0804.1375/ms.tex
1: %\documentclass{emulateapj}
2: \documentclass[12pt,preprint]{aastex}
3: %% manuscript produces a one-column, double-spaced document:
4: 
5: %% \documentclass[manuscript]{aastex}
6: 
7: %% preprint2 produces a double-column, single-spaced document:
8: 
9: %% \documentclass[preprint2]{aastex}
10: 
11: %% Sometimes a paper's abstract is too long to fit on the
12: %% title page in preprint2 mode. When that is the case,
13: %% use the longabstract style option.
14: 
15: %% \documentclass[preprint2,longabstract]{aastex}
16: \newcommand{\rxte}{{\em RXTE\ }}
17: \newcommand{\rxteno}{{\em RXTE}}
18: \newcommand{\exosat}{{\em EXOSAT\ }}
19: \newcommand{\exosatno}{{\em EXOSAT}}
20: \newcommand{\integral}{{\em INTEGRAL\ }}
21: \newcommand{\integralno}{{\em INTEGRAL}}
22: \newcommand{\swift}{{\em Swift\ }}
23: \newcommand{\swiftno}{{\em Swift}}
24: 
25: \newcommand{\Hzs}{Hz s$^{-1}$} 
26: \begin{document}
27: \title{Outbursts Large and Small from EXO 2030+375}
28: \author{Colleen A. Wilson, Mark. H. Finger}
29: \affil{VP62, National Space Science and Technology Center, 320 Sparkman Drive, Huntsville, AL 35805}
30: \email{colleen.wilson@nasa.gov}
31: \author{Ascensi{\'o}n Camero Arranz}
32: \affil{GACE/ICMUV, Universidad de Valencia, P.O. Box 20085, 46071 Valencia,
33: Spain}
34: \begin{abstract}
35: 
36: During the summer of 2006, the accreting X-ray pulsar EXO 2030+375 underwent
37: its first giant outburst since its discovery in 1985. Our 
38: observations include the first ever of the rise of a giant outburst of  EXO 
39: 2030+375. EXO 2030+375 was monitored daily with the 
40: {\em Rossi X-ray Timing Explorer (RXTE)} from 2006 June through 2007 May. During the 
41: giant outburst, we discovered evidence for a cyclotron
42: feature at $\sim 11$ keV. This feature was confidently detected for about 90
43: days, during the brighter portion of the outburst. Daily observations of the 
44: next five EXO 2030+375 orbits detected pulsations at all orbital phases and
45: normal outbursts shifted to a later orbital phase than before the giant 
46: outburst. An accretion disk appears to be present in both the normal and giant
47: outbursts, suggesting that the long-term behavior is a product of the state of
48: the Be star disk and the accretion disk. Here we will present flux and frequency histories from our detailed 
49: \rxte observations of the giant outburst and the normal outbursts that
50: surrounded it. A new orbital analysis is presented that includes observations from 1991 
51: through 2007 August. 
52: 
53: \end{abstract}
54: \keywords{accretion---stars:pulsars:individual:(EXO\ 2030+375)---X-rays:
55: binaries}
56: \section{Introduction}
57: Be/X-ray binaries are the most common type of accreting X-ray pulsar systems
58: observed.
59: They consist of a pulsar and a Be (or Oe) star, a main sequence star of
60: spectral type B (or O) that shows Balmer emission lines 
61: \citep[See e.g.,][for a  review.]{Porter03} The line emission is believed to be 
62: associated with circumstellar material shed by the Be star into its equatorial 
63: plane. The exact nature of the mass loss process is unknown, but it is thought to be related to
64: the rapid rotation of the Be star, typically near 70\% of the critical break-up
65: velocity \citep{Porter96}. The equatorial material forms a slow, dense outflow,
66: which is generally believed to fuel the X-ray outbursts. Near the Be star, the
67: equatorial outflow probably forms a quasi-Keplerian disk
68: \citep{Quirrenbach97,Hanuschik96}.
69: 
70: X-ray outbursts are produced when the pulsar interacts with the Be star's disk.
71: Be/X-ray binaries typically show two types of outburst behavior: (a) giant 
72: outbursts (or type II), characterized by high luminosities and high spin-up 
73: rates  (i.e., a significant increase in pulse frequency) and (b) normal 
74: outbursts (or type I), characterized  by lower luminosities, low spin-up rates
75: (if any), and repeated occurrence at the orbital period \citep{Stella86,Bildsten97}. As a
76: population Be/X-ray binaries show a  correlation between their spin and orbital
77: periods \citep{Corbet86,Waters89}.
78: 
79: For isolated Be stars, variations in the infrared bands (J,H,K) are believed to be good
80: indicators of the size of the Be star's disk. However, when the Be star is in a
81: binary system with a neutron star, the Be disk is truncated at a resonance 
82: radius by tidal forces from the orbit of the neutron star \citep{Okazaki01}. In
83: these systems, since the disk cannot easily change size because of the truncation 
84: radius, changes in mass loss from the Be star produce changes in the disk 
85: density, which can even become optically thick at infrared wavelengths 
86: \citep[see, e.g.,][]{Neg01a,Mir01}.  
87: 
88: EXO 2030+375 is a 42-s transient accreting X-ray pulsar discovered with {\em
89: EXOSAT} during a giant outburst in 1985 \citep{Parmar89a}. In this system, the
90: pulsar orbits a B0 Ve star \citep{Motch87, Janot88, Coe88} every 46 days
91: \citep{Wilson05}. A normal outburst has been detected for nearly every periastron
92: passage since 1991 \citep{Wilson02, Wilson05}. For these normal outbursts, the
93: outburst intensity and the global spin-up rate appeared to be tied to
94: the K-band intensity of the Be star. From 1992-1994 EXO 2030+375's outbursts
95: were bright and the pulsar was spinning-up. In 1994, shortly after a drop in the
96: K-band intensity of the Be disk, the X-ray intensity abruptly dropped and the
97: global trend changed to spin-down, indicating that the density of the Be disk
98: had dropped and less material was available for accretion. The pulsar continued
99: with faint X-ray outbursts and a global spin down trend until 2002, when again
100: the K-band intensity increased to the 1992-1994 level followed by brighter 
101: X-ray outbursts and a transition to global spin-up \citep{Wilson05}. The 
102: outbursts continued to brighten and show spin-up until 2006 June when EXO 
103: 2030+375 underwent its first giant outburst since its discovery in 1985 
104: \citep{Corbet06,Krimm06,McCollough06,Wilson06}. In this paper we present a
105: timing analysis including twelve normal outbursts leading up to the giant
106: outburst, the giant outburst, and seven normal outbursts after the giant
107: outburst combined with previously published observations. 
108: 
109: 
110: Accreting X-ray pulsars have strong surface magnetic fields of $\sim 10^{12}$ G. A
111: direct measurement of this field strength is provided by the energies of
112: cyclotron resonance scattering features in their X-ray spectra. The
113: magnetic field strength and the resonance energy are related as $E_{\rm cyc} = 11.6
114: B_{12} (1+z)^{-1}$, where $B_{12}$ is the magnetic field strength in units of
115: $10^{12}$ G and $z$ is the combined gravitational plus bulk motion Doppler shift 
116: \cite[and references therein]{Nakajima06}. The range of previously measured cyclotron
117: features is from about 11 keV in 4U0115+63 \cite{Nakajima06} to  about 50 keV in
118: A0535+262 \citep{Terada06}. Previously a tentative cyclotron at 36 keV feature was
119: reported in a 1996 normal outburst of EXO 2030+375 \citep{Reig99}; however, this 
120: feature has not been seen in other any other observations. Recent observations 
121: of EXO 2030+375 during its giant outburst have resulted in reports of a 
122: cyclotron feature near 10 keV \citep{Wilson06, Klochkov07}. In this paper we
123: present detailed spectral fits to \rxte observations from the giant 
124: outburst, demonstrating that the cyclotron feature was consistently detected 
125: over an extended period of time. 
126: 
127: %EXO 2030+375 is a 42-s transient accreting X-ray pulsar with a Be star companion
128: %\citep{Coe88, Janot88} that was discovered during a giant outburst with 
129: %{\em EXOSAT} \citep{Parmar89a,Parmar89b}. The pulsar orbits its companion in a
130: %46-day eccentric orbit and showed outbursts near every observed periastron
131: %passage since 1991 \citep{Wilson02,Wilson05}.
132: 
133: %Be/X-ray binaries such as EXO 2030+375 show two types of outbursts: (1) Normal
134: %or Type I outbursts that are periodic with the orbital period, resulting from
135: %interaction between the pulsar and the Be star's circumstellar disk and (2)
136: %Giant or Type II outbursts that exhibit much larger fluxes and spin up
137: %rates\citep{Stella86,Bildsten97}.
138: 
139: 
140: 
141: 
142: \section{Observations and Analysis}
143: 
144: Figure~\ref{fig:ltflx} shows the long-term flux history for EXO 2030+375 since
145: its discovery as measured with \exosat, BATSE, and \rxte ASM. This Figure is provided here to aid the reader
146: in placing the observations described below in context.
147: 
148: \begin{figure}
149: \plotone{f1.eps}
150: \caption{Long-term Flux history for EXO 2030+375. The pink diamonds and left
151: y-axis denote \exosat 1-20 keV total flux measurements from \citet{Parmar89a}.
152: Blue points and the y-axis near the center denote 20-50 keV pulsed fluxes
153: measured with BATSE. Typically 1-4 of these BATSE points correspond to a normal outburst.
154: The red histogram and right y-axis denote \rxte ASM 2-10 keV total flux 
155: measurements averaged over 4 days. Each spike corresponds to a normal outburst 
156: and the large peak is the 2006 giant outburst. 
157: \label{fig:ltflx}}
158: \end{figure}
159: \subsection{\rxte Observations}
160: 
161: The 2006 giant outburst and five normal outbursts that followed from EXO 2030+375
162: were observed daily from 2006 June 22 through 2007 May 25 (MJD 53,908-54,245) with
163: the {\em Rossi X-ray Timing Explorer (\rxteno)} Proportional Counter Array 
164: \citep[PCA]{Jahoda06} and the High-Energy X-ray Timing Experiment 
165: \citep[HEXTE]{Rothschild98}. Also included in this paper are analyses of previously
166: unpublished \rxte observations of eight normal outbursts in the year prior to the 
167: giant outburst (MJD 53,540-53,868) and \rxte observations of outbursts in 2007
168: June (MJD 54,276-54,283) and 2007 August (MJD 54,322-54,329). For these outbursts,
169: the typical 1-5 ks observations spanned about six days near the peak of the 
170: normal outburst. In this paper, timing analysis was performed using the Standard 1 
171: PCA data, which has 0.125 s time resolution and no spectral resolution. Phase 
172: averaged spectral analysis was performed using the Standard 2 PCA data with 129 
173: channel energy resolution and 16-s time resolution and the science data mode
174: E\_8us\_256\_DX1F HEXTE data for cluster B only, since cluster A was not rocking and
175: was not on-source during the early and late parts of the giant outburst.
176: 
177: \subsection{\integral Observations}
178: {\em International Gamma Ray Astrophysics Laboratory} (\integralno) ISGRI
179: \citep{Ubertini03} public data from revolutions 18-22 (MJD 52,615-52,627), 67
180: (MJD 52,661-52,662), 80 (MJD 52,800-52,802), 159-160 (MJD 53,036-53,041), and 
181: 185-193 (MJD 53,126-53,133) was also included in our timing analysis. Our software,
182: described in detail for the revolutions 18-22 observations in \citet{Camero05}, 
183: collected good events in the 20-60 keV band using the pixel information function 
184: when the source was in the partially or fully coded field-of view of ISGRI. The 
185: good events were then epoch folded using a simple phase model based on spin 
186: frequency measurements with \rxte and \integral. To avoid binning effects, each 
187: pulse profile was fit with a Fourier series of harmonic coefficients. A correction
188: described in \citet{Camero05} was applied to account for aperiodic noise due to 
189: pulse profile variations and due to nearby noisy sources. A template profile was 
190: estimated from the average profile for revolutions 18-22. To generate phase offsets
191: from the model, we then cross-correlated the individual profiles with the template
192: profile. The new phases for each outburst were then fit with a linear or
193: quadratic phase model and the process was repeated, creating new folded
194: profiles, new harmonic coefficients, and new phase offsets. The pulse profiles
195: were then combined over time to improve statistics and to allow the phase
196: measurements to constrain spin-up during each outburst. Pulse phase measurements
197: with \integral from five normal outbursts of EXO 2030+375 are used in this paper.
198: 
199: 
200: \subsection{Timing Analysis}
201: For each \rxte PCA observation, we generated a light curve file from Standard 1
202: data using FTOOLS v6.2. We corrected each light curve to one average PCU using
203: the FTOOL {\it correctlc}. Using {\it faxbary} we corrected the times on
204: each bin to the solar system barycenter. Then we corrected the times for
205: the pulsar's orbit using the orbital parameters from \citet{Wilson05}. Lastly, 
206: we fit a pulse profile model consisting of a 6th order Fourier expansion in 
207: the pulse phase model plus a constant background term. For outbursts before and
208: after the giant outburst where we did not have daily coverage, the phase
209: model initially consisted of a constant frequency and was iteratively improved
210: to a quadratic phase model for each outburst. For the daily 
211: measurements, the model was a quadratic spline with each $\sim 5$-day interval 
212: having an independent spin-up rate. To compensate for aperiodic noise due to pulse
213: profile variations within each \rxte observation, we first subtracted the pulse
214: profile model from the light curve and then computed a power spectrum. Within 
215: each power spectrum we computed the average noise power around each harmonic 
216: $\bar P_n$ in the frequency range $[(n-1/2)\nu,(n+1/2)\nu]$ where $n$ is the 
217: harmonic number and $\nu$ is the model frequency for that observation. The 
218: errors on the pulse profile harmonic coefficients were then inflated by 
219: $(\bar P_n/2)^{(1/2)}$ where 2 is the assumed Poisson level. Phase offsets were 
220: computed by cross-correlating the individual profiles with a template profile.
221: 
222: Examination of the measured spin-up rates within the contiguous daily
223: observations of the giant outburst and five normal outbursts showed a periodic 
224: dip in the spin-up rate, just before periastron, indicating that the orbital 
225: solution needed improvement. To improve the orbital solution, we combined our 
226: new phase measurements with previously published phase measurements from the 
227: Burst and Transient Source Experiment (BATSE)
228: \citep[51 outbursts: 1991-2000 (MJD 48,385-51,657)]{Wilson02} \rxte 
229: \citep[4 outbursts; 1996 (MJD 50,266-50,275), 1998 (MJD 50,821-50,828), 2002
230: (MJD 52,432-52,443), and 2003 (MJD 52,894-52,899)]{Wilson02, Wilson05}, and 
231: \integral \citep[1 outburst; 2002]{Camero05}. The phases were fit with a global
232: orbital model plus a quadratic spline. Prior to the giant outburst and for
233: outbursts from 2007 June through September, the quadratic spline had an 
234: independent frequency derivative for each outburst. Phase ``slips", jumps in the
235: pulse cycle count, were fit in large gaps between outbursts where no data were 
236: available to constrain the phase model. The large spin-up rates during the giant
237: outburst overwhelmed the orbital effects, so it was excluded from orbital 
238: analysis. 
239: 
240: We found, however, that our initial fits were much poorer than expected,
241: with reduced $\chi^2$ values of about 2.0. Neither the phase residuals nor the
242: phase errors showed any dependence on intensity or orbital phase, suggesting
243: that the phase errors were simply too small across the board, indicating that
244: the phase errors did not sufficiently reflect the effects of pulse profile
245: variations from observation to observation. We added a systematic error of 
246: $2.4 \times 10^{-3}$ cycles in quadrature to all of the phase measurements 
247: included in our fits. In addition we eliminated six BATSE points that were large
248: outliers despite the already large BATSE error bars. 
249: 
250: Table~\ref{tab:orb} lists the orbital parameters resulting from our fits: the orbital period $P_{\rm
251: orb}$, the epoch of periastron passage $T_{\rm peri}$, the projected semi-major
252: axis $x = a_x \sin i$, the eccentricity $e$, and the periapse angle $\omega$. The
253: largest differences between our new orbital fit (Fit 1) and published results
254: \citep{Wilson05} are in $e$ (2.9$\sigma$ smaller) and $a_x \sin i$ (2.5$\sigma$
255: larger). Next we searched for time dependent variations in the orbital
256: parameters, fitting the orbital period derivative $\dot P_{\rm orb}$, the 
257: derivative of the periapse angle $\dot \omega$, and the derivative of the 
258: projected semi-major axis $\dot x$. First we varied each parameter separately,
259: holding the other two fixed at zero, while allowing all other fit parameters to
260: vary. Varying $\dot \omega$ produced the most significant result (Fit 2), with
261: an F-test significance of 3.4$\sigma$. $\dot P_{\rm orb}$ and $\dot x$
262: were less significant with F-test values of 1.7$\sigma$ and 3.0$\sigma$,
263: respectively. Next we tried varying both $\dot \omega$ and $\dot x$ (Fit 3) while
264: holding $\dot P_{\rm orb}$ fixed at zero and allowing all other parameters
265: to vary. This fit had an F-test significance of 3.6$\sigma$ for two parameters
266: and 2.0$\sigma$ for the addition of $\dot x$. Lastly we tried varying all three
267: parameters (Fit 4). This fit had an F-test significance of 3.7$\sigma$ for the
268: addition of three parameters and 2.0$\sigma$ for just the addition of $\dot
269: P_{\rm orb}$.
270: 
271: \begin{deluxetable}{lllll}
272: \tablecaption{EXO 2030+375 Orbital Fits}
273: \tablewidth{0pt}
274: \tablehead{\colhead{Parameter} & 
275: \colhead{Fit 1} & \colhead{Fit 2} & \colhead{Fit 3} & \colhead{Fit 4}}
276: \startdata
277: $P_{\rm orb}$ (days) &  $46.0205 \pm 0.0002$ & $46.0213 \pm 0.0003$ & 
278:  $46.0211 \pm 0.0003$ & $46.0207 \pm 0.0004$ \\
279: $T_{\rm peri}$ &  $54044.73 \pm 0.01$ & $52756.17 \pm 0.01$
280:  & $52802.20 \pm 0.01$  & $53308.45 \pm 0.02$ \\ 
281: $x$ (lt-s) &  $244 \pm 2$ & $246 \pm 2$ & $248 \pm 2$ & $248 \pm 2$\\
282: $e$ &  $0.412 \pm 0.001$ & $0.410 \pm 0.001$ & $0.409 \pm 0.001$ & 
283:  $0.410 \pm 0.01$\\
284: $\omega$ (deg) &  $211.3 \pm 0.3$ & $211.9 \pm 0.4$ & 
285:  $212.0 \pm 0.4$ & $212.6 \pm 0.4$\\
286: $\dot P_{\rm orb}$ (days/day) & \nodata & \nodata & \nodata & 
287:  $(-4 \pm 2) \times 10^{-7}$\\
288: $\dot \omega$ (deg/year)  & \nodata & $0.18 \pm 0.05$ & $0.15 \pm 0.05$ & 
289:  $0.17 \pm 0.06$ \\
290: $\dot x$ (lt-s/year)  & \nodata & \nodata & $0.2 \pm 0.1$ & $0.2 \pm 0.1$\\
291: $\chi^2/$dof &  689.8/656 & 677.5/655 & 673.4/654 & 669.4/653 \\
292: \enddata
293: \label{tab:orb}
294: \end{deluxetable}
295: 
296: Figure~\ref{fig:postbatseres} shows the phase residuals for all phase
297: measurements after BATSE was de-orbited in May 2000. Phase residuals prior to
298: this date are shown in \citet{Wilson02}.
299: Figure~\ref{fig:postgiantres} zooms in to show the 2-60 keV rms pulsed flux 
300: overlaid with the frequency derivatives estimated from the orbit fitting and the phase
301: residuals. From this figure, we note that the spin-up rate and pulsed flux are
302: clearly correlated across all orbital phases. At the end of the giant outburst,
303: the pulsar spins down briefly. The minimum in the frequency derivative precedes
304: the pulsed flux minimum slightly. Although there are clearly defined
305: outbursts for each orbit, the pulsed flux never drops to zero between outbursts and we see
306: considerable flaring activity in the intra-outburst region. 
307: 
308: \begin{figure}
309: \plotone{f2.eps}
310: \caption{Phase residuals for all of the measurements after May 2000 included in
311: the fitting. Red filled circles denote \rxte measurements and blue open diamonds
312: denote \integral measurements. Phase residuals from BATSE and \rxte before 
313: May 2000 are shown in \citet{Wilson02}. The differences between this orbital 
314: fit and that one are too small to be visually apparent. \label{fig:postbatseres}}
315: \end{figure}
316: 
317: \begin{figure}
318: \plotone{f3.eps}
319: \caption{A close look at timing results after the giant outburst. {\it Top:} 
320: 2-60 keV rms pulsed flux measured with the \rxte PCA (left y-axis, black diamonds) 
321: overlaid with the estimated model frequency derivatives (right y-axis, red 
322: histogram). The red dashed line denotes a frequency derivative equal to zero. 
323: {\it Bottom:} Phase residuals after the giant outburst. \label{fig:postgiantres}}
324: \end{figure}
325: 
326: Using the orbit from Fit 3 in Table~\ref{tab:orb}, we fit the phase measurements
327: for the giant outburst with a new quadratic spline model in which each 5-day 
328: interval had an independent frequency derivative. 
329: Figure~\ref{fig:giantfreqfdot} shows the spin frequency, the spin-up rate, and
330: the 2-100 keV flux determined for the giant outburst. The giant outburst reached a peak
331: spin-up rate of $(1.815 \pm 0.006) \times 10^{-11}$ Hz s$^{-1}$ for the 5-day
332: interval 2006 August 1-6 (MJD 53948-53953). The 2-100 keV flux also peaked in
333: this interval at $(3.59 \pm 0.07) \times 10^{-8}$ erg cm$^{-2}$ s$^{-1}$.
334: 
335: \begin{figure}
336: \epsscale{0.8}
337: \plotone{f4.eps}
338: \caption{Measured spin-frequency, spin-up rate, and 2-100 keV flux for the 2006
339: giant outburst from \rxte 
340: data. Each frequency and spin-up rate point corresponds to the average 
341: barycentered, orbit corrected (using Fit 3), frequency and frequency derivative
342: for a 5-day interval of \rxte PCA observations. The bottom panel shows 3-day
343: averages of the total 2-100 keV flux measured with joint spectral fits to \rxte
344: PCA and HEXTE data.\label{fig:giantfreqfdot}}
345: \end{figure}
346: 
347: \subsection{X-ray Spectral Analysis}
348: Data analysis was performed using FTOOLS v6.1.2
349: \citep{Blackburn95}\footnote{\url{http://heasarc.gsfc.nasa.gov/ftools/}}.
350: For each \rxte PCA observation, we created a background spectrum using the
351: bright source model with the FTOOL {\it pcabackest} and extracted source and 
352: background spectra using the FTOOL {\it saextract}. In addition we corrected for
353: deadtime following the recipe
354: \footnote{\url{http://heasarc.gsfc.nasa.gov/docs/xte/recipes/cook\_book.html}}. 
355: A 0.5\% systematic error was included. 
356: For HEXTE cluster B data, detectors 0,1, and 3, we separated the data into 
357: source and background using the FTOOL {\it hxtback} and extracted source and 
358: background spectra using {\it seextrct}. HEXTE spectra were
359: corrected for deadtime using {\it hxtdead}. An analysis that includes HEXTE
360: cluster A data is described in \citet{Camero07}. The results were consistent
361: with ours. Unfortunately cluster A was not on-source during the early and late
362: parts of the giant outburst, so the data set is more limited. Six observations (obsids
363: 91089-01-01-04, 91089-01-02-10, 91890-01-03-10, 91089-01-04-02, 91089-01-12-01, 
364: 91089-01-15-04) were excluded from our analysis because they had less than 20 seconds
365: of HEXTE cluster B background exposure, i.e. HEXTE cluster B was not rocking. Only the last two of these observations were 
366: during the giant outburst.
367: 
368: From a joint fit to \rxte PCA and HEXTE data from August 15, 2006 (MJD 53962),
369: we found evidence for a cyclotron scattering feature near 10 keV
370: \citep{Wilson06}. The model
371: used was an absorbed power-law (PHABS*POWERLAW) with a high-energy cut-off
372: (HIGHECUT), an iron line (GAUSSIAN), and a
373: Gaussian cyclotron absorption line (GABS). The power-law had a photon index of
374: 1.53(2), with a cutoff energy at 12.4(4) keV, a folding energy of 27.4(5) keV.
375: The cyclotron energy was 10.1(2) keV, with a Gaussian width of 3.3(2) keV and a
376: peak depth of 1.1(1). This feature was significant at a 7.5 sigma level. Other 
377: continuum models, e.g., a Bremsstrahlung model,also showed evidence for this feature. 
378: 
379: %We then fit the PCA and HEXTE data from each observation with an absorbed 
380: %power-law with a Gaussian iron line and a high energy cutoff. We then added a 
381: %Gaussian absorption feature (GAUSSABS) to represent a cyclotron line. We 
382: %discovered a problem with the GABS model in XSPEC that affected the line depth,
383: %so we wrote our own local model using the parameterization of \citet{Coburn02}.
384: %Figure~\ref{fig:spec} shows the cyclotron line parameters, the f-test 
385: %probability of the cyclotron feature,and the 2-100 keV flux. A significant cyclotron feature was only detected during
386: %the giant outburst, at 2-100 keV fluxes larger than $10^{-8}$ erg cm$^{-2}$
387: %s$^{-1}$. Of the 102 available observations during the giant outburst, only
388: %eight are excluded from this plot. Six had less than 20 seconds of HEXTE
389: %background data and two near the end of the outburst had very large error bars
390: %for $E_{\rm cyc}$. 
391: Using Xspec v12.3, we first fit the PCA and HEXTE data from each observation with an absorbed
392: power-law with a high energy cutoff plus a Gaussian iron line
393: (PHABS (POWERLAW+GAUSSIAN) HIGHECUT). Based upon our previous detection of a
394: cyclotron feature at 10 keV \citep{Wilson06}, we included a Gaussian absorption
395: feature (GAUSSABS) to represent the cyclotron line. We discovered a problem with
396: the built-in XSPEC model GABS. This model calls the Gaussian line model and then
397: exponentiates it. However, the Gaussian line model returns the integral of the
398: Gaussian across the input bins rather than the value of the Gaussian, resulting in
399: GABS being incorrectly integrated. We wrote a local model GAUSSABS using
400: the parameterization of \citet{Coburn02}. Figure~\ref{fig:comparespec} shows the 
401: spectrum and residuals with and without the Gaussian absorption feature for an 
402: observation on 2006 August 7 (MJD 53954.5) near the peak of the outburst.
403: Dips are visible at 10 and 20 keV. Figure~\ref{fig:cycl} shows the
404: cyclotron feature parameters: the line energy $E$, the line width $\sigma$, and
405: the optical depth $\tau$ defined in Equations 6 and 7 in \citet{Coburn02}.
406: Dotted lines across each panel show the average value for that parameter 
407: ($\bar E = 11.44 \pm 0.02$ keV, $\bar \sigma = 3.08 \pm 0.02$ keV, and $\bar
408: \tau = 0.1263 \pm 0.0009$).
409: The second from the bottom panel in Figure~\ref{fig:cycl} shows the F-test probability
410: for including the cyclotron feature. The bottom panel shows the
411: 2-100 keV flux for each observation. Figure~\ref{fig:otherpar} shows the values
412: for the absorption $N_H$, the power-law photon index, the cutoff energy $E_{\rm
413: cut}$, and the folding energy $E_{\rm fold}$. The 2-100 keV flux is again shown
414: in the bottom panel for comparison. Lastly, Figure~\ref{fig:Fe} shows the iron
415: line parameters, the line energy, width, and normalization. In addition, we also
416: examined fits that included an additional Gaussian absorption feature at $E_{\rm
417: cut}$, the MPLCUT model of \citet{Coburn02}. An F-test showed that this
418: component was not significant. Lastly we tried adding a second Gaussian
419: absorption feature around 20 keV. We attempted two approaches, one where the
420: line energy was exactly twice that of the other feature and one where both line
421: energies were free to vary independently. F-tests showed that including a second
422: cyclotron feature (using either approach) did not significantly improve the fit.For a 
423: small number of spectra near the peak of the outburst, we also tried fitting a spectrum
424: including a bump near 15 keV after \citet{Klochkov07} instead of a Gaussian absorption
425: feature. For all six spectra we fit with this model, it was a significantly poorer
426: fit than the Gaussian absorption feature.
427: 
428: \begin{figure}
429: \epsscale{1.0}
430: \plotone{f5.eps}
431: \caption{Energy spectra from 2007 August 7, near the peak of the
432: outburst. In the top panel, the data were fit with a model consisting of an 
433: absorbed power law with a high energy cutoff with a Gaussian absorption feature
434: representing the cyclotron feature and a Gaussian iron line. The center panel 
435: shows the residuals in units of sigmas for a fit that excluded the cyclotron
436: feature. The bottom panel shows the residuals in units of sigmas with the
437: cyclotron feature included in the model. 
438: \label{fig:comparespec}}
439: \end{figure}
440: 
441: \begin{figure}
442: \epsscale{0.8}
443: \plotone{f6.eps}
444: \caption{From top to bottom: EX0 2030+375 Cyclotron line energy (keV), 
445: width (keV), and optical depth defined in Equations 6 and 7 in \citet{Coburn02}.
446: In the top panel, diamond symbols denote the line energy reported in 
447: \citet{Klochkov07}. Dotted lines show the average value for each parameter. The second from the
448: bottom panel shows the F-test value in sigmas for including the cyclotron
449: feature and the bottom panel shows the 2-100 keV flux.}
450: \label{fig:cycl}
451: \end{figure}
452: 
453: \begin{figure}
454: \plotone{f7.eps}
455: \caption{Additional model parameters, from top to bottom: the absorption $N_{\rm
456: H}$, power-law photon index, Cut-off energy $E_{\rm cut}$, e-folding energy
457: $E_{\rm fold}$, and the 2-100 keV flux.}
458: \label{fig:otherpar}
459: \end{figure}
460: 
461: \begin{figure}
462: \plotone{f8.eps}
463: \caption{Gaussian Iron line model parameters, from top to bottom: the line
464: energy (keV), line width (keV), the line normalization (photons cm$^{-2}$
465: s$^{-1}$), and the 2-100 keV Flux.}
466: \label{fig:Fe}
467: \end{figure}
468: 
469: 
470: 
471: %\subsection{Giant Outburst Details}
472: 
473: %\subsubsection{Timing}
474: 
475: %\begin{figure}
476: %\plotone{giant_outburst_freq.ps}
477: %\caption{Barycentered, orbit-corrected, spin frequencies measured with RXTE
478: %during the giant outburst.\label{fig:gifreq}}
479: %\end{figure}
480: 
481: %Spin frequencies were measured by fitting quadratic models to 5-day intervals
482: %of pulse phase measurements made using harmonic expansions of pulse profiles
483: %fit to RXTE PCA standard1 data. These frequencies have been corrected to the
484: %solar system barycenter and for EXO 2030+375's orbit. Periastron passages have
485: %occurred during this outburst on MJD53907, 53953, and 53999.
486: 
487: %The average spin-up rate during the 2006 giant outburst is $1.27 \times
488: %10^{-11}$ Hz/s. This is larger than the average rate of $1.1 \times 10^{-11}$ 
489: %Hz/s observed during the discovery outburst \citep{Wilson02}. However, the 
490: %discovery occurred at or near the peak of that outburst, so spin-up during the 
491: %outburst rise was not measured.
492: 
493: %\begin{figure}
494: %\plotone{avg_pulse_profiles.ps}
495: %\caption{EXO 2030+375 2-60 keV pulse profiles. Two cycles are shown for clarity.
496: %The profiles are phase aligned.\label{fig:giprof}}
497: %\end{figure}
498: 
499: %Pulse profiles were generated by fitting a 6th order Fourier expansion in pulse
500: %phase model to RXTE PCA Standard 1 data with times corrected to the solar
501: %system barycenter and for the pulsar's orbit. Phase offsets were computed by
502: %cross-correlation with a template profile. Then the total phases were divided
503: %in to 5-day intervals and fit with a quadratic spline model. The mean
504: %profiles from each 5-day interval are shown here. Time progresses from top to
505: %bottom, then left to right.
506: 
507: %At luminosities below about $10^{38}$ erg cm$^{-2}$ s$^{-1}$, the peak at
508: %phases 0.75-1.0 is quite prominent. As the luminosity increases, this peak
509: %becomes much less prominent. The behavior repeats as the luminosity again
510: %decreases. This is very similar to the pulse profile evolution observed during
511: %the fall of the 1985 discovery outburst \citep{Parmar89b}.
512: 
513: %\subsubsection{X-ray Spectroscopy}
514: 
515: 
516: %\begin{figure}
517: %\plotone{luminosities.ps}
518: %\caption{X-ray luminosity measured with RXTE PCA/HEXTE joint
519: %fits.\label{fig:gilum}}
520: %\end{figure}
521: 
522: %We measured the luminosity during the current giant outburst by fitting energy
523: %spectra from the RXTE PCA and HEXTE, using a model described above and assuming
524: %a distance of 7.1 kpc.
525: 
526: 
527: %\begin{figure}
528: %\epsscale{1.0}
529: %\plotone{fluxfdot_correlation.ps}
530: %\caption{2-100 keV flux versus the spin-up rate measured with RXTE
531: %PCA/HEXTE.\label{fig:fluxfdot}}
532: %\end{figure}
533: 
534: %Figure~\ref{fig:fluxfdot} shows the 2-100 keV flux measured using the RXTE PCA and HEXTE versus
535: %the spin-up rate. Red points denote the outburst rise and blue points denote
536: %the fall. A linear fit and a 6/7 power-law fit have been overlaid on the data. 
537: 
538: %Simple accretion theory predicts a correlation between spin-up and flux, if an
539: %accretion disk is present. For the EXOSAT data, we found a best fit power-law
540: %index of 1.17 (1), quite similar to what we see here.  
541: 
542: 
543: 
544: \section{Discussion}
545: \subsection{Evidence for a Cyclotron Feature}
546: The giant outburst of EXO 2030+375 began in early June 2006 and lasted
547: until the end of October 2006. From our spectral fitting, we consistently
548: detected a cyclotron feature at about 11 keV from 2006 June 22 (MJD 53908), when
549: pointed PCA and HEXTE observations began, until 2006 September 26 (MJD 54004).
550: From 2006 September 26 through 2006 October 2 (MJD 54010), we still detected the
551: feature in about half of the observations. After 2006 October 2, we did not
552: consistently detect the feature. This means that we consistently detected the
553: feature at luminosities greater than about $5 \times 10^{37}$ erg s$^{-1}$.
554: 
555: As a consistency check, we examine the expected
556: relationship between the magnetic field and the maximum spin up rate, $\dot \nu$, for EXO 2030+375.
557: \begin{eqnarray}
558: \dot \nu = 2.61 \times 10^{-11} {\rm Hz\ s}^{-1} k^{1/2} \alpha^{8/7} \beta^{6/7} 
559: \left( \frac{M_x}{1.4 M_\sun}\right)^{-3/7} I_{45}^{-1} R_6^{12/7} 
560: \left(\frac{d}{7.1\ {\rm kpc}}\right)^{12/7} \\
561: \left(\frac{E_{\rm cyc}}{11\ {\rm keV}}\right)^{8/7} \nonumber
562: \left(\frac{F_{\rm peak}}{3.5 \times 10^{-8}\ {\rm erg\ cm}^{-2}{\rm s}^{-1}}\right)^{6/7} \
563: \label{eqn:peaknudot}
564: \end{eqnarray}
565: where $k \simeq 0.5-1.0$ is a multiplier of the equation for the disk inner edge; $\alpha = (1.31 \times
566: 0.621) = 0.890$ accounts for the redshift of the cyclotron line $1+z \simeq 1.31$ and the general relativistic
567: correction for the dipole field, 0.621 \citep[See][]{Wasserman83}; $\beta = L/4\pi
568: d^2 F$ is the beaming factor,
569: estimated to be 0.7-1.3, based on the range seen by eye in the pulse profiles (A more detailed study of the
570: beaming factor based upon detailed modeling of pulse profiles is planned for a future paper.); $M_x$ is the 
571: neutron star mass; $I_{45}$ is the neutron star moment of inertia in units of
572: 10$^{45}$ g cm$^{2}$; $R_6$ is the
573: neutron star radius in units of $10^6$ cm; $d$ is the distance; $E_{\rm cyc}$ is the cyclotron feature 
574: energy; and $F_{\rm peak}$ is the peak flux. Our measured peak spin-up rate of $1.815 \times 10^{11}$ Hz 
575: s$^{-1}$ is consistent with Equation~\ref{eqn:peaknudot} for $k \simeq 0.5-0.9$. 
576: 
577: Examining Figure~\ref{fig:cycl}, we see that near the beginning of the outburst,
578: the cyclotron line energy was smaller for the first two observations, 10-11 keV,
579: then larger, about 13-14 keV. As the outburst brightened,
580: the line energy evolved to about 11 keV, where it remained stable for much of the
581: outburst. As the outburst began to fade, the cyclotron energy appeared to
582: decrease; however, these observations were shorter than those taken at the
583: beginning of the outburst, so the line parameters were less well-determined. The
584: line depth and width also showed some possible luminosity dependence. In 
585: Figure~\ref{fig:otherpar}, we see that the power-law photon index and the 
586: e-folding energy show evidence for a luminosity dependence. Lastly in 
587: Figure~\ref{fig:Fe}, only the Fe line normalization shows a luminosity dependence.
588: This is expected for a Fe line that is due to EXO 2030+375 and not a background
589: source.
590: 
591: \integral and \swift observations of EXO 2030+375 during the 2006 giant outburst
592: revealed evidence for two cyclotron features at 10 and 20 keV \citep{Klochkov07}.
593: The lower energy feature lies slightly below our \rxte results. We believe this
594: is related to problems with the GABS model in XSPEC discussed earlier. The 20 
595: keV feature is not detected with \rxteno. Other spectral parameters also differ between our \rxte measurements
596: and \integralno/\swiftno, likely due to inclusion of second line feature in the
597: \integralno/\swift spectrum and instrumental differences. 
598: 
599: During an outburst in 1998, the Be/X-ray binary XTE J1946+274 was found to have
600: cyclotron feature at about 35 keV \citep{Heindl01}. The cyclotron feature
601: appeared to have been consistently detected above $3\times 10^{-9}$ erg cm$^{-2}$
602: s$^{-1}$ (2-60 keV) corresponding to a luminosity of $(2-4) \times 10^{37}$
603: erg s$^{-1}$ for an assumed distance of 8-10 kpc \citep{Wilson03}, similar 
604: luminosities to where EXO 2030's feature was consistently detected. However, no 
605: evolution of the cyclotron feature with energy is seen in XTE J1946+274.
606: 
607: 4U0115+63 observations with \rxte of an outburst in 1999 showed two cyclotron
608: features at $\sim 11$ and $\sim 22$ keV for 3-50 keV luminosities of $(5-13)
609: \times 10^{37}$ erg s$^{-1}$, assuming a distance of 7 kpc. As the luminosity
610: decreased below $\sim 5 \times 10^{37}$ erg s$^{-1}$, the second resonance
611: disappeared and the fundamental resonance energy gradually increased, up to
612: $\sim 16$ keV at $0.16 \times 10^{37}$ erg s$^{-1}$ \citep{Nakajima06}. Similar
613: behavior was also observed in earlier outbursts. In 1990 February, two cyclotron
614: features were detected at 11.3 and 22.1 keV, when the luminosity was 
615: $1.4 \times 10^{38}$ erg s$^{-1}$. In 1991 March, when the luminosity was 
616: $2.0 \times 10^{37}$ erg s$^{-1}$, the fundamental cyclotron feature was 
617: detected at 15.6 keV \citep{Mihara04}. In both papers, the change in the
618: cyclotron energy is believed to be related to a decrease in the height of the
619: accretion shock in response to the reduction of the accretion rate; however,
620: the existing models only qualitatively describe the observations. Above $\sim 7
621: \times 10^{37}$ erg s$^{-1}$ the accretion shock height appears to saturate and
622: is no longer correlated with luminosity \citep{Nakajima06}. During the rise of
623: its 2006 giant outburst, EXO 2030+375's cyclotron energy appears to saturate 
624: above about $1.2 \times 10^{38}$ erg s$^{-1}$ (2-100 keV), assuming a distance 
625: of 7.1 kpc. 
626: \subsection{Long-Term Behavior}
627: 
628: \begin{figure}
629: \plotone{f9.eps}
630: \caption{Barycentered, orbit corrected, spin-frequency measurements for EXO 
631: 2030+375 measured with \exosatno, BATSE, \rxte PCA, and 
632: \integralno.}
633: \label{fig:ltfreq}
634: \end{figure}
635: 
636: 
637: 
638: 
639: \begin{figure}
640: \plotone{f10.eps}
641: \caption{Orbital phase (in days past perisastron) of EXO 2030+375 outburst peaks.
642: Normal outburst peaks measured with BATSE ({\em squares}) are taken from
643: \citet{Wilson02}. Normal ({\em open circles}) and giant ({\em star}) outburst peak
644: orbital phases were determined from Gaussian fits to \rxte ASM data. The giant
645: outburst peaked 1.9 orbits after the previous normal outburst peak.}
646: \label{fig:outbphase}
647: \end{figure}
648: 
649: \begin{figure}
650: \plotone{f11.eps}
651: \caption{A comparison between the giant outburst and normal outbursts before and
652: after it. The black line with asterisks denotes 4-day average \rxte ASM
653: measurements in the 5-12 keV band for the giant outburst and its precursor
654: normal outburst. Epoch-folded profiles, repeated 4.5 times, are shown for the
655: eight outbursts preceding the giant outburst (denoted green diamonds), four
656: outbursts immediately following the giant outburst (denoted by red triangles),
657: and next three outbursts following those (denoted by blue squares). Arrows
658: denote the approximate orbital phase of the initial rise of the giant outburst
659: and the approximate orbital phase of an abrupt jump in flux. Both occur at
660: orbital phase $\sim 0.75$.}
661: \label{fig:giantphase}
662: \end{figure}
663: 
664: 
665: Figure~\ref{fig:ltfreq} shows the long-term frequency history of EXO 2030+375
666: including measurements with {\em EXOSAT} \citep{Parmar89a}, BATSE, \integralno, and 
667: \rxte \citep{Wilson02,Wilson05}. Figure~\ref{fig:ltflx} shows the long-term flux 
668: history, including \exosat measurements from \citet{Parmar89a}. Each spike in the ASM history corresponds to a 
669: normal outburst. Typically, 1-4 points in the BATSE history \citep{Wilson02} correspond
670: to a normal outburst. The large peak is the 2006 giant outburst.
671: 
672: As of early September 2007, seven normal outbursts have
673: been observed with \rxte following the giant outburst. They continue to show 
674: considerable spin-up. During the first five of these outbursts, daily PCA
675: observations detected EXO 2030+375 pulsations throughout its orbit. For the last
676: two outbursts, PCA observations were taken only near the outburst peaks. 
677: Figure~\ref{fig:ltflx} shows that these outbursts are generally brighter than 
678: those just before the 2006 giant outburst. Further, the ASM observations 
679: continue to indicate that there is significant emission, even between the 
680: outbursts. Figure~\ref{fig:outbphase} shows the orbital phase of EXO 2030+375 
681: outbursts since 1991. The points for the BATSE outbursts are taken from 
682: \citet{Wilson02}. The \rxte ASM data were divided into 46-day intervals and each
683: interval was fit with a Gaussian to determine the outburst peak. Similarly, 
684: using a Gaussian fit to 120 days of ASM data, we found that the giant
685: outburst peaked at 2.5 days after periastron, 1.9 orbits after the previous
686: normal outburst. From 1991 to 1995, EXO 2030+375's outbursts consistently peaked
687: about 6-days after periastron. At some point between 1993 and 1995, a precessing
688: density perturbation developed in the Be star's disk. In late 1995, the density
689: perturbation interacted with the neutron star's orbit at a phase corresponding 
690: to 2.5 days before perisastron. The outburst peaks then quickly migrated in 
691: orbital phase until late 1997 when the density perturbation lost contact with 
692: the neutron star's orbit at a phase corresponding to 2.5 days after periastron 
693: \citep{Wilson02,Wilson05}. From 1997 until just before the giant outburst in 
694: 2006, the outburst orbital phases continued to slowly migrate. Just before the 
695: giant outburst in 2006, the outburst peaks had finally returned to approximately six days after periastron.
696: After the giant outburst, the first normal outburst peaked at about 14 days
697: after periastron as did the third and brightest of the normal outbursts. The remaining
698: five normal outbursts peaked about 11 days after periastron. It is interesting to note
699: that both sudden shifts in outburst phase are similar in magnitude, about 
700: 8-9 days, but opposite in sign. 
701: 
702: Figure~\ref{fig:giantphase} shows a comparison between the giant outburst and
703: the average profiles of normal outbursts before and after it. The giant outburst
704: and the small normal outburst that preceded it are denoted by black asterisks.
705: The normal outburst preceding the giant outburst was not particularly bright,
706: occurred at the same orbital phase as the average profile for the preceding
707: eight outbursts (green diamonds), but was unusual in that at about the point the
708: outburst should have started fading, the flux remained higher than expected and
709: then quickly rose into the giant outburst. Arrows on the plot denote orbital
710: phase 0.75, where the giant outburst appears to be starting to rise rapidly, and
711: where we see a rapid increase in flux one orbit later. The average profile from
712: the first four outbursts following the giant outburst (red triangles) shows that
713: those outbursts began later in orbital phase and had a higher intra-outburst 
714: flux than the pre-giant outbursts and the later post-giant outbursts (blue 
715: squares). From Figures~\ref{fig:giantphase} and \ref{fig:ltflx}, the 
716: intra-outburst flux is slowly fading. It appears to be a long-lived 
717: tail of the giant outburst. A long-lived tail of a giant outburst has not been 
718: observed before in Be/X-ray binaries.
719:  
720: From our orbital fits using 17 years of EXO 2030+375 data, we see marginal
721: evidence for apsidal precession. Our measurements are consistent with
722: predictions from simple calculations given below. In the supernova explosions 
723: that form Be X-ray binaries the neutron star will receive a kick out of the 
724: original orbital plane. If the orbit is misaligned with the companion's spin 
725: axis, precession of the orbital plane is expected. Using calculations described
726: in \citet{Lai95} we estimate the magnitude of the expected effect due to 
727: interactions between the companion's rotation period and the orbit. 
728: The apsidal motion and orbital plane precession change $\omega$ the observed
729: longitude of periastron and $i$ the orbital inclination angle. Using Equation~7
730: in \citet{Lai95} to calculate the apsidal rate $\dot \omega$ gives
731: \begin{eqnarray}
732: \dot \omega = 1.6 \times 10^{-3} 
733: \left( \frac{46.0208\ {\rm days}}{P_{\rm orb}} \right) 
734: \left( \frac{k}{0.01} \right)
735: \left( \frac{8.9 R_{\rm c} \sin i}{a_{\rm x}} \right)^2 \\
736: \times \left( \frac{\hat \Omega_{\rm s}}{0.5} \right) \nonumber 
737: \left( 1-\frac{3}{2} \sin^2 \theta \right) {\ \rm rad\ yr}^{-1}.
738: \end{eqnarray}
739: where $k \simeq 0.01$ is the apsidal motion constant for a $10 M_{\sun}$ 
740: main-sequence star \citep{Lai95}; $R_{\rm c} = 6 R_{\sun}$ is the assumed Be 
741: star radius; $\hat \Omega_{\rm s}$ is the dimensionless spin of the companion, 
742: assumed to be near break-up $\hat \Omega_{\rm s,max} = 0.5$; $a_{\rm x}$ is the
743: measured semi-major-axis; $P_{\rm orb}$ is the measured orbital period 
744: (see Table~\ref{tab:orb}); $i$ is the orbital inclination angle; and $\theta$ is
745: the angle between the orbital angular momentum and the spin angular momentum.
746: 
747: From our orbital fits, we measured apsidal rates ranging from 
748: $(2.6 \pm 0.9) \times 10^{-3}$ radians yr$^{-1}$ (Fit 3, Table~\ref{tab:orb}) 
749: to $(3.1 \pm 0.9) \times 10^{-3}$ radians yr$^{-1}$ (Fit 2, 
750: Table~\ref{tab:orb}). Our measurements are within a factor of about 2 of the 
751: simple calculations, suggesting that we are seeing apsidal motion in this 
752: system. The change in $i$ is given by Equation~8 in
753: \citet{Lai95} and has the same magnitude as $\dot \omega$
754: \begin{eqnarray}
755: \frac{{\rm d}i}{{\rm d}t} = 1.6 \times 10^{-3} 
756: \left( \frac{46.0208\ {\rm days}}{P_{\rm orb}} \right) 
757: \left( \frac{k}{0.01} \right)
758: \left( \frac{8.9 R_{\rm c} \sin i}{a_{\rm x}} \right)^2 \\
759: \times \left( \frac{\hat \Omega_{\rm s}}{0.5} \right) \nonumber 
760: \sin \theta \cos \theta \sin \Phi {\ \rm rad\ yr}^{-1}.
761: \end{eqnarray}
762: In Fits 2 and 3, we see a suggestion of a change in $i$, 
763: $di/dt = (0.8 \pm 0.4) \times 10^{-3} \sin^{-1} i$ radians yr$^{-1}$, also 
764: consistent with the calculation within a factor of about two. 
765: 
766: 
767:  
768: \subsection{Comparison with 1985 Giant outburst}
769: In the discovery outburst observed with EXOSAT in 1985, the maximum observed
770: luminosity in the 1-20 keV band was $2 \times 10^{38}$ erg cm$^{-2}$ s$^{-1}$
771: \citep{Parmar89a}. The maximum 2-20 keV luminosity observed with RXTE PCA was 
772: $1.44 \times 10^{38}$ erg cm$^{-2}$ s$^{-1}$, about 72\% of the EXOSAT value.
773: Both luminosities assume a distance of 7.1 kpc \citep{Wilson02}. 
774: The difference cannot be explained from absorption, suggesting that the 1985 outburst
775: was brighter. Further evidence for this comes from comparing the peak spin-up rates
776: from both outbursts. The peak spin-up rate measured in the 1985 outburst was $2.4
777: \times 10^{-11}$ Hz s$^{-1}$, while the peak spin-up rate measured in the 2006 outburst
778: was $(1.815 \pm 0.006) \times 10^{-11}$ Hz s$^{-1}$, also about 75\% of the EXOSAT
779: value. However, despite the fainter peak, the 2-20 flux minimum reached after 
780: the 2006 outburst was $(4.0 \pm 0.4) \times 10^{-10}$ erg cm$^{-2}$ s$^{-1}$ 
781: much brighter than the 1985 upper limit of $1.3 \times 10^{-11}$ erg cm$^{-2}$ 
782: s$^{-1}$ (1-20 keV). The 2006 flux minimum occurred about 90 days after the peak
783: and similarly the 1985 upper limit was measured 98 days after the initial and 
784: brightest detection. Presumably, the \exosat
785: detections of EXO 2030+375 from 1985 October 29 - November 3 \citep{Parmar89b}
786: corresponded to a normal outburst. Using our best ephemeris, we find that
787: these observations span orbital phases of 8.5 to 13.5 days past periastron, very
788: similar to the orbital phases where we see normal outbursts after the 2006 giant 
789: outburst. The first normal outburst after the 2006 giant outburst reached a peak 2-20
790: keV flux of about $6.6 \times 10^{-9}$ erg cm$^{-2}$ s$^{-1}$, approximately a factor
791: of two brighter than the non-flaring observation on 1985 November 3 and a factor of
792: about 3 fainter than the brightest flares on 1985 October 30-31 \citep{Parmar89b}.
793: From Figure~\ref{fig:postgiantres} it is clear that considerable day to day variability
794: was present especially in the first outburst after the 2006 giant outburst and between outbursts.
795: This variability may be related to the flaring seen on 1985 October 30-31. Both
796: flaring episodes occurred at similar orbital phases; however, our \rxte 
797: observations were too short to confirm if the same flaring behavior was 
798: occurring.
799: 
800: \section{Conclusions}
801: EXO 2030+375 has now been observed for more than 22 years. In this time, it underwent two giant
802: outbursts, in 1985 and 2006, and numerous normal outbursts. Weak evidence for evolution of the binary orbit is
803: presented in this paper. Perhaps by the next giant outburst this can be accurately determined. 
804: In 2006, we observed the onset of a giant outburst of EXO 2030+375 for the first time. A normal outburst, which peaked slightly 
805: later than the previous ones, preceded the 2006 giant outburst. Near orbital phase 0.5 (See Figure~\ref{fig:giantphase})
806: the flux from EXO 2030+375 began to increase instead of declining as it usually would after a normal outburst. The flux
807: and spin-up rate continued to increase for more than an entire orbit. Near orbital phase 0.75, the flux abruptly 
808: increased, accompanied by an abrupt increase in the spin-up rate. This was followed by a relatively flat-topped 
809: maximum during which the flux remained within 10\% of the maximum for about 25 days, from orbital phase --0.2 to 0.35.
810: After that, the flux declined to a minimum more than 1.5 orbits later. The entire giant outburst spanned more than 
811: three pulsar orbits.
812: 
813: During the 2006 giant outburst, we discovered evidence for a cyclotron feature with a mean value of
814: $11.44\pm0.02$ keV that was consistently detected for about 90 days, at 2-100 keV luminosities above 
815: $5\times 10^{37}$ erg s$^{-1}$. This feature translates into a magnetic field strength of $B = 9.9 \times
816: 10^{11} (1+z)$ G $\simeq 1.3 \times 10^{12}$ G. Predictions of the peak spin-up rate during the giant
817: outburst, using this magnetic field strength, are consistent with the measured peak spin-up rate during the
818: 2006 giant outburst. \citet{Klochkov07} proposed that the EXO 2030+375 spectrum
819: measured with \integral and \swift could also be explained by a model containing a bump at 15 keV. However, our
820: \rxte observations do not appear to support that model. The cyclotron features for EXO 2030+375 and 
821: 4U 0115+63 \citep{Nakajima06} have the lowest energies measured to date. Now we are left with the question: 
822: Is the lack of detections of $E_{\rm cyc} < 10$ keV a physical or observational bias ?
823: 
824: The two classes, normal (type I) and giant (type II), of Be/X-ray pulsar outbursts are clearly distinguishable
825: in the observations of EXO 2030+375. However, the evolution seen in the phasing, duration, and peak flux of
826: the normal outbursts, while apparently related to the Be star's disk \citep{Wilson02}, is poorly understood.
827: The lack of gaps in detection between the normal outbursts after the 2006 giant outburst implies
828: that the accretion disk present in the giant outburst persists throughout these normal outbursts.
829: Many have believed that the giant outbursts and normal outbursts were distinguished by the presence of an
830: accretion disk in the giant outburst, while the normal outbursts proceeded by wind accretion. Possibly the
831: normal outbursts following a giant outburst \citep[as in the ``mother duck/baby duck" complexes seen with
832: BATSE in][]{Bildsten97} are a distinct class of normal outbursts, with the remainder proceeding by wind accretion.
833: However, XTE J1946+274 \citep{Wilson03} had a series of normal outbursts that also had no detection gaps
834: between them, but these outbursts were not preceded by a giant outburst. \citet{Ikhsanov01} show for longer 
835: period sources, e.g. A0535+26, disk accretion is required for the normal
836: outbursts to occur. \citet{Okazaki} conclude from simulations that a transient disk forms in normal outbursts
837: in shorter period systems, e.g. 4U0115+63. Measurements of significant spin-up during normal outbursts of EXO
838: 2030+375 prior to the 2006 giant outburst \citep{Wilson02, Wilson05} suggested that at least a transient 
839: accretion disk was also present in those outbursts.  Further, \citet{Hayasaki06} conclude from simulations that accretion disks
840: in Be/X-ray binaries evolve through three phases: a `developing phase' where the mass accretion rate is double-peaked, but 
841: dominated by direct accretion at periastron; a 'transition phase' where the mass accretion rate evolves from double-peaked to
842: single peaked as the approximately Keplerian disk grows with time; and finally a quasi-steady state where the mass accretion rate
843: has a single peak induced by a one-armed spiral wave and is on average balanced with the mass-transfer rate from the Be disk.
844: \citet{Camero05} reported multiple detections of an initial spike preceding the main peak in outbursts prior to the giant
845: outburst. This initial spike was usually smaller than the main outburst peak, suggesting that EXO 2030+375 was in the `transition
846: phase.' After the giant outburst, the initial spike preceding the main peak has disappeared from the normal outbursts, suggesting
847: that the accretion disk in EXO 2030+375 has reached the final quasi-steady state. Therefore, the long-term behavior seen in the 
848: normal outbursts of EXO 2030+375 appears to be product of the state of the Be disk and the accretion disk, allowing for a wide 
849: range of variations. 
850: 
851: Giant outbursts from EXO 2030+375 also vary in brightness and in orbital phase. The 2006 giant outburst peaked at 
852: a 2-100 keV luminosity of $2.2 \times 10^{38}$ erg s$^{-1}$. In the 2006 giant outburst, the 2-20 keV luminosity and 
853: the peak spin-up rate were about 72\% and 75\%, respectively, of that measured in the the 1985 outburst, suggesting 
854: that the 1985 outburst had a 2-100 keV peak luminosity of $(2.9-3.0) \times 10^{38}$ erg s$^{-1}$, which was 
855: super-Eddington unless the neutron star mass was greater than about $2.2 M_{\sun}$.  The 1985 outburst was shifted by 
856: 0.65 in orbital phase relative to the decline of the 2006 outburst. Interestingly, the decay constants of both outbursts
857: were very similar. Giant outbursts in EXO 2030+375 do not appear to be locked in orbital phase, unless the onsets are.
858: The 2006 minimum spanned orbital phase --0.2 to --0.1, very similar to the orbital phase of the 23 August 1985 
859: minimum; however, the 2006 minimum was more than 30 times brighter than in 1985. Despite the differences between the 
860: giant outbursts, the normal outburst following the 1985 giant outburst lined up quite well in orbital phase with the 
861: second normal outburst following the 2006 giant outburst. Both normal outbursts also showed flaring activity.
862: 
863: 
864: % with the second about 72\% as bright as the observed maximum of the 1985
865: %outburst. During the 2006 giant outburst, we discovered evidence for a cyclotron feature at about 11
866: %keV that was consistently detected for about 90 days, at luminosities above $5\times 10^{37}$ erg s$^{-1}$. Between these giant outbursts, normal outbursts were observed at nearly every periastron 
867: %passage from 1991 up until the giant outburst began in 2006 June. The global trend was spin-up from
868: %1991 to 1993, then spin-down with fainter normal until 2003, when the pulsar began to spin-up again
869: %and the outbursts brightened and then a giant outburst occurred. From 1991 until 1995, the normal 
870: %outbursts consistently peaked at 6 days after periastron. In 1995, due to an apparent density 
871: %perturbation, the outburst peaks abruptly shifted to 2.5 days before periastron and then gradually 
872: %migrated in phase until they were back to peaking at approximately 6 days before periastron just 
873: %before the giant outburst. After the giant outburst, the normal outbursts changed. They were 
874: %brighter and had more spin-up than before. They peaked 11-14 days after periastron, a significantly
875: %later orbital phase than just before periastron. Since 1991, EXO 2030+375 has had an usually long
876: %series of normal outbursts for a Be/X-ray binary, with an outburst detected at nearly every periastron
877: %passage for 15 years. After the giant outburst, EXO 2030+375, has become a persistent system,
878: %remaining detectable throughout its orbit for at least seven orbits after the giant outburst. This
879: %suggests that a major reconfiguration of the Be star's disk has occurred so that material is
880: %available throughout the neutron star's orbit.
