arXiv:0907.4348v1 [astro-ph.SR] 24 Jul 2009
A Possible Period for the K-band Brightening Episodes of GX 17+2
Jillian Bornak1†, Bernard J. McNamara1†‡, Thomas E. Harrison1†, Michael P. Rupen2, Reba M.
Bandyopadhyay3‡, Stefanie Wachter4
The low mass X-ray binary and Z source GX 17+2 undergoes infrared K-band
brightening episodes of at least 3.5 magnitudes. The source of these episodes is not
known. Prior published K-band magnitudes and new K-band measurements acquired
between 2006 and 2008 suggest that the episodes last at least 4 hours and have a period
of 3.01254 ± 0.00002 days. Future bright episodes can be predicted using the ephemeris
JDmax(n) = 2454550.79829+(3.01254±0.00002)(n) days. A growing body of evidence
suggests that the GX 17+2 could have a synchrotron jet, which could cause this activity.
Subject headings: radiation mechanisms: non-thermal — stars: neutron — X-rays:
binaries — stars: individual (GX 17+2) — ISM: Jets and Outflows
The low mass X-ray binary (LMXB) GX 17+2 is one of the brightest persistent X-ray sources
in the sky and its high energy behavior has been extensively monitored by X-ray satellites. It is
one of only eight objects known as Z sources, so-called for the track they trace out in an X-ray
color-color diagram. Z sources are thought to be neutron star binaries accreting near or at the
Eddington rate, and position along the Z is thought to reflect a change in mass accretion rate.
1Department of Astronomy, New Mexico State University, 1320 Frenger Mall, Las Cruces, NM 88003; jbor-
email@example.com, firstname.lastname@example.org, email@example.com
2NRAO, 1003 Lopezville Road, Socorro, NM 87801; firstname.lastname@example.org
3Department of Astronomy, University of Florida, Gainesville, FL 32611; email@example.com
ScienceCenter, CaltechM/S220-6,1200E.CaliforniaBlvd., Pasadena,CA91125;
†Visiting astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatory, which is operated
by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National
‡Visiting astronomer, Cerro Tololo Inter-American Observatory, National Optical Astronomy Observatory, which
is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with
the National Science Foundation
– 2 –
The radio emission of GX 17+2 is already known to be connected with position in this diagram
(Penninx et al. 1988; Migliari et al. 2007). Associating the multiwavelength behavior of a LMXB
with its Z position provides an excellent way in which to test theoretical models of these systems
(Psaltis et al. 1995).
In contrast to the well-studied X-ray emission from GX 17+2, relatively little is known about
this object’s optical/infrared counterpart (Davidsen et al. 1976; Naylor et al. 1991; Deutsch et al.
1999). It was only recently that Callanan et al. (2002) determined the IR counterpart of GX
17+2 is blended with a G type field star 0.9” away, which was erroneously given the variable
star name NP Ser. These authors also observed a remarkable 3.5 mag difference between their
two K-band observations. The underlying cause of these bright episodes remains a mystery. The
X-ray data of GX 17+2, extending back to the early 1980s, do not show any evidence for eclips-
ing or dipping behavior, so these phenomena cannot be responsible for the variable IR emission
(Kuulkers et al. 1997; Penninx et al. 1988; Homan et al. 2002). The IR and soft X-ray fluxes also
do not appear to be correlated, making it unlikely that they arise from the reprocessing of X-rays
in an accretion disk surrounding the system’s neutron star or on the surface of the donor star
(Bandyopadhyay et al. 2002). Finally, free-free emission from an X-ray driven wind would also
produce a correlation between the X-ray and IR emission; therefore, this process appears to be
ruled out as well (Callanan et al. 2002).
Some attempts have been made to explain the perplexing nature of the GX 17+2 K-band bright
episodes, but they are based on very few measurements. Bandyopadhyay et al. (2002) used the Ohio
State Infrared IMager/Spectrometer on the Perkins 1.8m telescope at the Lowell Observatory to
obtain eighteen K-band magnitudes of the blended image of GX 17+2/NP Ser between July 16
and August 20, 1997. These magnitudes were compared to RXTE All-Sky-Monitor (ASM) one-day
average X-ray flux measurements to determine whether they are connected, but no correlations
were found. These K-band magnitudes and several hundred days of ASM data were then used to
conduct a search for periods between 4 and 200 days. No periodic behavior was found.
