The is a report on the recent visit to MIT/Lincoln Labs. The purpose of the visit was to assess Lincoln's capacity to fabricate CCDs of the types needed by Keck, and to discuss with Lincoln personnel some of the details of the desired devices. I have tried to include the views of the other members of the KCCDWG in this report. The visit occurred on July 7, 1994. The agenda was divided into three parts. First, we were given a presentation on the Lincoln Lab CCD program. Next, we were given a tour of their CCD fabrication facilities. Third, we discussed some of the technical details of the Keck CCD. The following people were the principle attendees:
|UCO/Lick||Richard Stover (KCCDWG member)|
|Bill Brown (thinning work)|
|Caltech||Jim McCarthy (KCCDWG member)|
|Rick Wilson (CCD packaging)|
|Shouleh Nikzad (JPL, MBE work)|
|UC Santa Barbara||Chris Stubbs (KCCDWG member)|
|MIT/Lincoln Labs||Dick Savoy|
Bill Brown, who will be doing much of the thinning work at Lick, was present to help assess Lincoln in this area. Rick Wilson and Shouleh Nikzad attended to help Jim McCarthy assess Lincoln's ability to produce the devices needed for LRIS, as well as DEIMOS. In particular, Shouleh Nikzad has been doing much of the CCD backside MBE work at JPL, and her presence helped in the discussions of possible Lincoln/JPL joint work in this area. At Lincoln Dick Savoy is the Microelectronics Group Leader and Bernie Kosicki is the Assistant Group Leader. Barry Burke runs the silicon fabrication line and does much of the detailed CCD design. Jim Gregory's expertise seemed to be in semiconductor fabrication. Note that Gerry Luppino did not attend this meeting because he is already very familiar with Lincoln Labs and because we wanted an independent check on Gerry's glowing reports.
2.0 The Bottom Line
Based on what we saw during the visit Lincoln Labs appears to have an excellent facility and competent people, and it was the conclusion of all of the KCCDWG members that Lincoln Labs is likely to be able to produce the CCDs needed by Keck. The remainder of this report provides some of the details which helped us reach this conclusion, lists some of the concerns we have, and provides a recommendation for the next steps.
3.0 What we learned about Lincoln Labs' CCDs
Lincoln Labs, although affiliated with MIT, is administered by the Air Force. Their primary mission over the years has been to develop technology for military applications. This continues to be their primary mission, but Lincoln has not escaped the effects of the cutback in military spending. This is probably one reason they are interested in our project. The Air Force requires Lincoln Labs to be strictly non-competitive. If industry can provide the devices we need, then they can't. If it turns out that SITe, for instance, can provide similar devices this restriction may still impact us. If the SSC approves further effort with Lincoln it may be important to move as quickly as possible.
Lincoln got into the business of CCD fabrication to provide imagers for use in surveillance from space and missile defense. Their first small CCD was produced in 1977. Their current project for the Air Force, and the largest CCD they've done to date, is a 1960x2560 frame-transfer back-illuminated device. This device just about fills the standard 4-inch silicon wafer. They are now in the process of thinning their first devices of this type. They didn't have a thinned device to show us.
Lincoln has done several focal plane mosaics. In one, they aligned the rows and columns of the individual CCDs to a few microns, but in a non-repairable, non-replaceable fashion. In another, the CCD co-alignment was not critical, but the CCDs were arranged in a non-flat focal plane.
Lincoln has done work on high-resistivity deep depletion silicon CCDs. High-resistivity CCDs can be made significantly thicker than standard CCDs. At the red end of the visible spectrum this has the potential to both increase RQE and decrease fringing.
Lincoln has developed very low-noise amplifiers. Their current designs are reported to be as low as 2-4 electrons at a megahertz read rate. Interestingly, Lincoln's customers apparently report the lowest noise figure, which Lincoln has not been able to reproduce themselves. Lincoln is also working on a non-destructive readout structure that may have even lower read noise. Gerry has a sample CCD from Lincoln. He reports less than 3 electrons read noise using the Leach CCD controller. The low noise at high readout rates could provide several benefits: 1) reducing the number of readout channels per CCD simplifies the controller electronics and reduces potential cross-talk problems between channels, and 2) higher readout rates could mean faster overall readout of the CCDs which translates directly into improved observing efficiency. These potential benefits have CCD controller implications that must be explored. Lincoln has been getting relatively good yields on finished, high quality CCDs. We should have a very good estimate of expected yields on our devices once Lincoln has had time to thin a reasonable number of their wafer-scale Air Force devices.
Lincoln has developed a stable backside treatment for CCDs. It is a high temperature boron doping process which they call their `refractory process.' We didn't get may details on the process itself (patents pending). But they did show us some results. It is claimed to be highly stable, with no UV hysteresis. However, the resulting RQE may not be as good as MBE processed CCDs in the far UV. They also have a more conventional boron implant with laser anneal process (Like Tektronix CCDs). This treatment is probably suitable for the DEIMOS CCDs, and it is compatible with hafnium oxide AR coatings. The refractory process is probably better in the UV than the implant, so would be better for Jim McCarthy's needs. Lincoln is willing to process some CCDs with one process and some CCDs with the other.
The anti-reflection coatings shown to us were designed for wavelengths below 3000 so the resulting RQE in the optical range was not very impressive. While they said they had done a little work with hafnium oxide, it wasn't clear they had used it on CCDs, or at least on CCDs that also use their refractory process. In fact, I had the distinct impression that the success of the refractory process depended to some extent on their AR coating, which is a simple layer of SiO2.
All of the KCCDWG members support continued work with Lincoln. However, there are a number of important issues we must address. None of these points are `show stoppers' but some may imply extra NRE costs.
4.1 Lincoln Technology
They started their presentation by telling us that Lincoln has no internal funds. Everything they do is funded by a project. They have no funding from other sources to develop any new technology. This is rather different than the other organizations we have dealt with in the past (Loral, Reticon, Orbit, JPL, SITe, etc.) where the development of the technology was driven by needs in addition to our own. On the one hand this could be good because we can get exactly what we want. On the other hand it means that we get nothing for free. Any deviation from their current process will cost us money. This will be important to keep in mind.
4.2 MBE Work
Lincoln has an MBE machine, so potentially could do MBE backside treatments similar to JPL. However, they have never done any work on silicon and this could be a costly development effort. We explored the possibility of using JPL to do MBE treatments. After further discussion between Jim McCarthy and Lincoln it appears that Lincoln might be open to shipping a few wafers to JPL for experiments. It's not yet clear how many wafers would be available for this work, since Lincoln is envisioning doing a single wafer run (about 20 wafers) and perhaps a backup run during the NRE phase.
4.3 Anti-reflection coatings
As stated above, Lincoln may have to do more work if we want them to provide good anti-reflection coatings for visible wavelengths. It is not clear if this work was included in the original NRE estimate from them. I think not. The small number of wafers available for this work may be a more significant problem, and Lincoln has suggested putting some small CCDs on the wafer, for AR coating experiments. The cost of designing the small CCDs would no doubt be included in a revised NRE. In the end we might decide to develop the AR coatings ourselves, since this can be done on any thinned CCD.
4.4 Mosaicking and packaging
Lincoln's original NRE estimate did not include any development of a package for the individual CCDs, nor any development of the DEIMOS mosaic, nor any fabrication of the packages or mosaic. Jim McCarthy wasn't impressed by the mosaics Lincoln showed us, and Gerry Luppino thinks the cost for Lincoln to do the design and fab would be very expensive. Some additional investigation of this point needs to be done with Lincoln, but it may be best for us to design and fabricate packages and mosaics ourselves. Also, if Keck wants Lincoln to develop the mosaic then Keck might have to support the entire cost of this activity if the other NRE consortium members are not interested.
4.5 Consortium members
Lincoln must get approval from the Air Force before they can proceed. Lincoln felt that approval was likely, but that the foreign members of the consortium might cause a problem. Lincoln has supplied CCDs to foreign customers in the past, but under restricted conditions. For instance, they supplied CCDs for the Japanese X-ray satellite payload. But they supplied a finished, sealed camera that was never opened in Japan before the satellite was launched. We don't expect such restrictions to be placed on us, but it is a possibility the SSC needs to be aware of.
I'm not sure Lincoln can meet our low fringing amplitude specification. At the meeting they said they didn't know if they could meet the requirements because we hadn't specified the bandpass of the illuminating light. This is a valid criticism, and one we will correct. But still fringing is not a problem ever addressed by Lincoln before and I don't think any NRE effort was figured for this. There are certainly no experiments with high resistivity CCDs envisioned in their original NRE.
It is the recommendation of the KCCDWG members that the SSC approve continued work with Lincoln Labs to develop our CCDs. This approval must be contingent on SSC approval of the NRE consortium as set up by Gerry Luppino. Upon approval by the SSC Lincoln will send a letter to the appropriate Air Force office to request their approval of the project. It takes about two months for Lincoln to get a response from the Air Force, so it is important for Lincoln to get our approval as soon as possible. While Lincoln is waiting for the Air Force reply we will refine our CCD specifications including packaging and other requirements, and Lincoln will be preparing a revised proposal based on our specifications. As soon as they receive Air Force approval Lincoln can formally present the new proposal to us. At that time the SSC can review the proposal and make a decision regarding participation in the NRE phase and the expenditure of funds. We expect each consortium member will have to provide about $100K towards the effort. The NRE phase of the work should take about one year from the time the contract is signed.