881: 
882: 
883: \acknowledgements
884: This research has made use of data obtained from the High Energy Astrophysics
885: Science Archive Research Center (HEASARC), provided by NASA's Goddard Space
886: Flight Center (GSFC). \rxte ASM quick-look results were provided by the \rxte
887: ASM teams at MIT and at the GSFC SOF and GOF. We thank Morgan Dwyer, a summer
888: intern from Yale who performed the early timing analyses as the observations originally
889: came in. We also thank Evan Smith and Jean Swank for their help scheduling the
890: daily \rxte observations of the giant outburst and of the normal outbursts that
891: followed.
892: 
893: \begin{thebibliography}{}
894: \bibitem[Arnaud (1996)]{Arnaud96} 
895: Arnaud, K.A. 1996, Astronomical Data Analysis Software and Systems V, eds.
896: G. Jacoby and J. Barnes, ASP Conf. Series, Vol 101, 17
897: \bibitem[Bildsten et al.(1997)]{Bildsten97}
898: Bildsten, L. et al. 1997, ApJS, 113, 367
899: \bibitem[Blackburn (1995)]{Blackburn95}
900: Blackburn, J.K. 1995, in ASP Conf. Ser., Vol. 77, Astronomical Data Analysis
901: Software and Systems IV, ed. R.A. Shaw, H.E. Payne, and J.J.E. Haynes (San
902: Francisco: ASP), 367
903: \bibitem[Camero-Arranz et al.(2005)]{Camero05}
904: Camero-Arranz, A. et al. 2005, A\&A, 441, 261
905: \bibitem[Camero-Arranz(2007)]{Camero07}
906: Camero-Arranz, A. 2007, Ph.D. Thesis, Universidad de Valencia
907: \bibitem[Coburn et al.(2002)]{Coburn02}
908: Coburn, W. et al. 2002, ApJ, 580, 394
909: \bibitem[Coe et al.(1988)]{Coe88}
910: Coe, M.J. et al. 1988, MNRAS, 232, 865.
911: \bibitem[Corbet (1986)]{Corbet86} 
912: Corbet, R.H.D. 1986, MNRAS, 220, 1047
913: \bibitem[Corbet \& Levine (2006)]{Corbet06}
914: Corbet, R.H.D. \& Levine, A.M. 2006, Atel \#843
915: \bibitem[Hanuschik (1996)]{Hanuschik96}
916: Hanuschik, R.W. 1996, A\&A, 308, 170
917: \bibitem[Hayasaki \& Okazaki (2004)]{Okazaki}
918: Hayasaki, K. \& Okazaki 2004, MNRAS, 350, 971
919: \bibitem[Hayasaki \& Okazaki (2006)]{Hayasaki06}
920: Hayasaki, K. \& Okazaki 2006, MNRAS, 372, 1140
921: \bibitem[Heindl et al.(2001)]{Heindl01}
922: Heindl, W.A. et al.2001, ApJ, 563,L35
923: \bibitem[Ikhsanov (2001)]{Ikhsanov01}
924: Ikhsanov, N.R. 2001, A\&A, 372, 227
925: \bibitem[Jahoda et al.(2006)]{Jahoda06} Jahoda, K., Markwardt, 
926: C.~B., Radeva, Y., Rots, A.~H., Stark, M.~J., Swank, J.~H., Strohmayer, 
927: T.~E., \& Zhang, W.\ 2006, \apjs, 163, 401 
928: \bibitem[Janot-Pacheco, Motch, \& Pakull(1988)]{Janot88}
929: Janot-Pacheco, E., Motch, C., \& Pakull, M.W. 1988, A\&A, 202, 81.
930: \bibitem[Klochkov et al.(2007)]{Klochkov07}
931: Klochkov, D. et al. 2007, A\&A, in press, astro-ph/0701791
932: \bibitem[Krimm et al.(2006)]{Krimm06}
933: Krimm, H. et al. 2006, Atel \# 861
934: \bibitem[Lai, Bildsten, \& Kaspi (1995)]{Lai95}
935: Lai, D., Bildsten, L., \& Kaspi, V.M. 1995, ApJ, 452, 819
936: \bibitem[Levine et al.(1996)]{Levine96} Levine, A.~M., Bradt, 
937: H., Cui, W., Jernigan, J.~G., Morgan, E.~H., Remillard, R., Shirey, R.~E., 
938: \& Smith, D.~A.\ 1996, \apjl, 469, L33
939: \bibitem[McCollough et al.(2006)]{McCollough06}
940: McCollough, M.L. et al. 2006, Atel \# 868
941: \bibitem[Mihara, Makishima, \& Nagase (2004)]{Mihara04}
942: Mihara, T., Makishima, K., \& Nagase, F. 2004, ApJ, 610, 390
943: \bibitem[Miroshnichenko et al.(2001)]{Mir01}
944: Miroshnichenko, A.S. et al. 2001, \aap, 377, 485
945: \bibitem[Motch \& Janot-Pacheco (1987)]{Motch87}
946: Motch, C. \& Janot-Pacheco 1987, A\&A, 182, L55.
947: \bibitem[Nakajima et al.(2006)]{Nakajima06}
948: Nakajima, M., Mihara, T., Makishima, K., \& Niko, H. 2006, ApJ, 646, 1125
949: \bibitem[Negueruela et al.(2001)]{Neg01a}
950: Negueruela, I. et al. 2001, \aap, 369, 117
951: \bibitem[Okazaki \& Negueruela(2001)]{Okazaki01}
952: Okazaki, A.T. \& Negueruela, I. 2001, \aap, 377, 161
953: \bibitem[Pamar et al. (1989a)]{Parmar89a}
954: Parmar, A.N. et al. 1989, ApJ, 338, 358
955: \bibitem[Parmar, White, \& Stella (1989)]{Parmar89b}
956: Parmar, A.N., White, N.E., Stella, L, 1989, 338, 373
957: \bibitem[Porter \& Rivinus (2003)]{Porter03}
958: Porter, J.M., Rivinus, T. 2003, PASP 115, 1153
959: \bibitem[Porter (1996)]{Porter96}
960: Porter, J.M. 1996, MNRAS, 280, L31
961: \bibitem[Quirrenbach et al.(1997)]{Quirrenbach97}
962: Quirrenbach, A. et al. 1997, ApJ479, 477
963: \bibitem[Reig \& Coe (1999)]{Reig99}
964: Reig, P. \& Coe, M.J. 1999, MNRAS, 302, 700
965: \bibitem[Rothschild et al.(1998)]{Rothschild98} 
966: Rothschild, R.~E., et al. 1998, \apj, 496, 538
967: \bibitem[Stella, White, \& Rosner (1986)]{Stella86}
968: Stella, L., White, N.E. \& Rosner, R. 1986, ApJ, 308, 669
969: \bibitem[Terada et al.(2006)]{Terada06}
970: Terada, Y. et al. 2006, ApJ, 648, L139
971: \bibitem[Ubertini et al.(2003)]{Ubertini03}
972: Ubertini, P. et al. (2003), A\&A, 411, 131
973: \bibitem[Wasserman \& Shapiro (1983)]{Wasserman83}
974: Wasserman, I. \& Shapiro, S.L. 1983, ApJ, 265, 1036
975: \bibitem[Waters \& van Kerkwijk (1989)]{Waters89} 
976: Waters, L.B.F.M. \& van Kerkwijk, M.H. 1989, A\&A, 223, 196
977: \bibitem[Wilson et al.(2002)]{Wilson02}
978: Wilson, C.A. et al. 2002, ApJ, 570, 287.
979: \bibitem[Wilson et al.(2003)]{Wilson03}
980: Wilson, C.A., Finger, M.H., Coe, M.J., Negueruela, I. 2003, ApJ, 584, 996
981: \bibitem[Wilson, Fabregat, \& Coburn (2005)]{Wilson05}
982: Wilson, C.A., Fabregat, J. \& Coburn, W.2005, ApJ, 620, L99
983: \bibitem[Wilson \& Finger (2006)]{Wilson06}
984: Wilson, C.A. \& Finger, M.H. 2006, Atel\#877.
985: \end{thebibliography}
986: 
987: \end{document}
988: