astro-ph0602354/ms.tex
1: \documentclass{mn2e}
2: %% mnsample.tex
3: % v1.2 released 5th September 1994 (M. Reed)
4: % v1.1 released 18th July 1994
5: % v1.0 released 28th January 1994
6: 
7: % IASOS CRC
8: %\documentclass[useAMS,usenatbib]{mn2e}
9: %\documentclass{mn2e}
10: 
11: % xray02 physics
12: % \documentclass{mn}
13: 
14: % If your system has the AMS fonts version 2.0 installed, MN.sty can be
15: % made to use them by uncommenting the line %\AMStwofontstrue
16: %
17: % By doing this, you will be able to obtain upright Greek characters.
18: % e.g. \umu, \upi etc.  See the section on "Upright Greek characters" in
19: % this guide for further information.
20: %
21: % If you are using AMS 2.0 fonts, bold math letters/symbols are available
22: % at a larger range of sizes for NFSS release 1 and 2 (using \boldmath or
23: % preferably \bmath).
24: 
25: 
26: %  my line added to incorporate .eps figures in IASOS CRC
27:  \usepackage[dvips]{epsfig}
28:  \usepackage{amsfonts,amsbsy}
29: % \usepackage{amsfonts,amsbsy,amssymb}
30: %\usepackage{natbib}
31: 
32: %  upwards shift for astro-ph submission
33: % \voffset-0.85in
34:  \voffset-0.7in
35: 
36: %  lines on xray02 in physics
37:  %\usepackage{epsfig}
38:  %\usepackage{amssymb}
39: \title [{\it I\/} band flaring in SAX J1808.4-3658]
40: {A transient {\it I\/} band excess in the optical spectrum of the accreting millisecond pulsar SAX J1808.4--3658}
41: \author [J.G.Greenhill {\it et al.}]       
42: { J. G. Greenhill$^{1}$, A. B. Giles$^{1,2}$ and C Coutures$^{3}$\\
43:       $^{1}$ School of Mathematics and Physics, University of Tasmania, 
44:              Private bag 37, Hobart, Tasmania 7001, Australia \\
45:       $^{2}$ Spurion Technology Pty. Ltd., 200 Mt. Rumney Road, Mt. Rumney, 
46:              Tasmania 7170, Australia \\
47:       $^{3}$ CEA/Saclay, 91191 Gif-sur-Yvette cedex, France } 
48: 
49: % \date{Accepted .
50: %       Received ;
51: %      in original form }
52: 
53: % \date{Draft 27 September 2005        }
54: % \date{Draft 11 Dec  2003 -  submitted to MNRAS }
55: 
56: % \date{Draft 10 May 2006}
57: \date{Accepted 10 May 2006}
58: 
59: \pagerange{\pageref{firstpage}--\pageref{lastpage}}
60: \pubyear{2005}
61: 
62: \begin{document}
63: \maketitle
64: 
65: \label{firstpage}
66: 
67: \begin{abstract}
68: The optical counterpart of the transient, millisecond X-ray pulsar SAX J1808.4-3658
69: was observed in four colours ({\it BVRI\/}) for five weeks during the 2005 June-July outburst.
70: The optical fluxes declined by $\sim 2$ magnitudes during the first 16 days and then commenced 
71: quasi-periodic secondary outbursts, with time-scales of several days, similar to those seen in 
72: 2000 and 2002. The broadband spectra derived from these measurements were 
73: generally consistent with emission from an X-ray heated accretion disc. During the first 16 days 
74: decline in intensity the spectrum became redder. We suggest that the primary
75: outburst was initiated by a viscosity change driven instability in the inner disc and 
76: note the contrast with another accreting 
77: millisecond pulsar, XTE J0929--314, for which the spectrum becomes bluer during the decline. 
78: On the night of 2005 June 5 (HJD 2453527) the {\it I\/} band flux was $\sim 0.45$ magnitudes brighter 
79: than on the preceding or following nights whereas the {\it BV\/} \& {\it R\/} bands showed no obvious 
80: enhancement. A Type I X-ray burst was detected by the {\it RXTE} spacecraft during this {\it I\/} band 
81: integration. It seems unlikely that reprocessed radiation from the burst was sufficient to explain 
82: the observed increase. We suggest that a major part of the {\it I\/} band excess was due to
83: synchrotron emission triggered by the X-ray burst.  Several other significant short duration changes 
84: in {\it V-I\/} were detected. One occurred at about HJD 2453546 in the early phase of the first 
85: secondary outburst and may be due to a mass transfer instability or to another synchrotron emission event.
86:  
87: 
88: \end{abstract}
89: \begin{keywords}
90: pulsars: general -- pulsars: individual: SAX J1808.4--3658
91: -- accretion, accretion discs -- X-rays: binaries -- X-rays: bursts -- radiation mechanisms: non-thermal
92: \end{keywords}
93: 
94: \section{Introduction} 
95: SAX J1808.4--3658 was the first transient millisecond X-ray pulsar to be 
96: discovered and has been studied  extensively at 
97: all wavelengths (Chakrabarty \& Morgan 1998; in 't Zand et al. 1998; Wijnands \& van der Klis 1998; Giles, Hill \& Greenhill 1999;  Wachter et al. 2000; 
98: Wang et al. 2001; Wijnands et al. 2001;  Homer et al. 2002; Markwardt, Miller 
99: \& Wijnands 2002;  Chakrabarty et al. 2003; Wijnands et al. 2003). 
100: Study of these systems is providing important 
101: information on the evolutionary path by which a conventional LMXB system 
102: becomes a millisecond radio pulsar (Campana et al. 2004; Bogdanov, Grindlay \& van den Berg 2005; Wijnands 2005). 
103: 
104: The optical spectrum of SAX J1808.4--3658 during the 1998 outburst was found by Wang et al. (2001) to be 
105: consistent with emission from an X-ray irradiated accretion disc although there
106: was a clear IR ({{\it JHK\/})excess on one occasion. It is possible that this 
107: extended into the optical bands (Giles et al. 2005). A 2 hour orbital period modulation in {\it V\/} 
108: with amplitude $ 0.12 \pm 0.02$ magnitudes peak to peak  was detected
109: during the 1998 outburst (Giles et al. 1999).  The phase was consistent with X-ray
110: heating of the companion. A much deeper orbital modulation, too large to be generated by X-ray
111: heating, was present in the quiescent state (Campana et al. 2004). They suggested that heating of the 
112: companion by a pulsar generated relativistic particle wind was responsible.
113: 
114: SAX J1808.4--3658 is the only member of the 7 known accreting millisecond pulsars known to 
115: undergo extended low level activity states. These follow the initial outburst and are characterised 
116: by erratic, large amplitude variability on time-scales of hours to days. Repeated secondary X-ray outburts 
117: with durations of several days are sometimes accompanied by optical outbursts (Wijnands, 2004). Disc 
118: instability and mass transfer instability models have been suggested as possible mechanisms 
119: (Wijnands et al. 2001). 
120: 
121: Recently it has become apparent that synchrotron emission, probably from matter 
122: flowing out of the system via bipolar jets, makes a highly variable 
123: contribution to radio and IR emission from many different classes of X-ray 
124: binaries (Fender 2003). In some cases this emission may extend into the 
125: optical region (Hynes et al. 2000). Krauss et al. (2005) detected
126: an {\it I\/} band excess during the discovery outburst of the accretion-powered 
127: millisecond pulsar XTE J1814--318 and Giles et al. (2005) found evidence
128: of a variable {\it I\/} band excess from another accretion-powered 
129: millisecond pulsar XTE J0929--314. 
130: 
131: In this paper we describe the broad-band optical spectra of SAX J1808.4--3658 
132: measured in 4 colours during the 2005 June outburst. 
133: 
134: 
135: 
136: 
137: 
138: \section{Observations}
139: All the observations described in this paper were made using the 1-m 
140: telescope at the University of Tasmania, Mt. Canopus Observatory. The CCD 
141: camera, its operating software (CICADA), the image reduction and analysis 
142: tools (MIDAS and DoPHOT) were identical to that described in Giles et al. 
143: (1999). The data were reduced using the PLANET microlensing collaboration 
144: pipeline QUYLLUR WASI. 
145: The CCD camera contains an SITe 512 x 512 pixel
146: thinned back illuminated chip with an image scale of 
147: $0.434 \arcsec $ pixel$^{-1}$. Cousins standard {\it BVRI\/} filters (Bessell 1990) 
148: were used for the observations. The magnitudes of three stable local reference stars 
149: (Fig. 1) were determined using a 
150: sequence of observations of three standard stars within the E7 field of  
151: Graham (1982). Both source and standard star fields were observed at 
152: virtually identical airmasses (1.01--1.03) and experience with this camera has shown that 
153: differential colour corrections are generally negligible.
154: From these observations we derived 
155: the magnitudes of three local secondary standards close to SAX J1808.4--3658 
156: and within the CCD frame. These local standards are marked as stars 1--3 on 
157: the finder chart in Fig. 1 and we tabulate their derived magnitudes in 
158: Table 1. The magnitudes for SAX J1808.4--3658 were then obtained using 
159: differential photometry relative to these local secondary standards. 
160: The complete data set from 2005 June 2 to 
161: July 6 (HJD 245[3524] -- 245[3558]) is detailed in Table 2 and 
162: forms the subject of this paper. 
163: 
164: 
165: \begin{table}
166:  \caption{The magnitudes of local standard stars 1--3 in Fig. 1.}
167:  \label{symbols}
168:  \begin{tabular}{@{}ccccc}
169:  \hline
170:  Star No. &  {\it B\/}  &  {\it V\/}  &  {\it R\/}  &  {\it I\/}  \\
171:  \hline
172:     1     &   15.60    &   14.76    &   14.23    &   13.74    \\
173:     2     &   16.89    &   15.31    &   14.50    &   13.71    \\
174:     3     &   15.93    &   15.18    &   14.71    &   14.34    \\    
175:   Mean    &  16.02(4)  &  15.06(3)  &   14.46(3) &   13.89(3) \\   
176:  \hline
177:  \end{tabular}
178:  \medskip
179: \end{table} 
180: 
181: \begin{table*}
182: \centering
183: \begin{minipage}{175mm}
184: \begin{center}
185:  \bf Table 2. \rm A journal of the observations. 
