Time Dependent Electric Field
May 25, 2007
I propose investigating the fate of our electron-hole pairs using a time dependent electric field. At the Super CDMS detector meeting there were questions about the phonons due to recombination. Are these responsible for the long tails in our phonon signals? Is the band-gap energy lost from our detectors? Furthermore, if we can make long lived free charges, then, combined with a time dependent field, it may be possible to answer other questions. Can we readout ionization from the crystal faces with QETs? Can we improve our charge resolution?
Š To understand our detectors. If ~25% of the energy of electronic recoils goes into promoting carriers into the conduction band, then ~8.4% of the energy of nuclear recoils promotes the carriers. We assume that 100% of this energy is readout through the phonon sensors. If 0% of this energy reaches our phonon sensors, then this could represent a ~17% systematic offset between our recoil energy scales, and we don’t have any mono-energetic neutron runs with which we could set an absolute nuclear recoil energy scale.
Š To improve our analysis. If we could monte carlo our detectors then we might be able to better use some of the subtleties in our pulse shapes to improve background rejection and event understanding. Matt Pyle’s monte carlo is a very nice initiative and understanding when and how ‘recombination phonons’ enter our pulses is an open question. Imagine a modulation of the electric field. If we can ‘play ping-pong’ with the electrons and/or holes, then the decay of the associated Luke phonon pulse should be related to the demotion of the carriers to the non-conducting state.
Š To improve our detectors. There are two problems that “voltage-assisted calorimetric ionization detection” could solve. First, it is theoretically “possible to obtain a sensitivity level sufficient for single-carrier detection.”  Second, it could be a way to read out ionization in a detector with QETs on both faces. Now imagine the detector kept with a low (zero) voltage such that Luke phonons do not contribute to the phonon signal. Some time after a phonon signal is detected the field is switched to a higher field where Luke phonons are produced by any free charges left from the event. If I take some terminology from our esteemed competitors, the first signal (P1) would be proportional to the recoil energy. The second signal (P2) would be proportional to the free charges. So the charge would not only be read out of the same face as the phonon detectors, it would be read out of the same sensors. The noise from the QET-SQUID sensor is historically much easier to isolate from various pickups and the noise is lower in general. Additionally, it would be possible to adjust the field (amplitude and delay) based on the recoil energy so that there would be more/less gain for lower/higher energy events. This could have the advantages of keeping the electronic noise contribution a constant over some energy range, and increasing the dynamic range of the ionization measurement.
Š Practically, it may be difficult to change the field rapidly enough without heating the TESs and/or losing lock on the SQUIDs. Care should be taken with translating from these gedanken experiments to real experiments. Starting with slow and small changes is probably advisable. Special QET geometries (no floating fins) may be helpful (or necessary).
Š The carriers may not be free long enough. Research into better charge traps may be necessary. (Special doping, special coatings, special fields…)
Š The carriers may be free for too long. If the free charge builds up in the detector, then small signals will become harder to detect over time. Research into clearing the trap might be needed. More flashing may be necessary.
Š If the ionization were read out in a delayed fashion, then the detectors would have some dead time. Calibration would have to be done at a lower rate, but WIMP searching at ~ 1 Hz or less should not be a problem. If it is necessary to flash after every event, then even WIMP search rates may be a problem.
Š The easiest thing to try would be shorting out the charge bias resistors, and removing the filtering from the charge bias to enable fast charging of the detector.
Š The buffering op-amps could easily be replaced with the AD8513. This is roughly the same speed as the charge bias DACs and so charge biasing could be accomplished in about 200 ns.
Š Certainly shorter striplines would be helpful. Currently the capacitance of the striplines is an order of magnitude larger than the detector. I don’t know if these currently exist, or if they would be easy to install at a test stand.
Š To readout double-sided phonon detectors would require charging the crystal faces instrumented with QETs. It would be possible to start floating the QETs, even with modest modifications of the current electronics. In fact, the ‘Guys from Berkeley’ have already solved this problem with their work on the T5Z2 Qinner bias short. Basically, the returns of the QET bias could be connected to a charge bias DAC. The other charge bias DAC could be used to charge the other side of the detector. Differential biasing of Qinner and Qouter of the detector would not be possible, but this could be the more symmetric way to put electrons on one face of the detector and remove them from the other.
 P.N. Luke, “Voltage-assisted calorimetric ionization detector”, J. Appl. Phys. 64 (12), 15 December 1988, pp. 6858-6860