MOIRCS R1300 Grism: Image Degradation
2020-02-01 Update
The R1300 grism has been decomissioned from S20A.
The grism has a lot of micro-cracking, which might be the cause of the image degradation. The use of the grism is not recommended for science (shared-risk operation).
It has been recognized that our R1300 grisms show an image degradation. What we have initially found is that the degradation shows the position dependence on channel-2, as you can see in the investigation history below. The symptom has been already there at least before 2010.
Currently the most likely cause of the spectral shift and degradation is the problem of the grism itself. If the grism material has the inhomogeneity (by crack etc), its affection will appear on the line profile, such as the shift/blur of the synthesized line profile (and the time-evolution as reported). Indeed, the inspection by microscope has revealed serious micro-cracking on the surface of the grism. The measurement of the wavefront by Zygo also shows significant (at least more than several lamda) wavefront degradation. Thus, the use of R1300 grism is not recommended.
The most recent (April 2016: by new moircs detector) measurement of the total throughput of the R1300 grism is roughly 15 to 20% in K-band, which is only 60-70 % of HK500 grism and half of the VPHK grism. We may replace it to new Hybrid-type Grism in the future (see Ebizuka et al. 2016 for the plan).
(Update) We have decided to discontinue the R1300 grism from S20A. The table below is kept for historical purpose.
| Grism name |
Operating range [um] |
Resolution (0.5'' slit) (*4) |
Straight-through Wavelength [um](*5) |
Dispersion [A/pixel] |
Sensitivity (Vega magnitude) [mag/arcsec2] |
|---|---|---|---|---|---|
| R1300 (*1) | 1.18 - 1.33 @ J 1.49 - 1.78 @ H 2.0 - 2.4 @ K |
1100 @ J 1050 @ H 920 @ K |
1.14 @ J 1.50 @ H 2.24 @ K |
1.91 @ J 2.61 @ H 3.88 @ K |
J=17.8(*6) H=16.7 K=17.1 |
(*1) For R1300 we use the 2nd-, 3rd-, 4th-order spectra for K, H, and J band observation, respectively. We use the imaging broadband filters for the order-sorting purpose. For K-band region you can use Ks-band filter instead (1.99 - 2.31 um range, with lower thermal background). The spectral resolution here is for 0.7-arcsec width slit. Note that due to the R1300 grism problem, ch2 grism resolution would not go better than this. The sensitivity is measured in 2006, not the current values. We do not recommend to use the grism due to the problem on the grism.
(*6)The depth for R1300 is still for the previous detectors (for data before June 2015).
(*7)Note that some spatial shift of the spectra (with respect to the slit position) is observed for the channel-1 R1300 grism spectra. The shift is +22 pixels to the spatial direction (+y direction on raw image, -x direction on preimage: in wmdp_moircs software it is not displayed!).
Figure: Total system throughput (telescope + instrument) for R1300 grism with STD_K filter and comparison with the throughput of HK500 grism (Apr 2016 data). The system throughput of the grism with H-band or J-band filters are each roughly 40-60% (H) and 20-25% (J) of the STD_K case. Note that the use of the R1300 is currently not recommended due to the grism defects as described here in detail.
Investigation History: Some discussions are old and may be based on a wrong assumption. Careful!
2014-10-14 update:
In Oct 2014 we have had a chance to do the direct inspection by removing the grism from the main dewar. The inspection by Zygo and Microscope has done, and the results clearly suggest the problem of the grism itself. The wavefront of the light through the grism has unusual distortion. The surface of the grism has a lot of micro-cracking. The substrate of the grism is KRS5, which is known to cause such micro-cracking by some shock. However we do not know when it has become significant. It may be the result of the repeated thermal cycle. It might be caused by the shock during the T3 troubles we had long time ago. We have also observed that the surface is in general very dirty, with a lot of scratches. We knew that they were there since these grisms were brought to MOIRCS in 2006.
The report of the investigation can be find from link below. Sorry but it is written in Japanese.
The use of the ThAr data by R1300 is not recommended, especially for channel-2. Those who need higher resolution you could use the VPH grisms.
2014-08-14:
Advanc of the R1300 line blurring over time.
In order to see how the line blurring has advanced over time, I have checked the fwhm of the OH night emission line observed. The observed fwhm is compared with the slit width used. If the intrinsic image fwhm of a grism is enough narrow, the observed fwhm should be propotional to the slit width, so the data should follow the y=x relation. The result is blow.
First of all, the degration of the ch2 data compared with the ch1 data is evident. The channel-1 grism (shown brown asterisks) seems healthy in terms of the FWHM distribution.
As for the channel 2 data, the difference of degradation between 2006 and 2014 is small but definitely exists, though the degree of progress is not so significant as I have guessed by its apparance. The data between 2006 and 2008 show fairly consistent distribution, suggesting that the grism has not changed its character at least until March 2008. However, the data points for 2010 (cyan open circle) already shows a different fwhm behavior around the slit width of 4 pix. It indicates that the degradation started between Apr 2008 and Apr 2010.
Though it's the only single point, the difference of the FWHM at narrowest slit width (<2pix) between 2012 and 2014 is clear. We need more data to confirm the difference. At larger slit widths (<4pix), the data points around 2010-2014 show relatively simiar behavior. As the most scientific observations use such wide slits, we may say that the change is relatively slow over the last a couple of years under the normal observing conditon.
2014-08-13:
zJ500/HK500/VPHK investigation: the edge-illumination effect hypothesis is not supported.
We have checked the ThAr data with zJ500 and HK500 taken during 2013-2014, as well as the VPH-K data taken in 2012. No indication supporting the edge-brightening effect was seen, again suggesting the reported R1300 issue is specific to that grism. In the first place, the edge-brightening by side-illuminated slit should not show the time evolution. Thus for now we can say that the reported line profile degradation is probably due to something which is the grism-specific.
Clearly, a more thorough investigation is necessary to addresss the cause. For now, we do not close the grism, but again the use will be under the "shared-risk" policy from now on. The users who plan to use R1300 grism should pay enough caution for the data analysis. You should not use the ThAr data for the wavelength calibration.
2014-08-11:
Observed Change of Line Blurring by Arc Lamp Position
An example of the change of line blurring with Arc-Lamp position. The images above are for channel-2 (Apr 2014 data). See the panner window for the area zoomed, as well as a rough idea for which part of the MOS mask it is illuminated. When the Arc Lamp become near, the edge illumination become significant and generally shows more blurring. Such dependence on the cal probe position cannot be explained by the problems on the grism itself. More investigation will follow.
The same images for channel-1 data (Apr 2014 data). The above example may be understood if both edges are illuminated when the slits have more light.
2014-08-05
Recently it is recognized that the R1300 grism shows slight image degradation. What we have found is that the degradation shows the position dependence on channel-2. The symptom has been already there at least since early 2013. Before that, only broad wing of the spectral image were recognized. As the advance speed is apparently slow, we don't close the grism for now. The investiation is still on-going. The use of the R1300 grism will be under shared-risk policy.
The dull appearance of the spectra has been recognized at least before 2012. The top figure below shows the comparison of the slit profile (width=0.8arcsec), Arc Lamp profile from R1300 grism and from low-resolution zJ500 grism. Usually the intrinsic profile of grism is enough narrow, so the profile of arc lamp resembles the slit profile, just like the case for zJ500 grism. On the other hand, R1300 grism line profle is more like Gaussian, implying the significant degradation by the intrinsic profile. But on the FWHM the effect of blurring is not evident for the slit width over 5 pixels (bottom figure). The R1300 users should be careful about existence of the wing on the intrinsic line profile when you interpret the science data.
The new issue recognized is the position dependence of the additional blurring. As is evident in the 2nd and 4th figures, some part of the spectra show significant broadening. A close look of the blurred line profile shows a hint of two closely-separated lines, implying a cracking of a grism substrate.
The comparison of the line profile (zJ500, R1300, and slit profile).
The comparison of the ThAr line profile of R1300 grism at two different position (y=1100pix and 1500 pix) as well as the slit profile (from long-slit data taken in Nov 2013). The measured direct FWHM of each profile is 4.3pix, 5.7pix, and 4.0 pix for y=1100, y=1500, and slit profile, respectively. Significant broadening of R1300 profile for y=1500pix is evident.
The relation of the slit width and the arc lamp line profile by the slits (channel 2). The longslit data from late 2012 (blue) and early 2014 (red) is shown. Significant broadening of the arc lamp fwhm for 2014 data is seen for narrow-slit side. The deviant data point at 4pix is from the slit at upper side on the detector.
An example of the line blurring on channel-2 ThAr arc lamp spectra. The ThAr line (near 2.4um) show a clear and gradual blurring from bottom to top in a slit, as seen in the measured FWHM. The displayed region is near middle bottom (near [730:940,230:550] on raw image) of channel-2 data.
Any questions should be directed to the Support Astronomer (Ichi Tanaka: e-address is below).
Last update: 2016-07-05
by Ichi Tanaka (ichi$naoj.org [change $ to @])