A SPITZER SURVEY OF NOVAE IN M31

Novae are all semi-detached binary systems where a late-type star transfers mass to a white dwarf companion (Warner 1995, 2008). A thermonuclear runaway eventually ensues in the material accreted onto the surface of the white dwarf resulting in the nova explosion (Starrfield et al. 2008). The resulting release of energy (∼ 10⁴⁴–10⁴⁵ erg) is sufficient to expel the accreted envelope and drive substantial mass loss (10⁻⁴ to 10⁻⁵ M_☉) from the system at high velocities (∼ several hundreds to a few thousand km s⁻¹; Bode 2010). Novae exhibit outburst amplitudes of roughly 10–20 mag, and can reach peak luminosities as high as M_V ∼ −10, second only to GRBs and supernovae in the energetics of their outbursts, but far more frequent in a given galaxy than either. Their high luminosities and rates (∼50 yr⁻¹ in a galaxy like M31; Shafter & Irby 2001; Darnley et al. 2006) make novae powerful probes of the properties of binary star systems in different (extragalactic) stellar populations. The rapid formation of copious amounts of dust in many novae post-outburst also makes them unique laboratories in which to explore cosmic dust grain formation (Bode & Evans 1989; Evans & Rawlings 2008; Gehrz 2008).

Models show that properties of the eruption (e.g., the peak luminosity and the decline rate) are strongly dependent on parameters such as the accretion rate, and the mass and luminosity of the white dwarf (e.g., Livio 1992), some or all of which may vary systematically with the underlying stellar population. The strength of the nova outburst is most sensitive to the mass of the accreting white dwarf, with the increased surface gravity of a more massive white dwarf resulting in a higher pressure at the base of the accreted envelope at the time of thermonuclear runaway and a more violent outburst. In addition, since a smaller mass of accreted material is required to achieve the critical temperature and density necessary for a runaway, nova outbursts produced on massive white dwarfs are expected to have shorter recurrence times and faster light curve evolution (Hachisu & Kato 2006). Further, since the mean white dwarf mass in a nova system is expected to decrease as a function of the time elapsed since the formation of the progenitor binary (e.g., Tutukov & Yungelson 1995; Politano 1996), the proportion of fast and bright novae is expected to be higher in a younger stellar population, which should contain on average more massive white dwarfs.

Duerbeck (1990) became the first to formally postulate the existence of two populations of novae: a relatively young population of "disk novae," which are found in the solar neighborhood and in the Large Magellanic Cloud (LMC), and "bulge novae," which are concentrated toward the Galactic center and in the bulge of M31 and are characterized by generally slower outburst development. The argument in favor of two nova populations was further developed by Della Valle et al. (1992), who showed that the average scale height above the Galactic plane for "fast" novae (t₂ < 13 d ⁵) is smaller than for novae with slower rates of decline. At about the same time, Williams (1992) was proposing that classical novae could be divided into two classes based on their spectral properties: specifically, the relative strengths of either their Fe ii or He and N emission lines. Novae with prominent Fe ii lines ("Fe ii novae") usually show P Cygni absorption profiles, tend to evolve more slowly, have lower expansion velocities, and have a lower level of ionization compared with novae that exhibit strong lines of He and N ("He/N novae"). In addition, the latter novae display very strong neon lines, but not the forbidden lines that are often seen in the Fe ii novae. Following up on their earlier work, Della Valle & Livio (1998) noted that Galactic novae with well-determined distances that were classified as He/N were concentrated near the Galactic plane, and tended to be faster and more luminous compared with their Fe ii counterparts.

Infrared observations offer another avenue for the study of nova populations. Models for thermonuclear runaways on the surfaces of white dwarfs predict that the ejecta will contain not only accreted gas, but material dredged up from the white dwarf as well. Thus, spectroscopic analysis of the ejecta offers an opportunity to distinguish between eruptions that occur on CO white dwarfs and those occurring on more massive ONe white dwarfs (Gehrz et al. 1998). Eruptions occurring on CO white dwarfs are expected to produce a significant amount of dust, mainly in the form of carbon grains that shroud the central source, and result in warm (∼ 1000 K) blackbody emission that peaks in the near-IR (∼3μm) 1–4 months after eruption (e.g., Geisel et al. 1970; Ney & Hatfield 1978; Bode & Evans 1982; Gehrz et al. 1995; Gehrz 2008). On the other hand, eruptions seated on the more massive ONe white dwarfs are believed to produce strong forbidden lines of neon ([Ne ii] 12.8 μm and [Ne vi] 7.6 μm), but little or no dust. The first observational support for a class of novae occurring on massive ONe white dwarfs was found more than 20 years ago when Gehrz et al. (1985) discovered that the [Ne ii] 12.8 μm emission line in QU Vul (λ/Δλ = 67) reached a flux level ∼60 times that of the free–free continuum more than four months after eruption. At that time, this was the strongest 12.8 μm emission line seen in any astrophysical object. Since that time several additional Galactic novae with an overabundance of Ne have been discovered, and they are collectively referred to as "neon" novae (Gehrz et al. 1985). It appears that up to one-third of novae are of this type, with the remainder being CO novae.

Although much can and has been learned from the study of Galactic novae, the nearby spiral M31, where more than 800 nova candidates have been discovered over the past century (e.g., Pietsch et al. 2007; Shafter 2008, and references therein), offers a unique opportunity to study an equidistant sample of novae from differing stellar environments (bulge and disk) while minimizing some of the uncertainties that plague Galactic observations (e.g., highly variable extinction along the lines of sight to different novae). With this motivation in mind, we are currently involved in a multi-wavelength program of photometric and spectroscopic follow-up observations of novae discovered in M31 (A. W. Shafter et al. 2011, in preparation). As a part of this effort focusing on the nature of the white dwarfs (CO versus ONe), we have obtained both IR photometric and spectroscopic observations of a sample of 10 novae in M31 using the IRAC and IRS instruments on board the Spitzer Space Telescope. Here, we present the results of our survey.

In order to search for dust formation in a sample of M31 novae, we initiated a program of Spitzer IRAC (Fazio et al. 2004) and IRS (Houck et al. 2004) observations in early 2007 and 2008. Our observing strategy called for the observations to be triggered approximately three months after discovery of a nova. However, due to the vagaries of Spitzer scheduling, our observations were typically executed anywhere from three to as long as seven months after eruption.

During the first year of observations, the same four novae were observed with both IRAC and the IRS. The results from the IRS observations were disappointing, with no line emission detected in any nova and continuous emission detected in just two of the novae. We concluded that most novae had likely faded beyond detectability by the time the IRS observations were scheduled. In order to increase the probability of detection, in the second epoch of observations, we were able to modify our IRS target list to include two more recent novae for observation: M31N 2007-11d and 2007-11e resulting in line emission being detected in one of these novae. Our Spitzer observations are summarized in Table 1.

2.1. IRAC Observations

IRAC data were obtained in all four IRAC bands (3.6 μm, 4.5 μm, 5.8 μm, and 8.0 μm). The IRAC observations consisted of a 12-point Reuleaux dither pattern with a medium scale factor. A single 100 s frame was obtained at each dither position, resulting in a total of ∼ 1200 s of exposure in each band. Source extraction and photometry were performed on the post-basic-calibrated-data (BCD) mosaics produced by the Spitzer Science Center (SSC), pipeline version 18.5. The extractions and determination of uncertainties were performed using version 2.1 of the IDL program ATV, written and maintained by Barth (2001). Due to the crowded nature of the observed fields along with rapidly varying background levels, it was necessary to use apertures of minimal size to avoid contamination from neighboring objects. Therefore, we chose an aperture size of radius 2 pixels for each extraction. We used the nominal positions of the novae as determined by visual photometric observations, along with an ATV 3 pixel-radius centering box, to determine the center for each aperture used in extraction. Aperture correction factors, derived by the SSC and provided in the IRAC Data Handbook 3.0, were then used to estimate the true flux of the novae based on the measured flux in each aperture. Finally, we determined the background level for each aperture using an annulus of inner radius 2 pixels and outer radius 6 pixels. The energy distributions for the six novae detected by IRAC are shown in Figure 1 and summarized in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Spitzer IRAC Photometry

Nova	Flux (10−14 erg cm−2 s−1 μm−1)
	3.6 μm	4.5 μm	5.8 μm	8.0 μm
M31N 2006-09c	11.4 ± 3.37	9.14 ± 2.39	7.07 ± 1.72	4.78 ± 1.11
M31N 2006-10a	13.8 ± 3.83	13.7 ± 2.96	10.9 ± 2.14	5.94 ± 1.23
M31N 2006-10b	0.92 ± 0.94	0.30 ± 0.43	0.15 ± 0.25	⋅⋅⋅
M31N 2006-11a	4.93 ± 2.22	2.00 ± 1.13	0.87 ± 0.63	⋅⋅⋅
M31N 2007-07f	5.63 ± 2.31	5.42 ± 1.83	4.99 ± 1.44	3.11 ± 0.88
M31N 2007-08a	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
M31N 2007-08d	0.99 ± 0.97	0.41 ± 0.51	0.23 ± 0.31	0.14 ± 0.19
M31N 2007-10a	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅

Download table as:  ASCIITypeset image

2.2. IRS Observations

IRS data were obtained in the low-resolution SL1 module, covering a wavelength range of 7.4–14.5 μm. The IRS data were obtained using 24 cycles of 60 s ramps. As each cycle of an IRS observation consists of two nod positions, this resulted in a total of ∼ 2880 s of on-source integration time per target.

The IRS data were analyzed in the SMART package (Higdon et al. 2004) using optimal extraction (Lebouteiller et al. 2010). The BCD products were cleaned using the IRSCLEAN tool with campaign-specific masks before all two-dimensional spectra at a given nod position were combined into a single image. For our observations, this resulted in two ∼ 1440 s cleaned two-dimensional spectra. Prior to spectral extraction, the two nod positions were subtracted from each other to remove the background and mitigate rogue pixels not present in the masks. The spectral extraction was performed on these background-subtracted images. For each object, we selected the "manual optimal extraction" option in SMART. The position of the IRS slit was projected on our IRAC images to verify that we were extracting spectra for the correct object. A visual inspection of the two-dimensional spectra revealed that only two objects (M31N 2006-09c and 2006-10a) had clear traces at the nominal source position. For the remaining objects (with the exception of M31N 2007-11e, see below), we extracted spectra at the nominal source position in each nod position. As expected from the visual inspection of the two-dimensional spectra, no sources were detected. In the case of M31N 2006-09c, the nova was the only detected source. The spectrum of M31N 2006-09c was extracted using the optimal extraction algorithm with a background order of one; even though background subtraction was accomplished to first order by subtracting nod positions, residual structure was still present owing to the variation of the local background between the nod positions. In the case of M31N 2006-10a, there were two objects detected, the brighter nova at the nominal position in the slit as well as a fainter source offset by ∼ 3 pixels (∼ 5''). To extract M31N 2006-10a, two sources were specified and the extraction algorithm was allowed to fit the trace within 1 pixel of the nominal position (for the nova) and between 1 and 4 pixels from the nominal position for the contaminating source. In the case of M31N 2007-11e, no source was apparent save for a faint line at the position of [Ne ii] 12.8 μm. For M31N 2007-11e, we forced the extraction aperture to be centered on the nominal source position for each nod. For all three objects, spectra extracted at each nod position were clipped at the order edges and combined. The spectra extracted for M31N 2006-09c and 2006-10a are shown in Figure 2 and the extracted spectrum of M31N 2007-11e is shown in Figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

2.3. Optical Photometric Observations

To complement our Spitzer observations, we attempted to obtain optical photometric time series of several novae in our survey using the Liverpool Telescope (LT; Steele et al. 2004), primarily to measure the peak nova brightness and rate of decline (t₂). The LT data were reduced using a combination of IRAF⁶ and Starlink software. These data were calibrated with standard stars from Landolt (1992) and checked against secondary standards from Magnier et al. (1992), Haiman et al. (1994), and Massey et al. (2006). These observations were augmented by the extensive photometric database compiled by one of us (K.H.) as part of an ongoing program to monitor nova light curves in M31. The latter observations include both survey and targeted images taken with the 0.35 m telescope of the private observatory of K.H. at Lelekovice, the 0.65 m telescope of the Ondřejov Observatory (operated partly by the Charles University, Prague), and the 0.28 m telescope of the Zlín Observatory. Standard reduction procedures for raw CCD images were applied (bias and dark-frame subtraction and flat-field correction) using SIMS⁷ and Munipack⁸ programs. Reduced images of the same series were co-added to improve the signal-to-noise ratio (S/N; total exposure time varied from 10 minutes up to about 1 hr). The gradient of the galaxy background of co-added images was flattened by a spatial median filter using SIMS. These processed images were then used to search for novae. Photometry was performed using "Optimal Photometry" (based on fitting of point-spread function, PSF, profiles) in GAIA,⁹ and calibrated with standard stars from Landolt (1992).

Of the 10 novae observed by Spitzer, we were successful in obtaining a series of photometric observations for five of the novae. The results of these observations are summarized in Tables 3–7, with the resulting light curves shown in Figures 4–6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. Photometric Observations of M31N 2006-09c

HJD	Mag	Filter	Observer(s)	Telescope
(2,450,000+)
3993.376	>19.1	R	K.H.	Lelekovice 0.35 m
3996.404	18.1 ± 0.2	R	P. Kušnirák	Ondřejov 0.65 m
3999.609	17.15 ± 0.1	R	P. Kušnirák	Ondřejov 0.65 m
4000.317	17.0 ± 0.1	R	K.H.	Lelekovice 0.35 m
4000.590	17.35 ± 0.1	R	M. Wolf	Ondřejov 0.65 m
4001.360	17.3 ± 0.1	R	K.H.	Lelekovice 0.35 m
4002.302	17.4 ± 0.1	R	K.H.	Lelekovice 0.35 m
4002.329	17.5 ± 0.1	R	K.H.	Lelekovice 0.35 m
4005.312	17.8 ± 0.1	R	K.H.	Lelekovice 0.35 m
4007.312	17.9 ± 0.1	R	K.H.	Lelekovice 0.35 m
4014.295	19.0 ± 0.2	R	K.H.	Lelekovice 0.35 m
4017.258	19.1 ± 0.2	R	K.H.	Lelekovice 0.35 m
4019.319	18.8 ± 0.2	R	K.H.	Lelekovice 0.35 m
4024.383	19.3 ± 0.2	R	K.H.	Lelekovice 0.35 m
4026.330	19.3 ± 0.2	R	K.H.	Lelekovice 0.35 m
4026.364	19.3 ± 0.2	R	K.H.	Lelekovice 0.35 m
4034.312	19.9 ± 0.3	R	K.H.	Lelekovice 0.35 m
4256.677	>21.2 ± 0.2	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4260.700	>21.7 ± 0.3	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4260.705	>21.0 ± 0.3	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4260.690	>21.4 ± 0.2	r	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4260.695	21.8 ± 0.5	i	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m

