Princeton University Observatory

Princeton, New Jersey

Stray Light in PEP

D. G. York

Recently, a test was done on Lambda Ori to determine the cause of residual light at the bottom of saturated absorption lines. The basic procedure was to hold U1 fixed at the bottom of four lines: 1026 Å, 1077 Å, 1216 Å, and 1302 Å, while moving Carriage two. Far from the suspected stray light source (50 Å longward of the U1 slit), observations were made by moving U2 20 Å at a step, then doing a standard routine on U1 with U2 motionless. At Ly Alpha and Ly Beta, the standard routine did not take U1 out of the bottom of the line. For the 1302 line the standard routine included a point in the continuum shortward of the line wings. For the 1077 line, a point in the shortward wing was included.

These low resolution results are shown in Figures 1, 2, 3, and 4. In Figure 2, the regions where the slit and stray light hole are separately or simultaneously occulted are labeled. In Figure 3, the various background components are labeled as 1) particles, 2) stray light through the hole about 50 Å longward of the slit, and 3) scattered light. This last designation comes chiefly from the fact that a component of background remains after the dipping mirror occults the stray light hole and before the slit on U1 is occulted by the mirror, but the component disappears when the slit is occulted. This component may be due to grating scatter or to some other very localized scatter.

High resolution scans of the stray light occultation are shown in Figures 5 and 6 (with U1 at 1216 Å and 1026 Å, respectively). Counts on U1 are shown as a function of the U2 slit wavelength. The discontinuity after dump 1397 in Figure 5 may be partially explained by a high particle count on U1 (U3 showed a background level three times higher than normal). The discontinuity at dump 1415 in Figure 6 is not so easily explained.

The present position of the dipping mirror relative to the U2 slit is shown by occulting the U1 slit positioned at continuum points near Ly Alpha and Ly Beta in Figures 7 and 8.

Based on the various background components observed, the following scheme was derived for computing background corrections where saturated lines are not available, i.e., particularly in the case of molecular lines and unreddened stars.

Lambda	=	wavelength being observed on U1
S2,avg(Lambda + 55)	=	average continuum level on U2 measured from a point 55 Å greater than Lambda and averaged over ± 17Å.
S2,avg(Lambda ± 10)	=	average U2 continuum centered at Lambda averaged over ± 10Å.
a	=	a correction to derive S1,avg(Lambda ± 10), comparable to S2,avg(Lambda ± 10). a is plotted in Figure 9.
S1,avg(Lambda ± 10)	=	S2,avg(Lambda ± 10)/a

The stray light on U1 at Lambda in counts/14 sec on U1 is

S_t = 0.1 x C₁ x S_2,avg(Lambda + 55) where C₁ is defined below.

The scattered light is

S_c = 0.08 [S_1,avg(Lambda ± 10) - 0.1 x S_2,avg(Lambda + 55)]

To determine the state of occultation, compute the quantity Delta E = E_U2-E_U1, the wavelength difference of U2 and U1, in Å. Define the additional correction for occulted stray light as C1.

If Delta E < -170, C1 = 1.0.

If -170 < Delta E < -100, C1 ~ 0.75 - 1.0, a poorly defined effect which is not understood. This condition does not occur often.

If Delta E > 0, compute

z = Delta E - 0.143 Lambda_U1 + 35
If |z| <= 17, C1 =(17 + z)/35.
If |z| > 17, C1 = 1.

The equation for z is derived using a measured wavelength dependence in the separation of U1 and U2 when half the stray light is occulted.

The above formalism applied to Lambda Ori observations of saturated lines gives predicted values within about 10% of those observed for residual signal at the center of saturated lines. For Tau Sco (B0 V), Gamma Ara (B1 Ib) and Delta Per (B5 III), the answers given by the formalism must be increased by 30%. With this addition, 10 to 15% accuracy is achieved. The background particle counts may vary from 15/14 sec to 80/14 sec, and the counts at the line centers for all but the shortest lines in Tau Sco are less than 2000/14 sec. These have not been taken into account and probably account for a part of the error.

The above statements of accuracy do not apply to the regions Delta E <= |17|, or in fact Delta E <= |25|. There is some evidence that the apparent size of the hole changes as U1 moves to shorter wavelength, as is shown by the comparison of stray light occultations in Figures 2 and 4.

The present results are probably acceptable for U1 wavelengths 1100 Å < Lambda < 1400 Å for all z. At shorter wavelengths they are acceptable for |z| > 25, but for smaller |z|, they must be applied with caution. In particular, the features near 1036 Å are often affected by partial occultation of the stray light hole. L. Spitzer has obtained good predictions of background for these lines using the following corrections.

Compute EL - E2 where E2 = the Carriage 2 encoder reading for a particular observation on U1 and EL = the corresponding longward occultation point of the V1 slit. These numbers are available in the "LS" printouts.

Compute C₂ = 43000 - (EL - E2) = distance of C2 mirror from edge of U1 beam, in encoder bits.

If C₂ < 15 K, C1 = 0.

If C₂ > 30 K, C1 = 1.

If 15 K < C₂ < 30 K, C1 = (C2 - 15 K)/15 K.