NanoWiki

It should be possible to sputter MoGe in the ATC. Also Nb Tc should be > 8K. NbN can be better. These issues should be investigated.

Good Au recipes are available in ATC. Check dependence of Au growth on some parameters. Hypothesis of flattening by O2 because of 'reverse sputtering' might be unlikely but this can be tested in the ATC. Try DC bias. I think adatoms might be important.

This process is not working well due to a change in developer concentration. Buy old developer. As a patch an O2 descum process is used to remove resist 'tails' but a good development process is much better.

RIE etching opens lots of new possibilities. Develop recipes for Nb, NbGe, MoGe, NbN, etc.

Stripping of resists is problematic due to the absence of a 'bain Marie'. Tamson heater is available. Build something to make strippers boil. (typical strippers form highly flammable/explosive mixtures at/ slightly above the boiling temp so a hotplate cannot be used, acetone vapour sinks below the hotplate where temp is higher…) Baths must be covered since boiling acetone evaporates very fast.

Enter measured thicknesses immediately in wiki.

msm09 should be centrally administrated. Access from outside to some parts would be handy. There's a lot of important information there.

For epitaxial Au growth it is beneficial to have a well defined pretreatment (degassing) of the mica and controlled cooldown. Temp controller does not provide this internally but is remote controllable by 0-10V signals. A National Instruments usb6008 USB DAQ module is available. Maybe signals have to be scaled with an opamp, this can possibly avoided. Code needs to be written so that a temp profile can be programmed. Labview would be good, can it run on the Dell PC that is available for this?

The oxide sputtering system has a control system that contains an 8 input datalogger. Now, 3 channels are used. Wiring for these 3 channels should be improved by making converter cables to BNC. 1 more pressure gauge can be logged, a converter cable must be made. The 4 remaining channels can log voltage and current of 2 power supplies. Check signal levels before connecting.

Drive the heater feedthrough by a belt (buy/machine sprockets) connected to a stepper motor via a gearbox. Mount limit switches.(where?) The DAQ box needs to be extended with a digital IO module (no problem). Requires considerable software development.

Buy 2 turbine flowmeters to interlock the sources. Use a leftover rotameter to protect cooling spiral. Machine a panel for the 'old PC hole' in the 19“ rack (paint black), mount turbine meters there visibly. 18 MOhm trip relay (floor water sensor) does not work. Wiring or relay? There is a spare relay. Fix. Connect relay also to water sensor in feedthrough. Interlock heater.

Automate knob to sputter w. interrupts more reproducably / less annoying. Software available. Buy (very strong) steppermotor + controller. Measure required torque first (>10 Nm).

Buy damped pneumatics for shutters. Interface AE supplies to PC. Build shutter (design sketch available). Mount steppermotor (already bought). Develop software to automate everything.

New and improved Eurotherm software is available, should be installed.

Install better thickness monitor.

Buy parts / build parts to enable sample transfer / tilt / rotation in tili plane / heating

Mount e-gun in bolt-on chamber (expensive, not urgent)

Build a better loadlock and transfer system, allowing shadow evaporation. Expensive.

Frame should be mounted to floor or wall. According to the manual of the turbo the 1600 liter pump with several 10's of kg of spinning mass generates a torque of 4500 nm when crashing. Extra brackets to mount vacuum chamber on frame should be built.

The etch gases should be injected in the exhaust, as far as possible downstream, with a reliably mounted construction. N2 flush should be connected to each stage of the 2 stage rotary pumps. Pressure? Flow: stay below LEL, reccomendations from Leybold?

Right now there is no 'refrigerant temperature high' protection, the H20/ethylene glycol can boil if cooler is switched of / fails.

The RIE etcher sends a 0-10V signal to a chiller connection and expects a 0-10V signal back, these signals correspond linearly to a temperature (1V=0C, 10V=90C) (right now the output voltage is looped back to the input, resulting in 'instant perfect control'). The chiller does not have a connection for these signals / regulator with such a capability. A process regulator with 'analog process variable retransmission' and 'analog remote setpoint in' is needed. The regulator needs to read the temperature from a pt-100 (temp coeff 0.00385, 108 Ohm or so at room temp) and control a solid state relay with voltage pulses (3-30V).

Mount Cl sensors (www.draeger.com) in cleanroom, 611 and in ventilation system. Install trip level relays (for critical applications, can't have enough approvals). Place an Cl2 compatible electrical valve (20 bar working pressure, 200 bar rated, normally closed) Implement a circuit with the following logic (independently):

1. if power down: close Cl2 valve, switch off etcher via PL7 interlock (2 normally open relay contacts)
2. if fire in 611: close Cl2 valve, switch off etcher via PL7 interlock
3. if flow < 10m3/h (that's 30 Pa underpressure or so): close Cl2 valve, switch off etcher via PL7 interlock
4. if Cl2 > 1ppm in a room or 3.5 ppm in line: close Cl2 valve, switch off etcher via PL7 interlock
5. if Cl2 > 15 ppm page fire brigade
6. if Cl2 > 50 ppm in a room: trigger fire alarm, evacuate HL

The beam stabilizer electronics works fine but the software does not work with patterns that regularly blank the beam and has a long built-in time delay after a stage move that makes it not work for some job lists. It also appears that the regulation parameters are valid only in a part of the range. After a stage move (regulation has been off) the accumulated error should be eliminated rapidly, now it takes 40 sec before anything happens. Above issues result in sample loss, sometimes after a lot of processing. Perhaps we should replace the programmable AD/DA chip with an old PC, install linux and connect it to a USB-DUX.

The regulator should employ adaptive control. This can be simple because when the gun alignment is adjusted properly (and this is normally done before writing) the current should increase monotonically with potentiometer setting/voltage and therefore the error function is convex. For small errors the error function is a parabola and regulation can be done in 1 iteration. Drift during a stage move can be extrapolated from the drift just before the stage move. Before/after a blank the time constant of the current source (constant) should always be taken into account.

Larger errors can be corrected with 2 or 3 iterations of Newton's method to cope with the varying and drifting derivative. Newton's method generalizes the parabola method above. For large errors damping should be used (it's a good idea anyway because Newton's method does not necessarily converge) so that large errors will be corrected with a bisection method.

The motor stage of the e-beam has an accuracy of 1 um after a move, several um accumulated after many moves. Accurate stage movement in closed loop would greatly improve alignments. A piezo substage with laser interferometers for x and y would accomplish this. See i.e. http://www.renishaw.com for laser interferometers (can those work in vacuum?). Motorization on Z would be a big improvement as well. It does not need to be closed loop, accuracy: 1 um. This would allow focussing with the stage motor. (focussing with the focus knob changes the objective lens field gradient and therefore the objective lens axial field, creating image rotation, so that alignment between stage motion coordinate system and beam scan coordinate system is lost; moreover the raith software can perform a 3 point plane correction for a tilted sample).

The following equipment would make sample fabrication much easier and efficient:

• Good optical microscope for cleanroom (with CCD)
• Profilometer (veeco dektak, alphastep etc.) (can be combined with above)

We could look for a used profilometer that's still in a good state.