From jmartin@nuclear.uwinnipeg.ca Wed Sep 13 17:30:46 2006 Date: Wed, 13 Sep 2006 17:30:45 -0500 (CDT) From: Jeffery William Martin To: d.dutta@msstate.edu Subject: Re: diamond (fwd) Hi Dipangkar, Well I talked to William Trischuk (U Toronto) today. This guy is very reasonable and very knowledgable. I also talked to my U Winnipeg colleague Steen Dannefaer (who is a solid state physicist - not a detector person). As it happens, Dannefaer is an expert on diamond as well as silicon. So both meetings were very profitable! Dannefaer told me that diamond is just like HPGE crystals. You just put some electrodes on it and it will sense the charge. He said there is no such thing as an n-p junction. It is not depleted in any way. You just have an electric field on it in order to collect the charge. Another thing I learned from these discussions is that polycrystalline is bad. Dannefaer said that any "grain boundaries" are where charge gets trapped and hence electron-hole recombination occurs there. Trischuk told me the thing about grain boundaries as applied to planar detectors is that for thin detectors it usually doesn't matter - the grain boundary is parallel to the interface, and the charge deposition region is often of order larger than the grain boundary. Dannefaer also stated that charges tend to move along the grain boundaries, so it would make sense that generally it wouldn't affect planar detectors. However, Trischuk said that research done with Element6 showed that the grain boundaries are worse on the side of the detector where the CVD was first deposited (on the substrate). In subsequent and later layers, it gets better and better as the layers become more and more uniform and single-crystal-like. In collaboration with Element6, they developed a technique by which they machine off the diamond from the initially-grown side. I asked, "So if I ask Element6 for detector-quality diamond, what do I get?" His response is that he doesn't know but likely you'll get something good. Since a lot of the techniques are proprietary, Element6 won't even tell him what they do. Likely it is some combination of techniques that RD42 helped develop. Dannefaer mentioned that single-crystal diamond also exists - he even has unmetallized samples that he showed me! Trischuk stated that the biggest single-crystal he's heard of is 0.5 x 0.5 cm^2. So 2x2 cm^2 is likely out as far as that goes. But he said not to worry too much about it given that polycrystalline (again not "too" polycrystalline) is generally good. Dannefaer even HAS single-crystal diamond (unmetallized) which he showed me. Regarding metallization, both Dannefaer and Trischuk recommended not doing R&D on it. Trischuk said they were torn over having Element6 vs. the bump-bonding company doing the metallization, because the bump-bonders are very particular about surface quality. (I believe bump-bonding is a process for doing pixel detectors, connecting a pad through an insulating layer to a wire - but I'm not sure.) He said for strips and single-element stuff they generally use standard Au or Al wirebonding. He said he didn't know if TRIUMF had a wirebonding machine we could use - I said I was working on it. He implied that if you're doing wirebonding, there's no special surface quality required. Re using "standard silicon electronics", he said this is generally fine. He typically uses 1 us shaping time. I said I'd like to push it lower. He said that's generally ok, but that noise would become a problem at some point, especially since for diamond there are fewer primaries. Now of course you can just make the detector thicker, but it gets more costly. He also said there are some potential advantages to diamond in this case because there is no "leakage current" like there can be in silicon. For this reason, you can use e.g. DC-coupled preamps instead of the usual AC-coupled ones for silicon. Or at least you don't have to compensate for detector leakage current increasing with time (as LHC people do for silicon). To summarize: you can use "standard silicon electronics" but probably you can do a bit better if you spend more time on it. He suggested looking into chips designed to be used with strip detectors. He knows more about chips for pixel detectors, but he advised not to go there: the strip chips more mature (6 years old as compared to 6 months old). Trischuk said that Element6, while not the only company, is the one that's been in the business the longest. Trischuk estimated $1000/cm^2 as a typical diamond cost. It sounded like he had both 700 um detectors and 500 um detectors. Re: interstrip dead regions - Trischuk said this is usually more of an issue of charge-sharing between strips. He said there is usually also some charge loss - he estimated 10-15%. Perhaps the issue I've heard of is therefore more one of having electronics smart enough to deal with charge sharing. He said some complicated stuff about how he usually has a strip pitch of 50 um, but it's always some kind of debate how wide a strip of metal you put down - whether 40 um down to 15 um or whatever. I didn't fully understand what the tradeoff was. But it sounded like it had to do with similar things as a wire chamber - equipotentials in the diamond etc. Anyway, based on these discussions, I'm even considering asking Dannefaer to sign up on the proposal! (That way we'll look like we know what we're doing since he is Mr. Diamond.) I should have asked Trischuk if he could lend me a detector, but I didn't have the guts. Maybe I'll call him back... or email him... after I've had a chance to think a bit more about things. Jeff -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Prof. Jeffery W. Martin Phone: (204)786-9443 % % Physics Department Fax: (204)774-4134 % % University of Winnipeg URL: http://nuclear.uwinnipeg.ca % % 515 Portage Avenue % % Winnipeg, MB R3B 2E9 CANADA % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ---------- Forwarded message ---------- Date: Mon, 11 Sep 2006 12:30:35 -0400 From: William Trischuk To: Jeffery William Martin Subject: Re: diamond Hi Jeff, All good questions. I think we have reasonable answers for most of them. I wish I had a single reference I could point you to, but I don't. Perhaps a phone call would be the best place to start. I'm in my office most of the rest of the day today. Tuesday's are full of teaching and my research group meeting/lunch. So the next chance will be Wednesday. In general I can usually be reached at 416-978-8095 between (9:30 and 4:00 weekdays -- except Tuesday's). Of course your mileage may vary as I will be out talking to grad-students etc. But should be able to get back to you fairly promptly if you don't reach me. William Jeffery William Martin wrote: > Dear Prof. Trischuk, > > I'm considering diamond as a technology for a strip detector for a > Compton polarimeter project at Jefferson Lab. > > I've worked with silicon before, and I was wondering if you could help > me with a few questions and concerns I have about migrating to diamond. > > I want to build four planes of 2 cm x 2 cm detectors, with a strip width > of 200 um. Per-plane thickness of 500 um. The detector has to be > placed about 5 mm from a 1 GeV, 180 uA electron beam, which is primarily > why I'm considering diamond over silicon (rad hardness). The per strip > signal rate could be as high as 100 kHz, and I want to count at that rate. > > My questions are, for diamond: > 1. What are the timing properties like, what's the energy resolution like? > 2. Are there interstrip dead regions and how big are they? > 3. Will typical silicon electronics generally work with such detectors? > 4. Dealing with diamond companies, and general fabrication issues. > 5. Any other tips you might have. > > Could I get you on the phone sometime to discuss it? If so, please > suggest a time for me to call. > > Regards, > > Jeff Martin >