Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.eso.org/~gfinger/hawaii_1Kx1K/hawaii1k_sci2_sci3/hawaii1k_sci2_sci3.pdf Äàòà èçìåíåíèÿ: Mon Sep 21 15:49:00 1998 Äàòà èíäåêñèðîâàíèÿ: Sun Dec 23 01:45:09 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ð ï ï ï ï ï ï ï ï ï

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #3
Gert Finger, G. Nicolini European Southern Observatory

1.0 Scope
Two detectors are available for the short wavelength arm of ISAAC, namely the science grade #1 and the science grade #2 Rockwell 1Kx1K MCT arrays. The science grade #1 is mounted in ISAAC and was operational and aligned in ISAAC during the July test on Paranal. The science grade #2 is presently being tested in IRATEC. An exchange of detectors can only be justified by superior performance of the science grade #2. For this reason a careful comparison of the quantum efficiency in broadband K was made. Noise and dark current will also be compared during the ongoing tests.

2.0 Cosmetic quality
Figure 1shows efficiency. The has a cosmetic of the detector. the cosmetic quality of both arrays displayed as images of the quantum cut levels are set to 0 and 1. The science grade #2 mounted in ISAAC defect in upper right quadrant due to hole in the IR sensitive LPE layer The science grade #3 has no defects and an increase of QE of ~10%.

3.0 Quantum efficiency
A comparison of quantum efficiency for both detectors was made, which does not rely on any assumptions of gain in the acquisition chain. The procedure was identical for both detectors. The temperature of extended blackbody filling the detector field of view was the varied between 10 C and 70 C. Both signal and rms noise were measured for detector integration times of 1.5 to 4 seconds. The shot noise method yields the conversion factor electrons per ADU by plotting the variance as function of signal which is shown in Figure 2and Figure 3. The same set of data is used to derive the QE. The photon flux at the detector is calculated taking into account the measured filter transmission at nitrogen temperature. The focal ratio at the detector is f/11. This is the most uncertain figure in the determination of the absolute value of the QE. Since both arrays have been measured in the same setup of the test camera, the ratio of the quantum efficiencies is more accurate than the absolute value. The quantum efficiency given here does not take into account optical losses of the test camera and assumes an optical efficiency of 100%. Figure 4 shows the measured average number of integrated electrons within the rectangles marked in the images of Figure 1 as function of calculated number of integrated photons for the K-prime broad band filter. Circles indicate measurements obtained with science grade #3, squares show data for science grade #2. The QE is 80.2% for science grade #3 and 68.1% for science grade #2.
Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998 1


FIGURE 1. Quantum efficiency. Cut levels: 0,1

Rockwell science grade #2 mounted in ISAAC. Cosmetic defect in upper right quadrant due to hole in LPE layer.Mean QE in rectangle is 68.1%.

Rockwell science grade #3. Candidate for ISAAC. No cosmetic defect. Mean QE in rectangle is 80.2%.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

2


FIGURE 2. Gain conversion factor of science grade #2 determined by shot noise method. Gain is 2.19 electrons/ADU. Gain on the ADC board was set to 3.

FIGURE 3. Gain conversion factor of science grade #3 determined by shot noise method. Gain is 5.57 electrons/ADU. Gain on the ADC board was set to 1. The mean QE within the rectangles marked in the images of Figure 1is 68.1% for the science grade #2 and 80.2% for the science grade #3. The histogram of QE is shown inFigure 5. The science grade #2 has a more uniform but 10% lower quantum efficiency.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

3


FIGURE 4. Quantum efficiency of Rockwell 1024x1024 MCT arrays with Kprime broad band filter. Circles: QE of science grade #3 is 80.2%. Squares: QE of science grade #2 is 68.1%. detector.

FIGURE 5. Histogram of quantum efficiency in K-band of Rockwell 1Kx1K MCT arrays: peak QE of science grade #2 is 68%, peak QE of science grade #3 is 77%.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

4


FIGURE 6. QE map of science grade #3 in K-band measured with IRATEC. Cut levels 65% to 95%. Contour interval 2.5%.

It is very surprising that the quantum efficiency reaches peak values of more than 93% as shown The same QE contour map for science grade #2 to 50%-80% is shown for comparison in Figure

in the center of science grade #3 by the contour map of QE in Figure 6. with cutlevels shifted from 65%-95% 7.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

5


FIGURE 7. QE map of science grade #2 in K-band measured with IRATEC. Cut levels 50% to 80%. Contour interval 2.5%.

4.0 Noise
Both double correlated and multiple nondestructive sampling have been applied. Noise histograms are shown in Figure 8. New low noise cryogenic operational amplifiers ( 8 nV / Hz ) have been used. The analog bandwidth has been reduced from 176KHz (rise time 1.88 µs) to 22 KHz (rise time 15 µs).