186:  \vspace*{0.125cm}
187:  \label{symbols}
188:  \begin{tabular}{@{}llrllrllrllr}
189:  \hline
190:             &                & Int.    &           &                & Int.    &           &                & Int.    &           &                & Int.     \\ 
191:   $HJD^{a}$ & {\it I\/} mag. &  (s)    & $HJD^{a}$ & {\it R\/} mag. &  (s)    & $HJD^{a}$ & {\it V\/} mag. &  (s)    & $HJD^{a}$ & {\it B\/} mag. &  (s)     \\ 
192:  \hline
193:    24.0918  &  17.12(1)      &  300    &           &                &         &  24.0978  &  17.43(1)      &  300    &           &                &          \\
194:    25.1991  &  17.15(2)      &  300    &  25.2076  &  17.14(1)      &  300    &  25.2119  &  17.31(1)      &  300    &  25.2183  &  17.29(1)      &  600     \\
195:    27.1449  &  16.72(1)      &  300    &  27.2305  &  17.19(1)      &  300    &  27.1673  &  17.29(2)      &  300    &  27.2367  &  17.36(1)      &  600     \\
196:    29.2256  &  17.20(4)      &  300    &  29.2324  &  17.33(2)      &  300    &  29.2371  &  17.45(2)      &  300    &  29.2485  &  17.48(2)      &  600     \\
197:    31.1353  &  17.25(4)      &  300    &  31.1433  &  17.33(2)      &  300    &  31.1393  &  17.48(2)      &  300    &  31.1492  &  17.60(1)      &  600     \\
198:    34.1917  &  17.36(3)      &  300    &  34.1960  &  17.56(2)      &  300    &  34.1999  &  17.82(3)      &  300    &  34.2058  &  17.79(2)      &  600     \\
199:    36.2103  &  17.65(3)      &  300    &  36.2189  &  17.86(2)      &  300    &  36.2146  &  18.03(2)      &  300    &  36.2249  &  18.17(1)      &  600     \\
200:    36.2811  &  17.63(2)      &  300    &  36.2887  &  17.80(2)      &  300    &  36.2849  &  18.01(2)      &  300    &  36.2944  &  18.04(4)      &  600     \\
201:    37.1149  &  17.76(2)      &  300    &  37.1235  &  17.98(2)      &  300    &  37.1196  &  18.19(1)      &  300    &  37.1292  &  18.23(1)      &  600     \\
202:    37.3128  &  17.73(3)      &  300    &  37.3217  &  17.95(3)      &  300    &  37.3178  &  18.21(2)      &  300    &  37.3278  &  18.36(2)      &  600     \\
203:    38.2559  &  18.00(4)      &  300    &  38.2603  &  18.26(2)      &  300    &  38.2645  &  18.53(3)      &  300    &  38.2704  &  18.63(3)      &  600     \\
204:    39.2164  &  18.19(4)      &  300    &  39.2206  &  18.40(3)      &  300    &  39.2250  &  18.58(3)      &  300    &           &                &          \\
205:    40.0592  &  18.40(4)      &  480    &  40.0738  &  18.65(3)      &  480    &  40.0675  &  18.88(3)      &  480    &  40.0807  &  19.09(4)      &  600     \\
206:    45.2344  &  18.38(9)      &  300    &  45.2387  &  18.48(8)      &  300    &  45.2425  &  18.85(14)     &  300    &  45.2482  &  18.83(12)     &  600     \\
207:    46.0150  &  18.16(4)      &  500    &           &                &         &  46.0064  &  18.81(4)      &  600    &           &                &          \\
208:    47.1002  &  17.73(2)      &  720    &  47.2723  &  17.98(2)      &  720    &  47.1142  &  18.12(6)      &  720    &  47.2822  &  18.37(3)      &  900     \\
209:    48.0390  &  18.02(3)      &  600    &  48.1593  &  18.28(2)      &  900    &  48.0490  &  18.47(2)      &  900    &  48.1720  &  18.64(2)      & 1200     \\
210:    49.1236  &  18.43(5)      &  900    &           &                &         &  49.1764  &  18.79(4)      &  900    &           &                &          \\
211:    51.0626  &  17.85(2)      &  720    &           &                &         &  51.0714  &  18.20(1)      &  720    &           &                &          \\
212:    55.1715  &  18.70(16)     &  600      &           &                &         &  55.1920  &  19.17(6)    &  600    &           &                &          \\
213:    58.1532  &  18.53(4)      &  600    &           &                &         &  58.1606  &  19.01(2)      &  600    &           &                &          \\
214:  \hline
215:  \end{tabular}
216: \end{center}
217: \hspace*{1.0cm} {\em a} Times of mid integration (-2453500).
218: \end{minipage}
219: \end{table*}
220: 
221: 
222: 
223: 
224: 
225: \begin{figure}
226: \epsfig{file=fig_1.eps, width=8.4cm } 
227:   \caption{A chart showing local standards used for SAX J1808.4-3658. This  
228:   image was taken when there was an {\it I\/} band excess on 2005 June 5 (HJD 2453527) 
229:   The source magnitude was  {\it I\/} = 16.72. The three local secondary standards listed in 
230:   Table 1 are marked with the numbers 1-3.  }
231: \end{figure}
232: 
233: 
234: \begin{figure}
235: \epsfig{file=fig_2.eps, width=8.4cm }
236:   \caption{{\it RXTE} ASM light curve for SAX J1808.4-3658
237:   (upper panel) and {\it BVRI\/} band light curves (lower panel). } 
238: \end{figure}
239: 
240: 
241: \section{Results}
242: \subsection{The light curves}
243: The optical {\it BVRI\/} band and {\it Rossi X-ray Timing Explorer (RXTE)}, 
244: All Sky Monitor (ASM) (Levine et al. 1996) X-ray light curves 
245: are shown in Fig. 2. Both show a similar declining trend until about HJD 
246: 3540 followed by two shorter duration flares similar to those 
247: seen in the 2000 and 2002 outbursts (Wachter et al. 2000; Wijnands et al. 2001, 
248: Wijnands, 2004).
249: There was also a large excess ($\sim 0.45$ mag) in the {\it I \/} 
250: band but not in the other colours on HJD 3527.
251: 
252: 
253: 
254: The decay time-scale during the first outburst (HJD 3522 to 3540) was 
255: $\sim 7$ days for both the optical and the X-ray emission. This is very 
256: similar to that during the discovery outburst in 1998 (Giles et al. 1999)
257: but much shorter than the $\sim 22 $ day time-scale for XTE J0929--314 (Giles
258: et al. 2005). Visual inspection of the light curves suggests that there is
259: a weak correlation between the optical 
260: and X-ray fluxes with the optical emission preceding the X-ray by a few
261: days as in the 1998 outburst (Giles et al. 1999). To test this hypothesis we converted the 
262: optical magnitudes to fluxes normalised to the peak ASM count rate and calculated the 
263: sum of the squares of the differences between the the two fluxes for delays varying between -10 and +10 days.
264: There is a broad minimum, for the whole dataset, corresponding to the optical emission leading the X-ray by -1 
265: to +3 days. When this analysis is restricted to the data during the secondary outbursts commencing at HJD 3542 
266: there is no significant minimum.  This lack of correlation may well be a consequence of inadequate 
267: sampling and the low statistical significance of the ASM data at these times.  
268: 
269: \subsection{Spectral changes}
270: 
271: 
272: In Fig. 3 we plot the {\it B-I\/} and {\it V-I\/} colour index changes during the 
273: observations. Where two measurements in the same colour were made on the same night 
274: we use the mean of the two. The solid line represents a linear fit to the data during the first 
275: steady decline in intensity ending at HJD 3540.  The anomalous point at HJD 3527 is discussed 
276: in section 3.3. Neither it nor the points measured after HJD 3540 are included in the linear fits. 
277: 
278: 
279: 
280: Many of our nightly measurements were made consecutively and therefore
281: cluster within a binary phase duration of  $< 0.25$ but this was not always 
282: the case. In the absence of a continuous set of measurements we are unable to confirm 
283: the continued existence of the 0.12 mag binary modulation in {\it V\/} observed 
284: during the 1998 outburst by Giles et al. (1999).
285: Larger {\it VRI\/} modulations are reported during the quiescent state (Campana et al. 1998)
286: but no {\it B\/} modulation value is known for either state. We have taken the Campana et al. values 
287: (scaled down to the 1998 outburst {\it V\/} modulation), assumed a {\it B\/} modulation a 
288: little smaller than at {\it V\/}, and then used the precise ephemeris of Chakrabarty 
289: \& Morgan (1998) to examine the {\it B-I\/} and {\it V-I\/} shifts expected if any such 
290: modulations were actually present. We find that for {\it V-I\/} the introduced 
291: simulated modulation 'errors' are  $< 0.05$  mag. The scatter in the 
292: lower panel of fig 3 indicates that a modulation of this order may be present although we 
293: have no positive indication for it. For the {\it B-I\/} data the modulation 
294: 'errors' are perhaps twice the size for {\it V-I\/} and of course rather dependent 
295: on the assumed {\it B\/} modulation. However, the small {\it B-I\/} scatter about the 
296: linear fit in the upper panel of fig 3 suggests that minimal modulation 
297: was actually present.
298: 
299: 
300: The most obvious features in both plots is the large increase due to the 
301: flare in {\it I\/} on HJD 3527 and two smaller but significant increases on HJD 3624 and 3546. We 
302: discuss these anomalies below. Overall {\it B-I\/} increased linearly
303: (became redder) until the intensity had decreased by about two magnitudes at the 
304: end of the primary outburst at HJD 3540.  This is illustrated by the solid lines in Fig. 3. The fact that the 
305: decline in emission is greatest in the blue end of the spectrum is suggestive of a viscosity driven 
306: ``inside out'' transition leading to a decline in inner disc emission. 
307: 
308: During the secondary outburst peaking at HJD $\sim 3547$ the spectrum was at first significantly redder 
309: at HJD 3546 and thereafter {\it V-I\/} returned to the values ($0.4 \pm 0.1$) observed at the end of the 
310: initial intensity decline. Inspection of the light curves in Fig. 2 suggests that the increase in {\it V-I\/} on 
311: HJD 3546 was due to a brightening in {\it I\/} 
312: rather than a dimming in {\it V\/}. Another sudden change in colour occurred between the first and second 
313: nights of our observations (HJD 3524 and 3525).  These anomalous changes may be similar in nature to the 
314: {\it I\/} band flare on HJD 3527 but this cannot be confirmed in the absence of {\it R\/} \& {\it I\/} measurements 
315: at those times. An alternative explanation is that they are due to a mass 
316: transfer instability in which cool matter is dumped into the outer disc. This is followed by an increase in 
317: optical emission and colour temperature as the matter diffuses inwards.
318: 
319: 
320: 
321: \begin{figure}
322: \epsfig{file=fig_3.eps, width=8.4 cm}
323:   \caption{Time dependence of the colour indices {\it B - I\/}  
324:   (upper panel) and  {\it V - I\/} (lower panel) for the SAX J1808.4--3658 observations. The solid 
325:   lines represent linear fits to the points marked by filled circles.} 
326: \end{figure}
327: 
328: On 14 occasions within Table 2 we have 4 colour {\it BVRI\/} 
329: measurements taken over a short interval (typically $\sim 1$ hr) on the same night.
330: Using these data and the bandwidth specifications for each filter we have derived
331: broadband {\it BVRI\/} spectra for these nights. In Fig. 4 we plot the spectra for
332: 5 representative nights (HJD 3525, 3527, 3536, 3540 \& 3547) corresponding to the 
333: early bright state, the {\it I\/} band flare, the declining phase, the low intensity
334: phase and peak phase of the secondary outburst respectively. We also include the 
335: {\it V\/} \&{\it I\/} points taken on HJD 3546 during the early rising phase of the 
336: first secondary outburst.
337: 
338: 
339: \begin{figure}
340: \epsfig{file=fig_4.eps, width=8.4cm }
341:   \caption{The {\it BVRI\/} broadband spectra for SAX J1808.4-3658. Each day's 
342:   spectrum is identified by the last two digits of the HJD of the observations. 
343:   The lines connect to the mean flux values 
344:   on each night. Some error limits are smaller than the points representing 
345:   the measurements. The dotted line represents a simple power law disc 
346:   emission model with exponent -3, arbitrary amplitude and interstellar 
347:   reddening corresponding to $A_{V}=0.68$. } 
348: \end{figure}
349: 
350: Also shown is a solid curve representing a power law approximation to the 
351: emission from an optically thick, X-ray heated disc. The distribution is 
352: given by the equation $F_{\lambda} \propto \lambda^{-3}e^{-A_{\lambda } 
353: /1.086}$ where $F_{\lambda }$ (Rieke \& Lebofsky  1985) is the reddened flux at wavelength $\lambda $ 
354: and $A_{\lambda }$ is the wavelength dependent reddening correction toward 
355: the source.  The spectrum is reddened assuming 
356: interstellar extinction $A_{V}=0.68$ (Wang et al. 2001). The amplitude is arbitrary.
357: 
358: 
359: The anomalous spectrum on HJD 3527 will be discussed in the next section. 
360: On most other nights the spectra are consistent with the reddened power law 
361: approximation expected for an X-ray heated accretion disc with interstellar 
362: extinction. The spectra became redder (cooler) with time as noted from the {\it B-I\/} colour index
363: plot (Fig. 3). A similar cooling trend was also apparent in broadband spectra 
364: taken during the 1998 outburst (Wang et al. 2001).  Overall the spectra 
365: were similar (spectral index $\sim 2.5-3.5$) to that in the 1998 outburst
366: (Wang et al. 2001)A somewhat steeper spectrum was reported for the 2003 outburst of the 
367: accretion-powered millisecond pulsar XTE J1814--338  (Krauss et al. 2005).
368: 
369: Except on the night of HJD 3527, there is no {\it I\/} band excess 
370: of the kind reported by Giles et al (2005) for XTE J0929--314 although, in the absence of {\it B\/} 
371: \& {\it R\/} measurements, we cannot exclude the possibility that there was a single colour {\it I\/} 
372: band excess on HJD 3524 and in the early phase of the secondary outburst at HJD 3546. 
373: 
374: 
375: \subsection{{\it I\/} band flare}
376: 
377: On the night of 2005, June 5 the {\it I\/} band measurement at HJD 3527.1449 
378: revealed a $\sim 0.45$ magnitude excess above the trend from previous and succeeding 
379: nights. There was no evidence of any excess above the trend-lines for any of the other
380: colours all of which were measured within $\sim 2$ hrs of the {\it I\/} band. The 
381: existence of an {\it I\/} band excess is also obvious in the broadband 
382: spectrum for the night as shown in Fig. 4.  
383: 
384: The absence of any perceptible increase in {\it R,\ V\/} or {\it B\/} indicates that the enhanced 
385: radiation in {\it I\/} was sharply cut-off in wavelength or was of duration less
386: than 32 minutes when the next measurement (in {\it V\/}) was made. An enhanced orbital modulation can be 
387: ruled out since there was no increase in {\it R\/} or {\it B\/} measured one orbital cycle 
388: after {\it I\/}. 
389: 
390: 
391: {\it RXTE} Proportional Counter Array (PCA) measurements show that a typical Type I X-ray burst occurred 
392: almost in the middle of  our 300 sec {\it I\/} band integration (private 
393: communication, Wijnands \& Klein-Wolt). The peak amplitude was $\sim 40$ times the baseline flux and 
394: the duration (to $ 5\% $ of baseline) was $\sim 35$ s. The associated optical burst generated by 
395: reprocessing of the burst X-rays undoubtedly contributed to the increase in {\it I\/}. 
396: 
397: Type I X-ray bursts have been observed from many neutron star binaries and typically last 
398: 10--20 sec although a few have lasted 
399: up to 150 seconds ( Kong et al. 2000). For the few examples available with simultaneous X-ray 
400: and optical data, mostly from 4U 1636--53 (Pedersen et al. 1982; Lawrence et al. 1983; 
401: Matsuoka et al. 1984) and GS 1826--24 ( Kong et al. 2000), the optical and X-ray profiles are similar 
402: in shape and duration  There is usually evidence for an optical lag of a few seconds corresponding to 
403: the light travel time to the reprocessing site. 
404: 
405: Preliminary calculations suggest that reprocessed radiation contributed only a small fraction of the 
406: increase seen in the {\it I\/} band. A paper on this event, combining the X-ray, optical and radio data, 
407: is in preparation. This will include detailed calculations setting limits on the 
408: optical flux expected from reprocessed radiation. For now we rely on approximate
409: estimates based on scaling arguments using published simultaneous optical and X-ray observations 
410: of the X-ray burst source 4U 1636--53. The peak amplitudes of the bursts were $\sim 40 $
411: times the baseline X-ray flux in both systems although the duration of the SAX J1808.4--3658
412: burst was about twice as long as typical bursts seen in 4U 1636--53.  
413: 
414: Optical bursts are generated by reprocessing of burst X-rays incident on the disc and the companion.
415: The fraction observed from the disc is expected to be strongly dependent on the inclination of the system. 
416: The component coming from the companion star will also be dependent on on the inclination (though not 
417: so strongly) and will be proportional to the solid angle subtended by the companion Roche lobe at the burst 
418: source. Its amplitude will depend on the orbital phase at which the burst occurred. We assume in the calculations 
419: below that the fraction arising from the companion is equal to the fractional orbital modulation observed during 
420: transient outbursts.
421: 
422: The inclination angles are poorly known in both SAX J1808.4--3658 and 4U 1636--53 but  are believed 
423: to be $\sim 60 \degr $ in both systems (Chakrabarty \& Morgan 1998; Homer et al. 2002; Frank, 
424: King \& Lasota 1987). The 
425: X-ray burst occurred at binary phase $\phi = 0.08$  (phase zero corresponds to the time when 
426: the companion star is at its maximum distance from the observer), almost optimum time for observing reprocessed 
427: radiation from the companion. Using the estimated dimensions of the two systems (Frank, 
428: King \& Lasota 1987; Chakrabarty \& Morgan 1998) we find that the solid angle of the Roche lobe
429: in the  4U 1636--53 system is $\sim 4$ times that in  SAX J1808.4--3658. Hence the peak optical burst 
430: flux from the companion in 4U 1636--53 will be 4 times that in SAX J1808.4--3658 assuming that the X-ray 
431: bursts are similar in peak flux. Orbital modulation measurements show, however, that the fraction of 
432: radiation reprocessed on the companion to 4U 1636--53 is $ \sim 25$ per cent (Giles et al. 2002) which 
433: is only twice that in SAX J1808.4--3658 (Giles et al. 1999). Hence, for the same 
434: peak X-ray burst flux, the observed reprocessed radiation from the disc is also larger in 4U 1636--53 
435: than in SAX J1808.4--3658.  This may be a consequence of a larger disc size or perhaps of a smaller 
436: inclination angle for 4U 1636--53. The two conditions above are satisfied if we assume that, for the 
437: same peak X-ray burst flux, optical bursts in 4U 1636--53 are twice as large as those in SAX J1808.4--3658.
438: 
439: The typical peak optical burst flux to baseline ratio in 4U 1636--53 is $\sim 1.5$ 
440: (Pedersen et al. 1982; Lawrence et al. 1983). Bursts in GS 1826--24 are much longer in 
441: duration but the ratio of optical to X-ray burst height is less (Kong et al. 2000). 
442: Given that the X-ray bursts in SAX J1808.4--3658  and 4U 1636--53 are of similar peak amplitude 
443: we expect the optical burst in SAX J1808.4--3658 to have half the  relative amplitude ($\sim 0.75$ times 
444: baseline). The total {\it I\/} band integration 
445: time (300 s) was much longer than the burst duration and this will reduce the amplitude of 
446: the reprocessing signal by the ratio of the integrated optical burst flux to the 'normal' optical flux. 
447: This ratio is $\sim 0.5\times20/300$ assuming the optical burst profile was triangular with effective duration 
448: $ \sim 20$ s. Hence the increase in the integrated {\it I\/} band flux due to reprocessing 
449: would be $\sim 2.5$ per cent above baseline. This is a factor 20 less than 
450: the observed $\sim 50 $ per cent increase in {\it I\/}. There are many uncertainties in the above argument but 
451: it seems unlikely that all the {\it I\/} band excess can be due to reprocessed X-ray burst emission.  
452: The burst may have been the trigger for an on-going synchrotron emission event. 
453: 
454: 
455: 
456: Rupen et al. (2005) detected weak 4.86 and 8.46 GHz radio emission from SAX J1808.4--3658
457: on 2005 June 7, 11 \&16 (HJD 3529, 3533 \& 3538) and suggested it was due to synchrotron emission.  
458: Rea et al. (2005) measured {\it V,\ R\/}  \& {\it I\/} magnitudes on 2005 June 5 (HJD $ \sim 3526.5 $)
459: and set a $5\sigma$ upper limit of 16.5 on {\it H\/} band IR emission. Their measurements of the 
460: optical magnitudes are of low precision but are more consistent
461: with our 'normal' spectrum of June 3 than with the 'anomalous' data of June 5. Their {\it H\/} band upper
462: limit is slightly above an extrapolation of the 'normal' spectrum but is more consistent with it than with 
463: the 'anomalous' one. The measurements by Rea et al. (2005) were made $\sim 12$ hrs before our 
464: observations of June 5. It seems likely that the IR excess was not present at that time and we note also that no 
465: radio emission was detected on 2005 June 4 the day before our observation
466: (Rupen et al. 2005). We conclude that the {\it I\/} band enhancement was due, at least in part, to synchrotron 
467: emission. It commenced less than 12 hours before our observation and was probably triggered by the 
468: type I X-ray burst. Enhanced {\it I\/} band emission lasted less than 60 hours 
469: since it was not present on June 7 (HJD 3529). This implies also that the synchrotron spectrum cut-off frequency decreased since radio synchrotron emission was detected on that day and continued for at least another 9 days 
470: (Rupen et al. 2005). 
471: 
472: Giles et al. (2005) reported a variable {\it I\/} band excess in another
473: accreting millisecond pulsar, XTE J0929--314, and suggested that the anomalous 
474: optical and IR spectra in SAX J1808.4--3658 reported by Wang et al. (2001) may be similar in 
475: nature. Krauss et al. (2005) observed 
476: an {\it I\/} band excess in a third accreting millisecond pulsar XTE J1814--318 and suggested
477: that it was probably due to synchrotron emission. 
478: Hence we conclude that variable synchrotron emission is common in these systems. 
479: 
480: 
481: 
482: 
483: \section{Discussion}
484: Our observations show that on most nights the optical spectra in outburst were consistent with
485: emission from an X-ray heated accretion disc.  The
486: spectra became redder as the intensity decreased. A similar decrease in colour
487: temperature was also apparent in the 1998 outburst of this source (Wang et al. 2001). 
488: These changes  are consistent with an 
489: 'inside-out' transition in the disc i.e. one triggered by a viscosity change driven instability
490: in the inner disc. In this scenario the $\sim 7$ day time-scale for the outburst decline corresponds to
491: the time-scale for depletion of matter in the inner disc.
492: 
493: \begin{figure}
494: \epsfig{file=fig_5.eps, width=8.4 cm}
495:   \caption{Time dependence of the colour indices {\it B - I\/} (upper panel) and {\it V - I\/} 
496:   (lower panel) for the accreting millisecond pulsar XTE J0929--314 measured by Giles et al. (2005). The 
497:   solid lines represent linear fits to the points marked by filled circles.}
498: \end{figure}
499: 
500: 
501: 
502: The spectral changes in this system are remarkably different from that in the accreting millisecond 
503: pulsar XTE J0929--314. In Fig.5 we show the time dependence of {\it B-I\/} and {\it V-I\/} during 
504: the 2002 outburst in XTE J0929--314 (Giles et al. 2005). These are drawn to the same scale as in Fig. 3.
505: Again we note the strong influence of the changes in {\it B\/} (predominately inner disc emission) 
506: on the spectrum. The change in spectral colour is, however, opposite in direction to that seen in SAX 
507: J1808.4--3658 suggesting different outburst mechanisms in the two systems. In XTE J0929--314 the disc became 
508: hotter as the intensity decreased during the 2002 outburst (Giles et al. 2005). This is suggestive 
509: of a mass transfer (outside-in) instability. It is not clear why these systems should differ in this way
510: although we note that the companion in XTE J0929--314 is thought to be a degenerate helium core with
511: remnant envelope and a mass $\sim 0.01M_\odot $ (Galloway et al. 2002). The companion to SAX J1808.4--3658 
512: is a brown dwarf at least 5 times as massive (Bildsten \& Chakrabarty, 2001). This may affect the 
513: stability of the atmospheres of the companions while undergoing X-ray heating. We note however that a
514: mass transfer instability may have contributed to the quasi-periodic secondary outbursts in 
515: SAX J1808.4--3658 commencing at HJD $\sim 3545$ during our observation. 
516: 
517: On several occasions we have detected highly significant changes from night to night in the colour 
518: indices suggestive of sudden changes in disc structure. These were particularly strong in {\it B - I\/}. 
519: Giles et al. (2005) reported similar anomalous changes in XTE J0929--314. During future outbursts four 
520: colour ({\it BVRI\/}) sampling several times per night will be required to adequately characterise 
521: the evolution of these changes.
522: 
523: The {\it I\/} band excess seen on 2005 June 5 was due in part to an optical burst generated by 
524: reprocessing of the X-rays from a Type I burst which occurred during our {\it I\/} band integration. 
525: Preliminary calculations indicate, 
526: however, that reprocessing alone is insufficient to explain the observed increase. We suggest that the X-ray burst 
527: triggered synchrotron emission which had disappeared at optical wavelengths within two days but 
528: which persisted at radio frequencies until at least 2005 June 16 (Rupen et al. 2005). If this scenario is correct 
529: the increase in {\it I\/} band flux due to synchrotron radiation was at least twice the normal flux since synchrotron 
530: emission did not commence until after the middle of the exposure.  We are not aware of 
531: any previous evidence for synchrotron emission being triggered by an X-ray burst. 
532: 
533: Of the seven known accreting millisecond pulsars, three (SAX J1808.4--3658, 
534: XTE J0929--314 \& XTE J1814--338) are now known to have transient near IR emission.
535: It seems likely that this is due to synchrotron emission which extends at times to {\it I\/} band
536: wavelengths. Synchrotron emission at radio and IR wavelengths has been detected from 
537: a number of other X-ray binaries during outburst (Fender, 2001) but it seems it is particularly
538: common in accreting millisecond pulsars. 
539: 
540: 
541: 
542: 
543: 
544: \section{Acknowledgments}
545: This research has made use of data obtained through the High Energy Astrophysics Science 
546: Archive Research Center Online Service, provided by the NASA / Goddard 
547: Space Flight Center. We are grateful to Rudy Wijnands, Marc Klein-Wolt and Manu Linares for 
548: providing us with unpublished {\it RXTE} PCA data. 
549: We thank Rossi Corrales for assistance with the observations, 
550: Kym Hill  and Stefan Dieters for helpful comments and  gratefully acknowledge 
551: financial support for the Mt Canopus Observatory by David Warren. ABG thanks the University of 
552: Tasmania Antarctic CRC for the use of computer facilities.
553: 
554: 
555: 
556: 
557: 
558: 
559: 
560: \begin{thebibliography}{99}
561: 
562: \bibitem{b1}  Bessell M.S., 1990, PASP, 102, 1181
563: \bibitem{b2}  Bildsten L., Chakrabarty D., 2001, ApJ, 557, 292
564: \bibitem{b3}	Bogdanov S., Grindlay J.E., van den Berg M., 2005, ApJ, 630, 1029
565: %\bibitem{b4}  Campana S., D'Avanzo P.,Casares J.,Covino s., Israel G.L., Marconi G., 
566: %			Hynes R., Charles P., Stella L., 2004, ApJ, 614, L49
567: \bibitem{b4}  Campana S. et al., 2004, ApJ, 614, L49
568: \bibitem{b5}  Chakrabarty D., Morgan E.H., 1998, Nature, 394, 346 
569: \bibitem{b6}  Chakrabarty D., Morgan E.H., Muno M.P., Galloway D.K.,
570:                 Wijnands R., van der Klis M., Markwardt C.B.,
571:                 2003, Nature, 424, 42
572: \bibitem{b7} Fender R., 2001, Ap \& SS, 276, 69
573: \bibitem{8} Fender R.P., 2003, in Fender R.P., Macquart J-P.,
574:                 eds, Circular Polarisation from Relativistic Jet Sources, 
575:                 Ap\&SS, 288, 79 
576: \bibitem{b9}  Frank J., King A.R., Lasota J.-P., 1987, A\&A, 178, 137               
577: \bibitem{b10} Galloway D.K., Chakrabarty D., Morgan E.H., Remillard R.A., 
578:                 2002, ApJ, 576, L137
579: \bibitem{b11} Giles A.B., Hill K.M., Greenhill J.G., 1999, MNRAS, 304, 47
580: \bibitem{b12} Giles A.B., Hill K.M., Strohmayer T.E., Cummings N., 2002, ApJ, 568, 279
581: \bibitem{b13} Giles A.B.,  Greenhill J.G., Hill K.M., Sanders E., 2005, MNRAS, 361, 1180
582: \bibitem{b14}	Graham J.A., 1982, PASP, 94, 244
583: \bibitem{b15} Homer L., Charles P.A., Chakrabarty G., van Zyl L., 2002, 
584:                 MNRAS, 325, 147                
585: \bibitem{b16} Hynes R.I., Mauche C.W., Haswell C.A., Shrader C.A., Cui W., 
586:                 Chaty S., 2000, ApJ, 539, L37
587: \bibitem{b17} in 't Zand J.J.M., Heise J., Muller J.M., Bazzano A., Cocchi M., 
588:                 Natalucci I., Ubertini P., 1998, A\&A, 331, L25
589: \bibitem{b18} Kong A.K.H., Homer L., Kuulkers E., Charles P.A., Smale A.P., 2000, 			MNRAS, 311, 405                
590: \bibitem{b19} Krauss M.I. et al., 2005, ApJ, 627, 910
591: \bibitem{b20} Lawrence A. et al., 1983, ApJ, 271, 793 
592: 
593: \bibitem{b21} Levine A.M., Bradt H., Cui W., Jernigan J.G., Morgan E.H., 
594:                Remillard R., Shirey R. E., Smith D.A., 1996, ApJ, 469, L33
595: \bibitem{b22} Markwardt C.B., Miller J.M., Wijnands R., 2002, IAU Circ. 7993
596: \bibitem{b23} Matsuoka M. et al., 1984, ApJ, 283, 774 
597: \bibitem{b24} Pedersen H. et al., 1982, ApJ, 263, 340 
598: \bibitem{b25} Rea N. et al., Atel \#515
599: \bibitem{b23} Rieke G.H., Lebofsky M.J., 1985, ApJ, 288, 618 
600: \bibitem{b26} Rupen M.P., Dhawan V., Mioduszewski A.J., 2005, ATel \#524
601: \bibitem{b27} Wachter S., Hoard D.W., Bailyn C., Jain R., Kaaret P., 
602:                 Corbel S., Wijnands R., 2000, HEAD, 32, 24.15 
603: \bibitem{b28} Wang Z. et al., 2001, ApJ, 563, L61
604: \bibitem{b29} Wijnands R., van der Klis M., 1998, Nature, 394, 344
605: \bibitem{b30} Wijnands R., M\'{e}ndez M., Markwardt C., van der Klis M., 
606:                 Chakrabarty D., Morgan E., 2001, ApJ, 560, 892
607: \bibitem{b31} Wijnands R., 2004, in Kaaret P., Lamb F.K., Swank J.H., eds,  AIP Conference Proceedings, 714, 209 				%						 astro-ph/0403409
608: \bibitem{b32} Wijnands R., 2005, astro-ph/0501264; To appear in Nova Science Publishers (NY) volume 'Pulsars New 											Research'                
609: \end{thebibliography}
610: 
611: 
612: 
613: \label{lastpage}
614: 
615: \end{document}
616: Content-Type: APPLICATION/postscript; name="ms.ps"; x-unix-mode=0644
617: Content-Description: ms.ps
618: Content-MD5: TZbeJqVXxJIwF8mo2DT5Dg==