Astrophysical X-Ray Spectroscopy: Then,
Then Again, and Now

Steven M. Kahn

Columbia University

Introduction

For the last three years, the grating experiments on Chandra and XMM-Newton have been providing magnificent spectra of nearly all classes of cosmic X-ray sources. In most cases, these are the very first high resolution X-ray spectra we have had available.

As an introduction to this meeting, I thought it might be fun to look back to where we came from, to get to where we are today in this field.

I have chosen three snapshots in time:

Then: Summer 1980 - I was driving across the U.S. (Berkeley to Boston). We had results from Uhuru, Ariel V, OSO 8, HEAO 1, and were just starting to get results from Einstein.

Then Again: Summer 1995 - I was driving across the U.S. (Berkeley to NY). EXOSAT, BBXRT and ROSAT had flown in the interim. We were starting to get results from ASCA.

Now: Summer 2002 - Chandra has been up for three years, XMM-Newton for 2 1/2. Astro-E tragically lost, but Astro-E2 in development.

Late-Type Stars: Then

Capella was the first extrasolar stellar coronal source detected in X-rays (rocket expt: Catura, Acton, and Johnson 1975; ANS: Mewe et al. 1975)

The HEAO 1 A-2 LEDs were the first to provide decent spectra at low energies.

Cash et al. showed that the spectra were significantly better fit (c²_red ~ 1) for models with line emission, then for simple continuum models (c²_red ~ 5).

They found T ~ 4 x 10⁶ - 2 x 10⁷ K.

Late-Type Stars: Then Again

The launch of ASCA enabled the first X-ray spectra of cosmic sources with CCD resolution - roughly 5 - 6 times better than the resolution achievable with the earlier prop counter experiments.

Observations of stellar coronae indicated the presence of multiple temperature components, as well as anomalous abundances for O, Ne, Mg, Si, S, and Fe.

This spectrum of RZ Cas (an Algol binary) indicated subsolar abundances by factors of 2 - 5, very similar to what had been found for RS CVn’s, but not for less active single stars.

Late-Type Stars: Now

Early-Type Stars: Then

Early-type stars were not known to be X-ray sources prior to the launch of the Einstein Observatory.

This class of sources was “discovered” serendipitously through extended observations of the binary Cyg X-3. Five of the brightest O and B stars in the nearby VI Cygni association showed up as point sources.

They could only infer spectral properties using broadband countrates. They typically found: T ~ 5 - 10 MK, N_H ~ 10²² cm^-2, and L_X ~ few x 10³³ ergs/s ~ 10^-4 L_V.

Early-Type Stars: Then Again

The first moderate spectral resolution observations of early-type stars were performed with the Solid State Spectrometer on Einstein, but a problem with ice buildup on those detectors reduced their effectiveness for soft sources.

The Broad-Band X-Ray Telescope, flown in 1990 on the Shuttle obtained somewhat better data.

The spectra of z Puppis, were used to infer the existence of ionized oxygen absorption intrinsinc to the stellar wind. In addition, the detection of Mg XI line emission was claimed.

Early-Type Stars: Now

X-Ray Binary Sources: Then

Most of the early rocket and satellite observations of X-ray binary sources concentrated on timing studies. The proportional counters had a limited number of pulse height channels suitable for spectral analysis.

Experiment C on Ariel V was designed to provided detailed X-ray spectra.

Cyg X-3 is a highly variable bright source with a 4.8 hr. period.

The Ariel V observation confirmed an earlier rocket detection of narrow Fe K emission, modulated with the continuum. The EW is ~ 0.33 keV.

No other features were detected. This was ascribed to the high fluorescence yield of iron.

X-Ray Binary Sources: Then Again

The ASCA Solid State Imaging Spectrometer observation of Cyg X-3 did indeed confirm the presence of Fe K emission in the spectrum, but indicated that it was far from the most prominent feature.

Liedahl & Paerels showed that in addition to H-like and He-like line emission from lower-Z elements, we also see narrow radiative recombination continua (RRCs).

These indicate that the line emitting gas is photoionized - suspected, but never previously demonstrated for accretion-powered sources.

X-Ray Binary Sources: Now

Supernova Remnants: Then

For older SNRs, shock heat gas to temperatures ~ few x 10⁶ K, where the plasma should radiate primarily via line emission from K-shell O and L-shell Fe.

Previous searches with crystal spectrometers had been inconclusive.

The HEAO A-2 LEDs were the best calibrated low energy prop counters yet flown. Although the resolution was not sufficient to see line emission in the raw data, Kahn et al. invoked a deconvolution procedure to demonstrate its presence conclusively.

The derived temperature was ~ 3 x 10⁶ K, with some evidence for multitemperature emission and/or nonequilibrium effects.

Supernova Remnants: Then

Supernova Remnants: Then Again

Although significantly lower in spectral resolution than the earlier FPCS, the ASCA SIS covered a much broader spectral range, which opened up new diagnostic opportunities.

Hughes et al. showed that the raw ASCA spectra of young LMC remnants could be used to distinguish Type Ia explosions from Type II’s. The Type Ia’s, as illustrated below are dominated by Fe L-shell and Si and S K emission, as opposed to the primary products of alpha-burning reactions (O, Ne, Mg).

Supernova Remnants: Then Again

For the bright galactic remnants, ASCA offered the opportunity to map emission line intensities within the remnant, and to look for systematic Doppler shifts.

The first such study was performed with Cas A, the youngest remnant in the Milky Way. The spectrum is of very high quality, and exhibits Ne, Fe-L, Si, S, Ar, Ca, and Fe-K emission.

The deconvolved Doppler map, derived primarily from the Si K lines, indicates a systematic flow. This confirmed earlier conclusions drawn from FPCS observations.

Supernova Remnants: Now

Interstellar Absorption: Then

At soft X-ray energies, interstellar attenuation is primarily due to K-shell photoelectric absorption by He, C, N, O, and Ne.

The most prominent “feature” expected is the neutral O K-edge at 0.532 keV. Although this was included in standard models at the time, it had never been explicitly detected.

Charles et al. used the HEAO A-2 LED observations of the Crab Nebula to study this feature. The Kahn & Blissett deconvolution procedure showed that the edge was there. Spectral fits indicated consistency with the standard cosmic abundance of oxygen, given the inferred hydrogen column to the source.

Interstellar Absorption: Then Again

The detection of the O K-edge was later explicitly confirmed in observations of the Crab performed with the FPCS on Einstein.

In addition, to the edge itself, Schattenburg & Canizares also found weak evidence for the 1s-2p absorption line in atomic oxygen.

The derived oxygen abundance agreed well with the earlier Charles et al. result, however the FPCS indicated a significantly higher total column density to the source.

Interstellar Absorption: Now

Cluster Cooling Flows: Then

The first imaging observations of clusters of galaxies with Einstein confirmed predictions that the cooling time of gas at the core of the intracluster media of many clusters should be less than the Hubble time, requiring the presence of a cooling flow.

One expected the presence of copious soft X-ray line emission from these cluster cores, incompatible with the background temperature of the cluster.

Canizares et al. detected O K-shell and Fe L-shell line emission, indicative of a range of temperatures, with the FPCS for M87 in the Virgo cluster.

Cluster Cooling Flows: Then Again

ASCA provided lower spectral resolution data, over a broader energy range, for a larger sample of cooling flow clusters.

The spectra exhibited some systematic discrepancies from standard cooling flow predictions.

This was ascribed partially to uncertainties in the atomic database (especially for the Fe L-shell transitions), and partially to excess absorption within the cluster.

Nevertheless, the data clearly indicated a soft excess for those clusters expected to exhibit cooling flows.

Cluster Cooling Flows: Now

The Reflection Grating Spectrometer on XMM-Newton is the first instrument capable of quantitatively testing the explicit cooling flow spectral predictions for a broad sample of clusters.

Surprisingly, we have found a systematic deficit of low temperature emission in all cases, starting at roughly half the background temperature.

This effect has been difficult to reconcile with any of the standard theoretical models.

Active Galactic Nuclei: Then

Proportional counter observations of AGNs had generally shown that their spectra were dominated by power law continua, attenuated at low energies by interstellar and/or circumsource absorption.

The first weak evidence for Fe K line emission came from higher statistics observations obtained with OSO 8 and HEAO 1 in the late 1970’s.

NGC 4151 was one of the most intensively studied sources. Mushotzky et al. were the first to report the Fe K line, although it was only a 2s detection.

Active Galactic Nuclei: Then Again

From earlier proportional counter observations, Seyfert 2 galaxies were found to be much more highly absorbed than Seyfert 1s, but not much else was known about the details of their spectral complexity.

The prototype object in this class was observed with BBXRT and found to exhibit discrete residuals, but the spectrum proved difficult to interpret. Marshall claimed a significant underabundance of oxygen.

ASCA provided a much higher quality spectrum, exhibiting narrow Fe K emission, and a host of lower energy features.

Ueno interpreted this in terms of a power law continuum, with an additional thermal component at low energies due to starburst activity.

Active Galactic Nuclei: Now

Summary and Conclusions

It is illuminating to see how our understanding of astrophysical sources evolves as the quality of our X-ray spectral data improves.

In some cases, the very early measurements and interpretations were remarkably accurate, even though the data were quite crude (e.g. stellar coronae, interstellar absorption).

In others, the higher resolution spectra have shown that our earlier spectral models were completely wrong (e.g. Seyfert 2’s, early-type stars).

I do not expect to see such a significant “jump” with future experiments. Hopefully, X-ray spectroscopy will then become more of a standard tool, and less of a novelty.

In any case, the last three years have been very exciting. I suspect we can look forward to at least a few more years in the same vein.


	For the last three years, the grating experiments on Chandra and XMM-Newton have been providing magnificent spectra of nearly all classes of cosmic X-ray sources. In most cases, these are the very first high resolution X-ray spectra we have had available.

	As an introduction to this meeting, I thought it might be fun to look back to where we came from, to get to where we are today in this field.

	I have chosen three snapshots in time:
		Then: Summer 1980 - I was driving across the U.S. (Berkeley to Boston). We had results from Uhuru, Ariel V, OSO 8, HEAO 1, and were just starting to get results from Einstein.
		Then Again: Summer 1995 - I was driving across the U.S. (Berkeley to NY). EXOSAT, BBXRT and ROSAT had flown in the interim. We were starting to get results from ASCA.
		Now: Summer 2002 - Chandra has been up for three years, XMM-Newton for 2 1/2. Astro-E tragically lost, but Astro-E2 in development.


	Capella was the first extrasolar stellar coronal source detected in X-rays (rocket expt: Catura, Acton, and Johnson 1975; ANS: Mewe et al. 1975)

	The HEAO 1 A-2 LEDs were the first to provide decent spectra at low energies.

	Cash et al. showed that the spectra were significantly better fit (c²_red ~ 1) for models with line emission, then for simple continuum models (c²_red ~ 5).

	They found T ~ 4 x 10⁶ - 2 x 10⁷ K.


	The launch of ASCA enabled the first X-ray spectra of cosmic sources with CCD resolution - roughly 5 - 6 times better than the resolution achievable with the earlier prop counter experiments.

	Observations of stellar coronae indicated the presence of multiple temperature components, as well as anomalous abundances for O, Ne, Mg, Si, S, and Fe.

	This spectrum of RZ Cas (an Algol binary) indicated subsolar abundances by factors of 2 - 5, very similar to what had been found for RS CVn’s, but not for less active single stars.


	Early-type stars were not known to be X-ray sources prior to the launch of the Einstein Observatory.

	This class of sources was “discovered” serendipitously through extended observations of the binary Cyg X-3. Five of the brightest O and B stars in the nearby VI Cygni association showed up as point sources.

	They could only infer spectral properties using broadband countrates. They typically found: T ~ 5 - 10 MK, N_H ~ 10²² cm^-2, and L_X ~ few x 10³³ ergs/s ~ 10^-4 L_V.


	The first moderate spectral resolution observations of early-type stars were performed with the Solid State Spectrometer on Einstein, but a problem with ice buildup on those detectors reduced their effectiveness for soft sources.

	The Broad-Band X-Ray Telescope, flown in 1990 on the Shuttle obtained somewhat better data.

	The spectra of z Puppis, were used to infer the existence of ionized oxygen absorption intrinsinc to the stellar wind. In addition, the detection of Mg XI line emission was claimed.


	Most of the early rocket and satellite observations of X-ray binary sources concentrated on timing studies. The proportional counters had a limited number of pulse height channels suitable for spectral analysis.

	Experiment C on Ariel V was designed to provided detailed X-ray spectra.

	Cyg X-3 is a highly variable bright source with a 4.8 hr. period.

	The Ariel V observation confirmed an earlier rocket detection of narrow Fe K emission, modulated with the continuum. The EW is ~ 0.33 keV.

	No other features were detected. This was ascribed to the high fluorescence yield of iron.


	The ASCA Solid State Imaging Spectrometer observation of Cyg X-3 did indeed confirm the presence of Fe K emission in the spectrum, but indicated that it was far from the most prominent feature.

	Liedahl & Paerels showed that in addition to H-like and He-like line emission from lower-Z elements, we also see narrow radiative recombination continua (RRCs).

	These indicate that the line emitting gas is photoionized - suspected, but never previously demonstrated for accretion-powered sources.


	For older SNRs, shock heat gas to temperatures ~ few x 10⁶ K, where the plasma should radiate primarily via line emission from K-shell O and L-shell Fe.

	Previous searches with crystal spectrometers had been inconclusive.

	The HEAO A-2 LEDs were the best calibrated low energy prop counters yet flown. Although the resolution was not sufficient to see line emission in the raw data, Kahn et al. invoked a deconvolution procedure to demonstrate its presence conclusively.

	The derived temperature was ~ 3 x 10⁶ K, with some evidence for multitemperature emission and/or nonequilibrium effects.


	Although significantly lower in spectral resolution than the earlier FPCS, the ASCA SIS covered a much broader spectral range, which opened up new diagnostic opportunities.

	Hughes et al. showed that the raw ASCA spectra of young LMC remnants could be used to distinguish Type Ia explosions from Type II’s. The Type Ia’s, as illustrated below are dominated by Fe L-shell and Si and S K emission, as opposed to the primary products of alpha-burning reactions (O, Ne, Mg).


	For the bright galactic remnants, ASCA offered the opportunity to map emission line intensities within the remnant, and to look for systematic Doppler shifts.

	The first such study was performed with Cas A, the youngest remnant in the Milky Way. The spectrum is of very high quality, and exhibits Ne, Fe-L, Si, S, Ar, Ca, and Fe-K emission.

	The deconvolved Doppler map, derived primarily from the Si K lines, indicates a systematic flow. This confirmed earlier conclusions drawn from FPCS observations.


	At soft X-ray energies, interstellar attenuation is primarily due to K-shell photoelectric absorption by He, C, N, O, and Ne.

	The most prominent “feature” expected is the neutral O K-edge at 0.532 keV. Although this was included in standard models at the time, it had never been explicitly detected.

	Charles et al. used the HEAO A-2 LED observations of the Crab Nebula to study this feature. The Kahn & Blissett deconvolution procedure showed that the edge was there. Spectral fits indicated consistency with the standard cosmic abundance of oxygen, given the inferred hydrogen column to the source.


	The detection of the O K-edge was later explicitly confirmed in observations of the Crab performed with the FPCS on Einstein.

	In addition, to the edge itself, Schattenburg & Canizares also found weak evidence for the 1s-2p absorption line in atomic oxygen.

	The derived oxygen abundance agreed well with the earlier Charles et al. result, however the FPCS indicated a significantly higher total column density to the source.


	The first imaging observations of clusters of galaxies with Einstein confirmed predictions that the cooling time of gas at the core of the intracluster media of many clusters should be less than the Hubble time, requiring the presence of a cooling flow.

	One expected the presence of copious soft X-ray line emission from these cluster cores, incompatible with the background temperature of the cluster.

	Canizares et al. detected O K-shell and Fe L-shell line emission, indicative of a range of temperatures, with the FPCS for M87 in the Virgo cluster.


	ASCA provided lower spectral resolution data, over a broader energy range, for a larger sample of cooling flow clusters.

	The spectra exhibited some systematic discrepancies from standard cooling flow predictions.

	This was ascribed partially to uncertainties in the atomic database (especially for the Fe L-shell transitions), and partially to excess absorption within the cluster.

	Nevertheless, the data clearly indicated a soft excess for those clusters expected to exhibit cooling flows.


	The Reflection Grating Spectrometer on XMM-Newton is the first instrument capable of quantitatively testing the explicit cooling flow spectral predictions for a broad sample of clusters.

	Surprisingly, we have found a systematic deficit of low temperature emission in all cases, starting at roughly half the background temperature.

	This effect has been difficult to reconcile with any of the standard theoretical models.


	Proportional counter observations of AGNs had generally shown that their spectra were dominated by power law continua, attenuated at low energies by interstellar and/or circumsource absorption.

	The first weak evidence for Fe K line emission came from higher statistics observations obtained with OSO 8 and HEAO 1 in the late 1970’s.

	NGC 4151 was one of the most intensively studied sources. Mushotzky et al. were the first to report the Fe K line, although it was only a 2s detection.


	From earlier proportional counter observations, Seyfert 2 galaxies were found to be much more highly absorbed than Seyfert 1s, but not much else was known about the details of their spectral complexity.

	The prototype object in this class was observed with BBXRT and found to exhibit discrete residuals, but the spectrum proved difficult to interpret. Marshall claimed a significant underabundance of oxygen.

	ASCA provided a much higher quality spectrum, exhibiting narrow Fe K emission, and a host of lower energy features.

	Ueno interpreted this in terms of a power law continuum, with an additional thermal component at low energies due to starburst activity.


	It is illuminating to see how our understanding of astrophysical sources evolves as the quality of our X-ray spectral data improves.

	In some cases, the very early measurements and interpretations were remarkably accurate, even though the data were quite crude (e.g. stellar coronae, interstellar absorption).

	In others, the higher resolution spectra have shown that our earlier spectral models were completely wrong (e.g. Seyfert 2’s, early-type stars).

	I do not expect to see such a significant “jump” with future experiments. Hopefully, X-ray spectroscopy will then become more of a standard tool, and less of a novelty.

	In any case, the last three years have been very exciting. I suspect we can look forward to at least a few more years in the same vein.