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

6


FIGURE 8. Histogram of readout noise for double correlated and multiple nondestructive sampling (16 nondestructive samples per integration ramp). Detector integration time 1.4 seconds for double correlated and 162.8 sec for nondestructive sampling. ADC gain = 3.

5.0 Darkcurrent
The cryogenic concept of the NIRMOS test camera is based on a continuous flow system using liquid nitrogen. The minimum temperature to which the detector can be cooled is 80 K. For this reason the temperature dependence of the darkcurrent has been re-investigated using science grade #3. To eliminate drift effects and instabilities of the detector immediately after the reset pulse all measurements were carried out in non-destructive readout mode by storing the individual non-destructive readouts. The darkcurrent is derived from the slope of the linear part of the integration ramp as shown in Figure 9. The temperature dependence of the darkcurrent is shown in Figure 10. The logarithm of the darkcurrent is plotted versus the reciprocal temperature. At higher temperatures the result is a straight line characteristic of a generation recombination darkcurrent process. The temperature dependence of a g-r current mechanism is nii/2 exp(-Eg/mKT) exp(-Teff/T) with ni being the intrinsic carrier density and m equal 2. The slope of the linear fit to the experimental data yields an effective temperature of Teff = 2558 K. For a cutoff wavelength of c=2.5 µm the measured parameter m is 2.25 which is close to the theoretical value. At temperatures below 90 K measured data represented by triangles deviate from the theoretical curve as shown in Figure 11. A radiation leak was suspected in the experimental setup and the darkcurrent measurement was repeated with the detector

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

7


blinded in addition to using dark positions in the two filter wheels. Data taken with the blinded detector are represented by squares in Figure 11. At a temperature of 42 K darkcurrents as low as 3 electrons per hour have been measured. For comparison previous measurements with science grade #2 are also plotted in Figure 11 as circles.

FIGURE 9. Dark current at 65 K. Measured points represent the mean signal of subsequent nondestructive readouts applied during a single integration ramp. The temperature dependence of the darkcurrent is shown in Figure 10. The logarithm of the darkcurrent is plotted versus the reciprocal temperature. At higher temperatures the result is a straight line characteristic of a generation recombination darkcurrent process. The temperature dependence of a g-r current mechanism is nii/2 exp(-Eg/ mKT) exp(-Teff/T) with ni being the intrinsic carrier density and m equal 2. The slope of the linear fit to the experimental data yields an effective temperature of Teff = 2558 K. For a cutoff wavelength of c=2.5 µm the measured parameter m is 2.25 which is close to the theoretical value. At temperatures below 90 K measured data represented by triangles deviate from the theoretical curve as shown in Figure 11. A radiation leak was suspected in the experimental setup and the darkcurrent measurement was repeated with the detector blinded in addition to using dark positions in the two filter wheels. Data taken with the blinded detector are displayed by squares in Figure 11. At a temperature of 42 K darkcurrents as low as 3 electrons per hour have been measured. For comparison previous measurements with science grade #2 are also plotted in Figure 11 as circles.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

8


FIGURE 10. Temperature dependence of darkcurrent.

FIGURE 11. Temperature dependence of darkcurrent. Triangle: science grade #3. Square: science grade #3 blind. Circle: science grade #2. In the center of the array where all quadrants touch the darkcurrent at a temperature of 65 K is 30 electrons per second within a very localized region of ~3 pixels in diameter.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

9


For comarison our darkcurrent measurements are plotted in Figure 12 together with darkcurrent measurements by differnt authors obtained with both InSb and MCT detectors having different cutoff wavelengths.

FIGURE 12. Comparison of different detector materials in terms current density versus temperature. Triangles: InSb. Crosses: c=5.2 µm LPE HgCdTe. Squares: c=5.2 µm MBE HgCdTe. Stars: c=2.5 µm LPE HgCdTe. Double triangles: c=2.2 µm MBE HgCdTe. Circles: ESO measurements of InSb and c=2.5 µm LPE HgCdTe. At a temperature of 75 K MBE grown HgCdTe having a cutoff wavelength of c=5.2 µm can achieve dark currents more than 3 orders of magnitude lower than dark currents achieved with InSb. Also for short cutoff wavelength c=2.5 µm materials MBE grown HgCdTe outperforms LPE grown material.

6.0 Conclusions
Both QE and cosmetic quality of science #3 are better than science #2. No low frequency noise was observed with science #3 in IRATEC. A replacement of the ISAAC detector is recommended. The lowest darkcurrent measured with science grade #3 in IRATEC is ~ 3 electrons/hour. Cooling the detector to 80 K in a NIRMOS type continuous flow cryostat is not sufficient if LPE grown detector material is used. For MBE grown material a detector temperature of 80K may be sufficient.

Comparison of Rockwell 1024x1024 MCT arrays science grade #2 and science grade #321 September 1998

10