S17_FOXSI

Characterizing the Low Energy Response of the FOXSI Silicon X-Ray Detectors

Kendra Bergstedt

University of Minnesota School of Physics and Astronomy

Abstract

The purpose of our experiment was to improve the characterization of the energy response of the Focusing Optics X-ray Solar Imager (FOXSI) silicon X-ray detectors below 5 keV, as there have been observed discrepancies in this regime. We did so by using a 5.9 keV Fe-55 X-ray source to measure the variation in transmission of low energy X-rays through the detectors' nonuniformly crinkled thermal blanketing. We found that at 5.9 keV, a smooth blanket had an average transmission of

and a crinkled blanket had an average transmission of

, indicating that there was significant change in transmission with crinkling and that the crinkling cannot have caused the observed discrepancy.

Introduction

FOXSI is a sounding rocket project that has taken X-ray spectra of the Sun. The project has launched twice, and has another launch planned for 2018. It was noticed that the data from different detectors aboard the FOXSI-2 launch disagreed. For X-ray energies below 5 keV, the different detectors aboard FOXSI-2 had spectra that differed by as much as 50%. This experiment determined whether the crinkling of the thermal blanket in front of the detectors could have caused fluctuations the X-ray transmission large enough to explain the discrepancies in spectra between detectors. To do so, we took a sample of the blanket material and measured its transmission in both smooth and crinkled states and compared them.

Theory

The transmission of a beam of monoenergetic photons through an object is the fraction of photons in the beam that pass through the object without absorbing or reflecting. As transmissions are independent probabilities for a photon to traverse a layer, they can be multiplied: if one layer of a material has transmission

and another has transmission

, the transmission through both layers is . The transmission T of a particular photon energy through a material is given by a decaying exponential:

Here

is the thickness of the material, and is the attenuation length, which is a parameter that depends both on the energy of the incident photon and the particular material. We used a database of attenuation lengths by material and incident photon energy to calculate the transmission of our blanket.

The above is a photo of the blanket sample used, with examples of both crinkled and uncrinkled layers. The blanket consisted of two parts: ten layers of aluminized mylar with a mylar thickness of

and an aluminum thickness of , and a single layer of Kapton with a Kapton thickness of and an aluminum thickness of

. The aluminized mylar layers were the part of the blanket that was crinkled while the Kapton was kept smooth. Using these parameters and our database of attenuation lengths, we predicted the transmission through the entire smooth blanket to be approximately 0.78. We then computed the transmission through a blanket with the aluminized mylar layers at 60 degrees to the horizontal to simulate severe crinkling, which had a transmission of approximately 0.67. Therefore, we expect a transmission of no more than 0.78, and possibly as small as 0.67.

One of the additional tests we did of the material was a determination of the attenuation length of the aluminized mylar. Since aluminized mylar is made of layers of material, we had to calculate the expected attenuation length, which we did using the following formula, which is simple to derive from the exponential definition of transmission above:

Where and are the thickness and attenuation length of the aluminum layer and and are the thickness and attenuation length of the mylar layer. is the total thickness of the sheet, and

to be based on values found in the database for and , which were and .

Experimental Setup

The experimental setup used was the payload from the FOXSI-2 rocket. It consisted of the seven detectors onboard the FOXSI-2 flight, as well as the power supply and all of the readout electronics used during the flight. The detectors were kept thermally insulated from the rest of the electronics so that they could be cooled by liquid nitrogen, which decreased their noise. The detectors themselves were silicon X-ray detectors. They detected incident X-rays by measuring current pulses caused by incoming photons ejecting electrons in the material via the photoelectric effect. The ejected electrons then liberated other electrons, with the total number of electrons freed proportional to the energy of the incident photon.

The X-ray source we used to measure the transmission of the blankets was a Fe-55 sealed source, which emits at 5.9 keV. The source slid beneath the payload, and there were windows in the thermal blanketing to allow the source's X-rays to be incident on the detectors inside.

Results

Determination of the Attenuation Length of Aluminized Mylar

We measured the decrease in flux with increasing layers of aluminized mylar between the source and the detector. The data was then fitted to a decreasing exponential in order to find the attenuation length of the aluminized mylar sheets.

The fit (shown above) had an attenuation length of , which is about from the predicted value of . This discrepancy could indicate that we do not have the correct thicknesses for the materials of the blanket. The blankets were handed down to the project from a previous space mission and so there is no documentation from the manufacturer, so it is possible the thicknesses are not what we think they are.

Fitting the Fe-55 5.9 keV Peak

One effect that increased noise has on the DSSDs is an increase in peak width, while the total counts in the peak stay the same. Therefore, if the detector noise is fluctuating from run to run, peak height will not be an accurate indicator of the total flux through a peak. Therefore we used total peak counts for our transmission measurements. To do so, we fitted the 5.9 keV peak to a Gaussian and found the area under that Gaussian. For our detectors, this fit was complicated by the existence of a 6.5 keV Fe-55 emission peak. Both peaks were broadened by the DSSDs to the point that they overlapped, so a fit using a sum of two Gaussians was necessary to adequately model the peak.

The fit is shown above. The low energy tail was broadened for unknown reasons, so it was neglected in the fit. The chi_i in the 7-8 keV portion of the fit are large due to the nonzero value of the fit, while there were no counts in the data, and has no bearing on the fit of the peak itself. The chi_i plot indicates that the fit of the peak region is adequate.

The parameters of this fit can then be used to calculate the area under the 5.9 keV peak, which is used later in the analysis.

Blanket Transmissions

Data was taken in an alternating fashion, between runs with the blanket and without it. The summary of the peak areas for both the smooth and crinkled blankets are below.

There are clear systematic changes in the total flux through time, both in the smooth and crinkled blanket runs. It was suspected that the main cause of the systematic changes in flux was accidental motion of the X-ray source while inserting the blanket between the source and the detector, and that the source motion while removing the source was minimal. Therefore, to minimize the effect of this source motion on our data we computed transmissions pairwise between a blanketed run and the next nonblanketed run. The results of this are shown next.

It is clear from the summary of computed pairwise transmissions that the systematic changes in flux seen in earlier data were effectively removed by computing the transmissions pairwise. It is also clear that there is not a visually significant difference between the transmissions of the crinkled and smooth blankets. We calculated that the average transmission for each run was

for the smooth blanket and for the crinkled blanket. There was no significant drop in transmission from the uncrinkled to the crinkled blanket. We also did not observe any increase in the variation of the transmission values from the crinkled blanket, which we would expect to see if the crinkling was causing variations in transmission throughout the blanket. These two results together strongly indicate that the crinkling of the blanket cannot have accounted for the spectral discrepancies between detectors, which were as large as 50%. It is also notable that both of the blankets produced an average transmission that was larger than the calculated maximum possible transmission, 0.78. Due to the large errors on our calculations the deviation is only

, but as it was observed for both blankets it is worth further investigation.

Conclusions

We found the average transmissions for the smooth and crinkled blankets to be

for the smooth blanket and for the crinkled blanket. These measurements are in strong agreement with each other, so we concluded that the crinkling of the flight blanket could not have accounted for the variations in spectra between the detectors on the FOXSI flights. This indicates that there was some other cause for the detector disagreements, so further investigations into the issue are necessary. In particular, investigating our result of the attenuation length of the aluminized mylar would be a good starting point, as that result was

from our prediction. Another measurement with a larger maximum amount of layers would indicate whether the material's composition was significantly different than expected.

References

[1] L. Glesener, S. Krucker, S. Christe, S.-n. Ishikawa, J. C. Buitrago-Casas, B. Ramsey, M. Gubarev, T. Takahashi, S. Watanabe, S. Takeda, S. Courtade, P. Turin, S. McBride, V. Shourt, J. Hoberman, N. Foster, and J. Vievering, "The FOXSI solar sounding rocket campaigns," in Space Telescopes and Instrumentation 2016: Ultraviolet to Gamma Ray, 2016.

[2] L. E. Glesener, Faint Coronal Hard X-rays From Accelerated Electrons in Solar Flares. PhD dissertation, University of California, Berkeley, 2012.

[3] B. L. Henke, E. M. Gullikson, and J. C. Davis, "X-ray interactions: photoabsorption, scattering, transmission, and reflection at e=50-30000 ev, z=1,92," Atomic Data and Nuclear Data Tables, vol. 54, no. 2, pp. 181-342, July 1993.

[4] E. Segre, Nuclei and Particles. Reading, MA.: W. A. Benjamin, 2 ed., 1977.

[5] H. Spieler, Semiconductor Detecctor Systems, vol. 12. Oxford University Press, 2005.

[6] G. F. Knoll, Radiation Detection and Measurement. John Wiley & Sons, Inc, 3 ed., 2000.