bad multilayer ceramic capacitor at 4 Kelvin?

ouch, thats cold, my only bad experience with ceramic caps is end caps coming off SMD parts. does it work again once its warmed up ?

Colin =^.^=

My experience is that capacitance of ceramic capacitors with high-epsilon dielectrics (such as X7R or Z5U) just go down but the capacitor in general would not fail. Ones with NPO type dielectric seem to perform quite nicely at liquid He. This is a rule-of-thumb style observation only: designations such as X7R don't define the chemical composition, and I find it conceivable that some X7R or other caps might show increasing capacitance towards cryogenic temperatures even though the ones I've experimented with do not. I seem to recall that strontium titanate is an example of an dielectric with such properties.

You can't get high capacitances out of NPO, unfortunately; I once had a fun hypothesis that whatever the capacitor value is at room temperature, it will be 100pF at 4.2K . For higher capacitances the Panasonic ECPU type plastic film caps are available up to 1uF in a 1210-sized case and they perform fine at 4.2K .

Different thermal expansion coefficients of the capacitor case and the substrate increase the risk of the capacitor mechanically breaking, but even then I would expect an open circuit rather than a short. What sometimes happens is that the thin metal layer (diffusion barrier?) covering the contacts of an SMD cap gets detached because of the mechanical stress. Still, off-the-shelf NP0 and ECPU type SMD caps soldered on ordinary FR4 board seem to tolerate thermal cycling surprisingly well. I have used those on my own cryogenic amplifier (which now appears to show 70 pV/rtHz white voltage noise and 1/f corner approximately at 5 kHz, by the way).

Regards, Mikko

I can't check it right now, the amplifier is a part of a vacuum system which has been cooled to 4 Kelvin. I am still trying to see if these is an implicit short and trying to avoid shutting down of the complete cryogenic system.

Thanks Mikko, I have used the X7R and also as i needed power supply bypassing upto high frequency (upto 10GHz) i have used RF capacitors from Vishay. HPC0402A type:

They have very high SRF, and i used 10pF and 1.2pF from them. Though, just now i have realized they are manufactured on a Silicon wafer. I know Silicon transistors fail at 4 Kelvin due to carrier freeze-out but it won't explain the short circuit. Do you know of any other high frequency good quality capacitors which i can use for cryogenic appications.

I have used those on my own cryogenic amplifier

I would really like to have a look at ur amplifier, if you want to share. Is it published, already?

Many possible things could be happening. First, warm them back up to verify proper operation. Second, wire up one capacitor, feeding the 2 leads to the outside world and measure the capacitance as a function of temperature. There is a good possibility that this will also show the same abrupt transition to a short circuit near/at 77K. If warming the capacitor back up maked the cap good again, then the ceramic might be changing to a superconductor below 77K. Depending on how much capacitance you need in various parts of the circuit, you might try mica capacitors or mylar capacitors.

Eh? I know (I think I know..) perovskite(sp) superconductors are relatively similar in crystal structure, but I never heard of titanates superconducting.

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

I did something similar. I slowly increased my VDD from 0 in very small steps. At first, the current drawn was high (say 0.05 Amps), however slowly it decreased to 0.01Amps. And basically with time i could put VDD = 2.2 volts and VSS = -1 volts, with Out(+) and (-) at 1 volts. This implying that there is around 1.2 mA of current in each half of the circuit and seems i am also getting some amplification out of it. But i want to do some more tests to confirm. The power supply still says 10 mA of current being drawn i want to know where it is going. We have spent something around US$2000 to cool this whole system and if i want to change something on the amplifier that money goes in drain. Thats why i am hoping to avoid this. Right now i only have access to VDD, VSS, GND and Out+,- pins which are on a connector outside the vacuum system.

I am going to try this, the first thing in the morning. Hopefully, this is the case.

If you w'ld look at my circuit Bob, i need a range of capacitors for bypassing the power rails, 1.2pF to something 1 uF. the pFcapacitors have to be of high frequency types with SRF in GHz regime. I am trying to find some which are non-silicon.

besides, 18 Ohms is a pretty shitty superconductor.

Cheers Terry

The 18 ohms i think is coming from the magnet wire (Thin 36 Gauge phosphorus bronze wire from lakeshore), that piece of crap had 6 ohm from the PC board to the connector at room temperature (almost 5ft long). So i don't wonder if cooled to 4 K it would show that value of resistance. Infact they do quote 8 ohm per meter at 4 Kelvins:

So i think the capacitor has become superconducting on the PCB and when i measure resistance between the pins VDD and GND on the connector all i get is the resistance of the 2 connecting wires.

[snip]

Two ideas: 1) Get some capacitor candidates and cool ferociously, external to the cyro setup, to see what they do. 2) Time-domain reflectometry of the cryo setup would tell you the location and nature of the short, if it exists. If you don't have a TDR, just make one--a simple rig suffices.

James Arthur

The capacitance of X7Rs I've fiddled with fall to 1/10 when cooled to

4.2K. NP0s keep their capcitance better. But of course this depends on the actual chemical composition of the dielectric.

I've been happy with small-value NP0s in a 0402 case, which can also have quite high SRFs. I have no experience on Si caps but would expect them to be OK.

I presented it in the WOLTE-7 workshop, a preprint is found in

Regards, Mikko

Everything i know about multilayer ceramic capacitors tells me that they will not fail in this way. All the failure modes that i know of are high temperature types. Unfortunately i have scant data at 77K and none below.

--
 JosephKK
 Gegen dummheit kampfen die Gotter Selbst, vergebens.  
  --Schiller

could you acheive anything usefull by pulling up the out+/- pins externally via 1k resistors to VDD, or maybe more to compensate for short, ? or has it already warmed upto room temp by now ?

Colin =^.^=

Its common sense that as an object goes towards 0 Kelvin, the brownian motion of its atoms decrease dramatically. If anything, I would assume a decrease in capacity for a capacitor at temps of 77 Kelvin and lower. As the brownian motion slows down, so does the number of collisions, decreasing electron transfers.