1: \documentclass{article}

2: \usepackage{graphicx}

3: \oddsidemargin +0.5cm

4: \evensidemargin -2.mm

5: \topmargin=-2cm

6: \textheight 24.0cm

7: \textwidth 16.5cm

8: \begin{document}

9: \rm

10: \let\sc=\sf

11: 

12: \begin{center}

13: \LARGE {\bf The Unique Eclipsing System KH 15D: New Photometric Data}

14: 

15: \vspace{1cm}

16: %\begin{center}

17: \Large {\bf O.Yu.\,Barsunova$^{1,3}$, V.P.\,Grinin$^{1,2,3}$,

18: S.G.\,Sergeev$^2$}

19: \end{center}

20: % \vspace{5mm}

21: \large

22: 

23: 1 - Central Astronomical Observatory Pulkovo, St. Petersburg, Russia,

24: 

25: 2 - Crimean Astrophysical Observatory, Crimea, Nauchny, Ukraine,

26: 

27: 3 - The Sobolev Astronomical Institute, St. Petersburg University, Russia\\

28: 

29: %

30: %

31: 

32: \begin{center}

33: \LARGE{}

34: \end{center}

35: 

36: \begin{abstract}

37: We present results of the photometric observations of the young

38: eclipsing binary KH 15D obtained during the two observational

39: seasons of 2002-2004 years. A comparison of our data with those by

40: Hamilton et al. (2001) and Herbst et al. (2002) reveals an

41: existence of the long-term photometrical brightness trend: during

42: 5 years a system brightness in the I band decreased by about one

43: stellar magnitude. It is also shown that a systematic change of

44: the eclipse parameters found by Herbst et al. (2002) is continuing up

45: to now. The shape of the light curve did not change essentially

46: and is characterized with a small brightening in the central part

47: of the eclipses. The results obtained are discussed in the context

48: of the current models of KH 15D.

49: \end{abstract}

50: 

51: \section{Introduction}

52: An object KH 15D (Sp = K7; V = 18 - 21.5) has been discovered in

53: 1997 by Kearns \& Herbst (1998) during a photometric study of the

54: young cluster NGC 2264. According to Hamilton et al. (2001) this

55: is a weak T Tauri star (WTTS). Its brightness fails by

56: approximately of 3.5 stellar magnitude with a periodicity of 48.35

57: d and in the failed state the object spends about one-third of the

58: period. Such eclipse parameters have no analog among the eclipsing

59: binaries that permits us to refer KH 15D to the unique

60: astrophysical objects.  During the last year it has been published

61: a series of important results which essentially completed an

62: information about this object. First of all, the fact of binarity

63: has been established: using such telescopes as Keck, Magellan II

64: and some others, Johnson et al. (2004) revealed a periodical

65: changes of the radial velocity of the object with an amplitude of

66: 10.7 km/s and estimated the most probable ranges of the system

67: mass function ($0.125 \le f_M/\sin^3 i \le 0.5M_\odot$) and an

68: eccentricity $(0.68 \le e \le 0.80)$. Using the Harvard collection

69: of the photographic observations of NGC 2264, Winn et al. (2004)

70: found that at the beginning of 20th century the light curve of

71: this object strongly differed from that at the present time: an

72: effective duration of eclipses was essentially shorter or they

73: were not observed at all.

74: 

75: From the spectra in the region of 2 $\mu$m Tokunaga et al. (2004)

76: and Deming et al.(2004) found a bipolar outflow in the $H_2$

77: molecular lines. The latter authors concluded that the outflow

78: axis is noticeably inclined to the line-of-sight. Spectral

79: observations of Hamilton et al.(2003) at the bright state and

80: during eclipses showed that an equivalent width of the $H_2$ line

81: as well as those of forbidden lines [OI] 6300/6363 increases

82: synchronously with the decrease of the system brightness at the

83: moments of eclipses. The linear polarization has a similar

84: behavior (Agol et al. 2003). Such a behavior is typical for

85: objects whose source of the continuous radiation completely

86: screened from the observer at the moment of the eclipse, while a

87: more extended region forming a scattered polarized and the

88: emission line radiation is not screened or screened only partly. A

89: similar coronagraphic effect is observed in a clearest form in UX

90: Ori type stars and caused by episodic obscurations of the stars by

91: circumstellar gas and dust clouds (see the review by Grinin 2000,

92: Rodgers 2002 and the literature cited there).

93: 

94: To explain the unusual eclipses of KH 15D several models have been

95: proposed; their review is given by Winn et al. (2003) (see also

96: below). They give different predictions of the photometric

97: behavior of the system, that is why it is interesting to

98: investigate the photometry of this binary for a long time. With

99: this purpose, in 2002 we began the photometric observations of KH

100: 15D. In this paper we present results obtained during the two

101: observational seasons of 2002-2004 years.

102: 

103: \section{Observations}

104: Observations were fulfilled at the AZT-8 0.7 m telescope in the

105: Crimean Astrophysical Observatory (CrAO). The detector was the CCD

106: camera AP7p (512 $\times$ 512 pixels, a pixel size was 24 $\times$

107: 24$\mu$m ) established in the primary telescope focus. The

108: observations were carried out in the tree bands $v, r, i$ and

109: reduced into the Johnson-Cousins photometric system $V, R_c, I_c$.

110: Since the object of interest is in the young cluster NGC 2264,

111: practically all stars in its vicinity demonstrate brightness

112: fluctuations with the different amplitude. We have chosen four

113: stars for the photometric reference set which gave a minimal mean

114: square amplitude of the variation relatively to each other during

115: the time range considered. Table 1 contents numbers of these stars

116: according to the catalogue by Flaccomio et al. (1999) as well as

117: stellar magnitudes in the $I_c$  band. The rest columns give mean

118: $I_c$ values and their dispersions which we obtained in our

119: observations for two subsequent seasons. It is seen that these

120: values agree well with those published by Flaccomio et al. (1999).

121: Using all the four comparison stars permits us to degrade an

122: influence of the brightness fluctuation for each of them and

123: assure an accuracy of the photometry in the $I_c$ band not less

124: than $0.^m03$.

125: 

126: %

127: \begin{table}

128: %\centering

129: \caption[]{Comparison stars} \label{ ta: one}

130: \begin{flushleft}

131: \begin{tabular}[center]{cccccc}

132: \hline\noalign{\smallskip}

133: N&$I_{Fl}$ & $I_1$&$\sigma$&$I_2$&$\sigma$  \\

134: %& & m&m&m   \\

135: \noalign{\smallskip}

136: \hline\noalign{\smallskip}

137: \noalign{\smallskip}

138: 281&13.60&13.57&0.03&13.62&0.04 \\

139: 328&14.53&14.48&0.04&14.47&0.04 \\

140: 353&13.63&13.74&0.03&13.70&0.04 \\

141: 422&13.66&13.62&0.05&13.62&0.05 \\

142: \noalign{\smallskip}

143: \hline\noalign{\smallskip}

144: \hline

145: \end{tabular}

146: \end{flushleft}

147: {\small Note: meanings of $I_{Fl}$ are taken from Flaccomio et al.

148: (1999).}

149: \end{table}

150: %

151: 

152: \subsection{Accounting the sky background}

153: One can see from Fig. 1 that near KH 15D there is a bright star

154: ($V \approx 6$) whose scattered light is characterized with a

155: strong gradient at the location of the object of interest. This is

156: a serious obstacle in the correct accounting the sky background,

157: especially in the brightness minima. Therefore, a standard

158: procedure of the aperture photometry is not valid in this case

159: because the brightness of the sky background under the object can

160: noticeably differs from that we determine using the ring around

161: it.

162: 

163: \begin{figure*}

164: \vspace{-5.5cm}

165: %\hspace{3.5cm}

166: \centering

167: \includegraphics[width=14cm]{FIG1.EPS}

168: \vspace{-6.5cm} \caption{ Image of KH 15D in the $I_c$ band in the

169: bright state (left) and during the eclipse (right).}

170: \end{figure*}

171: %

172: 

173: This problem can be solved by several ways from which we chose the

174: photometry with the instrumental profile PSF (Point Spread

175: Function). Let us suppose, that in the close vicinity of KH 15D

176: the scattered light is described with some plane defined by three

177: free parameters, say a central intensity and gradients on $X$ and

178: $Y$. Also, let us have the PSF for the image. Then, selecting the

179: position and intensity of the PSF as well as the parameters of the

180: plane yields the best approach to the brightness distribution

181: observed in this small region and hence, permits to find the

182: brightness of the object itself using the fitted PSF intensity.

183: The task is reduced to the multidimensional optimization of the

184: objective function. For this we developed and carefully tested a

185: special algorithm. In practice we varied only the position and

186: intensity of the PSF; the plane parameters were shaped due to the

187: plane approximation of the residual after a substraction of the

188: PSF. The size of the locality was $5 \times 5$ pixels or $8.8

189: \times 8.8$ arcseconds. The PSF itself was automatically

190: determined beforehand for each image by selection of the PSF for

191: the single stars. After averaging the PSF over all these stars we

192: condemned those of them which did not correspond to the mean PSF

193: due to the criterium of the mean square dispersion.

194: 

195: Comparison of the two methods (an aperture photometry and PSF)

196: showed that an essential discrepancy between them took place only

197: in the brightness minimum of KH 15D, and the first method often

198: gives formally negative brightness of the object thus resulting

199: the brightness of the sky background. In the bright state both

200: tools lead to close results with the mean square dispersion of

201: 0.03 of the stellar magnitude.

202: 

203: An accuracy of the photometry was determined by with the

204: Monte-Carlo method i.e., frequentative introduction of the Gauss

205: noise into the image. The error in each pixel was found from the

206: quanta statistically accumulated, taking into account the pickup

207: noises and operations made with the certain image during its

208: treating. To this error we added (in the quadrature) the error of

209: the stellar magnitude caused by the method inaccuracy mentioned

210: above (0.03 of the stellar magnitude). The resulting accuracy of

211: the KH 15D photometry in its bright state was conditioned mainly

212: by the brightness fluctuations of the comparison stars and amounts

213: to $0.^{m}$03; in the weak state it is not worse than $0.^{m}$2.

214: 

215: \section{Results}

216: Figs. 2 and 3 demonstrate results of the KH 15D photometry in the

217: $I_c$ band based on our data and also on the data by Hamilton et

218: al. (2001) and Herbst et al. (2002). The photometric observations

219: in these two papers are given in the Cousins system like ours. An

220: accuracy of the measurements was about $0.^{m}01 - 0.^{m}02$ in

221: the bright state and $0.^{m}1 - 0.^{m}2$ in the weak state.

222: 

223: We see the photometric trend: the brightness of KH 15D

224: monotonously decreases with time. We discovered this trend

225: analyzing our observations: as it is seen from Fig. 2, during the

226: observational season of 2002-2003 yrs and partially in the next

227: season the object's brightness was droningly declining out of

228: eclipses. Supplementing the data kindly provided by W. Herbst to

229: our data we ascertained that this trend accorded well to the

230: brightness behavior of the object in the previous observational

231: season 2001-2002 yrs (as we can see from Fig. 2, it is also in

232: evidence in the behavior of the bright state of the object in

233: season of 2001-2002 yrs, when observations of KH 15D were carried

234: out by several groups (Herbst et al. (2002)) and were closest in

235: time.

236: 

237: \begin{figure*}

238: \vspace{-1.5cm}

239: \centering

240: \includegraphics[angle=-90,width=12cm]{FIG2.EPS}

241: \caption{ Light curve of KH 15D in the Cousins I band based on the

242: data of Herbst et al. (2002) (dots) and ours (triangles).}

243: \end{figure*}

244: %

245: 

246: Further supplementing these data by those obtained in 1999-2000

247: yrs by Hamilton et al. (2001) yields (see Fig. 3a) that the

248: earlier observations of this objects also reveal a brightness

249: behavior of analogous tendency.

250: 

251: As one can see from Fig. 3a, the brightness of KH 15D out of

252: eclipses has diminished by approximately $0.^{m}$7 for the two

253: seasons of 1999-2001 while during next three observational seasons

254: (2001-2004) the object has failed by about $0.^{m}$3. This implies

255: a retardation of the photometric trend and some points on the

256: light curve obtained at the end of the observational season of

257: 2003-2004 yrs (Fig. 2) indicate to its possible completion and

258: beginning of the brightness increase of KH 15D. Such a light

259: behavior (if it will be confirmed by farther observations) can

260: serve an important clue for understanding a mechanism of the

261: eclipses of this object (see below).

262: 

263: On Fig. 3 we give convolutions of the light curve of the object

264: with the period of 48.35 days for the three time ranges mentioned

265: above. The brightness trend of KH 15D is distinctly seen. One can

266: see also that the increase of the eclipse duration noticed firstly

267: by Hamilton et al. (2001) occurs up to date. From Fig. 3b, where

268: the light curves are given after the trend removal we see that

269: diminution of the eclipse depth, observed by Hamilton et al.

270: (2001) and Herbst et al. (2002), has decelerated  or fully stopped

271: in 2002-2004. But the shape of the light curve in total did not

272: change. In the central part of the eclipse a weak brightening

273: takes place as before apparently being a special feature of the

274: light curves of the object (Hamilton et al. (2001),Herbst et al.

275: (2002)). Analyzing our observations in the $V$ and $R_c$ bands we

276: proved the conclusion of Hamilton et al. (2001) about a neutral

277: character of the eclipses of KH 15D.

278: 

279: 

280: \begin{figure*}

281: \vspace{-1.5cm}

282: \hspace{-3.5cm}

283: \includegraphics[angle=-90,width=23cm]{FIG3.EPS}

284: \vspace{-6cm} \caption{ a) Light curve of KH 15D â in the $I_c$

285: band convolved with a period of 48.36 d; b) the same as a) after

286: accounting the trend.  Circles denote data of Hamilton et al.

287: (2001) for 1999-2000 yrs, dots are the data of Herbst et al.

288: (2002) for 2001-2002 yrs, triangles are our data for 2002-2004

289: yrs.}

290: \end{figure*}

291: %

292: 

293: As we can see from Fig. 3, the radiation flux from the object

294: fluctuates at the moment of eclipse with an amplitude of about one

295: stellar magnitude. It follows from the light curve obtained by

296: Herbst et al. (2002) in 2001-2002 yrs when such fluctuations were

297: observed rather often due to the dense monitoring. We also fixed

298: one episode when the flux strongly deviated from the average light

299: curve during the eclipse. Since in the light minima the radiation

300: scattered by the circumstellar dust mainly contributes to the flux

301: observed (a growth of the linear polarization in the minima

302: justifies this (Agol et al. (2003)), the special behavior of the

303: object's brightness during the eclipse pointed out above testifies

304: that parameters of the scattered radiation of KH 15D are undergone

305: strong fluctuations.

306: 

307: \section{Discussion}

308: The new and the most important result of our analysis is a

309: conclusion about an existence of the trend of the KH 15D

310: brightness: in total, for the last 5 years the object faded by

311: about one stellar magnitude. When we have already been obtained

312: this result, Johnson and Winn (2003) published results of their

313: analysis of the photographic observations of NGC 2264 cluster

314: carried out at the observatory Asiago in 1967 - 1982. They found

315: that at that epoch KH 15D in the $I_c$ band was brighter by $\sim

316: 0.^{m}9$ of stellar magnitude compared to the data of Herbst et

317: al. (2002) in 2001-2002. Taking into consideration that the

318: brightness trend we discovered is comparable in amplitude with

319: that determined by Johnson and Winn (2003) but is reached during

320: the essentially shorter time, one can suppose that it has a cyclic

321: character and, apparently, duration of the cycle significantly

322: less then 30 years. The deceleration of the trend noted above

323: indirectly confirms this suggestion.

324: 

325: How to explain this trend of the brightness in the framework of

326: the models elaborated for this eclipsing system?

327: 

328: Obviously, one of the possible reasons for the slow changes of the

329: object's brightness out of eclipse can be a change of the

330: luminosity of the primary component of the binary system. Although

331: we cannot to specify a concrete mechanism caused such variations,

332: nevertheless, one cannot reject completely such an explanation.

333: Another possible cause is an additional, slowly varied with time

334: extinction of the radiation from the primary on the way to an

335: observer.

336: 

337: Based on the observed variations of the eclipse parameters, Winn

338: et al. (2004) and Chiang and Murray-Clay (2004) proposed the model

339: of KH 15D in which the binary system consists of two components

340: with approximately the same parameters and is surrounded with and

341: opaque gas and dust ring or disk (farther for brevity we shall

342: call it circumbinary (CB) disk. It is assumed that the system is

343: seen nearly edge-on while the CB disk is inclined to the

344: line-of-sight on about 20$^{\circ}$ and slowly precesses

345: relatively to the mass center of the system. Eclipses occur when

346: one or both companions turn up in "the shadow" of the CB disk

347: during their orbital motion. In order to explain the sharp ingress

348: and egress from the minimum, the authors claim a rather steep

349: internal boundary of the CB disk.

350: 

351: The changes of the parameters of the eclipse in this model is

352: attributed to the CB disk precession. It is suggested that in the

353: modern epoch one of the component has already completely closed by

354: the CB disk and only near the periastron its radiation slightly

355: shines trough the disk (that explains a small rise of the

356: brightness in the central part of the eclipse). Eclipses befall

357: when the secondary sets behind the opaque boundary of the disk in

358: its orbital motion.

359: 

360: Grinin and Tambovtseva (2002) suggested that the disk wind of the

361: secondary, opaque due to presence of the dust can cause the long

362: lasting eclipses. They supposed that the plane of the binary

363: (coinciding with the plane of the CB disk) is inclined to the

364: line-of-sight. Calculations show (Grinin and Tambovtseva (2002);

365: Grinin et al. (2004)) that if the dust in the wind has parameters

366: typical for the circumstellar disks and well mixed with the gas in

367: a proportion 1:100 as in the interstellar medium, then the

368: amplitude of the brightness decrease observed in KH 15D is

369: obtained at the mass loss rate of the disk wind (in the low

370: velocity component) of about $3 \cdot 10^{-8} \dot{M_{\odot}}$

371: yr$^{-1}$. A change of the eclipse parameters in this model takes

372: place as a result of the motion of the apsis line of the

373: secondary's orbit (Grinin et al. (2004)) but can be also caused by

374: the wind instabilities. Brightening observed in the central part

375: of an eclipse is due to low density of the matter in the interiors

376: of the disk wind cone.

377: 

378: Since CB disks in the binaries with eccentric orbits are described

379: with a global asymmetry (Artymowicz and Lubow (2000)), then, under

380: a small inclination to the line-of-sight, in such systems along

381: with eclipses caused by the disk wind of the secondary one can

382: also observe essentially slower brightness variations (similar to

383: that we found in KH 15D) attributed to changes of extinction on

384: the line-of-sight due to the CB disk precession.

385: 

386: \section{Conclusion}

387: Thus, we showed that in the modern epoch there is trend of the

388: brightness in the light variations of KH 15D: during 5 years this

389: object has failed by about one stellar magnitude. Analyzing the

390: bright state in the last observational season we suspect that this

391: trend slowed down and some cue on the opposite tendency in the

392: brightness behavior appeared (Fig. 2). If this tendency will be

393: confirmed by future observations it will be possible to certify

394: that "the end of light" predicted by the models of Winn et al.

395: (2004) and Chiang and Murray-Clay (2004), when both components of

396: the system will be completely closed from an observer by the CB

397: disk (according to estimates of Winn et al. it will be in 2012)

398: will not take place.

399: 

400: The results of the analysis given above show that in order to

401: clarify the essence of unusual eclipses of KH 15D it is necessary

402: to have an information about light curves not only at the moment

403: of eclipses (which is mainly interested by observers) but also out

404: of them. It would be highly interesting also to carry out a

405: photometric monitoring of this unique object in the near infrared

406: region of spectrum.\\

407: 

408: 

409: \textbf{Acknowledgements}\\

410: We thank V. Doroshenko and E. Sergeeva for the assistance in

411: observations, Bill Herbst for providing us with the data of the

412: photometric observations of KH 15D and also V.M. Larionov, the

413: referee of the paper for his useful comments. The work is

414: supported by the grant of Presidium of RAS "Non-stationary

415: phenomena in Astronomy" and the grant INTAS N 03-51-6311.

416: 

417: \begin{center}

418: {\bf References}

419: \end{center}

420: \large

421: E. Agol, A. Barth, S. Wolf, D. Charbonneau, 2003, astro-ph 0309309\\

422: P. Artymowicz, S.H. Lubow, 1996, Astrophys. J. {\bf 467} L77 \\

423: E.I. Chiang, R.A. Murray-Clay, 2004, astro-ph 0312515\\

424: D. Deming, D. Charbonneau, J. Harrington, 2004, Astrophys. J.,

425: {\bf 601}, L87\\

426: E. Flaccomio, G. Micela, S. Sciortino, F. Favata, C. Corbally, A.

427: Tomaney, 1999, A\&A, {\bf 345}, 521\\

428: V.P. Grinin, 2000, in "Disk, Planetesimal and Planets", Eds. by

429: F.Garzon, et al. ASPC, {\bf 219}, p. 216 \\

430: V.P.Grinin, L.V. Tambovtseva, 2002, Astron. Letters, {\bf 28},

431: 601\\

432: V.P.Grinin, L.V. Tambovtseva, N.Ya. Sotnikova, 2004, Astron.

433: Letters. \textbf{30}, 694\\

434: C.M. Hamilton, W. Herbst, C. Shih, A.J. Ferro, 2001, ApJ

435: \textbf{554}, L201 \\

436: C.M. Hamilton, W. Herbst, R. Mundt, C.A.L. Bailer-Jones, C.M.

437: Johns-Krull, 2003, Astrophys. J., {\bf 591}, L45 \\

438: W. Herbst, C.M. Hamilton, F.J. Vrba et al. 2002, PASP \textbf{114}, 1167 \\

439: J.A. Johnson, J.N. Winn, 2003, astro-ph 0312428\\

440: J.A. Johnson, G.W. Marcy, C.M. Hamilton, et al. 2004, astro-ph 0403099\\

441: K.E. Kearns, N.L. Eaton, W. Herbst, C.J. Mazzurco 1997, AJ \textbf{114}, 1098\\

442: B. Rodgers, D. Wooden, V.P. Grinin, D. Shakhovskoj, A. Natta,

443: 2002, Astrophys. J., {\bf 564}, 405\\

444: A.T. Tokunaga, S. Dahm, W. G\"assler et al. 2004, astro-ph 0401177\\

445: J.N. Winn, P.M. Garnavich, K.Z. Stanek, D.D. Sasselov, 2003, ApJ

446: {\bf 593}, L121\\

447: J.N. Winn, M.J. Holman, J.A. Johnson et al. 2004, astro-ph 0312458\\

448: 

449: \end{document}

450: