Spectrum

For the following analysis, we have used data from the TRW thermal vacuum tests (also referred to as AXAFTV) at TRW facilities during May-June, 1998. As the TV tests were primarily intended for a systems check of the entire spacecraft, ACIS was given limited time in telemetry format 2, the format used for acquiring science data. Instead, ICM data was collected in graded histogram mode. In this configuration the instrument finds and grades events, disregarding individual frames. Acceptable events (ASCA g02346) are written to a histogram, and after a specified number of frames, the histogram is telemetered.

The better response of the FI chip (relative to the BI chip) makes it the best choice for detailed studies of the ICM spectrum. Unless otherwise noted the data is from S2, quadrant D. At the same time, since most of the ICM flux falls on S3, we also discuss the data from the BI chip. In both cases, the data results from a total of 38,400 frames of normal integration (3.3 sec) data. Using a number of easily identified lines (see below), gains and offsets were calculated to convert energy units (ADU) to energy (keV). To facilitate the discussion, the spectral range is divided into three different energy bands, with only descriptive significance given to the names of each band.

Low Energy Band (0-1 keV): An examination of Figure 4.106 reveals a rich diversity of lines below 1 keV. There exists three main complexes of lines. The highest energy complex is clearly due to the three most common Ni L lines. The spectral resolution of the device does not allow separation of the individual lines. Instead, we have computed an average Ni L-line, using the expected intensities of the lines: $\alpha_{1}$:$\alpha_{1}$:$\beta{1}$ 100:10:50. Table 4.76 lists the various low energy lines and computed average line energy for all the low energy features. The next complex, labeled the ``L Complex'' is centered around 0.67 keV. Although a gaussian nicely fits this feature, two problems exist with identifying it with a single emission line. First, the only line at this energy if F K$\alpha_{1}$ (677 eV) and no fluorine is present in the ICM. Second, the width of the best-fit gaussian to this features is twice as large as that measured for the width of the F K$\alpha_{1}$ during sub-assembly calibration. Thus, we conclude that this feature is a blend of at least two distinct lines, and most probably a combination of the expected Mn and Fe L lines. Referring to Table 4.76 and assuming equal contributions of the average Mn and Fe L lines, the identification of this feature as the ``L Complex'' becomes clear. Finally, there exists a complex centered around 0.55 keV. Due to the low count rate, a clear identification is not possible. At the same time, the tentative identification of the complex as a blend of O K$\alpha _{1,2}$ and Cr L seems rather reasonable.
 
 

Figure 4.106: FI chip response to the ICM spectrum in the range 0-1 keV.

The poorer spectral resolution of the BI device and its slight non-linearity in gain blends the low-energy lines together and displaces the features with respect to their expected energy location as seen in Figure 4.107. Instead of the three complexes seen by the FI chip, there are only two visible in the BI chip. The Ni L line feature is still discernible, as is the Mn-Fe L complex, but it now contains the O K$\alpha$ and Cr L lines. When a functional form is fit to the L complex to determine the K:L fluorescence yield, what constitutes the L complex will differ between a BI and FI chip. Thus, even after accounting for the different quantum efficiencies for the BI and FI chip, we expect K:L ratio as determined by the FI, will be higher than the K:L ratio measured by the BI chip.
 
 

Table 4.76: ICM lines in the 0-1 keV band

 

Line	Energy (keV)	Line	Energy (keV)
O K$\alpha _{1,2}$	0.525
Cr L$\alpha _{1,2}$	0.571	Fe L$\alpha _{1,2}$	0.704
Cr L$\beta_{1}$	0.581	Fe L$\beta_{1}$	0.717
<Cr L>	0.574	<Fe L>	0.708
Mn L$\alpha _{1,2}$	0.636	Ni L$\alpha _{1,2}$	0.849
Mn L$\beta_{1}$	0.647	Ni L$\beta_{1}$	0.866
<Mn L>	0.639	<Ni L>	0.854

 
 

Figure 4.107: BI chip response to the ICM spectrum in the range 0-1 keV.

Middle Energy Band (1-7 keV): Figure 4.108 shows the FI response to the ICM source in the 1-7 keV band. In this energy range Mn emission lines are directly responsible for most of the spectral features. The two brightest lines are Mn K$\alpha _{1,2}$ and Mn K$\beta_{1}$. The corresponding escape features of these two lines are also present in the spectrum. The energies of the escape lines equal the emission line energy minus the energy of a Si K$\alpha _{1,2}$ fluorescence photon (1.74 keV). The Si K$\alpha _{1,2}$ fluorescence peak is also clearly visible. The relatively flat continuum at the $\sim$10 count level is the ``shoulder'' of both the Mn K$\alpha _{1,2}$ and Mn K$\beta_{1}$ lines that is due to the incomplete collection of the charge generated by the incident photons.
 
 

Figure 4.108: FI chip response to the ICM spectrum in the range 1-7 keV.

In addition to the the Mn lines, there is also weak evidence for a Cr K$\alpha _{1,2}$ line at the beginning of the Mn K$\alpha _{1,2}$ shoulder. Taken alone, the existence of the line seems marginal, but when coupled with detection of Cr L lines at lower energies, this identification becomes more secure. Finally, an Al K$\alpha _{1,2}$ line is clearly visible. Table 4.77 lists all the lines and energies for this band.
 
 

Table 4.77: ICM lines in the 1-7 keV band

 

Line	Energy (keV)	Line	Energy (keV)
<Al K$\alpha _{1,2}$>	1.487	<Mn K$\alpha _{1,2}$> esc	4.155
<Si K$\alpha _{1,2}$>	1.740	Mn K$\beta_{1}$ esc	4.750
Au M$\alpha _{1,2}$	2.112	<Cr K$\alpha _{1,2}$>	5.411
Au M$\beta$	2.205	<Mn K$\alpha _{1,2}$>	5.895
Au M$\gamma$	2.410	Mn K$\beta_{1}$	6.490

 
 

Figure 4.109: BI chip response to the ICM spectrum in the range 1-7 keV.

Figure 4.109 shows the BI response to the ICM source in the 1-7 keV band. In addition to the increased width of the spectral lines, the continuum level for the BI chip is much higher (a factor of $\sim$20) and has a slow energy dependence. Due to the high continuum level, the Al K$\alpha _{1,2}$ line is not visible in the BI spectrum, and the Si K$\alpha _{1,2}$ line and Au M complex are just above the continuum level.

High Energy Band (7-15.5 keV): Figure 4.110 shows the FI response to the ICM source in the 7-15.5 keV band. Since this band begins at an energy above that of the highest Mn K emission line, we a priori expect 1) few events and 2) that those events will have an unusual origin. The rich spectral features arise from three distinct processes: pile-up, excitation by cosmic rays, and instrumental saturation.

Figure 4.110: FI chip response to the ICM spectrum in the range 7-15.5 keV

Pile-up results when charge from two or more absorbed photons are collected in the same pixel during a single integration time. Since Mn K$\alpha _{1,2}$ is the most abundant line, it is not surprising that all but one of the pile-up ``lines'' involves the summation of a Mn K$\alpha _{1,2}$ with another line. Table 4.78 lists the seven most probable pile-up lines in this band and their energies. The counting statistics are too low to absolutely resolve all of the lines listed (Au M + Mn K$\alpha _{1,2}$, for example), but given a sufficiently long integration, their existence would be obvious. A completely different mechansim is responsible for the Au L lines. These lines arise not from the ICM source, but from cosmic rays that are absorbed by the Au covering of the frame store. The origin of these lines is confirmed by the existence of Au L and Au M lines in the spectrum of all CCDs, including ones that are not directly illuminated by the ICM. This means that every spectrum will contain some level of Au M and Au L contamination. The highest energy features are due to the saturation of the digital chain. The spike at 15.22 keV corresponds to the highest analog-to-digital (ADU) value allowed (4096). The spikes at $\sim$14.1 keV and $\sim$13.7 keV correspond to 4096 ADU minus the bias level. The extremely narrow width of these line clearly indicates a non-emission origin. As the gain and bias level varies quadrant to quadrant, these features will occur at a range of energies. Table 4.78 also lists all the instrumental and cosmic-ray produced lines and energies.
 
 

Table 4.78: ICM lines in the 7-15.5 keV band

 

Line	Energy (keV)	Line	Energy (keV)
Mn K$\alpha _{1,2}$ + Si K$\alpha _{1,2}$	7.635	Au L$\beta_{2}$	11.52
Mn K$\alpha _{1,2}$ + Au M$\alpha _{1,2}$	8.007	Mn K$\alpha _{1,2}$ + Mn K$\alpha _{1,2}$	11.79
Mn K$\alpha _{1,2}$ + Au M$\beta$	8.095	Mn K$\alpha _{1,2}$ + Mn K$\beta_{1}$	12.39
Au L$\alpha_{1}$	9.711	Mn K$\beta_{1}$ + Mn K$\beta_{1}$	12.98
Mn K$\alpha _{1,2}$ + Mn K$\alpha _{1,2}$ esc	10.05	4096 ADU - bias	13.6-14.2$^{\dagger}$
Au L$\beta_{1}$	11.44	4096 ADU	15.22$^{\dagger}$
$^{\dagger}$: these energies are dependent on the gain and bias levels and will vary from chip to chip

Figure 4.111 shows the BI response to the ICM in the high energy band. In general, there is very little difference between the BI and FI spectra at the these energies. In addition to the discrete pile-up features, a flat continuum (few count level) extends from 7 keV up to highest pile-up line, a result of the summation of Mn K$\alpha _{1,2}$ photons with the low-energy tail from the Mn K lines. The energy response in the BI chips is higher than the FI chips ($\sim$ 18 keV versus $\sim$ 16 keV) due to the higher gain in the BI. The different gain and bias levels shifts the saturation features to higher energies than in the FI chip.
 
 

Figure 4.111: BI chip response to the ICM spectrum in the range 7-18 keV