The study by (Callanan et al. 2002) used four K-band measurements, only one of which was
obtained during an IR bright episode. Callanan et al. (2002) argue the case for a jet interpretation
for GX 17+2. They note that the levels of the radio and K-band variability in GX 17+2 are
comparable suggesting that they have a common origin and that, if an optically thick synchrotron
power law index is assumed, an extrapolation of the maximum GX 17+2 radio flux density predicts
a peak magnitude of K = 14. This is close to the observed value that GX 17+2 reaches during an
IR bright state. Additionally, Russell et al. (2007) supports the idea that the GX 17+2 K-band
bright episodes are related to the presence of a jet. Based on an examination of 19 low magnetic
field neutron star binaries, they conclude that jets are necessary to explain the quasi-simultaneous
optical/IR/X-ray emission from high luminosity systems such as GX 17+2.
The fact that GX 17+2 is a Z-source might allow studies of its X-ray/IR behavior to further
our understanding of the multi-wavelength emission processes operating in other peculiar LMXBs.
– 3 –
The unusual nature of the IR emission from GX 17+2 and its very limited observational data base
led us to conduct a multi-year campaign to better quantify its K-band properties. In this Letter we
present the first evidence that the IR bright episodes of GX 17+2 last at least 4 hours and repeat
with a period of 3.01245 days. We also find that the shape of this system’s K-band light curve and
its peak brightness are highly variable.
2. Observations and Data Reduction
This study is based on new K-band measurements of GX 17+2 acquired between July 2006
and July 2009 as listed in Table 2. Measurements were obtained at the Astrophysical Research
Consortium’s Apache Peak Observatory (APO), and at the National Optical Astronomy Observa-
tory’s Kitt Peak National Observatory (KPNO). Additionally, we had 102 nights of unpublished
observations at the Cerro Tololo Inter-American Observatory (CTIO). The K-band photometry of
GX 17+2 reported by Callanan et al. (2002) and Bandyopadhyay et al. (2002) were also used, as
well as a Spitzer MIPS 24 µm observation (Wachter 2009).
K-band photometry was obtained in 2006 on the nights of UT July 1, 12, 14 and in 2009 on
the night of UT July 1 using the Near-Infrared Camera and Fabry-Perot Spectrometer (NICFPS)1
on the APO 3.5 meter telescope. NICFPS has a 0.273 arcsec/pixel scale, giving a 4.58 arcmin
square field of view. Exposures were 8s and a five point dither pattern was used. While NICFPS is
capable of resolving GX 17+2 from NP Ser, poor seeing prevented us from doing so. On October 4,
2007 GX 17+2 was observed using the Cornell Massachusetts Slit Spectrograph (CorMASS) in slit
view mode. CorMASS is a 256x256 pixel detector with a scale of 0.375 arcsec/pixel in slit mode
giving a 1.6 arcmin square field of view.
GX 17+2 was observed in 2007 from May 25 to 30 and in 2008 on July 8 and 13 with the
Simultaneous Quad Infrared Imaging Device (SQIID)2on the KPNO 2.1m telescope. SQIID si-
multaneously collects data in the JHK filters. SQIID has a scale of 0.69 arcsec/pixel providing a
roughly 5 arcmin square field of view. On these images GX 17+2 is blended with the image of
NP Ser. The data were taken with 8s exposures using a two point dither co-adding six images per
dither position for 2006 and a five point dither co-adding eight images per dither position for 2007.
The above datasets were analyzed using standard IRAF3reduction routines. Aperture photom-
etry was performed on the blend of GX 17+2 and NP Ser and relative photometry was performed
with respect to four bright field stars, whose K-band magnitudes were taken from 2MASS Point
Source Catalog. Dome flats were used for flat fields and the dithered images were median combined
3IRAF is distributed by the NOAO, which is operated by AURA under cooperative agreement with the NSF.
– 4 –
to create sky images. Flat-fielded and sky subtracted images for each position in the dither were
then combined to increase the S/N. Except for the extra care needed to select uncontaminated
calibration stars in the crowded region of GX 17+2, no unusual reduction issues were encountered.
CTIO K-band measurements of GX 17+2 were obtained during 51 nights in 2001 extending
from March 1 to June 9 and for 51 nights in 2007 from March 15 to July 11 with A Novel Dual
Imaging CAMera (ANDICAM)4on the 1.3m telescope operated by the Small and Moderate Aper-
ture Research Telescope System (SMARTS) consortium. ANDICAM has a 0.369 arcsec/pixel scale
giving a 6.29 arcmin square field of view. The 2001 data were taken with 30s exposures and a seven
point dither co-adding two images per dither position, and the 2007 have 60s exposures and a five
point dither pattern.
The reduction of the SMARTS data sets was considerably more challenging. The K-band
images possess a variable fringing pattern that cannot be totally removed. Therefore, although the
formal K-band measurement errors were ±0.07 mag, the transient noise pattern problem produced
unreliable photometry for sources as faint as NP Ser and GX 17+2. Therefore, these data were not
included in our period-searching analysis.
Unfortunately, the position of GX 17+2 along its Z was unknown during the IR bright observa-
tions. X-ray data was taken an hour after the 2006 July 12 observation with the Rossi X-ray Timing
Explorer (RXTE)5satellite’s Proportional Counter Array (PCA), where GX 17+2 was found to
be on the Normal Branch (McNamara et al. 2009, in prep.). Even though we have simultaneous
RXTE ASM data for our observations, we cannot recover the Z position due to the low X-ray flux
and the resultant uncertainties.
3.The K-band Period Search
Table 1 lists the six dates when a bright K-band episode of GX 17+2 was detected. During
four of the observing sessions at the APO and KPNO, light curves extending for approximately
4 hours were acquired, as shown in Figure 1. The 2007-10-04 entry in Table 1 refers to the APO
CorMass detection during which only four short exposures were obtained; the 1999-06-26 entry
refers to the IR bright observation by Callanan et al. (2002); and the 2008-10-21 entry refers to
a Spitzer MIPS 24 µm observation (Wachter 2009). These three observations are single images
rather than full light curves, and so the time of observation is shown rather than the time of peak
IR brightness. The estimated times given in Table 1 for the center of the bright period are much
more uncertain than the size of the given heliocentric correction.
The extent of a GX 17+2 bright episode is defined as the time interval during which the GX
– 5 –
17+2 and NP Ser blended image had a K-band magnitude brighter than 14.5 (see Callanan et al.
2002 for measurements of NP Ser). An entire bright episode is not covered by any of our light
curves, but our longest-duration light curve shows it can last for at least four hours (see Figure
1). The differing shapes of these light curves pose obvious problems for the determination of any
periodicity. First, the variety in shape does not permit the light curves to be folded as is done to
determine the period of an eclipsing system. Secondly, the maximum K-band magnitude from one
bright episode to another is not a constant. As shown in Table 1, the peak magnitude can vary by
at least ±0.28 magnitudes. These two properties do not allow snap-shot K-band magnitudes to be
associated with a particular phase.
Given the above limitations, the following procedure was used to determine the period of the
IR bright episodes of GX 17+2. The KPNO observations of two IR bright light curves (middle
two curves shown in Figure 1) were separated by about three days. Observations prior to and
between these two IR bright times showed GX 17+2 persistently faint, implying a three day period.
Therefore, a grid of phase points was constructed for candidate periods between 2.8 days and 3.2
days for the times during July 2006 and May 2007 listed in Table 1. The Callanan et al. (2002)
observation was not used in this analysis step. The standard deviation of the five phase points at
each epoch was plotted versus period. This is shown in the top panel of Figure 2. If the system’s
IR bright episodes are periodic, each of the times in Table 1 should have nearly the same phase.
This would then produce a minimum in this figure. In a system whose period is days long, 1-2 hour
uncertainties in the mid points of the IR bright times will broaden this minimum. The result of
this computation indicates that the IR bright episodes of GX 17+2 have a period of about 3.012
To better define this value, the bright Callanan et al. (2002) K-band observation of GX 17+2
was then included (see their Figure 1), and the above procedure was repeated. A finer grid step
in period was used and the searched interval was reduced to 3.0 to 3.1 days. The results of this
exercise are shown in Figure 2 (bottom). The minimum dispersion is located at a period of 3.01254
± 0.00002 days. The dates of our observations were constrained by the small window in which GX
17+2 was visible from APO and KPNO. The fact that our observations were taken nearly the same
time each year causes the multi-cusp pattern in the bottom of Figure 2.
The validity of our period was tested in three different ways. First, to test the robustness of
the period, the center of the bright episode times in Table 1 were randomly changed by 1-2 hours.
The period was then recomputed using the above procedure. Second, the period was calculated
using the time when the system’s brightness stopped its rise. For the first two tests, the derived
period changed by less than 0.00002 days. Lastly, since this period was initially computed using
only the July 2006 and May 2007 data, the ephemeris was used to predict IR bright and faint
episodes in October 2007, July 2008, and July 2009. The observations of GX 17+2 in IR bright
and IR faint states confirmed our period and eliminated the possibility of any aliasing. The period
calculation was then redone with the additional times included, and we find that this period was
consistent with all of our APO and KPNO non-detections.
– 6 –
We also attempted to use the K-band observations reported by Bandyopadhyay et al. (2002)
as a test. None of the SMARTS observations corresponded to an IR bright time, and no definitive
bright episodes were detected in that study. This result agrees with the brightening period suggested
in this Letter. Based on the KPNO, APO, and Keck data the ephemeris of the GX 17+2 IR bright
episodes is JDmax(n) = 2454550.79829 + 3.01254(n) ± 0.00002 days.
We have presented the first evidence that the K-band brightening episodes of GX 17+2 are
periodic and that its IR light curve and maximum brightness vary from event to event. Although
this result is consistent with all of our K-band data, we would like to emphasize that additional
observations are clearly desirable to confirm the nature of these IR bright episodes and to better
quantify the structure of the K-band light curve. As mentioned in the prior sections, although
several mechanisms have been ruled out as the likely source for the IR bright episodes in GX 17+2,
their origin remains a mystery. The data collected for this study is not adequate to answer this
question. However, we offer the following comments.
Migliari et al. (2007) interpret the correlated radio and X-ray fluctuations of GX 17+2 as
suggestive of a compact jet. These authors showed that the peak radio flux of GX 17+2 at 4.8 and
8.4 GHz varied smoothly with the location in the Z plot. Maximum fluxes of 3 mJy and 0.4 mJy,
respectively, were found when this system was on its horizontal branch. Contemporaneous RXTE
PCA data with our K-band bright episode of 2006 July 12 suggest that GX 17+2 was on its normal
branch (McNamara et al. 2009, in prep.). VLA observations obtained five hours after the IR peak
had fluxes of 7 mJy and 4.4 mJy at 4.8 and 8.4 GHz, respectively. These are the highest flux levels
ever observed for this system at these frequencies. The correlation between GX 17+2’s radio and
IR emission suggests the presence of a synchrotron jet.
The lack of a periodicity in the RXTE ASM soft X-ray data can be explained if the jet is
responsible for only a small amount of the soft X-ray flux. The periodic signal would then be lost
in the stronger X-ray emission from the accretion process. That GX 17+2 was detected at a flux
of ∼10 mJy at 24 µm supports this interpretation, for neither donor star nor accretion disk would
contribute to the emission at that wavelength. For example, STAR-PET6predicts a 24 µm flux of
0.011 mJy for NP Ser.
Assuming our interpretation of a jet is correct, one could ask whether this periodic IR ac-
tivity represents the orbital period of an eccentric binary, with the jet turning on at periastron
as used to explain Cir X-1, or whether it represents the precession period of the jet. However,
periastron passage implies an eccentric orbit. GX 17+2 is a persistent X-ray source, as are all the
Z sources. Continuous accretion by Roche lobe overflow requires these systems to have circular or-
– 7 –
bits. Therefore, we believe our observations support the possibility of GX 17+2 having a precessing
synchrotron jet, making GX 17+2 an obvious choice for a future simultaneous multi-wavelength
This material is based upon work supported by the National Aeronautics and Space Admin-
istration under Proposal No. 92042 issued through the Science Mission Directorate. Support for
this work was also provided by the New Mexico Space Grant Consortium, and the New Mex-
ico Higher Education Department. Results provided by the ASM/RXTE teams at MIT and at
the RXTE SOF and GOF at NASA’s GSFC, by the National Optical Astronomy Observatory
(NOAO)/Association of Universities for Research in Astronomy (AURA)/National Science Foun-
dation (NSF), and based on observations obtained with the Apache Point Observatory 3.5-meter
telescope, which is owned and operated by the Astrophysical Research Consortium. This research
has made use of the NASA/ IPAC Infrared Science Archive, which is operated by the Jet Propulsion
Laboratory, California Institute of Technology, under contract with the National Aeronautics and
Facilities: RXTE, CTIO:1.3m, KPNO:2.1m, APO:3.5m, VLA.
Bandyopadhyay, R. M., Charles, P. A., Shahbaz, T., & Wagner, R. M. 2002, ApJ, 570, 793
Callanan, P. J., Curran, P., Filippenko, A. V., Garcia, M. R., Margon, B., Deutsch, E., Anderson,
S., Homer, L., & Fender, R. P. 2002, ApJ, 574, L143
Davidsen, A., Malina, R., & Bowyer, S. 1976, ApJ, 203, 448
Deutsch, E. W., Margon, B., Anderson, S. F., Wachter, S., & Goss, W. M. 1999, ApJ, 524, 406
Homan, J., van der Klis, M., Jonker, P. G., Wijnands, R., Kuulkers, E., M´ endez, M., & Lewin,
W. H. G. 2002, ApJ, 568, 878
Kuulkers, E., van der Klis, M., Oosterbroek, T., van Paradijs, J., & Lewin, W. H. G. 1997, MNRAS,
Migliari, S., Miller-Jones, J. C. A., Fender, R. P., Homan, J., Di Salvo, T., Rothschild, R. E.,
Rupen, M. P., Tomsick, J. A., Wijnands, R., & van der Klis, M. 2007, ApJ, 671, 706
Naylor, T., Charles, P. A., & Longmore, A. J. 1991, MNRAS, 252, 203
Penninx, W., Lewin, W. H. G., Zijlstra, A. A., Mitsuda, K., & van Paradijs, J. 1988, Nature, 336,
Psaltis, D., Lamb, F. K., & Miller, G. S. 1995, ApJ, 454, 137
– 8 –
Russell, D. M., Fender, R. P., & Jonker, P. G. 2007, MNRAS, 379, 1108
Wachter, S. 2009, in Wild Stars in the Old West II: The 14th North American Workshop on
Cataclysmic Variables and Related Objects, ed. S. Howell, American Institute of Physics
This preprint was prepared with the AAS LATEX macros v5.2.
– 9 –
Table 1.GX 17+2 K-Band Bright Episode Midpoint Times
Date JD Hel.Cor. (days)Peak K magaDuration (hours)Site
aMagnitudes reported are of the blend of GX 17+2 and NP Ser.
1Callanan et al. (2002)
Table 2. GX 17+2 Observation Information
DateDuration (hours) IR State Instrument Site
– 10 –
Fig. 1.— K-band light curves for the blend of NP Ser and GX 17+2 are shown with 1σ error bars.
The blend is shown clearly brighter than the K magnitude of NP Ser as measured by Callanan et al.
(2002), indicated with a solid line in each panel. The shapes and peak levels of an IR bright episode
vary, making a unifying folded light curve impractical.
– 11 –
Fig. 2.— Period vs. the dispersion in phase for the times when GX 17+2 was IR bright. The top
figure includes bright episode times listed in Table 1 between 2006 and 2008. The bottom figure
includes the bright episode reported by Callanan et al. (2002) for June 1999. The minimum shown
in the bottom figure corresponds to the period when the dispersion in the bright time phases was
at its minimum. This occurs at a period of 3.01245 days. The multi-cusp pattern in the bottom
figure (and present but not visible in the top figure) is due to our data having been taken at nearly
the same time each year.