1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{amsmath, amssymb}
3: \shorttitle{Soft Component and QPO of SAX J2103.5+4545}
4: \shortauthors{Inam et al.}
5: %\received{2004 February 10}
6: \begin{document}
7: \title{Discovery of Soft Spectral Component and Transient 22.7s Quasi Periodic
8: Oscillations of SAX J2103.5+4545}
9: \author{S. \c{C}. \.{I}nam and A. Baykal}
10: \affil{Department of Physics, Middle East Technical University, Inonu
11: Bulvari, Balgat, Ankara 06531 Turkey}
12: \email{scinam@metu.edu.tr, altan@astroa.physics.metu.edu.tr}
13:
14: \author{J. Swank}
15: \affil{NASA Goddard Space Flight Center, Greenbelt, MD 20771 USA}
16: \email{swank@xenos.gsfc.nasa.gov}
17: \and
18:
19: \author{M.J.Stark}
20: \affil{Department of Physics, Lafayette College, Easton, PA 18042 USA}
21: \email{starkm@lafayette.edu}
22:
23: \begin{abstract}
24: XMM-Newton observed SAX J2103.5+4545 on January 6, 2003, while
25: RXTE was monitoring the source. Using RXTE-PCA dataset
26: between December 3, 2002 and January 29, 2003, the
27: spin period and average spin-up rate
28: during the XMM-Newton observations were found to be
29: $354.7940\pm0.0008$ s and $(7.4\pm0.9)\times10^{-13}$Hz s$^{-1}$ respectively.
30: In the power spectrum of the 0.9-11 keV EPIC-PN lightcurve, we found
31: quasi periodic oscillations around 0.044 Hz (22.7 s) with an rms fractional
32: amplitude $\sim $6.6 \%. We interpreted this QPO feature as the
33: Keplerian motion of inhomogenuities through the inner disk.
34: In the X-ray spectrum, in addition to the power law component with
35: high energy cutoff and $\sim6.4$ keV fluorescent iron emission line
36: (Baykal et al., 2002), we discovered a soft component consistent with a
37: blackbody emission with ${\rm{kT}}\sim1.9$keV. The pulse phase
38: spectroscopy of the source revealed that the blackbody flux peaked at the peak
39: of the pulse
40: with an emission radius $\sim 0.3$ km, suggesting the polar cap on the
41: neutron star surface as the source of blackbody emission. The flux of the iron
42: emission line at $\sim 6.42$ keV was shown to peak at the off-pulse phase,
43: supporting the idea that this feature arises from fluorescent emission of the
44: circumstellar material around the neutron star rather than the hot region in
45: the vicinity of the neutron star polar cap.
46: \end{abstract}
47: \keywords{stars: individual (SAX J2103.5+4545) --- stars: neutron
48: --- X-rays: binaries --- X-rays: stars}
49:
50: \section{Introduction}
51: The transient X-ray source SAX J2103.5+4545 was discovered by the
52: Wide Field Camera on the {\it{BeppoSAX}} X-ray observatory during its outburst
53: between 1997 February and September with 358.61s
54: pulsations and a spectrum consistent with an absorbed power law
55: model with the photon index of $\sim1.27$ and the absorption column
56: density of $\sim3.1\times 10^{22}$cm$^{-2}$
57: (Hulleman, in't Zand, \& Heise 1998).
58:
59: After detection of another outburst in November 1999 by the
60: {\it{all-sky monitor (ASM)}} on
61: the {\it{Rossi X-ray Timing Explorer (RXTE)}}, the source was found to
62: be active for more than a year, and was continously monitored through
63: regular pointed {\it{RXTE}} observations. Using pulse arrival times,
64: the orbital period and eccentricity of the orbit
65: were found to be
66: 12.68(25) days and 0.4(2) (Baykal, Stark, \& Swank 2000a,b).
67: In the timing analysis, the source was initially
68: found to be spinning up for $\sim 150$ days,
69: at which point the flux dropped quickly by
70: a factor of $\simeq 7$, and a weak spin-down began afterwards
71: (Baykal, Stark, \& Swank 2002). Strong
72: correlation between X-ray flux and spin-up rate was explained by using
73: Ghosh \& Lamb (1979) accretion disk model.
74: The X-ray spectra well fitted the
75: absorbed power law model with high energy cutoff and a $\sim $6.4 keV
76: fluorescent emission line (Baykal et al. 2002).
