0709.1095/ms.tex
1: \documentclass[12pt, preprint]{aastex}
2: \usepackage {graphicx}
3: \def\deg{$^{\circ}$}
4: \def\um{${\rm \mu m}$}
5: 
6: \begin{document}
7: 
8: \title{HST/NICMOS Observations of Fast Infrared Flickering in the
9: Microquasar GRS 1915+105}
10: 
11: \author{Stephen S. Eikenberry\altaffilmark{1}, Shannon
12: G. Patel\altaffilmark{1}, David M. Rothstein\altaffilmark{2}, Ronald
13: Remillard\altaffilmark{3}, Guy G. Pooley\altaffilmark{4}, Edward
14: H. Morgan\altaffilmark{3}}
15: 
16: \altaffiltext{1}{Department of Astronomy, University of Florida, Gainesville, FL  32611}
17: \altaffiltext{2}{Department of Astronomy, Cornell University, Ithaca, NY 14853}
18: \altaffiltext{3}{Department of Physics, Massachusetts Institute of Technology, Cambridge, MA  02138}
19: \altaffiltext{4}{Cavendish Laboratory, Cambridge University}
20: 
21: 
22: \begin{abstract}
23:   
24:   We report infrared observations of the microquasar GRS 1915+105 using the
25:   NICMOS instrument of the Hubble Space Telescope during 9 visits in April-June
26:   2003.  During epochs of high X-ray/radio activity near the beginning and end
27:   of this period, we find that the $1.87 $\um  infrared flux is generally low
28:   ($\sim 2$ mJy) and relatively steady.  However, during the X-ray/radio
29:   ``plateau'' state between these epochs, we find that the infrared flux is
30:   significantly higher ($\sim 4-6$ mJy), and strongly variable.  In
31:   particular, we find events with amplitudes $\sim 20-30$\% occurring on
32:   timescales of $\sim 10-20$s (e-folding timescales of $\sim 30$s).  These
33:   flickering timescales are several times faster than any previously-observed
34:   infrared variability in GRS 1915+105 and the IR variations exceed
35:   corresponding X-ray variations at the same ($\sim 8s$) timescale. These
36:   results suggest an entirely new type of infrared variability from this
37:   object.  Based on the properties of this flickering, we conclude that it
38:   arises in the plateau-state jet outflow itself, at a distance $<2.5$ AU from
39:   the accretion disk.  We discuss the implications of this work and the
40:   potential of further flickering observations for understanding jet formation
41:   around black holes.
42: 
43: \end{abstract}
44:  
45: \keywords{ accretion, accretion disks --- black hole physics --- infrared: stars --- stars: individual (GRS 1915+105) --- X-rays: binaries }
46: 
47: \section{Introduction}
48: 
49: As the archetypal microquasar, GRS 1915+105 offers great promise as the
50: potential key for understanding the formation of relativistic jets in black
51: hole systems.  GRS 1915+105 was first discovered as a hard X-ray transient
52: \citep{alberto}, and then later seen as the first-known superluminal jet
53: source in the Galaxy \citep{mr94}.  Further study revealed a rich variety of
54: X-ray, infrared (IR), and radio variability from this system (see \citet{MRaa}
55: and references therein).  Among these, \citet{fp97} discovered radio and
56: infrared flaring on the same day (in disjoint observations), each displaying a
57: $\sim 30$-minute quasi-periodicity.  Later simultaneous observations in the
58: X-ray/IR and X-ray/IR/radio showed that these events tied together activity in
59: the inner accretion disk (seen in the X-ray) with synchrotron jet ejections
60: (seen in the IR and radio) (\citet{eiken98a}; \citet{felix98}; \citet{fp99}).
61: Thus, GRS 1915+105 seemed to be living up to its potential as a prototype
62: source for investigations of the disk-jet connection in black hole binaries.
63: 
64: Since then, continued study has revealed that the disk-jet connection in GRS
65: 1915+105 takes multiple, apparently distinct, forms.  \citet{eiken00} describe
66: three types of long-wavelelength variability: large-amplitude superluminal
67: radio events (Class A); the $\sim 30$-minute IR/radio oscillations noted above
68: associated with hard X-ray dips (Class B); other IR/radio flares associated
69: with individual soft X-ray oscillations (\citet{eiken00}; \citet{rothstein04})
70: (Class C).  The Class B and C events share similar radio properties (e.g.
71: \citet{kleinwolt02}), but the Class B IR flares are significantly brighter
72: ($\sim 200-300$ mJy) than their class C counterparts ($\sim 5-10$ mJy), and
73: the X-ray properties of these differ (\citet{rothstein04}; \citet{mikles}).
74: Both of the observed IR event types have e-folding rise times of $\sim
75: 200-300$s.
76: 
77: To date, no IR counterparts to Class A events have been reported, though there
78: is some evidence for extended IR jet emission (\citet{sams}; \citet{eiken96}).
79: This has been due in part to the transient nature of these events, with
80: occurrence rates of only one every few years since 1994.  However, detailed
81: study has revealed an apparent pattern in these events which could allow the
82: prediction of Class A events with lead times of a few weeks.  Occasionally,
83: GRS 1915+105 enters a ``plateau state'' (\citet{Foster96}; \citet{Fender99}),
84: where the X-ray count rates are relatively low, dominated by a hard power-law
85: component, and with strong quasi-periodic oscillations at frequencies $\sim
86: 1-10$ Hz.  The radio emission from GRS1915+105 is generally elevated and
87: quasi-steady ($\sim 50-200$ mJy).  It seems that the beginning and end of these
88: plateaux are marked by large-amplitude radio flares \citep{kleinwolt02}, and
89: that just before they post-plateau flare the X-ray hardness ratio begins
90: curving towards softer values over $\sim 10-20$ days.  By identifying plateaux
91: via X-ray and radio monitoring observations, we can thus hope to schedule
92: ground- and space-based telescope resources catch the IR behavior of GRS
93: 1915+105 during these events.  With this goal in mind, we proposed Target of
94: Opportunity (TOO) observations of GRS 1915+105 using the Near-Infrared Camera
95: and Multi-Object Spectrograph (NICMOS) instrument \citep{NICMOS} on the Hubble
96: Space Telescope (HST), triggered by X-ray/radio observations of a plateau
97: state nearing its end, as observed by the All-Sky Monitor (ASM) of the Rossi
98: X-ray Timing Explorer (RXTE) and the Ryle Telescope.
99: 
100: In the sections below, we first describe the observations taken as
101: part of this program, and the data reduction techniques used to
102: analyze them.  We next describe the results of these analyses,
103: including the identification of fast IR flickering from GRS 1915+105
104: during the plateau state.  We then move on to discussion of the
105: implications of these observations for undertsanding the formation of
106: jets in GRS 1915+105.  Finally, we present the conclusions based on
107: these results.
108: 
109: \section{Observations and Data Analysis}
110: 
111: \subsection{Infrared Observations}
112: 
113: The NICMOS observations we report here were taken as TOO data with
114: HST.  We triggered the TOO on MJD 52732 (3 April 2003), based on the
115: combination of a ``plateau'' state in GRS 1915+105 lasting $> 10 $
116: days, and the beginning of a curving drop in the HR2 hardness ratio in
117: data from the RXTE ASM.  We then observed GRS 1915+105 using NICMOS
118: over the span of nine visits from 2003 April 7 to 2003 June 24 (see
119: Table 1).  For each visit, we used the F187W filter ($1.75-2.0 $\um)
120: with the NIC2 detector ($256 \times 256$-pixel HgCdTe array, $
121: 19.2\arcsec\ \times 19.2\arcsec$ field of view) for imaging
122: observations and the F240M ($2.3-2.5 $\um) and G206 ($1.4-2.5 $\um)
123: direct-grism filter pair with NIC3 ($256 \times 256$-pixel HgCdTe
124: array) for spectroscopic observations.  We also made polarimetric
125: observations with the long-wavelength polarizers ($1.9-2.1 $\um) and
126: NIC2.  All exposures were set to MULTIACCUM mode, with multiple
127: non-destructive reads allowing us to analyze the behavior of the
128: source on time scales shorter than the total exposure times.  For each
129: of the first three visits, we obtained a set of 5 dithered 72-s images
130: using the F187W filter, and a set of 12 dithered 56-s polarized images (4
131: in each of 3 polarization angles: 0\deg, 120\deg, 240\deg).
132: 
133: We processed data taken with the F187W and polarization (POL) filters
134: through the standard NICMOS calibration pipeline, CALNICA.  We also
135: processed the grism data with CALNICA, with the exception of the
136: flatfield correction, which was determined upon extraction of
137: individual spectra through CALNICC.  We further processed and combined
138: all dithered images with CALNICB.
139: 
140: For each F187W MULTIACCUM frame, we carried out photometry for GRS 1915+105
141: and for a nearby comparison star (Star A -- see e.g. \citep{eiken98a}) in the
142: same field of view.  We defined an aperture with a 6 pixel radius
143: ($\sim$0.5\arcsec) around each star, which just encompassed the first
144: diffraction ring, and subtracted the background flux from an annulus with
145: inner radius of 15 pixels ($\sim$1.1\arcsec) and outer radius of 18 pixels
146: ($\sim$1.4\arcsec) to calculate the flux.  We found typical background count
147: rates for the F187W filter of $\sim 0.15\ {\rm DN/s}$.  We replaced any
148: isolated bad pixels within the photometric aperture with the median of the
149: adjacent pixels.  We omitted results for a star in a given readout if groups
150: of bad pixels were located within its aperture.  We calculated the uncertainty
151: in background-subtracted flux using the associated error array for each
152: exposure, and then multiplied by the PHOTFNU keyword in the header to convert
153: to units of flux density.  We calculated values of flux density for MULTIACCUM
154: exposures at NICMOS STEP8 time intervals by first extracting individual
155: readouts from the IMA FITS file extensions, and then determining the count
156: rates during a given interval by finding the difference in total counts
157: between readouts and dividing by the time interval.
158: 
159: We also included the polarimetry data in our photometric results using the
160: same calculation procedure as above for the F187W data.  (Due to
161: highly-variable IR emission from GRS 1915+105 during this epoch (see below)
162: and the sequential procedure of the polarimetry measurements with NICMOS, we
163: could not extract meaningful polarimetric measurements for the core source.)
164: We found background count rates for the POL filters of $\sim 0.23 \ {\rm
165:   DN/s}$.  While differential polarization may affect absolute flux
166: comparisons between GRS 1915+105 and Star A, it is only on the order of a few
167: percent or less, and should not significantly impact measurements of
168: time-variability in flux within a given filter/polarization.
169: 
170: \subsection{Radio Observations}
171: 
172: The Ryle Telescope has been used to monitor a small number of variable and
173: transient sources, including GRS 1915+105, at 15 GHz (during gaps in its primary
174: CMB-related program).  The observational details are as described by
175: \citet{pf97}; typical runs are for an hour or so, with longer times for some
176: coordinated programs.  The data points shown in Fig 1 are 5-min averages, on
177: which the noise level is about 2 mJy rms, and the overall flux-density scale
178: has an uncertainty of about 4\%.
179: 
180: 
181: \subsection{X-Ray Observations}
182: 
183: For X-ray observations of GRS 1915+105, we used both the All-Sky Monitor (ASM)
184: and the Proportional Counter Array (PCA) instruments of the {\it Rossi X-ray
185:   Timing Explorer} (RXTE).  The ASM uses scanning masked proportional counters
186: to provide ``snapshot'' observations of X-ray sources distributed across the
187: sky, with typical sampling frequencies of several measurements per day (for
188: details on the ASM, see \citet{lev96}).  For monitoring the X-ray behavior of
189: GRS 1915+105 during the IR observations, we used the dwell-by-dwell count rate
190: and hardness ratio (HR2) downloaded from the ASM web site.
191: 
192: The PCA instrument on RXTE uses up to 5 proportional counter units (PCU) with
193: large collecting area for detailed pointed observations (for details on the
194: PCA, see \citet{jah2006}) .  For specific pointed observations, we analyzed
195: the light curves, energy spectra, and power density spectra of GRS 1915+105.
196: We used the FTOOLS package for RXTE to generate light curves with time
197: resolutions of 1-second (using the Standard1 data mode) and 0.020-seconds
198: (using the binned 8ms data mode).  For spectral analyses, we used the XSPEC
199: package together with FTOOLS v5.2 to analyze Standard2 data (16-second time
200: resolution and 129 energy channels over the 2-60 keV energy range).  We used
201: standard procedures for response matrix generation, background estimation and
202: subtraction, and correction for PCA deadtime.  We then used XSPEC v11.2 to fit
203: the 2.9-25 keV spectrum with a standard model for black hole candidates
204: consisting of a ``soft'' component (which peaks in the low-energy X-rays''
205: modeled as a multi-temperature disk blackbody (e.g. \citet{mitsuda}) and a
206: ``hard component'' (which extends to the higher-energy X-rays) modeled as a
207: power law, both modified by interstellar absorption corresponding to a fixed
208: hydrogen column density of $N_H = 6 \times 10^{22} {\rm cm^{-2}}$
209: \citep{muno99}.  We added a systematic error of 1\% to the spectrum before
210: performing the fit.
211: 
212: \section{Results}
213: 
214: \subsection{The Multi-Wavelength Context}
215: 
216: Figure 1 shows the multi-wavelength behavior of GRS 1915+105 during
217: the period of the Spring 2003 HST observations.  The first $\sim 25$
218: days of these lightcurves show GRS 1915+105 to be in a typical
219: ``plateau'' state, with low, steady, hard X-ray emission and elevated
220: radio flux.  Near MJD 52735, the X-rays reveal a sharp outburst
221: accompanied by radio variability -- the curvature in X-ray hardness
222: ratio leading up to this caused us to trigger the HST TOO as described
223: above, in anticipation if a plateau-ending outburst.  However, rather
224: than showing a very large outburst signalling the end of the plateau,
225: GRS 1915+105 showed a rather modest outburst and then {\it returned}
226: to the plateau state.  It then continued on in this state until near
227: MJD 52785.  Then, a second outburst led to the actual termination of
228: the plateau state.
229: 
230: In Table 1, we present the average IR flux densities for each visit, star and
231: filter type (F187W and POL).  Note that the nine data points in Figure 1c
232: indicate the mean flux density value for each visit, while the bars represent
233: the {\it range} of flux density values rather than uncertainty.  The light
234: curve in Figure 2 displays the range of IR flux density for GRS 1915+105 and
235: also presents the mean flux density of the comparison star for each visit.  In
236: Figure 2, the error bars represent the standard deviation in the measured flux
237: of the comparison Star A over the nine visits, which also gives a
238: reasonable estimate of the uncertainty in flux density for GRS 1915+105.
239: These errors are significantly larger than those derived from the CALNICA
240: pipeline error arrays, indicating some additional uncertainty in the
241: photometry.  However, when we use the CALNICA error estimates to fit a
242: constant flux density to Star A {\it within} a single visit, we obtain reduced
243: chi-squared values of $\sim 1$, indicating that the CALNICA error estimates
244: are valid within a single visit.  We attribute the additional between-visits
245: variations in the Star A photometry to possible PSF variations from the
246: telescope and/or NICMOS instrument.  We can see from Figure 2 that GRS
247: 1915+105 is higher in both flux and fractional variability in Visits 2-4, as
248: compared to Visit 1 and Visits 6-9.  Interestingly, these visits (2-4)
249: correspond to the time period {\it between} the major X-ray/radio variability
250: episodes described above.  When the X-ray flux is high/variable, the IR flux
251: density seems suppressed and relatively steady.
252: 
253: \subsection{Fast IR Flickering}
254: 
255: To further illustrate the intra-visit IR variability from GRS 1915+105, we
256: present selected segments of the light curves in Visits 2-4 in Figure 3, along
257: with the accompanying photometry of Star A for comparison.  At times, GRS
258: 1915+105 exhibits significant ($>20 \%$ amplitude) variability from one
259: 8-second MULTIACCUM readout to the next.  We note that the visit-to-visit
260: instability in the photometry is not an issue for these rapid fluctuations, as
261: shown by the stability of Star A, which has similar brightness to GRS
262: 1915+105, but is much more stable over these same timespans.  As noted above,
263: the residuals of the flux density values for Star A to a best-fit constant
264: value have $\chi^2_{\nu} \sim 1$, indicating that these uncertainties are
265: accurate within a given visit.  Typical events in Figure 3 have $\sim 5-10
266: \sigma$ statistical significance.  Thus, we conclude that this flickering is
267: in fact intrinsic variability in GRS 1915+105.  We quantify the mean IR flux
268: at $1.87 $\um and the sample RMS excess variability in that band (8 s bins) in
269: Table 1, so that the IR flickering rates can be compared within the set of HST
270: results and also with respect to results from other wavebands.
271: 
272: 
273: We characterized the timescale of these flickering events assuming an
274: exponential rise time for IR flares.  We derived characteristic
275: e-folding timescales using adjacent MULTIACCUM readouts (Table 2),
276: calculating the uncertainty by simple propagation of errors from the
277: flux density uncertainties.  The upper-limit timescale column in Table
278: 2 indicates the maximum characteristic rise time based on the highest
279: possible flux value (within uncertainties) at the beginning of a time
280: interval and the lowest possible flux value at the end of it.  Many
281: flickering events show best-estimate e-fold timescales of $\sim 30$s,
282: and several have firm upper limits on this number of $<40$s.  We note
283: that these timescales are nearly an order of magnitude shorter than
284: the fastest previously-reported timescales for long-wavelength
285: (IR/radio) variability from GRS 1915+105 (e.g. \citet{eiken98a};
286: \citet{eiken00}).
287: 
288: 
289: \subsection{X-ray Behavior Associated with the Fast IR Flickering}
290: 
291: We have examined the X-ray spectra and power density spectra (PDS)
292: derived from RXTE pointed observations of GRS~1915+105 obtained near
293: the times of the HST observations (Table 1).  Particularly relevant
294: are the RXTE observations made on 5 days (2003 April 10, 16, 23, May 1
295: and 7) that are interleaved with HST visits 2-5.  The X-ray spectral
296: properties during this epoch of rapid IR flickering are typical of the
297: radio plateau state, showing a power-law dominated spectrum with a
298: photon index, $\Gamma \sim$ 2.1--2.3 (see \citet{Muno01}).  The PDS
299: during these times are similar, and they are likewise typical of the
300: plateau state.  The first two of these PDS (April 16 = MJD 52745.1 and
301: April 23 = MJD 52752.2) are shown in Fig. 4, using the full PCA
302: bandpass, which is effectively 2--40 keV.  These PDS are subtracted
303: for deadtime-corrected Poisson noise, normalized to power density
304: ($P_{\nu}$) in units of $(rms/mean)^2$ Hz$^{-1}$, and then displayed
305: as log $(\nu \times P_{\nu}$ vs. log $\nu$.
306: 
307: The PDS shows a strong QPO just below 1 Hz, along with its second
308: harmonic.  These features are superposed on a strong and extremely
309: broad peak in the power continuum that is centered just above 1 Hz.
310: It has been argued that the broad power peak is a specific yet
311: unexplained X-ray signature of the steady jet in the hard state of
312: black hole binaries \citep{ron04}.  Since the X-ray PDS exhibits
313: little change over the interval of 2003 April-May, we can investigate
314: the X-ray and IR connection by comparing the flickering amplitudes at
315: the 8 s timescale of the IR light curve.  When the X-ray light curves
316: are examined in 8 s bins, the sample standard deviations for each of
317: the 5 relevant PCA observations are all in the range of 2.4--3.0 \% of
318: the mean count rate (2-40 keV).  This result represents the source
319: flickering amplitude, since the Poisson variations for these date are
320: well below 1\%.  These 8 s flickering amplitudes also represent the
321: X-ray behavior associated with the jet, since, as noted above, the
322: X-ray spectrum is dominated by nonthermal emission that is typical of
323: the plateau state, or more generally the X-ray hard state associated
324: with the presence of a steady radio jet \citep{jeffron06}.  One can
325: derive the analogous 8 s flickering amplitudes for the IR band using
326: the results given in Table 1.  The source RMS excess variability
327: amplitude (i.e. $\sqrt{\sigma_{sample}^2 - \sigma_{meas}^2} / mean$,
328: where $\sigma_{meas}$ are the measurement errors) is in the range of
329: 4-8 \% for HST visits 2--5.  This exceeds the equivalent flickering
330: amplitude in the X-ray band by a factor of 2--3.  In addition, the
331: baseline IR flux contributed from the donor star could be in the range
332: of $\sim$ 2 mJy, in which case the jet flickering at 8 s would be
333: elevated to the range of $\sim 10$ \%
334: 
335: 
336: \section{Discussion}
337: 
338: \subsection{The Diversity of IR Variability in GRS 1915+105}
339: 
340: The type of rapid IR variability described above is particularly intriguing in
341: that it seems to be unrelated to the previous types of IR variability observed
342: in GRS 1915+105 (e.g. \citet{rothstein04}; \citet{eiken00}; \citet{reba}).
343: First of all, this IR flickering, while small in amplitude, is significantly
344: faster than previous IR flares, which seem to have e-folding timescales of
345: $\sim 200-300$s, as compared to the $\sim 25-60$s timescales seen here.  Thus,
346: the flickering is $\sim 4-10$ times faster than any previously-reported IR
347: variability from GRS 1915+105.  Second, in these observations we see no
348: apparent association between IR flickering events and any similar X-ray
349: events.  This is also in stark contrast to previous IR flares of various types
350: in GRS 1915+105, all of which seem to be associated with significant activity
351: in the X-ray band as revealed through relatively large-amplitude changes in
352: the X-ray lightcurve and/or spectrum (e.g. \citet{rothstein04};
353: \citet{eiken00}; \citet{eiken98a}; \citet{eiken98b}; \citet{felix98};
354: \citet{fp97}).  We note that our observations above {\it do} reveal
355: significant X-ray variability on {\it faster} timescales than the IR
356: flickering.  However, any X-ray variability is much smaller in amplitude when
357: smoothed to the same time resolution as the IR data.  Thus, the rapid IR
358: flickering we report here seems to be an entirely new kind of IR variability
359: in GRS 1915+105.
360: 
361: 
362: \subsection{IR Emission and Relativistic Jets in the Plateau State}
363: 
364: As shown above (Figures 1-2), the IR flux density and fractional variability
365: from GRS 1915+105 are significantly enhanced during the plateau state, even
366: though the X-ray flux is lowest and steadiest during this time period.  A
367: number of observations indicate that during such plateaux, GRS 1915+105
368: exhibits a quasi-continuous radio synchrotron-emitting relativistic jet
369: outflow which also disappears after the plateau-ending X-ray/radio outburst
370: (see \citet{kleinwolt02} for a detailed discussion).  As also shown above, the
371: IR flickering does not seem to correlate well with any particular X-ray
372: variability from GRS 1915+105.  The typical timescales for X-ray variability
373: (e.g. the $\sim 1$ Hz QPO) are much less than the 8-second time resolution of
374: the IR observations, and there is very little X-ray variability on these
375: longer timescales -- when smoothed to 8-second resolution, the X-ray
376: lightcurve has an RMS of $\sim 3$\% about a constant value.  Thus, while there
377: is significant activity in the inner accretion disk in GRS 1915+105 at this
378: time, it is rather weak on the timescales seen in the IR.  We conclude that it
379: seems unlikely that the IR flickering arises from the accretion disk which is
380: producing the jet.  Rather, based on this apparent simultaneity of the plateau
381: jet and the IR flickering, along with the lack of IR/X-ray correlation, it
382: seems sensible to conclude that the enhanced IR emission of a few mJy arises
383: from the quasi-continuous jet itself.  Previous authors have also suggested
384: that emission from the jet may enhance IR emission during the plateau state
385: (\citet{ueda02} ; \citet{fuchs} ).
386: 
387: Radio observations of this plateau jet indicate that the radio synchrotron
388: emission begins in the optically thick regime, and the optical depth in the
389: radio decreases to $\sim 1$ in the expanding jet at distances $\sim 50$ AU
390: from the accretion disk and compact object (\citet{kleinwolt02};
391: \citet{vivek}).  If we assume that the opening angle of this jet is comparable
392: to or larger than the large-scale jets observed from GRS 1915+105, then the
393: diameter of the jet is $\ge 1$ AU, or equivalently $\sim 500$ lt-s, at this
394: distance.  Given that the IR flickering occurs on timescales as fast as $\sim
395: 25$s (20 times faster than the light travel time across the jet diameter at $d
396: \sim 50$ AU), we are forced to conclude that the IR flickering occurs much
397: closer to the source of the jet itself -- at a distance $\le 2.5$ AU.
398: 
399: We also note that the IR flux density excess seen here is $\sim 3-5$ mJy
400: (compared to observations outside the plateau state), corresponding to $\sim
401: 100-200$ mJy once we correct for the extinction towards GRS 1915+105
402: (\citet{fp97}).  Assuming a distance of $\sim 12$ kpc for GRS 1915+105
403: \citep{mr94}, this gives an excess IR luminosity from the jet of $\nu L_{\nu}
404: \sim 10^{36} \ {\rm ergs/s}$.  This is much larger than the same number for
405: the radio flux from the plateau jet ($\nu L_{\nu} \sim 10^{32} \ {\rm
406:   ergs/s}$), indicating that the IR jet luminosity dominates the radio jet
407: luminosity by several orders of magnitude.  The overall luminosity of the
408: system at this stage, however, remains dominated by the X-ray flux from the
409: inner accretion disks, at levels of $\sim 10^{38} \ {\rm ergs/s}$.  Thus, the
410: radiated power in the jet we observe here is $\sim 1$\% of the total
411: luminosity of GRS 1915+105.
412: 
413: 
414: \subsection{What Drives the Fast IR Flickering?}
415: 
416: If the rapid IR flickering, unlike the previously-reported Class B and Class C
417: flares, does {\it not} arise in the inner accretion disk but the jet, we must
418: consider the question, ``What drives the rapid IR flickering on $\sim 30-60$s
419: timescales?''.  Without the direct influence of the inner accretion disk, the
420: most obvious possibility is that the timescales seen in the IR flickering are
421: the natural timescales of some sort of instability in the jet itself, near its
422: base.  Several effects near the jet base are expected to generate
423: shocks/instabilities.  First of all, the actual jet base -- the location where
424: particles are initially accelerated away from the compact object -- is
425: expected to be a significant source of non-thermal radiation.  While it is
426: generally considered that the non-thermal X-ray emission from these systems
427: (which is essentially {\it all} of the X-ray emission from GRS 1915+105 in the
428: plateau state) may be the primary radiation from the particle acceleration
429: process, it is not impossible that significant IR radiation arises here as
430: well.  Secondly (and perhaps more likely), there may be a jet collimation
431: point, where the previously-accelerated outflow from the jet base is
432: constrained to move with much smaller opening angles ($\sim 1$\deg).  Such
433: behavior is predicted by several theoretical models for jet formation, and
434: apparently confirmed by the radio observations of the M87 jet near its base
435: \citep{junor}.  If the jet base emission is in fact dominated by the
436: non-thermal X-rays, the particles may have ``cooled'' enough at the
437: collimation point to emit predominantly in the IR waveband.  For either of
438: these scenarios, further support may come from the fact that several authors
439: have previously identified an ``IR break'' in the spectrum of the X-ray binary
440: jet source GX 339-4 (\citet{motch}; \citet{CorbelFender}; \citet{nowak}).  In
441: some models of jet formation this break is directly associated to the
442: shock-acceleration region of the jet \citep{markoff} -- applying this to GRS
443: 1915+105 would naturally lead to the interpretation of the flickering observed
444: here as coming from the jet launch region.  Finally, jet outflows from
445: supermassive black holes show clear evidence of internal shocks within the
446: collimated jet flow even further ``downstream'' from the collimation point.
447: If such shocks occur close enough to the compact object, they may also produce
448: rapid IR flickering as seen here.  Future observations of this flickering
449: which could further constrain these possibilities should give key insights
450: into the formation of the plateau jet in GRS 1915+105.
451: 
452: \section{Conclusion}
453: 
454: We have presented near-infrared observations of the microquasar GRS 1915+105
455: using the NICMOS instrument of the Hubble Space Telescope before, during, and
456: after a ``plateau'' state in March-May 2003.  During the X-ray/radio plateau,
457: we find that the infrared flux is significantly higher and more strongly
458: variable than outside the plateau.  In particular, we find ``flickering''
459: events with amplitudes $> 20$\% occurring on e-folding timescales of $\sim
460: 30$s.  These timescales are several times faster than any previously-observed
461: infrared variability in GRS 1915+105, and do not seem to be correlated with
462: any X-ray events of similar amplitude/timescale.  As such, they seem to
463: represent an entirely new mode of IR variability from GRS 1915+105.  Based on
464: the simultaneity of this flickering with the plateau jet and the lack of
465: IR/X-ray correlation on the flickering timescales, we conclude that the IR
466: flickering arises in the plateau jet itself.  If so, the IR jet luminosity is
467: much greater than the corresponding radio jet luminosity.  Also, the IR
468: flickering must occur very near the base of the jet ($<2.5$ AU from the
469: accretion disk).  The lack of correlation with inner disk activity implies
470: that $\sim 30-60$s is a natural timescale of the jet at this range of
471: locations, and that the flickering may be due to shocks at either the site of
472: particle acceleration, the site of jet collimation, or internal shocks in the
473: jet.  Further study could give important insights into the formation of the
474: plateau jet in GRS 1915+105.
475: 
476: \acknowledgements {The authors wish to thank the staff of the Space
477:   Telescope Science Institute, particularly, N. Schulz, for their help and
478:   support for this TOO program with NICMOS.  The authors also thank the
479:   anonymous referee for helpful suggestions.  SSE and SGP were supported in
480:   part by a STScI grant, by an NSF CAREER award (AST-0328522), and NSF
481:   grant AST-0507547.  DMR was supported in part by a National Science
482:   Foundation Graduate Research Fellowship}
483: 
484: \begin{thebibliography}
485: 
486: \bibitem[Bandyopadhyay, et al.(1998)]{reba} Bandyopadhyay,R., et al. 1998, \mnras, 295, 623
487: 
488: \bibitem[Castro-Tirado et al.(1994)]{alberto} Castro-Tirado,A., et al. 1994, \apjs, 92, 469
489: 
490: \bibitem[Corbel \& Fender (2002)]{CorbelFender} Corbel.S. \& Fender,R. 2002,
491:   \apj, 573, L35
492: 
493: \bibitem[Dhawan, et al. (2000)]{vivek} Dhawan,V., Mirabel,I.F., Rodr\'iguez,L.F. 2000, \apj, 543, 373
494: 
495: \bibitem[Eikenberry, et al.(1998a)]{eiken98a} Eikenberry,S.S., Matthews,K., Morgan,E.H., Remillard,R.A., Nelson,R.W. 1998a, \apj, 494, L61
496: 
497: \bibitem[Eikenberry, et al.(1998b)]{eiken98b} Eikenberry,S.S., Matthews,K., Murphy,T.W., Nelson,R., Morgan,E.H., Remillard,R.A., Muno,M.  1998b, \apj, 506, L31
498: 
499: \bibitem[Eikenberry, et al.(2000)]{eiken00} Eikenberry,S.S., Matthews,K., Muno,M., Blanco,P.R., Morgan,E.H., Remillard,R.A. 2000, \apj, 532, L33
500: 
501: \bibitem[Eikenberry \& Fazio(1997)]{eiken96} Eikenberry,S.S. \& Fazio,G.G. 1997, \apjl, 475, L53
502: 
503: \bibitem[Fender, et al.(1999)]{Fender99} Fender,R.P. et al. 1999, \mnras, 304, 865
504: 
505: \bibitem[Fender \& Pooley,(1998)]{fp99} Fender,R.P. \& Pooley,G.G. 1998, \mnras, 300, 573
506: 
507: \bibitem[Fender et al.,(1997)]{fp97} Fender,R.P., Pooley,G.G., Brocksopp,C., Newell,S.J. 1997, \mnras, 290, L65
508: 
509: \bibitem[Foster et al.,(1996)]{Foster96} Foster, et al. 1996, \apjl, 467, L81
510: 
511: \bibitem[Fuchs et al., (2003)]{fuchs} Fuchs,Y., Rodriguez,J., Mirabel,I.F,
512:   Chaty,S., Ribó,M., Dhawan,V., Goldoni,P., Sizun,P., Pooley,G.G.,
513:   Zdziarski,A.A., Hannikainen,D.C., Kretschmar,P., Cordier,B., Lund, N. , \aap,
514:   409, L35
515: 
516: \bibitem[Jahoda et al.,(2006)]{jah2006} Jahoda, K.,  et al., 2006 \apjs, in press
517: 
518: 
519: \bibitem[Junor, et al.(1999)]{junor} Junor,W., Biretta,J.A., Livio,M. 1999, \nat, 401, 891
520: 
521: \bibitem[Klein-Wolt et al.(2002)]{kleinwolt02} Klein-Wolt,M., Fender,R.P., Pooley,G.G., Belloni,T., Migliari,S., Morgan,E.H., van der Klis, M. 2002, \mnras, 331, 745
522: 
523: \bibitem[Levine et al.(1996)]{lev96} Levine, A. M., et al. 1996, \apjl, 469, L33
524: 
525: \bibitem[Markoff et al.(2003)]{markoff} Markoff,S., Nowak,M., Corbel,S.,
526:   Fender,R., Falcke,H. 2003, \aap, 397, 645
527: 
528: \bibitem[McClintock \& Remillard, (2006)]{jeffron06} McClintock, J.E., \& Remillard, R.A. 2006, in Compact Stellar X-ray Sources, eds. W.H.G Lewin \& M. van der Klis, Cambridge: Cambridge University Press, pp.157--214; astro-ph/0306213
529: 
530: \bibitem[Mikles, et al.(2006)]{mikles} Mikles,V., Eikenberry,S.S., Rothstein,D.M. 2006, \apj, 637, 978
531: 
532: \bibitem[Mirabel, et al.(1998)]{felix98} Mirabel,I.F., Dhawan,V., Chaty,S., Rodriguez,L.F., Robinson,C., Swank,J., Geballe,T. 1998, \aap, 330, L9
533: 
534: \bibitem[Mirabel \& Rodr\'iguez(1994)]{mr94} Mirabel,I.F. \& Rodr\'iguez,L.F. 1994, \nat, 371, 46
535: 
536: \bibitem[Mirabel \& Rodr\'iguez(1999)]{MRaa} Mirabel,I.F. \& Rodr\'iguez,L.F. 1999, \araa, 37, 409
537: 
538: \bibitem[Mitsuda, et al. (1984)]{mitsuda} Mitsuda, K., Inoue, H., Koyama, K., Makishima, K., Matsuoka, M., Ogawara, Y., Shibazaki, N., Suzuki, K., Tanaka, Y., \& Hirano, T. 1984, \pasj, 36, 741
539: 
540: \bibitem[Motch, et al.(1985)]{motch} Motch,C., Ilovaisky,S.S., Chevalier,C.,
541:   Angebault,P. 1985, Space Sci Rev, 40, 219
542: 
543: \bibitem[Muno, et al.(2001)]{Muno01} Muno,M.P. et al. 2001, \apj, 556, 515
544: 
545: \bibitem[Muno, et al.(1999)]{muno99} Muno, M. P., Morgan, E. H. \& Remillard, R. A. 1999, \apj, 527, 321
546: 
547: \bibitem[Nowak, et al.(2005)]{nowak} Nowak,M.A., Wilms,J., Heins,S.,
548:   Pooley,G., Pottschmidt,K. Corbel,S. 2005, \apj, 626, 1006
549: 
550: \bibitem[Pooley \& Fender(1997)]{pf97} Pooley,G.G. \& Fender,R.P. 1997, \mnras, 292, 925
551: 
552: \bibitem[Remillard (2004)]{ron04} Remillard, R.A. 2004, in Procs., Texas@Stanford 2004, ed., P. Chen, SLAC Electronic Conference Proceedings Archive; astro-ph/0504129
553: 
554: 
555: \bibitem[Rothstein, et al.(2005)]{rothstein04} Rothstein,D.M.,
556:   Eikenberry,S.S., Matthews,K. 2005, \apj, 626, 991
557: 
558: \bibitem[Sams, et al.(1996)]{sams} Sams,B.J., Eckart,A., Sunyaev,R. 1996,
559:   \nat, 382, 47
560: 
561: \bibitem[Thompson, et al.(1998)]{NICMOS} Thompson,R.I., Rieke,M.,
562:   Schneider,G., Hines,D.C., Corbin,M.R. 1998, \apjl, 492, L95
563: 
564: \bibitem[Ueda, et al.(2002)]{ueda02} Ueda,Y., Yamaoka,K.,
565:   Sánchez-Fernández,C., Dhawan,V., Chaty,S., Grove,J.E., McCollough,M.,
566:   Castro-Tirado,A.J., Mirabel,F., Kohno,K., Feroci,M., Casella,P.,
567:   Trushkin,S.A., Castaneda,H., Rodríguez,J., Durouchoux,P., Ebisawa,K.,
568:   Kotani,T., Swank,J., Inoue, H.  2002, \apj, 571, 918
569: 
570: \end{thebibliography}
571: 
572: 
573: %######## TABLES ##########
574: \begin{deluxetable}{lccc}
575: \tablewidth{0pt}
576: \tablehead{
577: \colhead{Visit}&\colhead{Date}&\colhead{F187W mean flux density}&\colhead{Excess Variability}\\
578: \colhead{}&\colhead{(MJD)}&\colhead{(mJy)}&\colhead{(\% RMS)\tablenotemark{a}} }
579: \startdata
580: 1& 52736.60290 & 2.54 & 1.0 \\
581: 2& 52741.67225 & 6.28 & 6.9 \\
582: 3& 52746.87292 & 4.24 & 4.8 \\
583: 4& 52752.87847 & 5.10 & 7.8 \\
584: 5& 52763.68303 & 3.66 & 4.0 \\
585: 6& 52796.43359 & 2.41 & 0.8 \\
586: 7& 52801.71115 & 2.11 & $<2.0$ \\
587: 8& 52808.31191 & 2.26 & 1.8 \\
588: 9& 52814.71761 & 2.31 & $<2.0$ \\
589: \enddata
590: 
591: \tablenotetext{a}{Excess variability RMS \% is defined as
592: $\sqrt{\sigma_{sample}^2 - \sigma_{meas}^2} / mean$ where
593: $\sigma_{sample}$ is the RMS scatter in the sample values,
594: $\sigma_{meas}$ is the pipeline-estimated measurement noise RMS, and
595: $mean$ is the mean value of the sample points.  Values reported as
596: $<2.0$ \% inidcate that the RMS sample scatter is less than the
597: estimated RMS measurement noise, and we take $\sigma_{meas}$ as an
598: upper limit on $\sigma_{sample}$.}
599: \end{deluxetable}
600: 
601: \begin{deluxetable}{cccccc}
602: \tablewidth{0pt}
603: \tablehead{
604: \colhead{Visit}&\colhead{Filter}&\colhead{Date}&\colhead{UT}&\colhead{Characteristic}&\colhead{Upper}\\
605: \colhead{}&\colhead{}&\colhead{(MJD)}&\colhead{(s)}&\colhead{Rise Time (s)}&\colhead{Limit (s)}
606: }
607: \startdata
608: 2 & F187W & 52741 & 58234 & $58\pm8$ & 71 \\
609: 2 & POL0L & 52741 & 59079 & $31\pm4$ & 38 \\
610: 2 & POL120L & 52741 & 59483 & $43\pm7$ & 55 \\
611: 3 & F187W & 52746 & 75663 & $27\pm9$  & 50 \\
612: 3 & F187W & 52746 & 75691 & $49\pm7$  & 61 \\
613: 3 & F187W & 52746 & 75818 & $64\pm13$  & 88 \\
614: 3 & POL120L & 52746 & 76755 & $52\pm9$  & 70 \\
615: 4 & F187W & 52752 & 75904 & $28\pm9$  & 48 \\
616: 4 & F187W & 52752 & 76289 & $45\pm5$  & 54 \\
617: 4 & F187W & 52752 & 76392 & $55\pm6$  & 65 \\
618: 4 & POL0L & 52752 & 76529 & $51\pm9$  & 69 \\
619: 4 & POL0L & 52752 & 76735 & $48\pm9$  & 65 \\
620: 4 & POL120L & 52752 & 77163 & $30\pm4$  & 37 \\
621: 4 & POL240L & 52752 & 77369 & $28\pm3$  & 34 \\
622: 4 & POL240L & 52752 & 77654 & $37\pm5$  & 45 \\
623: \enddata
624: \end{deluxetable}
625: 
626: 
627: %######## FIGURES #########
628: \begin{figure}
629: \plotone{figure1.eps}
630: 
631: \caption{Multiwavelength light curves of GRS 1915+105.  (Top) X-ray
632:   lightcurve from the RXTE ASM.  (Middle) Radio lightcurve from the Ryle
633:   Telescope.  (Bottom) Near-infrared ($1.87 $\um) lightcurve from
634:   $HST/NICMOS$.  For the near-IR lightcurve, note that the bars indicate the
635:   {\it range} of the variable flux density values during each observation,
636:   rather then the uncertainties.  Vertical lines provide a time reference for
637:   comparison among the different bands.}
638: 
639: \end{figure}
640: 
641: \begin{figure}
642: \plotone{figure2.eps}
643: 
644: \caption{Flux densities for GRS 1915+105 (Diamonds) and Star A
645: (Squares) in NICMOS F187W filter.  Note that Star A, which has similar
646: brightness to GRS 1915+105, is quite steady while GRS 1915+105 varies
647: both between visits and within individual visits.}
648: 
649: \end{figure}
650: 
651: \begin{figure}
652: \plottwo{im2_2_9-10.eps}{im3_3_5-6_9-10.eps}
653: \plottwo{im4_5_7-8.eps}{pol4_7_11-12.eps}
654: \plottwo{pol4_9_11-12.eps}{pol4_12_8-9.eps}
655: 
656: \caption{Selected high time-resolution near-IR lightcurves for GRS
657: 1915+105.}
658: 
659: \end{figure}
660: 
661: \begin{figure}
662: \plotone{figure4.eps}
663: 
664: \caption{X-ray power density spectra of GRS 1915+105 for 16 April 2003
665: (solid) and 23 April 2003 (dash).}
666: 
667: \end{figure}
668: 
669: 
670: 
671: \end{document}
672: