As covered in the section comparing the leak valve to the current jet the gas load on the system was only about 4 times larger than using the jet. Therefore the skimmer in conjunction with the cryoshroud pumping can be eliminated if the jet is moved as close to the substrate as the current leak valve. Note that the only purpose of the current skimmer and pumping arrangement is to reduce the gas load on the system, since the skimmer is placed in the molecular flow region of the expansion acting as a collimator, as discussed in the concluding section on jet theory. Therefore it can be eliminated without any problems. In the worst case a jet without skimmer and cryoshroud will put the same gas load on the system as the current leak valve, provided the same distance to the substrate is maintained. Some improvement is expected though due to the "jetting" action discussed in the concluding section on jet theory. Further the can and flap are unnecessary and undesired in this arrangement.
The advantage of eliminating the skimmer and cryoshroud is manifold. Beam attenuation due to a large background pressure P_b is completely eliminated since it was shown experimentally using the leak valve that the background pressure in the system is better than 3x10^-8 torr. This is due to NH_3 immediately sticking to the large cryoshroud in the growth chamber without bouncing around.
Next, NH_3 dissociating causes less of a problem in terms of beam attenuation due to an increase in background pressure. This is due to the fact that N_2 is pumped well by the cryopumps in the growth chamber. The only concern is H_2, which in the current jet is nearly expended completely into the growth chamber as well. Since no dramatic increase in pressure was noted this should not be a problem. Due to the larger volume of the growth chamber H_2 attenuation will also be minimized by orders of magnitude (neglectible).
NH_3 dissociating on the heater wires is also totally eliminated, since NH_3 missing the target (or bouncing off) will travel in a straight path to the growth chamber cryoshroud and stick to it immediately.
Also, the huge simplification of the design leaves less room for contamination and malfunctions, and performance should be much easier to characterize than the current jet due to the removal of many unknown factors, especially the hard to determine background pressure between skimmer and cryoshroud.
The only concern might be molecules backscattered from the substrate into the incoming "beam", but this effect should be much smaller than with the current skimmer arrangement.
The jet suggested is shown in Figure 16, which fits easily in any source port. There is a central feedthrough for the gas supply as well as 2 feedthroughs for the thermocouple and heater each. The nozzle reservoir suggested is made out of PBN with an integrated heater and slide in thermocouple, allowing for high temperatures to be obtained.
In summary there seems to be no disadvantages of designing a jet without skimmer and cryoshroud and moving it closer to the substrate, only advantages, compared to the current jet(s).
Figure 16: Sketch of Suggested Simplified Jet
Table of Contents