77:
78: Orbital parameters found by using {\it{RXTE}} observations of the source
79: (Baykal et al. 2000a,2000b) indicated that the source has a high mass
80: companion. Hullemann et al. (1998) pointed out a B8 type star within the
81: BeppoSAX error box, but its distance ($\sim 0.7$ kpc) implied a luminosity
82: too low to explain the spin-up that was seen in the RXTE observations. Recently,
83: a possible candidate for the optical companion of SAX J2103.5+4545 with
84: the visual magnitude of 14.2 was discovered
85: (Reig \& Mavromatakis, 2003).
86:
87: SAX J2103.5+4545 was also observed with the {\it{INTEGRAL}} observatory in the
88: 3-200 keV band with significant detection up to $\sim 100$ keV
89: (Lutovinov, Molkov,\& Revnivtsev 2003). The
90: spectral parameters found in the {\it{INTEGRAL}} observations of the source
91: were found to be compatible with those found by Baykal et al. (2002).
92:
93: Since the beginning of the most recent outburst in June 2002, SAX J2103.5+4545
94: has been monitored continously by {\it{RXTE}} through regular pointed
95: observations. It was possible to obtain some simultaneous
96: coverage with the {\it{XMM-Newton}} observatory on January 6, 2003.
97: The observation of {\it{XMM-Newton}} revealed
98: a soft spectral component of the source which was well-represented by a
99: blackbody model. This spectral model was verified by simultaneous fitting of
100: January 6, 2003 {\it{RXTE}}-PCA observation. Using {\it{XMM-Newton}} dataset,
101: we also discovered $\sim 22.7$s quasi periodic
102: oscillations (QPO's) of this source for the first time. In this paper, we
103: present our spectral and timing results of the analysis of RXTE and XMM
104: datasets of SAX J2103.5+4545.
105:
106: \section{Observations}
107: \subsection{RXTE}
108: We analyzed RXTE observations of SAX J2103.5+4545 between December 3, 2002
109: and January 29, 2003 with a total observation time of $\sim 52$ksec.
110: This set of observations is a subset of the RXTE observations
111: for the proposal number 70082.
112: The results presented here are based on data collected
113: with the Proportional Counter Array (PCA; Jahoda et al., 1996). The PCA
114: instrument consists of an array
115: of 5 proportional counters (PCU) operating in the 2-60 keV energy range, with
116: a total effective area of approximately 6250 cm$^{2}$ and a field of view
117: of $\sim 1^{\circ}$ FWHM. Although the number of active PCU's
118: varied between 2 and 5 during the observations, our observations belong to the
119: observational epoch for which background level
120: for one of the PCUs (PCU0) increased due to the fact that this PCU started
121: to operate without a propane layer. The latest combined background models
122: (CM) were used together with FTOOLS 5.2 to estimate the appropriate background.
123:
124: \subsection{XMM}
125: {\it{XMM-Newton}} observations took place on January 6, 2003 with an
126: 8.7 ksec continuous exposure.
127: Among the three EPIC detectors (Turner
128: et al. 2001; Str\"{u}der et al. 2001), the MOS1 and MOS2 detectors
129: were configured in Fast Uncompressed mode, while the PN was configured
130: in Fast Timing mode. Data collected by the EPIC detectors on
131: {\it{XMM-Newton}} were processed using version 5.4.1 of the
132: {\it{XMM-Newton}} Science Analysis System (SAS).
133: We did not include the data collected by the
134: two Reflection Grating Spectrometers (RGS1 and RGS2)
135: in data analysis since the count rates from these spectrometers were
136: too low.
137:
138: \section{Data Analysis}
139:
140: \subsection{Pulse Timing and Pulse profiles}
141:
142: In the timing analysis, we corrected the background subtracted
143: lightcurves of RXTE data to the barycenter of solar system.
144: The data have also been corrected for the orbit model using the eccentric
145: orbital parameters given by Baykal et al. (2000b) with the new orbital epoch
146: being MJD $52633.90\mp 0.05$. In order to estimate the pulse frequency and
147: pulse frequency derivative accurately, we used $\sim 57.5$ days time span of
148: RXTE observations between MJD 52611.48 and
149: MJD 52668.90 which covers $\sim 8.7$ ksec short
150: observation of XMM data starting at MJD 52645.85.
151: We obtained the nominal pulse frequency
152: by using Fourier transform and constructed 20 pulse profiles
153: (one pulse profile for each RXTE orbit) by folding the
154: lightcurve at this nominal pulse period. We found the pulse
155: arrival times (phase offsets)
156: by cross-correlating these pulse profiles with a template chosen as the most
157: statistically significant pulse profile. In the pulse timing analysis, we used
158: the harmonic representation
159: of the pulse profiles (Deeter \& Boynton 1985). In this technique, the pulse
160: profiles are expressed in terms of harmonic series and cross correlated
161: with the template pulse profile. The pulse phase offsets can be found of a
162: Taylor expansion,
163: \begin{equation}
164: \delta \phi = \delta \phi _{0} + \delta \nu (t-t_{0}) +
165: \frac{1}{2} \dot \nu (t-t _{0})^{2} ,
166: \end{equation}
167: where $\delta \phi $ is the pulse phase offset deduced from the
168: pulse timing analysis, $t_{0}$ is the midtime of the observation,
169: $\phi_{0}$ is the phase offset at $t_{0}$, $\delta \nu$ is the
170: deviation from the mean pulse frequency (or additive correction to the
171: pulse frequency), and $\dot{\nu}$ is the pulsar's pulse frequency
172: derivative.
173: We fitted the phase offsets to the Taylor expansion. From
174: the fit, we found the pulse period corresponding to the
175: XMM observation as $354.7940 \pm 0.0008$s and 57.5 days average spin-up rate
176: as $(7.4 \pm 0.9) \times 10^{-13}$ Hz s$^{-1}$.
177: We did not see any significiant timing noise in the residuals of arrival times,
178: which indicated that the spin up rate was stable through the observations.
179: The average 3-20 keV flux of RXTE observations was
180: $(5.5\mp 0.5) \times 10^{-10}{\rm{erg s}}^{-1}{\rm{ cm}}^{-2}$. The mean
181: spin-up rate and the average X-ray flux were found to be consistent with
182: the previously observed spin-up rate and
183: X-ray flux correlations during the 1999 outburst (see Fig. 7 in Baykal
184: et al. 2002). Detailed timing noise of RXTE observations is in progress
185: and is not the scope of this paper (Baykal et al. 2004).
186:
187: In Figure 1, we presented energy dependent pulse profiles of XMM-EPIC PN data.
188: The feature peaking at the phase
189: $\sim 0.25$ before the peak of the main pulse was a prominent
190: feature of the pulse profile. The pulse fraction was
191: $(50.9\mp 0.3)$\% at 0.9-11 keV, whereas it was found to be
192: slightly variable in the energy intervals shown in Figure 1 with a
193: minimum of $(48.5\mp 0.8)$\% at 7.5-11 keV and a maximum of
194: $(53.6\mp 0.7)$\% at 2.5-5.0 keV.
195:
196: \subsection{Transient 22.7 sec QPO}
197:
198: In the power spectra of the 0.9-11 keV EPIC-PN lightcurve we found
199: quasi periodic oscillations around 0.044 Hz (see Fig. 2).
200: In order to test the significance of these oscillations,
201: we averaged 9 power spectra and rebinned the frequencies by a factor 8.
202: Then we modeled the continuum power spectrum with a broken
203: power law model with the break value $(4.45\mp 0.16)\times 10^{-2}$ Hz, and
204: the power indices $-0.34\mp 0.08$ and $-2.14\mp 0.05$.
205: We modeled the transient oscillations with a Lorentzian centered at
206: $(4.40\mp 0.12)\times 10^{-2}$ Hz, with the full width
207: half maximum (FWHM) of $(6.1\mp 2.0)\times 10^{-3}$ Hz.
208:
209: To test the significance of this QPO feature, we normalized the
210: power spectrum by dividing it by the continuum and we multiplied this result by
211: 2 (van der Klis 1989). The resultant power spectrum would be consistent
212: with a Poisson distribution for a degree of freedom $2\times 8\times 9$ = 144.
213: As seen from the lower panel of Figure 2, there is a prominent peak at
214: $\sim 0.044$ Hz with a minimum of excess power (including the error of power)
215: 3.62 giving the total power as
216: $3.62\times 8\times 9$ = 260.64. This gives the probability of detecting
217: a false signal ${\rm{Q}}(260.64|144)= 9.39\times 10^{-9}$.
218: Since we have 512 frequencies in each power spectra, total probability of
219: having false signal becomes
220: $9\times 512 \times 9.39\times 10^{-9} = 4.3\times 10^{-5}$. Therefore
221: significance of transient oscillations is $1-4.3\times 10^{-5}\simeq0.99996$
222: which is consistent with more than 6 sigma level detection.
