Fracture processes studied in CRESST

https://doi.org/10.1016/j.nima.2005.12.127Get rights and content

Abstract

In the early stages of running of the CRESST dark matter search with sapphire crystals as detectors, an unexpectedly high rate of signal pulses appeared. Their origin was finally traced to fracture events in the sapphire due to the very tight clamping of the detectors. During extensive runs the energy and time of each event was recorded, providing large data sets for such phenomena. We believe this is the first time that the energy release in fracture has been accurately measured on a microscopic event-by-event basis. The energy distributions appear to follow a power law, dN/dEE-β, similar to the Gutenberg–Richter power law for earthquake magnitudes, and after appropriate translation, with a similar exponent. In the time domain, the autocorrelation function shows time correlations lasting for substantial parts of an hour. Some remarks are made concerning the possible role of such mechanical stress release processes in the noise of sensitive cryodetectors.

Introduction

In the first runs of the CRESST dark matter search in Gran Sasso, a phenomenon was observed which we believe may be of interest for the study of fracture in brittle materials. At the time, the detector elements were large (262 g) high-quality single crystals of sapphire, with a tungsten TES evaporated on one surface serving as the sensor. To eliminate possible microphonics, the crystals were held very tightly in their holders by small (mm) sapphire balls held against the crystal by plastic clamps. The plastic of the clamp, delrin, is known to contract substantially at low temperature, thus providing additional “tight holding”. When the system was first brought into operation, an unexpectedly high rate of signal pulses was observed. Initial fears that this might be due to an unexpected radioactive contamination were relieved by the observation that even an unknown radioactive contamination must be Poisson distributed in time, while the unexpected pulses appeared rather to come in “avalanches”. An analysis of the time distributions showed that they were indeed non-Poissonian. Pulse analysis was of no help in finding the source of the signals since any fast (μs) energy release leads to the same pulse shape [1].

An extensive search for the origin of the pulses was finally successful when it was noticed that there appeared to be markings or scratches where the sapphire balls contacted the crystal. When the sapphire balls were replaced by plastic stubs, the event rate immediately dropped from some thousands per hour to the expected few per hour.

These observations strongly suggest that the pulses were due to some kind of cracking or micro-fracturing phenomena in the sapphire crystal and/or the contact balls. Indeed, examination of the scratches under a microscope revealed a small crater with radiating irregular fissures extending sideways and down into the crystal. The sapphire balls was also damaged. Since the reduction in rate after the exchange of the sapphire balls was so large, we believe the data with the sapphire balls represent essentially 100% fracture events. If we accept this crack or fracture hypothesis, our data then represent a large sample of well measured fracture events, under low background conditions, and with good time and energy determination.

Section snippets

Energy distributions

From this large data sample, we can study some features of the microfracture process. One of these is the energy release in microfracture, which here seems to be measured on an event-by-event basis for the first time. In Fig. 1 we show the differential distribution dN/dE for the number of events N per unit energy for four data sets with two detectors from Run9. The straight line is the result of a power-law fitdN/dEE-βto the lowest curve which yields β1.9. Similar results are found for other

Correlations in time

We expect correlations in time, corresponding to the “avalanches” or non-Poissonian behavior. For this study, the calibration runs are particularly useful. In these runs a Cobalt source supplying 120keV photons is inserted in an external plug in the shielding. These photon-induced events can subsequently be selected by using the resulting 120keV peak in the data. Since a radioactive source produces statistically independent events, that is Poisson statistics, these events provide a useful check

Fractal statistics, unconventional noise

The fact that the energy spectrum—and also, as we find, the autocorrelations—can be fit with power laws is suggestive of a description without intrinsic dimensional parameters, as in fractal statistics. We have examined this by seeing if a consistent Hurst exponent Ha parameter often used in this subject—can be derived from the time series of the events [3]. This appears to be the case.

It is possible that this observation may be useful in the understanding of excess noise or certain

References (3)

  • For recent CRESST results see F. Petricca (this meeting), G. Angloher, et al., astro-ph/0408006, Astroparticle Phys. 23...
There are more references available in the full text version of this article.
View full text