Download table as:  ASCIITypeset image

Table 4. Photometric Observations of M31N 2006-10a

HJD	Mag	Filter	Observer(s)	Telescope
(2,450,000+)
4019.319	>19.7	R	K.H.	Lelekovice 0.35 m
4024.383	>19.0	R	K.H.	Lelekovice 0.35 m
4026.330	>20.1	R	K.H.	Lelekovice 0.35 m
4031.251	19.2 ± 0.25	R	K.H.	Lelekovice 0.35 m
4034.312	18.7 ± 0.15	R	K.H.	Lelekovice 0.35 m
4034.470	18.6 ± 0.15	R	P. Kušnirák	Ondřejov 0.65 m
4035.360	18.4 ± 0.3	R	K.H.	Lelekovice 0.35 m
4043.331	17.9 ± 0.1	R	K.H.	Lelekovice 0.35 m
4047.288	19.0 ± 0.3	R	K.H.	Lelekovice 0.35 m
4048.324	19.3 ± 0.2	R	K.H.	Lelekovice 0.35 m
4055.296	18.2 ± 0.15	R	K.H.	Lelekovice 0.35 m
4055.262	18.2 ± 0.1	R	P. Cagaš	Zlín 0.28 m
4070.308	19.2 ± 0.35	R	M. Wolf, P. Zasche	Ondřejov 0.65 m
4071.385	18.9 ± 0.2	R	P. Kušnirák	Ondřejov 0.65 m
4078.308	18.4 ± 0.3	R	K.H.	Lelekovice 0.35 m
4078.343	18.6 ± 0.2	R	K.H.	Lelekovice 0.35 m
4080.306	18.8 ± 0.25	R	K.H.	Lelekovice 0.35 m
4084.212	18.9 ± 0.2	R	K.H.	Lelekovice 0.35 m
4093.174	19.5:	R	K.H.	Lelekovice 0.35 m
4096.325	19.7 ± 0.25	R	K.H.	Lelekovice 0.35 m
4097.222	19.5 ± 0.25	R	K.H.	Lelekovice 0.35 m
4115.194	>20.0	R	K.H.	Lelekovice 0.35 m
4121.381	>19.8	R	K.H.	Lelekovice 0.35 m
4122.331	>19.5	R	K.H.	Lelekovice 0.35 m
4122.377	>19.4	R	K.H.	Lelekovice 0.35 m
4044.337	18.07 ± 0.04	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4050.589	19.06 ± 0.03	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4056.585	18.88 ± 0.03	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4062.606	18.58 ± 0.03	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4069.484	19.24 ± 0.04	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4071.549	19.06 ± 0.09	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4074.575	19.20 ± 0.11	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4077.542	19.26 ± 0.13	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4084.460	19.24 ± 0.03	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4092.454	19.93 ± 0.05	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4099.407	20.03 ± 0.11	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4101.393	19.78 ± 0.07	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4114.372	>21.90 ± 0.24	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4120.432	>20.89 ± 0.28	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4254.685	>17.6 ± 0.9	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4044.334	17.90 ± 0.03	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4050.586	19.10 ± 0.02	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4056.582	18.80 ± 0.03	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4062.604	18.48 ± 0.02	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4069.481	19.25 ± 0.04	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4071.546	19.02 ± 0.12	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4074.572	19.39 ± 0.10	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4077.539	19.30 ± 0.05	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4084.457	19.16 ± 0.03	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4092.451	19.67 ± 0.05	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4099.404	19.64 ± 0.10	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4101.391	19.73 ± 0.06	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4114.370	>21.55 ± 0.22	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4120.429	>21.22 ± 0.12	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4254.692	>18.4 ± 0.4	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4254.671	>19.0 ± 0.3	r	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m

Download table as:  ASCIITypeset image

Table 5. Photometric Observations of M31N 2006-10b

HJD	Mag	Filter	Observer(s)	Telescope
(2,450,000+)
4043.888	18.95 ± 0.06	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4049.001	19.55 ± 0.04	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4057.034	20.39 ± 0.06	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4062.963	20.89 ± 0.07	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4068.945	20.99 ± 0.13	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4071.940	>20.14 ± 0.21	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4074.973	>20.35 ± 0.20	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4083.911	21.41 ± 0.10	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4098.938	>20.65 ± 0.27	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4105.915	>19.74 ± 0.54	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4113.883	>22.91 ± 0.69	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4119.958	>21.83 ± 0.38	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4248.204	>22.42 ± 0.43	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4048.998	20.02 ± 0.05	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4053.817	>17.33 ± 0.29	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4057.031	20.88 ± 0.07	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4062.960	21.27 ± 0.08	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4068.942	>21.69 ± 0.23	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4071.937	>20.12 ± 0.21	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4074.970	>20.64 ± 0.19	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4083.908	>22.30 ± 0.22	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4098.936	>20.51 ± 0.22	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4101.885	>20.13 ± 0.33	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4105.912	>17.61 ± 0.60	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4107.986	>18.40 ± 0.28	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4113.880	>21.96 ± 0.39	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4119.955	>22.76 ± 0.65	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4248.209	>22.30 ± 0.43	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4248.193	>22.9 ± 0.4	r	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4248.199	>21.5 ± 0.3	i	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m

Download table as:  ASCIITypeset image

Table 6. Photometric Observations of M31N 2006-11a

HJD	Mag	Filter	Observer(s)	Telescope
(2,450,000+)
4048.324	>19.8	R	K.H.	Lelekovice 0.35 m
4055.296	>19.8	R	K.H.	Lelekovice 0.35 m
4070.263	16.9 ± 0.1	R	M. Wolf, P. Zasche	Ondřejov 0.65 m
4070.308	16.6 ± 0.15	R	M. Wolf, P. Zasche	Ondřejov 0.65 m
4078.308	16.1 ± 0.1	R	K.H.	Lelekovice 0.35 m
4078.343	16.0 ± 0.1	R	K.H.	Lelekovice 0.35 m
4080.306	16.3 ± 0.1	R	K.H.	Lelekovice 0.35 m
4084.212	16.9 ± 0.1	R	K.H.	Lelekovice 0.35 m
4093.174	17.7 ± 0.15	R	K.H.	Lelekovice 0.35 m
4096.325	17.8 ± 0.15	R	K.H.	Lelekovice 0.35 m
4097.222	17.8 ± 0.15	R	K.H.	Lelekovice 0.35 m
4115.194	18.8 ± 0.2	R	K.H.	Lelekovice 0.35 m
4121.381	18.9 ± 0.2	R	K.H.	Lelekovice 0.35 m
4122.331	19.2 ± 0.25	R	K.H.	Lelekovice 0.35 m
4122.377	19.1 ± 0.3	R	K.H.	Lelekovice 0.35 m
4126.289	19.3 ± 0.3	R	K.H.	Lelekovice 0.35 m
4126.339	19.1 ± 0.3	R	K.H.	Lelekovice 0.35 m
4128.275	19.2 ± 0.25	R	K.H.	Lelekovice 0.35 m
4135.298	19.4 ± 0.3	R	K.H.	Lelekovice 0.35 m
4135.334	19.3 ± 0.25	R	K.H.	Lelekovice 0.35 m
4141.362	19.2 ± 0.3	R	K.H.	Lelekovice 0.35 m
4146.295	19.4 ± 0.25	R	K.H.	Lelekovice 0.35 m
4149.273	19.4 ± 0.25	R	K.H.	Lelekovice 0.35 m
4166.247	19.7 ± 0.2	R	K.H.	Ondřejov 0.65 m
4167.366	20.0 ± 0.35	R	K.H.	Ondřejov 0.65 m
4170.264	20.0 ± 0.3	R	K.H.	Ondřejov 0.65 m
4173.269	19.8 ± 0.25	R	P. Kušnirák, T. Henych	Ondřejov 0.65 m
4174.268	>19.7	R	K.H.	Lelekovice 0.35 m
4240.561	>20.0	R	K.H.	Ondřejov 0.65 m
4140.868	20.12 ± 0.05	B	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4140.872	20.54 ± 0.06	V	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m
4140.856	19.22 ± 0.03	I	M.F.B., M.J.D.,A.W.S., K.A.M.	LT 2.0 m
4140.862	20.22 ± 0.08	i	M.F.B., M.J.D., A.W.S., K.A.M.	LT 2.0 m

Download table as:  ASCIITypeset image

Table 7. Photometric Observations of M31N 2007-08d

HJD	Mag	Filter	Observer(s)	Telescope
(2,450,000+)
4380.401	18.6 ± 0.2	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4380.424	18.5 ± 0.15	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4382.235	19.0 ± 0.25	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4387.219	>19.5	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4387.231	19.5 ± 0.3	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4387.561	19.6 ± 0.2	R	K.H.	Ondřejov 0.65 m
4388.227	19.8 ± 0.25	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4388.626	19.6 ± 0.35	R	K.H.	Ondřejov 0.65 m
4389.233	19.8 ± 0.3	R	K.H., P. Kušnirák	Ondřejov 0.65 m
4389.646	20.0 ± 0.3	R	K.H.	Ondřejov 0.65 m

Download table as:  ASCIITypeset image

2.4. Optical Spectroscopic Observations

In addition to our ground-based photometric observations, we also attempted to obtain optical spectroscopic observations of each nova using the Low-Resolution Spectrograph (LRS; Hill et al. 1998) on the Hobby–Eberly Telescope (HET). We used either the g1 grating with a 10 slit and the GG385 blocking filter covering 4150–11000 Å at a resolution of R ∼ 600 or the g2 grating with a 20 slit and the GG385 blocking filter, covering 4275–7250 Å at a resolution of R ∼ 650. When employing the lower-dispersion g1 grating, we limited our analysis to the 4150–9000 Å wavelength range where the effects of order overlap are minimal. The spectra were reduced using standard IRAF routines to flat-field the data and to optimally extract the spectra. A summary of all the HET observations is given in Table 8. Because the observations were made under a variety of atmospheric conditions with the stellar image typically overfilling the spectrograph slit, our data are not spectrophotometric. Thus, the data have been displayed on a relative flux scale.

Table 8. Summary of HET Spectroscopic Observations

Nova	R.A.	Decl.	UT Date	Exp.	Coverage	Weather
	(2000.0)	(2000.0)		(s)	(Å)
M31N 2006-09c	00 42 42.38	41 08 45.5	2006 Sep 24.18	1800	4275–7250	Spec
M31N 2006-10a	00 41 43.23	41 11 45.9	2006 Oct 30.31	1500	4275–7250	Spec
M31N 2006-10b	00 39 27.38	40 51 09.8	2006 Nov 23.24	1200	4275–7250	Phot
M31N 2006-11a	00 42 56.81	41 06 18.4	2006 Nov 28.23	1200	4275–7250	Spec
M31N 2007-08d	00 39 30.27	40 29 14.2	2007 Sep 13.44	1200	4275–7250	Spec
M31N 2007-10a	00 42 55.95	41 03 22.0	2007 Oct 19.13	1400	4275–7250	Phot
M31N 2007-11d	00 44 54.60	41 37 40.0	2007 Dec 04.22	1200	4100–8900	Phot
M31N 2007-11e	00 45 47.74	42 02 03.5	2007 Dec 05.23	1200	4100–8900	Phot

Download table as:  ASCIITypeset image

Spectra were successfully obtained for 8 of the 10 novae observed by Spitzer. These data, which were obtained primarily to ascertain the spectroscopic classes (Williams 1992) and ejection velocities of the novae, are shown in Figures 7–9 and are summarized in Tables 9 and 10. Spectroscopic classes for the two novae that we were unable to observe, M31N 2007-07f and 2007-08a, were available through spectra obtained by Quimby (2007) and Barsukova et al. (2007), respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 9. Balmer Emission Line Properties

Nova	EW (Å)		FWHM (km s−1)
	Hβ	Hα	Hβ	Hα
M31N 2006-09c	−127	−470	1910	1920
M31N 2006-10a	−44	−90	950	810
M31N 2006-10b	−133	−2130	3030	3562
M31N 2006-11a	−35	−57	1420	1120
M31N 2007-07fa	−56	−195	1340	1110
M31N 2007-08a	⋅⋅⋅	−66	⋅⋅⋅	1880
M31N 2007-08d	−68	−284	1160	1180
M31N 2007-10a	−37	−154	470	500
M31N 2007-11d	−297	−1223	2060	2260
M31N 2007-11e	−137	−306	1750	1600

Note. ^aMeasurements are from a spectrum kindly provided by R. Quimby (2009, private communication).

Download table as:  ASCIITypeset image

Table 10. Spatial Positions and Spectroscopic Types

Nova	Δαcos δ	Δδ	a	Type	References
	(')	(')	(')
M31N 2006-09c	−0.37	−7.39	11.61	Fe ii	1,2
M31N 2006-10a	−11.50	−4.36	17.31	Fe ii	1
M31N 2006-10b	−37.25	−24.81	63.14	He/N	1
M31N 2006-11a	2.35	−9.84	20.46	Fe ii	1
M31N 2007-07f	−45.77	−22.95	81.74	Fe ii	3
M31N 2007-08a	−20.78	−22.25	30.49	Fe ii?a	4
M31N 2007-08d	−36.91	−46.74	59.56	Fe ii	1
M31N 2007-10a	2.19	−12.78	27.57	Fe ii?b	1, 5
M31N 2007-11d	24.34	21.60	34.03	Fe ii	1, 6, 7
M31N 2007-11ec	34.06	46.07	57.52	Fe ii	1, 8

Notes. ^aPossible hybrid nova. ^bPossible peculiar He/N nova with narrow He i lines. ^cNeon nova. References. (1) This work; (2) Shafter et al. 2006; (3) Quimby 2007; (4) Barsukova et al. 2007; (5) Gal-Yam & Quimby 2007; (6) Quimby et al. 2007; (7) Shafter et al. 2009; (8) Di Mille et al. 2007.

Download table as:  ASCIITypeset image

Seven of the 10 novae in our sample clearly belong to the Fe ii class, which is characterized by relatively narrow Balmer and Fe ii emission. Of the remaining three, M31N 2006-10b was a "hybrid" nova (a broad-lined Fe ii system that later evolved into a He/N system), while M31N 2007-08a was described by Barsukova et al. (2007) as an Fe ii (or possibly a hybrid) nova. The large Balmer emission line widths seen in M31N 2006-10b are typical of what is found in He/N novae; however, the relatively narrow lines exhibited by 2007-08a are more characteristic of the Fe ii class. The remaining nova, M31N 2007-10a, is quite peculiar spectroscopically (see Figure 8). The spectrum displays narrow emission characteristic of the Fe ii novae, but in addition to Balmer emission, shows prominent He i lines rather than Fe ii emission.

The spatial positions of the 10 novae in our Spitzer sample are shown in Figure 10. As has been demonstrated by Shafter (2007) from a larger sample of novae, there is no evidence that the spatial distribution of Fe ii and He/N novae differ in M31 (see also A. W. Shafter et al. 2011, in preparation).

Zoom In Zoom Out Reset image size

Of the eight novae observed by IRAC, only M31N 2006-10a shows clear evidence of an infrared excess in the IRAC bands, peaking at λ ∼ 4 μm (see Figure 2). M31N 2007-07f shows evidence for a rather weaker, slightly longer wavelength excess (see Figure 1). We may associate such excess emission with that from dust grains condensing in the ejecta of each nova, as has been observed in many Galactic novae—see Section 1. Both M31N 2006-10a and 2007-07f belong to the Fe ii spectroscopic class and both can be considered relatively "slow" novae with decay times of t₂(V) = 83 and t₂(R)∼ 50 days, respectively.

To further explore the relationship between dust formation timescales and nova speed class, we have augmented our M31 sample with available data on Galactic dust-forming novae. A summary is presented in Table 11. In Figure 11, we have plotted the condensation time, t_cond, of dust grains in Galactic novae against t₂. Dust condensation times for this plot have been taken from Table 13.1 of Evans & Rawlings (2008) and times of dust breaks in D-type novae from Strope et al. (2010). In general, these are consistent within a few days, but we have taken the earlier quoted time in each case, if they are both available for a given nova. Speed class, in terms of t₂, has been taken from Strope et al. (2010). We have also plotted the two suspected dusty M31 novae on Figure 11, where t_cond is likely to be an upper limit in both instances (i.e., we have not necessarily observed the onset of the dust formation event). Note that for M31N 2006-10a and 2007-11d, we have used t₂(V) but as this was unavailable for the other M31 novae, we have used t₂(R).

Zoom In Zoom Out Reset image size

There is a clear trend for the Galactic novae in the sense that the faster novae tend to form grains earlier than the slow novae. An outlier is nova PW Vul. Gehrz et al. (1988) note that this nova had a very erratic early optical light curve which resembled DQ Her, but showed no deep minimum as did the latter, archetypal dust-forming nova. Indeed PW Vul's dust shell was also low mass and optically thin and the quoted value of t_cond = 154 days was calculated by Gehrz et al. (1988) from the angular expansion rate of the pseudo-photosphere and the maximum flux of the outburst. This is consistent with the time of first condensation being in the observational gap between 100 and 290 days post-outburst. Also shown as a lower limit on this plot is the dust condensation time for QU Vul, given by Evans & Rawlings (2008) as 40–200 days (we will return to this object below). Marking an extreme point on the plot is nova V445 Pup. A dust extinction dip is clearly seen in its light curve and (relatively sparse) infrared photometry thereafter confirms the presence of an extensive dust shell, but this is an example of a very rare helium nova outburst (see Woudt et al. 2009, and references therein).

The apparent correlation between t_cond and t₂ is perhaps surprising. If we assume a constant central source luminosity, L_*, and outflow velocity, v_ej, for a given nova, and that the grain condensation temperature, T_cond, is constant from nova to nova, we can write (e.g., Gehrz 2008)

$$\begin{equation} t_{\rm cond} \propto L_*^{0.5} v_{\rm ej}^{-1}. \end{equation} \tag{ 1 }$$

Adopting available empirical relations between v_ej and t₂, and between L_* and t₂ (the maximum magnitude versus rate of decline relation) such as those given by Warner (2008), we find that v_ej ∝ t^−0.5₂ and L ∝ t⁻¹₂. After inserting these relations into Equation (1), it appears that t_cond should be essentially independent of nova speed class! Clearly, the empirical correlation shown in Figure 11 justifies further investigation, but as pointed out by Evans & Rawlings (2008), dust grain nucleation and growth rely on a complex set of parameters. Finally, it should be noted that not all novae form detectable amounts of dust, with examples at both ends of the range of speed class (e.g., V1500 Cyg versus HR Del; Bode & Evans 1989). The reasons for this have been explored by various authors (e.g., Gallagher 1977; Bode & Evans 1983) but are not fully understood.

Figure 12 shows a plot of nova speed class versus the time of maximum infrared flux, t_IRmax, as observed in Galactic novae. Since t_IRmax is not available for our M31 novae, we have instead plotted the speed class and time after discovery (approximate optical maximum) for the novae observed with IRAC. Both M31N 2006-10a and 2007-07f are marked on the plot. Again there appears to be a trend of increasing t_IRmax with speed class for the Galactic novae, with the exception of QU Vul. We may note however that this is the time of the maximum flux observed in the 10 μm silicate feature in this nova by Gehrz et al. (1986) at t = 240 days, and not necessarily the time of maximum continuum emission from grains in the expanding shell, which may have been (much) earlier. Indeed, an observation on day 206 for this nova by Smith et al. (1995) showed that at this time emission from graphitic dust clearly dominated the 10 μm continuum in QU Vul.

Zoom In Zoom Out Reset image size

From Figure 12, it appears that M31N 2006-10a may have been observed by Spitzer nearer the time of the infrared emission maximum than 2007-07f, which again is consistent with the higher dust shell luminosity and inferred dust temperature in 2006-10a compared to 2007-07f. Indeed, M31 2006-10a might have been observed not too long after t_cond at around the time of t_IRmax. Evidence for this comes from its position on Figures 11 and 12, coupled with T_BB (∼720 ± 100 K) being much lower than T_cond (∼1200 K). There is in fact a steep drop in observed effective dust grain temperature in well-observed dusty novae coinciding approximately with infrared maximum arising from what Bode & Evans (1983) termed the "infrared pseudo-photosphere" in the optically thick (in the infrared) dust shells of prolific dust formers such as NQ Vul (Ney & Hatfield 1978). As the dust formation rate reduces in the expanding shell, the observed optical depth also reduces and T_BB increases again. Mitchell et al. (1983) found even better agreement with observations if grains were subject to some size reduction around the time of infrared maximum, possibly due to sputtering. In NQ Vul, the temperature minimum lasted from around 62 to 130 days post-outburst.

All but one of the faster M31 novae lie above the region occupied by the Galactic novae at infrared maximum (the exception being M31N 2006-11a). The non-detection of dust emission in these objects may then be due to them either being observed well after dust emission maximum, or the fact that they did not produce large amounts of dust in the first place. The latter appears to be the case for 2006-11a at least, as it does lie in the region of the plot occupied by the dusty Galactic novae.

We may estimate the total mass of dust in M31N 2006-10a, M_d, assuming an isothermal, optically thin dust shell of uniform spherical grains of radius a from the equation

$$\begin{equation} M_{\rm d} \approx a \rho _{\rm g} d^2 {F_\lambda } \frac{B(\lambda, T_{\rm g})}{Q_{\rm abs}(\lambda, a)}, \end{equation} \tag{ 2 }$$

where d is the distance, ρ_g is the bulk density of the grain material, F_λ is the observed flux density at wavelength λ, B is the Planck function at λ and grain temperature T_g, and Q_abs is the grain absorption coefficient. If Q_abs ∝ λ^−α for small dielectric absorbers, it can be shown that

$$\begin{equation} T_{\rm g} =\frac{2890}{\lambda _{\rm max}}\Big (\frac{5}{\alpha + 5}\Big), \end{equation} \tag{ 3 }$$

where λ_max is the wavelength of maximum emission in μm.

In the case of M31N 2006-10a, the IRS spectrum in Figure 2 shows no apparent 10 μm silicate feature and the emitting grains are more likely carbon based, as seen at some point in the evolution of most dust-forming novae (Evans & Rawlings 2008; Gehrz 2008). Taking for simplicity graphite spheres of size small compared with the wavelength of emission, then α ≈ 2 (Evans 1994). Thus, with λ_max = 4 μm, T_g = 516 K, and if a = 0.1 μm, then Q_abs = 4.72 × 10⁻² (Draine 1985, although this neglects any temperature dependence of the absorption coefficient). Taking ρ_g = 2000 kg m⁻³ (Love et al. 1992) and a distance to M31 of 780 kpc (Holland 1998; Stanek & Garnavich 1998), M_d ∼ 2 × 10⁻⁶ M_☉. This is consistent with the range of dust masses derived for Galactic novae, particularly the more prolific dust formers (see, e.g., Gehrz 2008).

We have conducted the first infrared survey of novae in the nearby spiral, M31, using the Spitzer Space Telescope. The primary motivation behind the survey was to determine the feasibility of using Spitzer observations of M31 novae to study dust formation property as a function of nova speed class and spectroscopic type. We observed a total of 10 novae in M31 with Spitzer over a two-year period: M31N 2006-09c, 2006-10a, 2006-10b, 2006-11a, 2007-07f, 2007-08a, 2007-08d, 2007-10a, 2007-11d, and 2007-11e. Eight of these novae were observed with IRAC (all but M31N 2007-11d and 2007-11e) and eight with the IRS (all but M31N 2007-07f and 2007-08a), with six novae observed with both instruments.

Observations of Galactic novae show that dust formation typically occurs with a timescale, t_cond, of between ∼1 and ∼5 months post-eruption (mean ∼2 months), depending on the speed class of the nova. For a typical nova, the peak infrared signature occurs shortly thereafter, with the time to infrared maximum, t_IRmax, averaging about three months post-eruption. Thus, our observing strategy was to schedule our Spitzer observations approximately three months post-discovery when the infrared signature due to dust formation was expected to reach a maximum. Unfortunately, the constraints imposed by the Spitzer scheduling process did not allow us to time our observations as precisely as we would have liked, and our observations occurred anywhere between ∼3 and ∼7 months post-eruption. Our principal conclusions can be summarized as follows.

• 1.  
  We were able to detect six of the eight novae observed with IRAC. Of these, only M31N 2006-10a showed clear evidence for an infrared excess peaking at λ ∼ 4 μm. The IRS spectrum of this nova showed no evidence of silicate emission features and thus we assume that the dust was carbon based in this case. We were then able to estimate the total mass of dust formed to be M_d ∼ 2 × 10⁻⁶ M_☉. This is comparable to the mass of dust found in Galactic novae forming the more extensive dust shells. Another nova, M31N 2007-07f, showed evidence of possible dust formation through a weaker infrared excess detected to peak at longer wavelength. Our observations of these two novae occurred 116 and 203 days post-discovery, respectively. Thus, it is plausible that our observations of M31N 2007-07f occurred after the peak IR flux was achieved as suggested by our comparison with the times of infrared maximum in Galactic dust-forming novae. Both novae are Fe ii systems with relatively slow light curve evolution: t₂(V) = 82 days and t₂(R) ∼ 50 days, respectively.
• 2.  
  We find a surprising correlation between the condensation time for dust grains (t_cond) and nova speed class (t₂) for Galactic novae (see Figure 11). Although upper limits, the values of t_cond for the two M31 novae with detected IR excesses are consistent with this correlation. Most of the M31 novae in our sample, which have t₂ ≲ 50, were likely observed too long after dust condensation for us to detect an IR excess in our data. One exception is M31N 2006-11a, which is a moderate-speed-class, Fe ii nova observed less than three months after discovery. A comparison with Galactic dust-forming novae (see Figure 12) shows that an IR excess likely would have been visible assuming dust formation had taken place.
• 3.  
  Three of the eight novae observed with the IRS were detected: M31N 2006-09c and 2006-10a showed only continuum emission, while 2007-11e revealed a [Ne ii] 12.8 μm emission feature characteristic of the class of "neon novae." Other than its neon emission, M31N 2007-11e, is fairly unremarkable in other respects, being neither a particularly fast or slow Fe ii system.

Our preliminary survey has demonstrated that IR observations can be used to detect dust formation in M31 novae. It is likely that those novae that went undetected were observed too long after maximum light when the dust had cooled sufficiently to render the IR flux below our limit of detection. Future observations, possibly from the ground, of a larger sample of novae will be required to further characterize the relationship between nova speed class, spectroscopic type, and dust formation timescales. The correlations we find between speed class and t_cond and t_IRmax will greatly assist observational planning.

The work presented here is based in part on observations obtained with the Hobby–Eberly Telescope, which is operated by McDonald Observatory on behalf of the University of Texas at Austin, the Pennsylvania State University, Stanford University, the Ludwig-Maximillians-Universitaet, Munich, and the George-August-Universitaet, Goettingen. Public Access time is available on the Hobby–Eberly Telescope through an agreement with the National Science Foundation. The Liverpool Telescope is operated on the island of La Palma by Liverpool John Moores University in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias with financial support from the UK Science and Technology Facilities Council (STFC). K.H. is grateful for obtaining and providing of M31 images to P. Kušnirák, M. Wolf, P. Zasche, P. Cagaš, and T. Henych. A.W.S. is grateful to the NSF for support through grant AST-0607682 and to the University of Victoria for hospitality during his sabbatical leave while this work was being completed.