223: The rms fractional amplitude associated with this QPO
224: feature was found to be $(6.6\mp 1.9)$\%.
225:
226: We searched transient oscillations using the RXTE-PCA lightcurves in the 3-20
227: keV energy range,
228: however we did not see any significant transient oscillations.
229: Then we extracted a $\sim 50$\% portion of the overall XMM-EPIC PN lightcurve at
230: the 3-10 keV energy range that coincided
231: exactly with the $\sim 4.5$ ksec part of the RXTE PCA lightcurve on January 6,
232: 2003 and performed power spectral analysis.
233: We found that the significances of 0.044 Hz oscillations for RXTE and XMM
234: lightcurves are $2.5\sigma$ and $2.8\sigma$ respectively.
235: We concluded that the QPO feature was originated mostly from the
236: soft component of the spectrum or highly transient. Future observations
237: are required to confirm these oscillations.
238:
239: \subsection{Spectral Analysis}
240: We fitted overall background subtracted 1-10 keV
241: spectra of PN, MOS1 and MOS2 to an absorbed power law model
242: (Morrison \& McCammon, 1983) with high energy cut-off (White, Swank, Holt
243: 1983). In addition, an iron line feature at 6.42 keV was required
244: in the spectral model (Case 1 in Table 1). However, this model
245: did not fit the spectrum of the source well, giving a reduced
246: $\chi^2$ of 2.69. Adding an additional blackbody component to the
247: model decreased the reduced $\chi^2$ to 1.23 (Case 2 in Table
248: 1). Joint fit including the PCA data on January 6, 2003 and adding 2\%
249: systematic errors (see Wilms et al. 1999; Coburn et al. 2000),
250: was possible to this model with a reduced
251: $\chi^2$ of 1.1 (Case 3 in Table 1 and Figure 3). Using Case 3 (i.e. the
252: joint fit to the model including blackbody component), 1-20 keV unabsorbed
253: flux was found to be $6.6\times 10^{-10}$ ergs s$^{-1}$ cm$^{-2}$.
254: Assuming a source distance of 3.2 kpc (Baykal et al. 2002), this value
255: corresponds to a luminosity of $7.5\times 10^{35}$ ergs s$^{-1}$. To compare
256: with the average X-ray flux, we found 3-20 keV flux of
257: the January 6 observation to be $6.1\times 10^{-10}$ ergs s$^{-1}$ cm$^{-2}$.
258: This value is approximately 10\% greater than the
259: 57.5 days average 3-20 keV RXTE-PCA flux. This is reasonable since the
260: XMM observation took place 3.55 days after the periastron passage where the
261: X-ray flux reaches approximately its maximum (see Fig. 5 and 8 of
262: Baykal et al. 2000b). It should be noted that line parameters
263: obtained from 3-20 keV RXTE PCA data agree with those
264: obtained from XMM EPIC data. However exclusion of the blackbody component,
265: while fitting RXTE PCA data only, increases the absorption column density
266: and equivalent width of line emission and make the power law index harder
267: as shown in Table 1, case 4.
268:
269: To study the spin-phase-resolved spectra using EPIC-PN
270: data, we divided the spin phase into 10 bins, and fitted 1-10 keV
271: spectrum of each bin with the model including blackbody component (i.e.
272: model in case 3). Figure 4 is a plot of the spectral parameters as a function
273: of the spin phase. For all the spin phases, we found that the model gives an
274: iron line peak energy consistent with $6.42\pm0.04$ keV within 1$\sigma$, so
275: we chose to fix this parameter. We also checked the consistence of freezing
276: the cut-off energy, iron line sigma, and e-folding energy parameters, and
277: found that these parameters did not vary significantly when they were
278: thawed. From Figure 4, strong modulation
279: (with a factor of $\sim 10$) of the blackbody flux with the spin phase is
280: evident. Similarly, the flux of the power law
281: component was shown to be varying with the spin phase, but the
282: variation was more moderate (with a factor of $\sim 3$) than the
283: that of the blackbody component.
284: The iron line feature at 6.42 keV was stronger for the off-pulse phases
285: where the X-ray flux was lower.
286:
287: \section{Discussion and Conclusion}
288: \subsection{QPO Feature of SAX J2103.5+4545}
289: Quasi-periodic oscillations in the X-ray band having periods in the
290: range of
291: $\sim 2.5-100$s have been observed in many accretion powered X-ray pulsars:
292: 4U 0115+63 (Soong \& Swank 1989), EXO 2030+375 (Angelini, Stella,\& Parmar 1989),
293: 4U 1626-67 (Shinoda et al. 1990),
294: SMC X-1 (Angelini, Stella,\& White 1991), Cen X-3 (Takeshima et al. 1991),
295: V0332+53 (Takeshima et al. 1994), A0535+262
296: (Finger, Wilson,\& Harmon 1996), GRO J1744-28 (Zhang, Morgan,\& Jahoda 1996;
297: Kommers et al. 1997),
298: X Per (Takeshima 1997), 4U 1907+09 (in't Zand, Baykal, \& Strohmayer 1998;
299: Mukerjee et al. 2001), XTE J1858+034 (Paul \& Rao 1998),
300: LMC X-4, and Her X-1 (Moon \& Eikenberry 2001a,b).
301: The QPO feature that we found in the
302: XMM-Newton EPIC-PN light curve of SAX J2103.5+4545 which has a
303: peak period of $22.7\mp 0.6$s and fractional rms amplitude
304: of $6.6 \pm 1.9$ percent is
305: quite typical (e.g. In't Zand et. al. 1998; Paul \& Rao 1998;
306: Takeshima et al. 1994).
307:
308: Models that explain the QPO phenomenon in accretion
309: powered X-ray pulsars fall basically into three categories:
310: In the Keplerian frequency model, QPOs are produced due to some inhomogenuities
311: at the inner edge of the Keplerian disk ($r_0$) and modulate the light curve
312: at the Keplerian frequency $\nu _{QPO}=\nu_{K}$
313: (van der Klis et al. 1987). In the beat frequency model,
314: the accretion flow onto the neutron star is modulated at the beat frequency
315: between the Keplerian frequency at the inner edge of the accretion disk
316: and the neutron star spin frequency
317: $\nu _{QPO}=\nu_{K}-\nu _{s}$ (Alpar \& Shaham
318: 1985). The third model involves
319: accretion flow instabilities (Fronter, Lamb,\& Miller 1989; Lamb 1988),
320: and applies only to the sources that have luminosities close to Eddington
321: limit, therefore it should not be applicable to our case for which the
322: luminosity is well below the Eddington limit.
323:
324: In our case,
325: QPO frequency $\nu _{QPO}= 4.4\times 10^{-2}$Hz is about one
326: order of magnitude greater than the spin frequency
327: $\nu _{s}= 2.8185\times 10^{-3}$ Hz.
328: Therefore, it is difficult
329: to distinguish between a Keplerian model and a beat frequency model.
330:
331: Assuming that the 22.7 s oscillation in SAX J2103.5+4545 is related to
332: Keplerian orbital motion via either Keplerian
333: frequency model or beat frequency model, and using the QPO and its FWHM values
334: we obtain the radius of inner disk as
335: \begin{equation}
336: r_{0}= \bigg(\frac{GM}{4\pi^2}\bigg)^{1/3}\nu_k^{-2/3} =
337: (1.32^{+0.13}_{-0.11}) \times 10^{9} {\rm{cm}},
338: \end{equation}
339: where M is 1.4 M$_{\odot}$ for a neutron star and G is the gravitational
340: constant.
341:
342: From the strong correlation between
343: pulse frequency derivatives and X-ray flux, Baykal et al. (2002)
344: obtained for the distance to the source
345: $3.2 \pm 0.8$ kpc and for the magnetic field $(12\pm 3)\times 10^{12}$ Gauss.
346: Using the distance and magnetic field values,
347: the inner edge of the Keplerian disk
348: $r_{0}$ can be found as (Ghosh \& Lamb 1979)
349: \begin{equation}
350: r_{0}\simeq 0.52\mu^{4/7}(2GM)^{-1/7}\dot M^{-2/7} =
351: (1.67_{-0.25}^{+0.23})\times 10^9 {\rm{cm}},
352: \end{equation}
353: where $\mu=BR^{3}$ is the neutron star magnetic moment with B the equatorial
354: magnetic field, R the neutron star radius, and $\dot M$ the mass accretion rate
355: having the value of $\simeq 4\times 10^{15}$g $s^{-1}$ for an accretion
356: luminosity of $\simeq 7.5\times 10^{35}$ erg s$^{-1}$ (as estimated in Section
357: 3.3).
