Chip carriers and sockets for low temperature

[hunt137]

I'm looking for chip carriers and sockets that will work well down to 4 K. Most aren't rated below -70 C, but I know in the past I've seen sets used down to 1.6 K. (Wish I could remember what they were!) Does anyone have specific recommendations? Or general advice, plastic vs ceramic, etc?

I've never purchased anything like this before, so I'm especially interested in carrier/socket pairings. (I guess they are never sold together...)

Any advice appreciated. Thanks!

 

[hunt137]

The quest continues.

I've been recommended to check out CMR's LCC carriers: http://www.cmr-direct.com/en/sample-mounting/cryogenic-sample-mounting/mpuck-system/mp-llcc20-nm The guy who recommended them says they work well in HV and down to cryo temperatures. However, he recommended against their "mP-body" mounting system because they seem to have outgassing issues.

So I'm still very much interested in advice on sockets that have good thermal contact and a comparable thermal expansion (to maintain good electrical contact). Or other mounting systems free of significant outgassing issues.

All advice appreciated, thanks.

 

[f95toli]

Most LCC sockets will work quite well.
However, so will just about everything else. We bond most of our samples to carriers we made ourselves; they are just small PCBs made from standard FR4 laminate. There are plenty of companies that can make these for you for a very reasonable cost (we typically something like £50-70 for maybe 25 holders) if you send them the Gerber files (you can use a free version of say Eagle for the design).

Hence, unless you have very specific requirements (low outgasing, non-magnetic) you can usually just buy something from Farnell.
I wouldn't bother with the CMR carriers unless you REALLY need non-magnetic holders, the latter is why they are so expensive. You can get identical carriers for a much lower price if you are prepared to live with the fact that they were made using a nickel sticking layer underneath the gold; unless you are doing an experiment where you are using lots of magntic shielding the Ni is unlikely to affect the results.

 

[hunt137]

Hi f95toli, thanks for the reply!

Good call about the CMR carriers. At this point, I don't think I'll need non-magnetic carriers.

I decided to go for a carrier/socket solution because I want to be able to switch in and out samples quickly without disturbing them or having to re-bond them. I need something low-outgassing because I'll be working in vacuum on a cold finger. So I'll need good electrical and thermal contact through the leads. (I'll possibly also be working in an He exchange gas environment at some point.) You can see how mismatched CTE's between the carrier and socket would be a problem. Anecdotally, I know of at least one case where PLCC's were a problem in an He exchange gas system. (As far as I can find, Farnell only carries plastic sockets.)

You mentioned you connect your samples directly to PCB's. How do you electrically connect the PCB's with your system? I'm open to alternatives to carriers/sockets.

 

[0xDEADBEEF]

I don't know if thermal expansion is really a big deal, contacts can bend. The main problem is, that all plastics become brittle and many crack. Normal epoxy board should be ok, the glass is a tough reinforcement layer. If you stick solderable sockets through them you can make your own carrier. POM is good, Torlon also there is a number of good plastics. But if I were you I would just take a normal DIL chip carrier, and dip it in liquid helium a few times. Chances are that you'll be fine. If all else fails take a DIL chip carrier, solder it to a PCB and remove the carriers plastic. Out gassing becomes a problem when you do surface physics. I am no expert in that, but just high vacuum with a few millitorr will be no problem.

 

[f95toli]

How do you electrically connect the PCB's with your system? I'm open to alternatives to carriers/sockets.

I varies. Sometimes we just use normal sockets, if you only need say a twisted pair you can always cut a DIL socket into pieces since they are stackable.

I do quite a bit of work with microwaves, and then I just solder/screw the SMA connectors to the carrier (which is either gold or silver covered duroid)

We also use surface mounted FPC sockets which then goes to a "breakout PCB" that you can solder to. FPCs sockets are nice because you can buy ribbon cables with very thin conductors, this means that the thermal conductivity is quite low which makes this type of ribbon quite good for e.g. connecting to cold fingers or cold stages in e.g. 300 mK systems. However, soldering the connector is a pain and we usually use a small local company to do that.

Btw, most -but not all- plastics are OK at low tempeatures. The best thing is usually just to buy off-the-shelf components and dip them a few times in liquid nitrogen, if they work then you can be reasonbly sure that they will be fine at lower temperatures as well.

Btw. it is usually quite easy to bond to normal PCBs, but do watch out for the laquer they tend to put on the pads. Also, Au wire is often easier to bond with than Al.

 

[hunt137]

Thanks for the feedback, 0xDEADBEEF and f95toli. Based on your advice and my own research, this is the solution I'm going with:

Ceramic LCC carriers from EverGreen Semiconductor:
http://www.evergreensemiconductor.com/ceramic-leadless-chip-carrier.html

PLCC sockets by 3M:
http://uk.farnell.com/3m/8420-21b1-rk-tp/socket-ic-plcc-20way/dp/1183018

If I end up working on a cold finger, I plan to modify the socket by drilling out the middle plastic and replacing it with sapphire. In principle, this should provide adequate thermal contact to to the carrier, which is "black alumina" and should also be thermally conductive enough.

I may also attempt to heat sink the wires for additional thermal contact. However, since my application involves ultrashort (ps-ns) single shot pulses, I'm worried about the broadening effects of this. We shall see.

If I end up working in an He-exchange environment, well, then I'll just cross my fingers that the PLCC socket will work well down to 4 K.

At any rate, I'll update this thread when something is implemented. If anyone has additional helpful comments, advice, or anecdotes, they are of course much appreciated.

 

[glientsc]

Hello hunt137,
faced with the same problem I wonder whether you gained some valuable experience that you want to share? How did your test setup go?