Physical conditions in BAL outflows

Sections below:

Instrument configuration

WHT/ISIS with the REDPLUS CCD on the red arm and EEV12 on the blue arm, spatial scale 0.20 arcsec/pixel. With the R1200 gratings, the red arm delivers a spectral range of 630 A unvignetted, scale 0.26 A/pixel. The corresponding numbers for the blue arm are 800 A and 0.23 A/pixel.

Good spectroscopic focus is crucial. The targets are all point sources.

Observations required

Calibrations 

Afternoon:
 Biases: 10 in each arm
 W flats: ~ 5 for each configuration, peak counts ~ 30k
   (We'll use many configurations, so will not attempt to get sky flats.)

During the night:
  Arcs: one red+blue for each target/configuration, (and another pair
    every 2 hours, if the observations last that long),
  Standard: one with each configuration e.g. during 31/12/05+1/1/06
    we used SP0305+261, SP0642+021, SP0946+139, SP1239+178

Configuration 

Slit - match to seeing, but maximum 1 arcsec, to avoid compromising
  the resolution
Dichroic - use the standard one, no need to change during the run
Dekker - clear8
Order-sorting filter - GG495 in red arm (NB on Jan 28 it was discovered
  that the GG495 filter was not physically present in the slide, so this
  could also have been the case Jan 6+7)
Gratings - R1200R, R1200B
Central wavelengths - target-specific
Exposure times - 1800 sec (or 2*1300 sec for 45-min exposures)
CCD
     window: 300 pixels in x, full range in y
     bin:    2 1 (i.e. bin x2 in spatial direction)
     speed:  slow

Observations made

We made ~ 180 observations of ~ 38 QSOs during (parts of) the 9 ITP WHT nights, and 5 earlier non-ITP WHT nights. Most of these observations are of 1800 sec and yield 2 spectra: one from the red arm, one from the blue arm (340 in total). The ~ 180 observations represent ~ 90 hours of integration, but a substantial fraction of this was obtained through light cloud or in poor seing.

Data reduction

Quick-look spectra (no debiasing or flat-fielding) have been extracted from all raw images showing significant counts. Paul, Sarah and Flori have made full reductions of some of the data.

The 14 (ITP + earlier) nights are listed below: 

  Night      X        Ywcen     Spec FWHM
           pixels     pixels      pixels
          red blue   red  blue   red blue
A 051231  130. 100. 2050. 2090.  3.6  4.0
B 060101  160. 140. 2050. 2050.  3.4  4.0
C 060621  100. 105. 2050. 2030.  3.5  3.4
D 060622   85.  70. 2110. 2070.  3.6  4.6
E 070310    0.   0. 1840. 2120.  0.0  0.0

F 080106   62. 105. 1850. 2260.  2.6  3.7
G 080107   65.  65. 1750. 1960.  2.8  3.7
H 080414    0.   0.    0.    0.  0.0  0.0
I 080415    0.  70.    0. 2080.  0.0  0.0
J 080424   69.  70. 1830. 1800.  2.8  0.0
K 080425   82.  65. 1790. 1960.  2.9  3.6
L 080426   75.  65. 1890. 1970.  2.9  4.2
M 080602    0.   0. 1780. 2050.  2.3  3.4
N 080603   67.   0. 1810. 2110.  2.3  3.6

0 = not measured.
The columns are as follows:
1 - reference letter used in the plots below
2 - date
3, 4 - typical X positions of the spectrum on the red and blue CCDs
5, 6 - approx (+- 50) Y value of the requested central wavelengths on the CCDs
7, 8 - spectroscopic FWHMs of sky lines on the CCDs, in pixels

Spectroscopic resolution
The median FWHM of sky lines for the different arm (R, B) and CCD combinations, i.e. R/MAR2 (up to 080622), R/RED+ (from 070310) and B/EEV12, are 3.5, 2.7 and 3.7 pixels, or 0.81, 0.70 and 0.85 A, i.e. resolution up to ~ 10000 in the red.
The measured FWHM each night are consistent with the expected values of 3.2, 2.9 and 3.3 pixels.

S:N achieved
Figs. 1, 2 and 3 below show S:N vs counts, S:N (quick-look) vs S:N (final) and counts vs SDSS r mag.


Fig. 1 - log 10 S:N/pixel vs log10 counts/pixel. Different letters indicate different nights (see above). Lowercase and uppercase indicate quick-look and final reductions respectively. The measured values are broadly consistent with the prediction (dotted curve), and confirm that we need ~ 200 counts/pixel to deliver a S:N ~ 10 per pixel. NB one resolution element is ~ 3 pixels.


Fig. 2 - log 10 S:N/pixel (quick-look reduction) vs log 10 S:N/pixel (final reduction). This shows that the final reduction delivers a median improvement of a factor of 1.3 in S:N (nights C, F, G). This suggests that the usefulness of the data for an individual QSO can largely be assessed by inspection of the quick-look spectra. The above plot also shows some problem with the data for night D (under investigation).


Fig. 3 - log10 counts/pixel vs SDSS r mag (NB this is not the r mag according to which the QSOs were selected). The predicted curve (from the ING's exposure-time calculator) is a resonable fit to the data from clear nights C D G N. It provides the expected upper envelope for data obtained on the partially-clouded nights A B F, and the bad-seeing nights J K L M, when we kept integrating on individual quasars in bad conditions.

Back to main BAL ITP page

Chris Benn (crb@ing.iac.es)

Version: 2008 Dec 10