358:
359: The radius of the inner disk inferred from
360: the Keplerian orbital motion of inhomogenuities
361: and the one inferred from the Ghosh Lamb disk accretion model agree each other.
362: This shows that the idea that the QPOs are formed due to the Keplerian motion
363: of inhomogenuities is indeed promising as the
364: explanation of the QPO of SAX J2103.5+4545 and the observed QPO frequency is
365: consistent with the distance and the
366: magnetic field values estimated by Baykal et al. (2002).
367:
368: \subsection{Blackbody and Iron Line Features of the Energy Spectrum}
369:
370: XMM-Newton observations of SAX J2103.5+4545 revealed for the first time that
371: the energy spectrum of the source has a blackbody component peaking at
372: $\sim 1.90$keV with the emission radius of $\sim 0.3$km.
373: The blackbody radiation may come
374: from the polar cap of the neutron star as it appears to for
375: the Be/X-ray pulsar system EXO
376: 2030+375 (Reig \& Coe 1999; Sun et al. 1994) and the millisecond X-ray pulsars
377: SAX J1808.4-3658 (Gierlinski, Done,\& Didier 2002) and
378: XTE J0920-314 (Juett et al. 2003). Blackbody emission radii on these X-ray
379: pulsars are reported to be greater than $\sim 1$km.
380: The relatively high surface magnetic field of
381: SAX J2103.5+4545 ($\sim 10^{13}$ Gauss) is
382: probably the reason for the relatively small blackbody emission radius ($\sim
383: 0.3$ km) compared to these X-ray pulsars. Although the contribution of
384: blackbody component is relatively more
385: significant for lower energies (i.e. energies smaller than $\sim 3$ keV),
386: power law flux is $\sim 3$ times greater than the blackbody flux even
387: at the 1-3 keV energy band.
388:
389: In our case, it is unlikely that the blackbody emission comes from the
390: reprocessed emission of the surrounding material or the accretion disk as in
391: the case of Her X-1 (Endo et al. 2000), Cen X-3 (Burderi et al. 2000),
392: SMC X-1 and LMC X-4 (Paul et al. 2002),
393: since blackbody component in such cases is expected to be softer
394: ($kT\sim 0.1$keV). Lower blackbody temperature and smaller blackbody emission
395: radius at the off-pulse phase shown in Figure 4 are also indications of the
396: plausibility of the polar cap emission interpretation, as the regions of the
397: soft polar cap emission must align with the peak of the
398: X-ray pulse of the pulsar.
399:
400: Using the spin-phase resolved spectroscopy, strength of the iron line feature
401: at $\sim 6.42$keV was also found to vary significantly with the spin phase as
402: seen in Figure 4. The peak energy of this feature clearly shows that it
403: corresponds to the fluorescent iron K-line complex. This line complex feature
404: is observed in the
405: spectra of most of the X-ray pulsars (White et al. 1983; Nagase 1989)
406: and is generally thought to be produced by the ions less ionized
407: than Fe XVIII in a relatively cool matter around the neutron star (e.g.
408: accretion disk, accretion disk corona) by fluorescent K$\alpha$ transition.
409:
410: Variation of the iron line feature with the spin phase can then be
411: interpreted as a sign that it is mainly produced outside
412: the polar cap region of the neutron star, thus should have a peak at
413: the off-pulse parts of the spin phase. From Figure 4, we see that
414: iron line flux and iron line equivalent width vary strongly
415: with the spin phase, peaking at the
416: off-pulse. Similar pulse phase dependence of the iron line feature is
417: also observed in Her X-1 (Choi et al. 1994).
418:
419: \acknowledgments
420:
421:
422: We thank anonymous referee for helpful comments. We also thank
423: Dr. T. Strohmayer and Dr. C. Markwardt for their useful
424: discussions. S.\c{C}.\.{I}nam acknowledges the Integrated Doctorate Program
425: scholarship from the Scientific and Technical Research Council of Turkey
426: (T\"{U}B\.{I}TAK).
