth sheets needs enough information for any one to be able to repeat the growth recipe with no prior knowledge. The run log is a chronological list of samples grown, with some fundamental growth parameters, characterisation parameters and importantly the final resting place of the wafer (which will hopefully be a wafer drawer close to hand). The maintenace/problem log simply tracks anything and everything out of the ordinary with regard to the system’s operation, it then represents a list of things to repair or improve in the next maintenance and development cycle.
Characterisation and growth are intimate friends. The best growth experiences of my life involve working in a grower-characteriser pair with another researcher. One person grows the samples, one person characterises the samples, the two discuss the results and plan the next samples. One produces growth papers, the other characterisation/device papers. Both share the publications (first author, second author switching) and everybody wins.
Finally we have system development. No system is perfect, not even a brand new system that you personally specified (though admittedly that should be close to perfect). There will always be additions and improvements. Technology is always advancing and that includes all aspects of MBE software and hardware. Perhaps I “shake my fist at the heavens” more than most during my MBE operating day, but noticing the parts to improve and actually spending the time to improve them are essential. The system should be working for us and not the other way around. If your system does not do what you want to, change it. If you do not know how to, ask. I would be delighted to discuss improvements and implementation thereof with you.
Dr. Faebian Bastiman 