427:
428: \begin{thebibliography}{}
429:
430: \bibitem[]{451}Alpar, M.A., Shaham, J. 1985, Nature, 316, 239
431: \bibitem[]{452}Angelini, L., Stella, L., Parmar, A.N. 1989, ApJ, 346, 906
432: \bibitem[]{453}Angelini, L., Stella, L., White, N.E. 1991, ApJ, 371, 332
433: \bibitem[]{454}Baykal, A., Stark, M., Swank J. 2000a, IAU Circ. 7355
434: \bibitem[]{455}Baykal, A., Stark, M., Swank J. 2000b, ApJ, 544, L129
435: \bibitem[]{456}Baykal, A., Stark, M., Swank J. 2002, ApJ, 569, 903
436: \bibitem[]{457}Baykal, A. et al. 2004, in preparation
437: \bibitem[]{458}Burderi, L., Di Salvo, T., Robba, N.R. et al. 2000, ApJ, 530, 429
438: \bibitem[]{459}Choi, C.S., Nagase, F., Makino, F. 1994, ApJ, 437, 449
439: \bibitem[]{460}Coburn, W., Heindl, W.A., Wilms, J. et al. 2000, ApJ, 543, 351
440: \bibitem[]{461}Deeter, J.E., Boynton, P.E., 1985, in Proc. Inuyama Workshop
441: on Timing Studies of X-Ray Sources, ed. S. Hayakawa $\&$ F. Nagase
442: (Nagoya: Nagoya Univ.), 29
443: \bibitem[]{464}Endo, T., Nagase, F., Mihara, T. 2000, PASJ, 52, 223
444: \bibitem[]{465}Finger, M.H., Wilson, R.B., Harmon, B.A. 1996, ApJ, 459, 288
445: \bibitem[]{466}Fronter, B., Lamb, F.K., Miller, G.S. 1989, Nature, 342, 775
446: \bibitem[]{467}Ghosh, P., Lamb F.K. 1979, ApJ, 234, 296
447: \bibitem[]{468}Gierlinski, M., Done, C., Didier, B. 2002, MNRAS, 331, 141
448: \bibitem[]{469}Hullemann, F., in't Zand, J.J.M., Heise, J. 1998, ApJ, 337, L25
449: \bibitem[]{470}in't Zand, J.J.M., Baykal, A., Strohmayer, T.E. 1998, ApJ, 496, 386
450: \bibitem[]{471}Jahoda, K., Swank J., Giles A.B. et al. 1996, Proc. SPIE, 2808, 59
451: \bibitem[]{472}Juett, A.M., Galloway, D.K., Chakrabarty, D. 2003, ApJ, 587, 757
452: \bibitem[]{473}Kommers, J.M., Fox, D.W., Lewin, W.H.G. et al. 1997, ApJ, 482, L53
453: \bibitem[]{474}Lamb, F.K. 1988, Adv. Space Res., 8, 421
454: \bibitem[]{475}Lutovinov, A.A., Molkov, S.V., Revnivtsev, M.G. 2003, AstL, 29, 713
455: \bibitem[]{476}Moon, D., Eikenberry, S.S. 2001a, ApJ, 549, L225
456: \bibitem[]{477}Moon, D., Eikenberry, S.S. 2001b, ApJ, 552, L135
457: \bibitem[]{478}Morrison, R., McCammon, D. 1983, ApJ, 270,119
458: \bibitem[]{479}Mukerjee, K., Agrawal, P.C., Paul, B. et al. 2001, ApJ, 548, 368
459: \bibitem[]{480}Nagase, F. 1989, PASJ, 41, 1
460: \bibitem[]{481}Paul, B., Rao, A.R. 1998, A\& A, 1998
461: \bibitem[]{482}Paul, B., Nagase, F., Endo, T. et al. 2002, ApJ, 579, 411
462: \bibitem[]{483}Reig, P., Coe, M.J. 1999, MNRAS, 302, 700
463: \bibitem[]{484}Reig, P., Mavromatakis, F. 2003, ATEL, 173
464: \bibitem[]{485}Shinoda, K., Kii, T., Mitsuda, K. 1990, PASJL, 42, 27
465: \bibitem[]{486}Soong, Y., Swank, J. 1989, in Proc. 23d ESLAB Symp., Two Topics in
466: X-ray Astronomy, Vol. 1, ed. N. White (Garching: ESA), 617
467: \bibitem[]{488}Sun, X.-J., Li, T.-P.., Wu, M. et al. 1994, A\& A, 289, 127
468: \bibitem[]{489}Struder, L., Briel, U., Dennerl K. et al. 2001, A\& AL, 365, 18
469: \bibitem[]{490}Takeshima, T., Dotani, T., Mitsuda, K. et al. 1991, PASJL, 43, 43
470: \bibitem[]{491}Takeshima, T., Dotani, T., Mitsuda, K. et al. 1994, ApJ, 436, 871
471: \bibitem[]{492}Takeshima, T. 1997, BAAS, 19111104
472: \bibitem[]{493}Turner, M.J.L., Abbey, A., Arnaud, M. et al. 2001, A\& AL, 365, 27
473: \bibitem[]{494}van der Klis, M., Stella, L., White, N. et al. 1987, ApJ, 316, 411
474: \bibitem[]{495}van der Klis, M. 1989, in Proc. NATO-ASI on Timing Neutron Stars,
475: ed. H.\"{O}gelman and E.P.J. van den Heuvel (Dordrecht:Kluwer), 27
476: \bibitem[]{497}White, N.E, Swank, J.H., Holt, S.S. 1983, ApJ, 270,711
477: \bibitem[]{498}Wilms, J., Nowak, M.A., Dove, J.B. et al. 1999, ApJ, 522, 460
478: \bibitem[]{499}Zhang, W., Morgan, E.H., Jahoda, K. 1996, ApJ, 469, L29
479:
480: \end{thebibliography}
481:
482: \clearpage
483: \begin{figure}
484: \epsscale{.80}
485: \plotone{f1.ps}
486: \caption{Energy dependence of the pulse profiles of SAX J2103.5+4545.
487: Pulse profiles were found by using EPIC-PN lightcurves that cover
488: the energy ranges from top to bottom respectively.}
489: \end{figure}
490:
491: \begin{figure}
492: %\vspace{-1.8cm}
493: %\begin{center}
494: %\begin{tabular}{cc}
495: \includegraphics[scale=.60]{f2.ps}
496: %\end{tabular} \\
497: %\end{center}
498: \caption{{\bf{(top)}} Power spectrum obtained from the 0.9-11.0 keV EPIC-PN
499: lightcurve and rebinned by a factor of 4. QPO feature centered
500: at 0.044 Hz is the prominent feature of the power spectrum.
501: {\bf{(bottom)}}
502: Power spectrum rebinned by a factor of 8, multiplied by 2, and divided by the
503: continuum fit consisting of a broken
504: power law model with the power indices $-0.34\mp 0.08$ and $-2.14\mp 0.05$.
505: Applying the method discussed by van der Klis (1989), significance of the QPO
506: feature was calculated to be more than $6\sigma$ confidence level
507: using the value of the peak at $\sim 0.044$ Hz as shown in this plot.}
508: \end{figure}
509: \clearpage
510: \begin{figure}
511: %\vspace{-1.8cm}
512: \begin{center}
513: \includegraphics[angle=-90,scale=.50]{f3.eps}
514: \end{center}
515: \caption{1-20 keV combined PN, MOS1, MOS2 and PCA spectrum of SAX J2103.5+4545
516: observed on January 6, 2003. The bottom panel shows the residuals of the fit
517: in terms of $\sigma$ values.}
518: \end{figure}
519: \clearpage
520: \begin{figure}
521: %\vspace{-1.8cm}
522: \epsscale{0.84}
523: \plotone{f4.ps}
524: \caption{Spin phase dependence of 1-10 keV X-ray flux in units of $10^{-10}$
525: ergs s$^{-1}$ cm$^{-2}$, Hydrogen column density in units
526: of $10^{22}$ cm$^{-2}$, blackbody kT in
527: units of keV, emission radius of the blackbody (assuming a
528: source distance of 3.2
529: kpc) in units of km, photon index of the power law component of the spectral
530: model, equivalent width of the iron line in units of eV, flux of the
531: iron line in units of $10^{-13}$ ergs s$^{-1}$ cm$^{-2}$, blackbody
532: flux in units of $10^{-10}$ ergs s$^{-1}$ cm$^{-2}$, power law flux in units of
533: $10^{-10}$ ergs s$^{-1}$ cm$^{-2}$, and reduced $\chi^2$. All of the errors show
534: $1\sigma$ confidence level. The spectral model used in fitting consists
535: of an absorbed power law model with high energy cut-off, a soft blackbody
536: component and an Iron line feature modeled as a Gaussian.
537: The values of iron line peak energy, iron line
538: sigma, cut-off and e-folding energies were found to be consistent
539: with the constant values 6.42 keV, 0,
540: 7.89 keV, and 27.1 keV respectively for all the spin phases, so these values
541: were kept fixed while the values of the other model parameters were
542: found.}
543: \end{figure}
544: \clearpage
545: \input{tab1}
546:
547: \end{document}
548: