MID-INFRARED SPECTRAL VARIABILITY ATLAS OF YOUNG STELLAR OBJECTS^*

At the early phases of their formation, stars are intimately linked to their circumstellar environment. Dust particles in the circumstellar disk and envelope are heated both internally by viscous energy released in the midplane of the accretion disk, and externally by irradiation from the central star. The absorbed energy is re-radiated as thermal emission at infrared (IR) wavelengths. The shape of the IR spectral energy distribution (SED) reflects the temperature and density structure of the circumstellar environment. Of particular importance is the mid-IR domain (3–25 μm), where the disk emission originates from the region where planets are formed. The same domain contains spectral features of polycyclic aromatic hydrocarbon (PAH) particles, small silicate grains, and molecular ices. PAH features give information about the UV radiation field and ionization balance in the circumstellar gas (Tielens 2008); the shape and amplitude of the 10 μm silicate feature is partly related to the degree of crystallinity of the silicate grains, indicating how far these grains evolved from the interstellar medium (Henning 2010), while ice features are used to study the organic chemistry and thermal history of the circumstellar material (van Dishoeck 2004).

Although optical and near-IR variability is a well-known property of young stellar objects (YSOs), a growing number of recent studies claim that a considerable fraction of them also exhibit mid-IR photometric changes (e.g., Barsony et al. 2005; Morales-Calderón et al. 2009, 2011). In young eruptive stars and some T Tauri stars, changing accretion is thought to be responsible for mid-IR variability (e.g., Ábrahám et al. 2004; Kóspál et al. 2007; van Boekel et al. 2010). There are claims that in some young eruptive stars the temporarily increased heat leads to structural changes in the innermost part of the system, resulting in decreased extinction and consequently in IR brightness changes (see, e.g., V1647 Ori in Muzerolle et al. 2005b, or PV Cep in Kun et al. 2011). In some intermediate-mass young stars, structural changes in the inner disk were invoked to explain their mid-IR flux changes (Juhász et al. 2007; Sitko et al. 2008). A similar reason, i.e., changing inner disk height, may also explain the variability of those low-mass objects whose disks contain holes and gaps (Muzerolle et al. 2009; Espaillat et al. 2011). These indications of dynamical processes suggest that disks are less steady-state than thought before.

The wavelength dependence of the mid-IR variability carries information on the disk and envelope structure because the emission at different wavelengths corresponds to disk regions with different temperatures and thus to different radial distances from the star. An interesting aspect is the exploration of possible changes in the strength and profile of IR spectral features. Several studies published so far concentrate on the variability of the 10 μm silicate feature. Time-dependent shadowing of disk areas by the inner gas disk could explain the spectral changes in the Herbig stars HD 31648 and HD 163296 (Sitko et al. 2008). In the highly accreting T Tauri stars DG Tau and XZ Tau (Bary et al. 2009), self-absorption may play a role in the variability of the 10 μm feature. Changes in the mineralogical composition of the emitting dust particles were also observed by Ábrahám et al. (2009) in the most recent outburst of EX Lupi, where they detected the appearance of crystalline silicates that were not present in quiescence. However, more detailed investigations are limited by the lack of multi-epoch mid-IR spectroscopic databases.

The timescale of mid-IR variability is related to the timescale on which the heating of the emitting region can vary and on which the dust particles in the disk can react to changes in the energy input. For example, almost instantaneous (seconds to minutes) responses to fluctuations in the stellar illumination may indicate emission from an optically thin medium, while longer timescales (years) may indicate an optically thick medium (Chiang & Goldreich 1997). Dynamical changes leading to rearrangement of the density structure can occur on the timescale of an orbital period. Thus, the timescale of the variability is an essential input for understanding the physics behind the flux changes. Recent studies carried out with the Spitzer Space Telescope sampled MIR variability on several different timescales. While hourly variations are rare and small in amplitude (Morales-Calderón et al. 2009), variability on daily, weekly, and yearly timescales is ubiquitous, and amplitudes range from a few tenths of magnitudes to >1 mag (Morales-Calderón et al. 2009, 2011; Luhman et al. 2010; Rebull 2012). These timescales, if caused by changes in the disk structure, indicate that the variability is connected to the innermost 0.1–1 AU of the disk. With Spitzer alone, however, one cannot investigate whether variability on a longer timescale exists. Such data would provide information on dynamical changes at larger radial distances from the star, typically several AUs. Such a study can only be performed by comparing data from two IR space missions separated in time by several years.

Motivated by the above-mentioned considerations, here we present a new atlas of 2.5–11.6 μm low-resolution spectra obtained with the ISOPHOT-S instrument on board the Infrared Space Observatory (ISO) between 1996 and 1998, as well as 5.2–14.5 μm low-resolution spectra obtained with the Infrared Spectrograph (IRS) instrument on board the Spitzer Space Telescope between 2004 and 2007. The overlapping wavelength range and different epochs of ISO and Spitzer make it possible to conduct a systematic mid-IR spectral variability study for a large sample of well-known YSOs both on an annual timescale (less than a few years, i.e., within the cryogenic lifetime of either telescope), and of nearly a decade (by comparing the results of the two telescopes). Although the beams of the two instruments significantly differ, beam confusion is usually not an issue, because at mid-IR wavelengths, the size of the emitting region (warm dust in the inner few AUs) is smaller than in either the optical (scattered light) or the far-IR/millimeter (thermal emission from cold dust). We identify typical variability patterns and hypothesize about their possible physical origin. Our results on monitoring and interpreting variability may provide a powerful "extra dimension" of information on the structure of the circumstellar material. Our sample may also serve as a starting point to plan follow-up mid-IR monitoring campaigns.

2.1. Sample Selection

2.2. ISO Data Reduction

2.3. Spitzer Data Reduction

2.4. Comparison of the ISOPHOT-S and Spitzer/IRS Calibration

The absolute spectrophotometric calibration of ISOPHOT-S was based both on empirical spectral templates of standard stars derived by Cohen (2003), and on stellar atmosphere models provided by Hammersley & Jourdain de Muizon (2003) using infrared observations. Spitzer/IRS spectra, on the other hand, were calibrated using standard stars for which model atmospheres are available in Decin et al. (2004). Nevertheless, Cohen templates were used to verify the calibration of IRS (Houck et al. 2004). Thus, despite the differences in calibration strategy, we do not expect significant systematic discrepancies between the ISOPHOT-S and the IRS spectra. In order to check this, we reduced the spectra of HD 181597 (HR 7341), a K1 III-type star that was used as a primary low-resolution standard for Spitzer and was also used in the calibration of ISOPHOT-S. In Figure 1 we plotted the observations along with the template and model spectra, and we also calculated their ratio. The figure shows that the template and model spectra are close to each other, and the observations are within 10% (for ISOPHOT-S) or 5% (for Spitzer/IRS) of the template/model spectra. We emphasize that no wavelength-dependent trend can be seen in the ratios, suggesting that the calibration uncertainties would not affect the observed colors of the targets. Thus, we conclude that the accuracies of the absolute calibration mentioned in Sections 2.2 and 2.3 are plausible and there is no need to increase the error bars when comparing spectra obtained with the two different telescopes.

Zoom In Zoom Out Reset image size

In Table 1 we present the catalog of spectroscopic observations. The table contains the following columns.

Column 1. Name of the target as common in the literature.

Column 2. L: low-mass pre-main-sequence (PMS) stars (T Tauri-type stars and embedded YSOs); I: intermediate-mass PMS stars (Herbig Ae/Be stars). B, A, and F stars are considered intermediate-mass stars, while G, K, and M stars are considered low-mass stars. Spectral types or stellar masses for sources where SIMBAD does not provide spectral types were taken from Straižys et al. (2002) for IRAS 03260+3111, Myers et al. (1987) for BARN 5 IRS 1, Brinch et al. (2007) for LDN 1489 IRS, Lim & Takakuwa (2006) for LDN 1551 IRS 5, Sandell & Weintraub (2001) for Reipurth 50 N IRS 1, Simon et al. (1995) for WL 16, Bontemps et al. (2001) for WL 6, Hodapp et al. (1996) for OO Ser, and Pontoppidan et al. (2004) for [SVS76] Ser 4.

Column 3. Spectral types from the SIMBAD database.

Columns 4 and 5. J2000 coordinates of the source from the Two Micron All Sky Survey (2MASS) All-Sky Catalog of Point Sources (Cutri et al. 2003), except where indicated in the table. Since the peak of the optical and IR emission might not coincide, we decided to use coordinates from 2MASS instead of optical positions. These coordinates are precise enough to carry out the offset correction if the source was off-center in the ISOPHOT-S aperture or in the Spitzer/IRS slit.

Column 6. Unique eight-digit identifier of the on-source ISOPHOT-S observation in the ISO Data Archive (TDT number).

Column 7. Corrections applied to the ISOPHOT-S observations: M: memory correction; B: in the lack of dedicated OFF measurement, zodiacal background was predicted and subtracted; O: position offset correction applied. An exclamation mark signals spectra where the uncertainty of the correction is above 10%, thus they are not used for variability analysis (for details, see Appendices A and B).

Column 8. Date of the ISOPHOT-S observation.

Column 9. Unique identifier of the Spitzer/IRS observation in the Spitzer Data Archive (AOR).

Column 10. Corrections applied to the Spitzer/IRS observations; O: position offset correction applied.

Column 11. Date of the Spitzer/IRS observation.

Columns 12–14. Inventory of clearly visible spectral features. Si-O: 10 μm silicate feature; ices: H₂O 3.1 μm, CO₂ 4.27 μm, H₂O 6.0 μm, CH₃OH 6.8 μm, or CH₄ 7.7 μm; PAH: 3.3, 6.2, 7.7, 8.6, 11.2, or 12.7 μm bands of polycyclic aromatic hydrocarbons; em: feature in emission; abs: feature in absorption; ...–feature not present.

Column 15. Type of the spectral shape: PAH dom.: the spectrum is dominated by PAH emission features, no 10 μm silicate feature is discernible; sil. em.: the 10 μm silicate feature is in emission; sil. abs.: the 10 μm silicate or ice bands are visible in absorption; ...–undecided (for details, see Section 4).

Column 16. Variability: yes: candidate variable source; no: constant source; ...–undecided.

The table and the spectra are also available electronically.

In Figure 2 we present the ISOPHOT-S and Spitzer/IRS spectra for each object. Where clearly detected, we marked with vertical dashed lines the wavelengths of molecular ice bands (H₂O 3.1 μm, CO₂ 4.27 μm, H₂O 6.0 μm, CH₃OH 6.8 μm, CH₄ 7.7 μm) and marked with vertical dotted lines the wavelengths of PAH features (3.3, 6.2, 7.7, 8.6, 11.2, 12.7 μm). The figures show both cases where the plotted spectra agree, and sources where the measurements at different epochs seem to differ. Some of the differences can be explained by serious instrumental artifacts or source confusion resulting from the different beam sizes of the ISOPHOT-S and IRS instruments. We studied the literature of each object in order to establish whether it is part of a multiple stellar system and whether there is any indication for extended near-IR or mid-IR emission in the vicinity. These information are presented in Appendix B for the affected objects. Based on this, we could decide whether it is meaningful to compare the different spectra in Figure 2 to look for temporal variability. For certain objects, Appendix B also contains a nonexhaustive comparison with other mid-IR spectra published in the literature.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In order to explore the significance of the observed flux changes and check their wavelength dependence, we computed the ratio of spectra taken at different epochs. In Figure 3 we plotted these flux ratios for sources where spectra at more than one epoch were available and their comparison was meaningful (47 stars). Where several spectra existed, we took the two most extreme cases. In case the ratio of a Spitzer/IRS and an ISOPHOT-S spectrum is plotted, the former was first resampled to the latter's wavelength resolution. Error bars plotted in this figure represent the quadratic sum of the errors of each spectrum. Where the ratio of two Spitzer spectra is computed, instead of the 10% absolute calibration uncertainty, we used the 5% uncertainty of the repeatability (BF Ori and HD 98800). We also overplotted with vertical dotted/dashed lines the wavelengths of PAH/ice features for sources where these features are present.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For the purpose of further analyses, we classified our sources into three groups based on the shape of the mid-IR SED and the presence of various spectral features as follows.

• 1.  
  PAH-dominated sources. The spectrum is dominated by PAH bands; no apparent 10 μm silicate emission band is visible (13 objects).
• 2.  
  Silicate emission sources. The spectrum is dominated by the 10 μm silicate emission band, otherwise it is rather featureless (36 objects). In a few sources, weak PAH emission is present.
• 3.  
  Silicate absorption sources. The spectrum is rising toward longer wavelengths; it displays 10 μm silicate absorption and/or several ice absorption bands (16 objects).

Our classification is presented in Column 15 of Table 1, as well as in Figure 2 in parentheses under the name of each source. We could classify all sources except DG Tau, Ced 112 IRS 4, and CK 1, which display an apparent combination of silicate emission and absorption.

4.1. Statistics and Wavelength Dependence of Mid-IR Variability

4.2. Synthetic Photometry and Variability Timescales

The smooth trends of the flux ratio curves in Figure 3 suggest that the wavelength dependence of the variability can be fully described, with high signal-to-noise ratio, via integrating the spectra over a few wavelength intervals and looking at the changes of these integrals. Following this idea, we derived synthetic photometry from the spectra. For objects with silicate emission or absorption, we calculated synthetic photometry for each source by averaging the ratio curves in Figure 3 between 6 and 8 μm and between 8 and 11.5 μm, and converting them to magnitude differences. The former wavelength range is representative of the continuum, while the latter one covers the 10 μm silicate feature and is close to the usual N photometric band. For PAH dominated objects, we defined a "continuum band" by averaging ratios at 6.5–7 and 9.5–10.5 μm and a "PAH feature band" containing the remaining parts of the spectrum.

The synthetic magnitudes can be used to quantitatively assess a fundamental question: whether the characteristics (frequency and amplitude) of the variability are the same on annual timescales (ISOPHOT-S versus ISOPHOT-S or Spitzer versus Spitzer) and on decadal timescales (ISOPHOT-S versus Spitzer), or there is a hint for long-term variability trends (higher frequency or larger amplitude) not measurable by ISOPHOT-S alone or Spitzer alone. In Figure 4 we compared the distribution of magnitude changes in the annual and decadal cases for the three object types defined above. Histograms are plotted separately for the "continuum" and the "feature" magnitude changes. For PAH-dominated objects, no systematic difference can be seen between the short-term and the long-term distributions, either in the continuum or the feature (Figures 4(a) and (e)). For objects with silicate emission, we plotted separately the intermediate-mass and the low-mass stars, but even with this separation, no significant differences can be seen between the two timescales (Figures 4(b), (c), (f), and (g)). The only weak hint for such long-term trends might be seen for silicate absorption sources, where both in the continuum and the feature, the decadal changes seem to be somewhat more common and exhibit somewhat larger magnitude differences than on annual scale (Figures 4(d) and (h)). Although the number statistics are limited, we can conclude that the characteristic timescales of the mid-IR variability we are observing are less than or equal to a few years, and in the following we do not differentiate between annual and decadal changes when studying the physical reasons for the variability.

Zoom In Zoom Out Reset image size

4.3. Trends in the Mid-IR Flux Changes

In Figure 5 we plot the obtained synthetic magnitude changes. As in Figure 3, we always considered the two most extreme spectra. Sources with constant flux in the whole 6–11.5 μm wavelength range (Figure 3) are situated close to the (0,0) point. Most of them fall into an area within 0.1 mag, indicated by a dark gray circle. Moreover, we checked the synthetic photometry for the calibration star HD 181597, which was observed by both ISOPHOT-S and Spitzer/IRS (Figure 1), and found Δmag(6–8 μm) = 0.067 mag, Δmag(8–11.5 μm) = 0.076 mag. For these reasons, we consider the value of 0.1 mag as a practical threshold between variable and constant sources. Objects with a wavelength-independent flux change (either brightening or fading) are distributed along the 45°diagonal line (for guiding the eye, this line, broadened by a ±0.1 mag typical uncertainty, is plotted with light gray). A similar, but vertical stripe is plotted for objects with silicate emission, marking the location of those sources where the continuum part was constant, but the 10 μm silicate emission feature varied significantly.

Zoom In Zoom Out Reset image size

Several trends in the distribution of the points can be recognized in Figure 5.

• 1.  
  Objects whose spectra are dominated by PAH emission are plotted in the upper left panel of Figure 5. Out of these eight objects, five are marked as constant in Column 15 of Table 1. These are situated within 0.1 mag of the (0,0) point. The remaining three sources (RR Tau, CU Cha, and HD 135344 B) are distributed along the diagonal stripe with a magnitude difference of 0.2–0.3 mag both in the continuum and in the PAH feature band. We suggest that these objects show real physical variability (RR Tau exhibited flux changes of 0.5 mag also between 2.5 and 5 μm, see Figure 2). The variability in the 6–11.5 μm range seems to be wavelength-independent, as indicated by the distribution of points along the diagonal stripe. This means that for each source the variation in the PAH features and in the continuum has similar amplitude.
• 2.  
  Stars with silicate emission are plotted in the upper right (Herbig stars) and lower left (T Tauri stars) panels of Figure 5. They exhibit the largest diversity in variability patterns within the whole sample. Three Herbig stars (HD 95881, HD 142527, and AK Sco) and no T Tauri stars are constant. Eleven stars show wavelength-independent flux changes of ⩾0.1 mag, located along the diagonal stripe (Herbig stars: BF Ori, HD 104237, HD 144432, VV Ser, WW Vul; T Tauri stars: DR Tau, VZ Cha, WX Cha, WW Cha, EX Lup, S CrA). The amplitude can be as high as 0.7 mag. Remarkably, another 11 sources change their brightness primarily in the 8–11.5 μm band (amplitude is ⩽0.6 mag) while at shorter wavelengths, they are less variable or constant (Herbig stars: HD 98800, HD 139614, HD 142666, 51 Oph; T Tauri stars: UZ Tau, VY Tau, CR Cha, CT Cha, VW Cha, Glass I, CV Cha). Interestingly, one T Tauri star does not follow any of the above-described trends. XX Cha became brighter at shorter wavelengths and fainter at longer wavelengths, exhibiting an anti-correlation between the two synthetic photometric bands. Although the amplitude of the variability of this source is small, the phenomenon is probably real, since it is the slope of the mid-IR SED that changed, rather than the more uncertain absolute flux level only.
• 3.  
  Stars with silicate and/or ice absorption (lower right panel in Figure 5) show frequent variability (only one source out of 10, Ced 111 IRS 5, is constant). The observed variations mostly follow the diagonal stripe, indicating wavelength-independent flux changes. The amplitude of the variability is at most 0.4 mag. Slight deviations from this trend can be seen in those objects where the 10 μm silicate absorption feature is very deep, thus, their 8–11.5 μm flux is close to zero and remains essentially constant (e.g., Haro 6-10, LDN 1551 IRS 5, OO Ser).

There are objects where the detected flux changes are only marginal (≳0.1 mag), and their association with the above-mentioned trends may not be unambiguous. However, for each trend, there are clear examples with magnitude changes >0.3 mag, which proves the existence of these trends. The association of a certain source with a variability trend might depend on which measurements are compared. For example, in the case of WW Cha, we have one ISOPHOT-S and three Spitzer/IRS spectra. The comparison of the Spitzer spectra indicates wavelength-independent changes, while the ratio of the ISOPHOT-S spectrum and the Spitzer/IRS spectrum obtained on 2006 March 7 provides a clear example where the flux variations above and below ≈9 μm anti-correlate, similarly to XX Cha. VY Tau, BF Ori, CR Cha, and Glass I are further examples for cases where certain spectral ratios show wavelength-independent flux changes while others indicate changes only in the silicate feature or hints for anti-correlation between the continuum and the feature. This suggests that the trend of variability may not be a unique characteristic of a given source, but may change in time.

For 15 sources where two or more ISOPHOT-S spectra were available, we could also study variability in the 2.5–4.9 μm wavelength range. We found that five sources (UZ Tau, VY Tau, HD 97300, CV Cha, HD 135344 B) are constant between 2.5 and 4.9 μm. Out of these, UZ Tau, VY Tau, and CV Cha are slightly variable at longer wavelengths, while HD 98300 and HD 135344 B are constant throughout the whole 2.5–11.6 μm domain. The two variable PAH-dominated objects in this sample (RR Tau and CU Cha) have larger flux changes at shorter wavelengths than at λ > 5 μm (0.2–0.5 mag). A number of silicate emission objects show variability amplitudes comparable to those at longer wavelengths (DR Tau, Glass I, HD 104237, and EX Lup). One silicate emission object, however, displays anti-correlation between the shorter and longer wavelength regimes (CR Cha). This behavior is similar to what we observed for WW Cha and XX Cha at longer wavelengths, except that the constant pivot point is at a somewhat shorter wavelength. The three variable silicate absorption objects (Haro 6-10, HH 100 IRS, and OO Ser) display slightly higher variability amplitude in the 2–5 μm range than at longer wavelengths.

In the previous sections we investigated the characteristics of variability for a sample of well-known YSOs. Based on the magnitude changes plotted in Figures 4 and 5, out of 47 objects where multi-epoch observations were available, we found 37 stars (79%) to vary ⩾0.1 mag and 20 stars (43%) to be variable at a ⩾0.3 mag level in either the continuum or the feature. Our results can be compared with those of earlier mid-infrared variability studies carried out on annual (typically 1–4 years) timescales. Barsony et al. (2005) surveyed the ρ Oph cloud and found that at least ∼20% of the YSOs are variable at 10 μm on a few years timescale. Based on multi-epoch Spitzer/IRAC photometry, Luhman et al. (2010) reported that 44% of Class 0/I/II sources in the Taurus, and 26% in the Chamaeleon I star-forming regions exhibit at least 0.05 mag flux changes. Recently, from the analysis of multi-epoch Spitzer/IRS spectra, Espaillat et al. (2011) concluded that 86% of the transitional and pre-transitional disks in the Taurus and Chamaeleon star-forming regions display larger than 10% flux variability in the 5–38 μm wavelength range. Our results confirm the high incidence rates found in these studies and demonstrate that in our—rather heterogeneous—sample of young objects, annual/decadal mid-IR variability is ubiquitous.

Our data set also offers a possibility to explore the characteristic timescales of the processes responsible for the observed variability. These timescales could place constraints on the nature of the underlying physics. In Section 4.2 and Figure 4 we statistically proved that the distribution of the variability amplitudes on both annual and decadal timescales do not differ significantly. This new result implies that on decadal timescales we observe the same variability process as on shorter timescales, and no extra long-term physical mechanism is evident from our data. This fact constrains the physical timescale of the variability to less than a few years. There are hints that this characteristic timescale may be even shorter. In our Spitzer/IRAC monitoring program (PID 60167, PI: P. Ábrahám), we obtained 14 day light curves at 3.6 and 4.5 μm of 38 selected low- and intermediate-mass YSOs and found that practically all targets were variable with amplitudes from a few tens of millimagnitudes to 0.4 mag. Morales-Calderón et al. (2009) observed 69 YSOs in the IC 1396A dark globule between 3.6 and 8.0 μm with Spitzer/IRAC. Their 14 day light curves indicated that more than half of the observed YSOs show daily variations, with amplitudes ranging from 0.05 mag to 0.2 mag. Finally, the initial results of the Spitzer YSOVAR program (Rebull 2012; Morales-Calderón et al. 2011), based on the analysis of 40 day light curves obtained for sources in the Orion Nebula Cluster and other well-known star-forming regions, indicate that 70% of the YSOs with IR excess are variable with timescales of days to weeks and with amplitudes up to higher than 1 mag. These papers suggest that the incidence and the amplitude of flux changes we found on the decadal timescale already seem to be present on the much shorter timescale of a few weeks. If confirmed, this result would strongly constrain the characteristic timescale of the physical variability process, possibly pointing to the innermost part of the system as the region of the true cause of variability.

In order to examine flux changes, we integrated our spectra in relatively broad synthetic photometric bands. Consequently, the amplitude of variability within limited wavelength ranges can be even higher. Moreover, since our conclusions are based on the comparison of only two or three different spectra, the incidence numbers are lower limits, and the fraction of mid-IR variable stars may very well be larger. A dedicated mid-IR monitoring program would probably reveal variability for many of those objects that we observed to be constant. The amplitude of variability might also turn out to be larger if more epochs are taken into account. Examples are HD 104237 in Figure 6. HD 142527 in Figure 7, and HD 169142 in Figure 8, where we plotted additional mid-infrared spectra from Sylvester et al. (1996) and van Boekel et al. (2005). Based on these results one might predict that a very high fraction of all low- and intermediate-mass young stars are variable at a certain level in the mid-IR wavelength regime. Nevertheless, our data also revealed significant differences in the incidence of variability among objects with different SED types. As an example, Figures 4 and 5 clearly show that the fraction of variable stars is lower among the PAH-dominated objects than among the T Tauri stars. In the following we discuss the possible physical reason for the variability of each object type separately.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

5.1. PAH-dominated Objects

5.2. Objects with Silicate Emission

5.3. Objects with Silicate/Ice Absorption

In this paper we present low-resolution mid-IR spectra of 68 low- and intermediate-mass young stars obtained with ISOPHOT-S between 1996 and 1998 and with Spitzer/IRS between 2004 and 2007. Utilizing self-developed software packages, we interactively re-processed the spectra for improved removal of instrumental artifacts. By comparing multi-epoch spectra of each object, we analyzed the statistics and the trends of mid-IR spectral variability and its implications on the geometry of the circumstellar material. Our study has three novel aspects: (1) the spectroscopic observations enable the study of spectral variations in a relatively extended mid-IR wavelength range, (2) instead of focusing on a certain star-forming region, individual well-known YSOs were observed, and (3) the timescale of our ISO/Spitzer comparison is longer than previous mid-IR variability studies, covering a full decade. Our main achievements are the following.

• 1.  
  The mid-infrared spectral atlas constructed for this project presents an improved, final data reduction of ISOPHOT-S spectra. The atlas contains all low- and intermediate-mass stars ever observed with the ISOPHOT-S instrument, thus, it constitutes the legacy of the ISO.
• 2.  
  Our spectra show that mid-IR variability among low- and intermediate-mass YSOs is ubiquitous. We calculated synthetic photometry in the 6–8 μm and the 8–11.5 μm wavelength range, and found that 79% of the sources vary more than 0.1 mag, while 43% is variable above the 0.3 mag level. A comparison of the variability characteristics on annual and decadal timescales revealed no significant differences, constraining the physical timescale of the variability to less than a few years.
• 3.  
  For intermediate-mass stars with spectra dominated by PAH emission, we found a relatively low incidence of variability. Flux changes in this group are mostly wavelength-independent, and can be interpreted in terms of nonsteady irradiation of the disk due to fluctuating accretion. We propose a simple flared disk geometry to model these sources.
• 4.  
  Intermediate-mass stars exhibiting silicate emission at 10 μm often show higher variability amplitude in the silicate feature than in the adjacent continuum. Shadowing the disk by vertical variations of a puffed-up inner disk rim is invoked to explain the wavelength dependence of the flux changes. The deduced geometry is a modestly flared or flat disk.
• 5.  
  T Tauri stars are the most frequent variables in our sample. For those that exhibit more pronounced variability in the 10 μm silicate feature than in the continuum we propose considering a new model: strong magnetohydrodynamical turbulence may intermittently lift clouds of small grains into the disk atmosphere resulting in extra silicate emission (Turner et al. 2010). Nevertheless, based on our spectra alone, we cannot exclude the possibility that other scenarios may also be applicable.
• 6.  
  Sources exhibiting silicate or ice absorption are objects embedded in a dense envelope. They typically show wavelength-independent flux changes, probably due to varying accretion rate.

Our results suggest that mid-IR variability is widespread among YSOs. Interpreting the amplitude, wavelength dependence, and timescales of these flux changes is a new and promising possibility to explore the structure of the circumstellar disks and their dynamical processes.

We thank the anonymous referee for providing a thorough and very helpful report. The ISOPHOT data presented in this paper were reduced using the ISOPHOT Interactive Analysis package PIA, which is a joint development by the ESA Astrophysics Division and the ISOPHOT Consortium, lead by the Max-Planck-Institut für Astronomie (MPIA). This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France, NASA's Astrophysics Data System, and the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA. Á.K. acknowledges support from the Netherlands Organization for Scientific Research (NWO). This work was partly supported by the grant OTKA-101393 of the Hungarian Scientific Research Fund.

Facilities: ISO (ISOPHOT-S) - Infrared Space Observatory satellite, Spitzer (IRS) - Spitzer Space Telescope satellite

In the following, we collected literature information on some of the sources in order to help the reader with the understanding and interpretation of the spectra plotted in Figure 2. We mostly focused on three points: information on (1) variability in the optical–IR wavelength regime, (2) multiplicity, and (3) extended emission at optical–IR wavelengths. Single sources with no extended emission, and whose observations are unaffected by technical artifacts, are not discussed here. Note that here we also included sources where variability could not be investigated.

IRAS 03260+3111 (also known as NGC 1333 SVS 3) is a young Herbig binary illuminating a small reflection nebula (Strom et al. 1976). The binary is composed of a B5 and an F2 star, with a separation of 362 (Straižys et al. 2002). The secondary is 32 times fainter than the primary in the K band and 24 times fainter in the L band (Haisch et al. 2004). The system appears extended with a fan-shaped nebulosity at 10 μm (Haisch et al. 2006). There is no evidence of emission from silicate grains in the mid-IR spectrum of IRAS 03260+3111. On the other hand, PAH emission is observed not only at the position of the Herbig star but also at various locations in the NGC 1333 nebula. The different PAH features change with changing distance from the central source due to the changing relative populations of ionized and neutral PAHs (Bregman et al. 1993; Roche et al. 1994; Joblin et al. 1996; Sloan et al. 1999). The MIR spectra published in these papers all have different flux levels, and they are also different from our ISOPHOT-S spectrum. These differences are due to the different beam sizes of the instruments and the exact pointing.

BARN 5 IRS 1 is a low-mass embedded YSO in the Barnard 5 dark cloud (Beichman et al. 1984). It is surrounded by a very faint reflection nebula detectable at near-infrared wavelengths (Moore & Emerson 1992). We found no evidence in the literature for extended mid-infrared emission. JHKL photometric monitoring revealed that the source faded significantly between 1983 and 1993 (Moore & Emerson 1992, 1994). In the case of the ISOPHOT-S spectrum, predicted background was subtracted, making the short-wavelength part more uncertain. However, the SED of the source rises steeply toward longer wavelengths, and the long-wavelength part of the ISOPHOT-S spectrum can be safely compared with the Spitzer/IRS spectrum.

LDN 1489 IRS is a single embedded YSO (Connelley et al. 2008). The source looks extended in the 2MASS J, H, and K images, as well as in the 3.6 μm Spitzer/IRAC image. Although it is pointlike in the IRAC images at ⩾5.8 μm, we cannot exclude the possibility of extended emission, which may be the reason for the slight difference between the ISOPHOT-S and the Spitzer/IRS flux levels. Thus, this source is not included in the discussion about variability.

T Tau is a triple system consisting of T Tau N, T Tau Sa, and T Tau Sb. The largest separation between the components is 07 (Köhler et al. 2008), therefore neither ISO nor Spitzer resolves the system: the observed flux represents the sum of all components. T Tau Sa is a Herbig star, while T Tau Sb and T Tau N are low-mass objects (Duchêne et al. 2006). Skemer et al. (2008) presented photometry at 8.7, 10.55, and 11.86 μm for each of the three components, while Ratzka et al. (2009) show 8–13 μm spectra for T Tau S and T Tau N. Based on these results, it seems that only T Tau Sa displays silicate absorption, while T Tau N and T Tau Sb have more or less featureless 10 μm spectra (T Tau N may display slight emission). Beck et al. (2004) resolve T Tau N and T Tau S in the 2–4 μm range, and claim that T Tau S shows deep, broad H₂O ice absorption at 3.1 μm, while T Tau N is featureless in this wavelength range. Photometric monitoring between 2 and 12 μm by Beck et al. (2004) and by van Boekel et al. (2010) indicates that T Tau N is constant within the uncertainties, while T Tau S shows significant variability.

DG Tau is a single classical T Tauri star (Leinert et al. 1991; Connelley et al. 2008). Wooden et al. (2000) report on multi-epoch mid-IR spectra where they found that the 10 μm silicate feature showed significant variability on timescales of months to years, while the underlying continuum remained relatively constant. They claim that prior to 1996, DG Tau had a featureless 10 μm spectrum, which then turned into silicate emission, then went into absorption at the end of 1998. Bary et al. (2009) also observed that the silicate feature displayed variability on monthly and yearly timescales. All our ISOPHOT-S and Spitzer spectra are featureless, but they indicate that the continuum is variable.

Haro 6-10 (also known as GV Tau) is a T Tauri binary with a separation of 12, position angle (P.A.) 355° (Leinert & Haas 1989). Both the ISO and the Spitzer spectra contain the full flux from both components (the Spitzer slit is oriented along the binary). At wavelengths shorter than 4 μm, the southern component dominates, while most of the emission above 4 μm comes from the extremely red northern component (Leinert & Haas 1989). Leinert et al. (2001a) reported J-, H-, K-, L'-, and M-band variability, as well as the variation of the 3.1 μm ice-band absorption. Their resolved observations prove that both components vary significantly on timescales of a few months. Our ISO and Spitzer spectra confirm that the system is variable.

LDN 1551 IRS 5 is a deeply embedded low-mass triple system consisting of a binary with projected separation of 03, and a third component at a distance of 009 of the northern component (Lim & Takakuwa 2006), each star harboring a small circumstellar disk. The whole system is surrounded by a circumbinary disk and a flattened envelope as well (Osorio et al. 2003). Liu et al. (1996) observed the source at 12 μm, and by comparing the observed flux with previous 12 μm and N-band photometry, reported significant mid-IR variability. Although LDN 1551 IRS 5 displays some extended emission in all Spitzer/IRAC images, the larger aperture ISO spectrum has a lower flux than the Spitzer spectrum, indicating real variability. We note that in the case of the ISOPHOT-S spectrum, predicted background was subtracted. The flux of LDN 1551 IRS 5 steeply rises toward longer wavelengths, but below 5 μm, the source is so faint that the uncertainty of the background subtraction dominates the error budget, and the overall error is above 10%.

HL Tau is a single classical T Tauri star. At optical wavelengths, the object is a compact reflection nebula with complicated morphology and diameter of ≈3''; no point source is visible (Stapelfeldt et al. 1995). In the JHK images of Murakawa et al. (2008), the central source can be seen, but it is still surrounded by extended nebulosity (about 3'' in extension). According to our knowledge, there is no information in the literature on whether the source is extended at mid-IR wavelengths. The fact that the larger aperture ISO spectrum has a lower flux level than the Spitzer spectrum indicates real variability. The 8–13 μm UKIRT/CGS3 spectrum obtained in 1993 by Bowey & Adamson (2001) is consistent with the ISOPHOT-S brightness level.

UZ Tau is a hierarchical quadruple T Tauri system. UZ Tau E is a spectroscopic binary with an orbital period of 19.1 days. At a distance of 38 can be found UZ Tau W, which itself is a 034 binary (Prato et al. 2002, , and references therein). Due to periodic accretion from the circumbinary disk, UZ Tau E shows optical variability of about 0.4–1.0 mag in the V, R, and I bands, while UZ Tau W is constant within 0.02 mag (Jensen et al. 2007; Kóspál et al. 2011). UZ Tau E is also variable at millimeter wavelengths, possibly related to nonthermal emission from magnetospheric reconnection events above the binary components' photospheres (Kóspál et al. 2011). The Spitzer/IRS slit was centered on UZ Tau E, although the W component also has some contribution to the observed flux. The ISOPHOT-S measurements were centered between the E and W components, and essentially measured the sum of the two components' fluxes. For this reason, the Spitzer/IRS and ISOPHOT-S spectra cannot be compared, but variability can be studied by comparing the two ISOPHOT-S spectra, which indicate some flux changes in the silicate feature around 10 μm.

VY Tau is a subarcsecond binary with a separation of 066, P.A. of 317°, K-band flux ratio of 3.8, and L-band flux ratio of 6.1 (Leinert et al. 1993; McCabe et al. 2006). Richichi et al. (1994) note that the K-band brightness ratio of the two stars is time variable. VY Tau is an EXor-type object, exhibiting sporadic 1–4 mag optical outbursts, but has been in quiescence at V ≈ 13.8 mag since about 1972 (Herbig 1977, 1990; Herbst et al. 1994). The source shows photospheric emission in the JHK bands (Shiba et al. 1993), but displays clear excess emission from about 3 μm and a clear 10 μm silicate emission feature (Furlan et al. 2006; Sargent et al. 2009; Watson et al. 2009). We found no indication in the literature for extended IR emission. Both the ISO and the Spitzer spectra contain the full flux from the binary components, thus they can be compared.

DR Tau is a highly accreting single classical T Tauri star with a long history of optical and near-IR photometric and spectroscopic variability, which point to magnetospheric accretion with a fluctuating accretion rate (e.g., Götz 1980; Isobe et al. 1988; Hessman & Guenther 1997; Smith et al. 1999; Alencar et al. 2001). Kenyon et al. (1994) present light curves from the B to the L band, and explain the observed flux variations with the presence of a hot spot (the base of the accretion flow) on the stellar surface. DR Tau is sometimes classified as an EXor, a young star exhibiting short accretion outbursts (Lorenzetti et al. 2009). Our observations indicate that DR Tau is also variable in the whole 2.5–15 μm wavelength range.

HD 34700 is a quadruple T Tauri system. The primary itself is a double-line spectroscopic binary (HD 34700 Aa and HD 34700 Ab), consisting of two equal-mass G0-type stars (Arellano Ferro & Giridhar 2003; Torres 2004). Two faint stars at a distance of 52 (HD 34700 B, spectral type: M1-2, classical T Tauri) and 92 (HD 34700 C, spectral type: M3-4, weak-line T Tauri) are also associated with HD 34700 A (Sterzik et al. 2005). The IRS slit only includes the spectroscopic binary HD 34700 A, while the ISOPHOT-S aperture includes all four components. Although HD 34700 B seems to be actively accreting, due to their faintness and distance from component A, components B and C have negligible contribution to the ISOPHOT-S flux. Thus, both instruments essentially measure HD 34700 A.

RR Tau is a Herbig Ae star. Grinin et al. (2002) observed a deep optical and near-IR minimum in the light curve of RR Tau in 2000–2001. They found that the UBVRIJH fluxes showed synchronous behavior, and the color changes indicated that the dimming was caused by a circumstellar dust cloud crossing the line of sight. However, the fading in the optical and JH bands were accompanied with a flux increase in the K and L bands, suggesting that flux changes at longer wavelengths have a different physical mechanism than variable extinction.

Reipurth 50 N IRS 1. Reipurth 50 was first reported as a nebulous object by Reipurth (1985). Reipurth & Bally (1986) reported the sudden appearance of a highly variable, conical nebula Reipurth 50 N, possibly due to an FU Ori-like eruption. Scarrott & Wolstencroft (1988) identified Reipurth 50 N IRS 1 (a.k.a. HBC 494) as the illuminating source of the Reipurth 50 N reflection nebula. Near-IR imaging, imaging polarimetry, and spectroscopy by Casali (1991) revealed that there is only a single source in the Reipurth 50 N cloud, IRS 1, which is a point source at 3.6 μm.

SX Cha is a T Tauri-type binary with a separation of 22, and P.A. of 288° (Reipurth & Zinnecker 1993). The east and west components have spectral types of M0.5 and M3.5, respectively (Muzerolle et al. 2005a). The 2MASS images show that while the west component is brighter in the J band, the east component is brighter in the K band. The Spitzer slit is centered on the east component, and the west component is 12–16 away from the slit centerline. This means that 55%–70% of the flux of the west component is also included in the Spitzer measurement. The ISOPHOT-S data, however, contain the total flux of both components. Thus, due to beam confusion, the spectra of the two instruments cannot be directly compared.

HD 95881 is a Herbig Ae star. Based on spectro-astrometric observations, Baines et al. (2006) consider it a possible subarcsecond binary. We found no indication in the literature for extended emission at any wavelength, thus the ISOPHOT-S and Spitzer spectra can be safely compared. The two spectra agree within the measurement uncertainties until about 10 μm. Between 10 and 11.5 μm, the ISOPHOT-S spectrum is higher than the Spitzer at a 1σ level. The TIMMI2 spectrum published by van Boekel et al. (2005) is even higher than the ISOPHOT-S, indicating possible variability.

CT Cha is a single T Tauri-type star (Guenther et al. 2007). Recently, Schmidt et al. (2008) found a candidate sub-stellar companion (brown dwarf or planet) by direct imaging, at a projected separation of 267. Since the companion is very faint (K-band flux ratio is 313), it does not affect our spectra, and the ISOPHOT-S and Spitzer/IRS data can be safely compared.

VW Cha is a triple system. The separation of the A and B components is 066, P.A. 177°, and their K-band flux ratio is 25.6. The separation of the B and C components is 010, P.A. 233°, and their K-band flux ratio is 1.34 (Brandeker et al. 2001). There is also an even fainter, fourth component at a distance of 1678 from VW Cha A (Correia et al. 2006). Both our ISO and Spitzer spectra contain the flux from the three brightest components (A, B, and C).

CU Cha (also known as HD 97048) is a single Herbig star with a spectral type of B9.5/A0 (Ghez et al. 1997; Bailey 1998; Doucet et al. 2007). Based on ISOCAM-CVF data, Siebenmorgen et al. (2000) found that the emission of CU Cha in the PAH bands is extended on scales of about 5''–10''. According to Doucet et al. (2007), this emission comes from transiently heated PAH molecules in an extended envelope around the star–disk system. Lagage et al. (2006) claimed that CU Cha is extended in the 8.6 μm PAH band, and the brightness profile is well reproduced by emission from the surface layer of a vertically optically thick, inclined, flared disk, extending at least up to 370 AU (or 21). Doucet et al. (2007) found that the object is also extended in the 11.3 μm PAH band, while van Boekel et al. (2004) could also resolve the disk at continuum wavelengths. While the differences between the ISOPHOT-S and Spitzer/IRS spectra might be due to the different beam sizes, the fact that the three ISOPHOT-S spectra differ by 20%–40% points to real variability. Long-term optical monitoring indicates that the amplitude of optical variability of CU Cha is <0.1–0.3 mag (Manfroid et al. 1991; Herbst & Shevchenko 1999).

Glass I is a T Tauri-type binary with a separation of 267, P.A. of 285°, and K-band flux ratio of 5.9 (Haisch et al. 2004). It is also known as Ced 111 IRS 4 or Cha I T33. The primary is brighter in the optical, but it is a "naked" T Tauri; most of the IR emission comes from the deeply embedded Class I secondary (Feigelson & Kriss 1989; Prusti et al. 1991). Gürtler et al. (1999) already reported strong mid-IR variability based on two ISO spectra. The Spitzer slit in 2006 was positioned halfway between the two stars, but the slit and the P.A. of the binary is misaligned, leading to some flux loss. We do not attempt to correct for this effect. The slit in 2008 was positioned exactly on the secondary component, which dominates the IR emission. Our variability study for this source is based on the ISOPHOT-S spectra, which contain the total flux of both components.

Ced 111 IRS 5 is a single T Tauri-type star (Haisch et al. 2004). It is extended on an arcminute scale, even in the 8 μm IRAC image. It is also called the Chamaeleon IR nebula. This explains the difference between the ISO and Spitzer spectrum. However, variability can be studied by comparing the long-wavelength parts of the two ISOPHOT-S spectra, which have nearly identical pointings and are not strongly affected by memory effects or uncertainties related to the subtraction of the predicted background.

HD 97300 is a B9-type binary Herbig star (Siebenmorgen et al. 1998). In the H-R diagram, it is located close to the zero-age main sequence, and it has no significant IR excess below 24 μm (Kóspál et al. 2012). The separation of the binary components is 08, and their K-band flux ratio is 17 (Ghez et al. 1997; Lafrenière et al. 2008). Siebenmorgen et al. (1998) noticed that at certain IR wavelengths, two emission peaks can be seen: one coinciding with HD 97300 and one being at a distance of about 3''. They also detected an extended, ringlike structure, whose spectrum is dominated by PAH emission. The size of the ring is about 50'' × 33''. It is not centered on the star, and is probably made of interstellar matter, whose density is enhanced by the interaction of the stellar wind from HD 97300 with the environment (see also Kóspál et al. 2012). The PAH molecules in the ring are transiently heated by HD 97300. This extended emission explains the factor of 3–4 difference we detect between the ISOPHOT-S and IRS spectra (the ISOCAM-CVF spectrum published by Siebenmorgen et al. 2000, representing the total flux of the system, is another factor of two higher than the ISOPHOT-S spectrum). The good agreement between the two ISOPHOT-S spectra indicates that the emission of the central source, and consequently that of the PAH molecules heated by it, is constant in time. Indeed, long-term optical and near-IR photometric monitoring indicates that the brightness of HD 97300 is constant within 0.1–0.2 mag (Manfroid et al. 1991; Davies et al. 1990).

Ced 112 IRS 4 (also known as FM Cha) is a single T Tauri-type star with no indication of extended near-IR or mid-IR emission in its immediate vicinity (Haisch et al. 2004, 2006). Alexander et al. (2003) presented an ISOCAM-CVF spectrum of the source and noticed that no ice absorption features are present, while the 10 μm region indicates the combination of silicate emission and absorption. Since our ISOPHOT-S and Spitzer/IRS spectra are consistent with this finding, we decided not to assign any type to this source. All the ISOCAM-CVF, ISOPHOT-S, and Spitzer spectra indicate significant flux changes and changes in the spectral slope.

WX Cha is a T Tauri-type binary with a separation of 079 and P.A. of 55° (Ghez et al. 1997). Both the ISO and Spitzer spectra contain all the flux from both components, thus, they can be safely compared.

CV Cha is a single T Tauri-type star (Guenther et al. 2007; Melo 2003), although it shows variable radial velocity (Reipurth et al. 2002). The Spitzer/IRS slit contained only CV Cha, while the ISOPHOT-S beam included also CW Cha (at a separation of 11''). Resolved IRAC photometry from Luhman et al. (2008) indicates that the flux ratio of CW Cha to CV Cha is 0.12–0.16 between 3.6 and 8 μm. This, and the fact that CW Cha falls at the very edge of the ISOPHOT-S beam ensures that CW Cha has negligible contribution to the ISOPHOT-S spectrum. Thus, we conclude that both the ISOPHOT-S and the Spitzer/IRS spectra represent the flux from CV Cha only, thus they can be compared.

HD 98800 (also known as TV Crt) is a quadruple T Tauri system. It consists of a binary with a projected separation of 08 (or 38 AU), P.A. 0°, and each component is a spectroscopic binary with a separation of about 1 AU (Prato et al. 2001; Boden et al. 2005). Both the ISO aperture and the IRS slit contain all four objects, so they can be compared. The IR excess is attributed to HD 98800 B (Prato et al. 2001). HD 98800 B displays no excess emission below about 5.5 μm, indicating an inner cleared-out region (Furlan et al. 2007).

HD 100453 is an isolated binary Herbig star. The primary has a spectral type of A9 Ve. The companion, HD 100453 B, is situated at a separation of 106 (or 120 AU), and is 110 times fainter than the primary in the K_s band (Habart et al. 2006). The companion is a M4.0–M4.5V star with no significant excess in the K or L band (Collins et al. 2009), thus it has no contribution to the plotted ISOPHOT-S or Spitzer/IRS spectra. Meeus et al. (2002) explain the lack of the 10 μm silicate feature with the depletion of small, hot silicate grains in the disk of HD 100453 (a factor of 100–500 depletion compared to AB Aur is needed in their model to be able to be consistent with the observed ISO-SWS (Short Wavelength Spectrometer) spectrum). Using higher signal-to-noise ratio TIMMI2 data, van Boekel et al. (2005) confirmed the lack of the 10 μm silicate emission, and attributed it to grain growth. The comparison of our ISOPHOT-S and IRS spectra reveals some differences in the 8–12 μm wavelength range, while the flux at λ < 8 μm is constant. It is possible that this variability is due to a weak, variable silicate emission feature (Figure 17). Confirmation of this idea would require a detailed modeling of the IRS spectrum, which has the highest signal-to-noise ratio among the 10 μm spectra published so far for HD 100453. The star shows no strong accretion activity and is photometrically stable (Collins et al. 2009; Garcia Lopez et al. 2006; Guimarães et al. 2006).

Zoom In Zoom Out Reset image size

HD 104237 is a spectroscopic binary consisting of a Herbig A4V primary and a K3-type companion (Böhm et al. 2004). Based on X-ray observations, Testa et al. (2008) identified four more low-mass stars in the vicinity, HD 104237 B-E, with separations between 1365 and 1488, while Grady et al. (2004) found five low-mass sources in their optical, near-IR, and mid-IR images. In addition to the Herbig Ae star itself, HD 104237-6 (corresponding to the X-ray source HD 104237 E) and HD 104237-2 have IR excesses, but the Herbig star HD 104237 dominates the mid-IR integrated light of the region (Grady et al. 2004). While HD 104237-6 falls outside of both the ISOPHOT-S beam and the Spitzer/IRS slit, HD 104237-2 is included in both. The Spitzer slit is centered between the Herbig star and HD 104237-2. Offset correction was applied using the position of the Herbig star. Although the observed Spitzer/IRS spectrum may contain 75%–100% of the flux of HD 104237-2, its contribution to the much brighter primary is negligible, thus we conclude that the ISOPHOT-S and Spitzer/IRS spectra can be compared. Nevertheless, the ISOPHOT-S spectra alone already show significant variability (see also Figure 6).

DK Cha is a deeply embedded active Herbig Ae star. It displays brightness variations of more than 1 mag in the K band (Hughes et al. 1991). The source is redder when fainter, and the amplitude of the brightness fluctuations decreases with increasing wavelength. They explain the observed flux changes with a combination of variable circumstellar extinction and intrinsic variability. Prusti et al. (1994) found no evidence for extended mid-IR emission (with multi-diaphragm observations between 3.8 and 18.6 μm, meaning that the emitting area is less than 54). We note that there is a significant discrepancy between the ISOPHOT-S (this work) and the ISO-SWS (Acke & van den Ancker 2004) spectra of the source, pointing to possible variability. The Spitzer data points are close to saturation and are not very reliable.

HD 135344 B is an F4 Ve-type Herbig star (van Boekel et al. 2005). Doucet et al. (2006) resolved the source at 20.5 μm, and modeled the emission with a disk with an outer radius of 200 AU (or 14, which is within both the ISOPHOT-S and the Spitzer/IRS aperture). Grady et al. (2009) observed significant changes between 1 and 10 μm, both in flux and in the shape of the SED. At wavelengths shortward of 4.75 μm, at least 13% variations are present, while at 10 μm, they found 60% variability.

HD 139614 is a Herbig Ae star (Dunkin et al. 1997). Yudin et al. (1999) detected polarimetric variability on a timescale of days, which they explain in terms of a model of a pole-on disk with dust clouds or cometlike bodies moving in a circumstellar envelope around the star. We found no evidence in the literature that the source might be extended at mid-IR wavelengths, thus we conclude that the ISOPHOT-S and the Spitzer spectra can be safely compared. The two spectra agree within the measurement uncertainties, moreover, the TIMMI2 spectrum of van Boekel et al. (2005) is also identical with them. Thus, HD 139614 is constant at mid-IR wavelengths (and does not show any significant variability in the optical; see Meeus et al. 1998).

HD 141569 is a B9.5/A0 Herbig star (Jaschek & Jaschek 1992; Dunkin et al. 1997). Coronagraphic observations of the scattered light by Augereau et al. (1999), Weinberger et al. (1999), and Mouillet et al. (2001) revealed that its disk is asymmetric, and consists of two rings with a gap between them. Weinberger et al. (2000) claim that the disk is composed of secondary debris material, although it still contains some gas (Zuckerman et al. 1995). HD 141569 has two M-type companions: HD 141569 B (M2V) and HD 141569 C (M4V) at projected distances of 76 (A-B) and 14 (B-C); K-band flux ratios are 5.3 (A/B) and 11.9 (A/C) (Pirzkal et al. 1997; Weinberger et al. 2000). The Spitzer/IRS slit included only the primary (A), while the ISOPHOT-S aperture included all three components. Due to their faintness and distance from component A, components B and C have negligible contribution to the ISOPHOT-S flux, thus the ISOPHOT-S and Spitzer/IRS fluxes can be compared.

HD 142666 is a Herbig Ae star (van Boekel et al. 2005). Mid-IR spectra of HD 142666 obtained with different instruments are published in Sylvester et al. (1997), Bouwman et al. (2001), and van Boekel et al. (2005). HD 142666 displays large photometric variations (Malfait et al. 1998 give a visual amplitude of about 1.2 mag, while Lecavelier Des Etangs et al. 2005 detected 0.3 mag variability with Stromgren photometry). The star appears redder when fainter, which they interpret as variable dust extinction in a close to edge-on disk with clumpy dust distribution. Since beam effects do not play a role here, we can safely compare the ISOPHOT-S and Spitzer spectra.

HD 142527 is an isolated Fe-type Herbig star (van Boekel et al. 2005, and references therein). Based on spectro-astrometric observations, Baines et al. (2006) consider it a possible subarcsecond binary. The circumstellar environment of the star was resolved at various wavelengths (see, e.g., near-IR images in Fukagawa et al. 2006, mid-IR images in Fujiwara et al. 2006, or millimeter images in Verhoeff et al. 2011). Based on these observations, the star is surrounded by an inner disk, an outer disk, and a halo around the inner disk regions. Spatially resolved 18 μm images in Fujiwara et al. (2006) and in Verhoeff et al. (2011) indicate that the emission is coming from within about 1'' of the star, which is fully included in both the ISOPHOT-S beam and the Spitzer/IRS slit. Thus, the spectra can be compared, and they indicate significant variability. An additional 8–13 μm spectrum from van Boekel et al. (2005) also confirms this (Figure 7).

EX Lup is the prototype of EXors, a class of young stars showing sporadic optical outburst separated by extended quiescent periods (Herbig 1977). The outbursts are powered by enhanced accretion from the circumstellar disk onto the star (Herbig et al. 2001). EX Lup itself brightened by ΔV ≈ 5 mag in 1955–1956 and again in 2008, and produced several smaller, 1–3 mag brightenings in-between. EX Lup was extensively studied during its outburst in 2008, and the observations indicate that it brightened in the whole 0.4–100 μm wavelength range (Juhász et al. 2012). By comparing pre-outburst and outburst Spitzer spectra, Ábrahám et al. (2009) observed a significant change in the shape of the 10 μm silicate feature: while before the outburst, the feature indicated mostly amorphous silicates, during outburst, several narrow peaks and shoulders appeared, indicating the presence of crystalline silicates. Ábrahám et al. (2009) explain the appearance of the crystalline silicates by annealing in the surface layer of the inner disk by heat from the outburst. In Figure 2, we plot ISOPHOT-S spectra from 1997 and Spitzer spectra from 2004 and 2005. These are all quiescent spectra, when the visual magnitude of EX Lup was between 12.5 mag and 14 mag.

HD 144432 is a PMS binary consisting of HD 144432 A, an A9/F0-type Herbig star primary and HD 144432 B, a K0-5-type T Tauri secondary, with a separation of 14, P.A. 0° (Pérez et al. 2004; Stelzer et al. 2009; Fukagawa et al. 2010). The secondary is a factor of 9–14 fainter than the primary in the K and H bands (Pérez et al. 2004; Fukagawa et al. 2010). The secondary displays photospheric emission until the K band, and the IR excess is attributed exclusively to the primary (Pérez et al. 2004). Both the ISOPHOT-S aperture and the Spitzer/IRS slit include both components and the observed spectra represent the total flux of the two objects (the Spitzer slit is oriented along the binary, north–south). Keck segment-tilting observations by Monnier et al. (2009) resolved the source at 10.7 μm, which is ≈40 mas. There is a ≈10% difference between the ISOPHOT-S and the Spitzer/IRS spectra plotted in Figure 2. Several other mid-IR spectra are available on HD 144432 in the literature, see, e.g., Sylvester et al. (1996), Leinert et al. (2004), and van Boekel et al. (2005). The first of these is consistent with the Spitzer/IRS flux level, the second one is consistent with the ISOPHOT-S flux level, and the third one is lower than the others by another ≈10%, supporting our finding that the source is variable in the mid-IR regime.

HR 5999 is a binary star consisting of a Herbig Ae primary, and a T Tauri secondary called Rossiter 3930 (Rossiter 1955). The separation is 14, the P.A. is 111°, and the K-band flux ratio is between 20 and 35 (Stecklum et al. 1995; Ghez et al. 1997; Leinert et al. 1997). The primary itself may be a spectroscopic binary (Tjin A Djie et al. 1989), although observations by Corporon & Lagrange (1999) do not confirm this. HR 5999 is a known optical variable, whose "bursts" are attributed to unsteady accretion (Perez et al. 1992). Multi-epoch BVRIJHKLMNQ band photometry by Hutchinson et al. (1994) indicated that the object is indeed variable at optical wavelengths but not much above 4.8 μm. An ISOCAM-CVF spectrum published in Acke & van den Ancker (2004) is consistent with our ISOPHOT-S spectrum until 8.5 μm, but is lower by about 20% at longer wavelengths, indicating a possibly variable silicate feature. Siebenmorgen et al. (2000) noted that HR 5999 is a point source at these wavelengths. Using VLTI/MIDI, Preibisch et al. (2006) resolved the mid-IR emission from HR 5999 and they give sizes between 5 and 25 mas (1–5 AU), depending on the fitted model.

WL 16 is a single B8-A7 Herbig star (Simon et al. 1995; Ratzka et al. 2005; Luhman & Rieke 1999). Images at different mid-IR wavelengths between 7.9 and 24.5 μm (both in the continuum and in the PAH bands) published by Ressler & Barsony (2003) revealed that the central star is surrounded by a bright disk with a size of 7'' × 35 (corresponding to a disk diameter of 900 AU), viewed at an inclination of 622. Geers et al. (2007) also resolve both the 8–13 μm continuum and the 8.6, 11.2, and 12.7 μm PAH features, but give smaller radial extents (50–60 AU or 04–05) than Ressler & Barsony (2003). In their study, the 3.3 μm feature is undetected, indicating the dominance of ionized PAHs. Our ISOPHOT-S spectrum of WL 16 in Figure 2 agrees well with the ISO-SWS spectrum, and the slight differences between these and the 10 μm spectrum published in Hanner et al. (1992) are probably due to the different beam sizes of the instruments.

MWC 863 is a PMS binary, consisting of a Herbig Ae primary, HD 150193 A, and an F9-type T Tauri secondary, HD 150193 B (Carmona et al. 2007). Their separation is 11, the P.A. is 227° (Reipurth & Zinnecker 1993), flux ratios are 6–8 and 4–17 in the K and L' bands, respectively (Zinnecker et al. 1991; Koresko 2002). The secondary component has no IR excess (Fukagawa et al. 2003, and references therein). Using coronagraphic images, Fukagawa et al. (2003) resolved the disk around HD 150193 A in the H band, and gave a radius of 13 (190 AU). During Keck segment-tilting observations by Monnier et al. (2009) the source remained unresolved at 10.7 μm, which gives an upper limit of FWHM = 35 mas for the emitting region. We note that while the ISO-SWS spectrum published in Acke & van den Ancker (2004) is identical with our ISOPHOT-S spectrum, the UKIRT/CGS3 spectrum taken by Sylvester & Mannings (2000) is significantly different both in shape and in flux level, pointing to possible changes in the composition and size distribution of the silicate dust grains.

AK Sco is a double-line spectroscopic binary consisting of approximately equal-mass F5-type stars with a projected separation of 0.14 AU (≈1 mas); the two stars are surrounded by a circumbinary disk (Alencar et al. 2003). Both the ISO and Spitzer spectra include both components. Its optical light curve shows a "roughly constant maximum light level being interrupted at irregular intervals by minima about a magnitude deep" (Andersen et al. 1989, p. 142). Our ISOPHOT-S and Spitzer/IRS spectra agree within the measurement uncertainties, and a TIMMI2 spectrum by Przygodda et al. (2003) also has the same flux level.

MWC 865 (also known as V921 Sco and CD−42°11721) is a B0IV-type star embedded in a small, thick dark cloud (Boersma et al. 2009, and references therein). It illuminates a reflection nebula (van den Bergh & Herbst 1975). Its distance is quite uncertain (values between 400 and 2500 pc appear in the literature, e.g., Boersma et al. 2009; Lopes et al. 1992), and there is also some debate about its PMS nature (Borges Fernandes et al. 2007), although it is generally considered to be a Herbig Be star. Habart et al. (2003) found four IR companions within 13'' of MWC 865, while Wang & Looney (2007) found about 10 YSO candidates in the vicinity (within about 70'') of MWC 865 using Spitzer/IRAC images. Boersma et al. (2009) show Spitzer/IRAC images between 3.6 and 8.0 μm and TIMMI2 images at 11.9 μm of MWC 865 and its surroundings. These images reveal two bright peaks, one coinciding with the star, the other one being a close-by patch 43 to the north. There is an arc stretching toward the southeast as well, at a distance of 11'' from the star. TIMMI2 long-slit spectra indicate that the central star has a featureless spectrum, while the close-by patch and the arc display strong PAH emission. Boersma et al. (2009) combine the spectra of these components and compare it with an ISO-SWS spectrum. They find that the ISO-SWS spectrum shows a similar shape, but has higher flux levels due to the larger aperture of ISO-SWS and the extent of the PAH emission. Our ISOPHOT-S spectrum of MWC 865 also has a very similar shape to the ISO-SWS spectrum, but even higher flux levels. We conclude that these differences are also due to the difference in the apertures and their exact position with respect to the PAH-emitting patches. We note that significant offset correction was applied to the ISOPHOT-S spectrum of MWC 865, introducing extra uncertainty in the absolute flux levels.

51 Oph is a single Herbig star (Baines et al. 2006), which is photometrically constant (Lecavelier Des Etangs et al. 2005). The source appeared pointlike in the 18 μm images of Jayawardhana et al. (2001), giving an upper limit of 05 (65 AU) for the radius of the emitting region. Leinert et al. (2004) resolved the source at 12.5 μm with VLTI/MIDI and gave a half-light radius of 7 mas (0.5 AU). Our conclusion is that the ISOPHOT-S and the Spitzer/IRS spectra can be compared and the observed variability is real. We note that ground-based mid-IR spectra published in Sylvester et al. (1996) and Leinert et al. (2004) are consistent with the Spitzer/IRS flux level.

HD 163296 is a well-known single (Corporon & Lagrange 1999; Pirzkal et al. 1997) Herbig Ae star. Using optical and near-IR coronagraphic images, Fukagawa et al. (2010) and Grady et al. (2000) could detect the disk in scattered light out to a radius of 36 (440 AU). Wisniewski et al. (2008) found that the optical scattered light is variable in time due to variable self-shadowing. During Keck segment-tilting observations by Monnier et al. (2009) the source remained unresolved at 10.7 μm, which gives an upper limit of FWHM = 35 mas for the emitting region. Doucet et al. (2006) resolved the disk thermal emission at 20.5 μm and derived an outer disk radius of 16 (200 AU). The source was resolved by Leinert et al. (2004) using VLTI/MIDI data, and the half-light radius at 12.5 μm is 7 mas (0.8 AU). The source appeared pointlike in the 18 μm images of Jayawardhana et al. (2001), giving an upper limit of 05 (60 AU) for the radius of the emitting region. Sitko et al. (2008) conducted a long-term 3–13 μm spectroscopic monitoring of HD 163296. Most of their data are consistent with no variability larger than 10%, except at one epoch when the 1–5 μm flux significantly increased, accompanied by a slight increase in the flux of the 10 μm silicate feature. Although no Spitzer/IRS low-resolution spectra are available for this source, our ISOPHOT-S spectrum is consistent with an ISO-SWS spectrum (taken on the same day), and also consistent with the TIMMI2 spectrum published in van Boekel et al. (2005), obtained in 2003 March, and the MIDI spectrum published in Leinert et al. (2004), obtained in 2003 June, while it is 10%–40% higher than that obtained by Kessler-Silacci et al. (2005) with the Keck/Long Wavelength Spectrometer in 1999 August and in 2000 June.

HD 169142 is a Herbig Ae star (Sylvester et al. 1996). High spatial resolution mid-IR observations at 3.3, 10.8, and 18.2 μm show that the emission comes from a disk with an outer radius of 150 AU (1'') at most (Habart et al. 2006; Jayawardhana et al. 2001). SED modeling gave disk radii in the range of 100–300 AU (07–2''; Dominik et al. 2003; Dent et al. 2006). HD 169142 has a possible companion, 2MASS J18242929−2946559, located at a distance of 93 (Grady et al. 2007). The 2MASS source is a binary weak-line T Tauri with 130 mas separation, and is 11 times fainter than HD 169142 in the K band. The Spitzer/IRS slit contains only HD 169142, while the ISOPHOT-S beam includes both components. However, the contribution of the 2MASS source at mid-IR wavelengths is probably negligible, thus the Spitzer and the ISO spectra can be compared. The spectra plotted in Figure 2 agree within the measurement uncertainties, thus we conclude that this source is constant in the 5–11.5 μm regime. The flux levels of the UKIRT/CGS3 spectrum from Sylvester et al. (1996) and the TIMMI2 spectrum from van Boekel et al. (2005) also agree with our spectra until ≈9 μm, while they differ by up to 50% above 9 μm (Figure 8). The source is photometrically stable at optical wavelengths (Guimarães et al. 2006).

VV Ser is a single (Pirzkal et al. 1997; Leinert et al. 1997) Herbig star. It displays UXor-type optical variability (Herbst & Shevchenko 1999; Rostopchina et al. 2001), i.e., 1–3 mag deep minima in the V band due to variable obscuration by dust clumps in a nearly edge-on circumstellar disk. Habart et al. (2003) found no other source emitting at 10 μm in the 1' × 1' vicinity of VV Ser. Li et al. (1994) found no extended emission around the source in JHK filters, while Leinert et al. (2001b) found faint extended emission with an FWHM of 06 in JHK with speckle interferometry. Eisner et al. (2004) resolved the source at 2.2 μm with the Palomar Testbed Interferometer, and determined a size of 1.5–4.5 mas (depending on the model used) for the emitting region. Pontoppidan et al. (2007) discovered nebulous mid-IR emission extending over 4' centered on VV Ser. The nebulosity is due to transiently heated dust grains. Although we cannot exclude the possibility that this extended emission has different contributions to the ISOPHOT-S and the Spitzer/IRS spectra, the fact that the Spitzer/IRS flux level is higher points to real variability.

OO Ser is a deeply embedded YSO that produced an outburst in 1995 and became 4.6 mag brighter in the K band (Hodapp et al. 1996), reaching its maximum brightness in 1995 October (Hodapp 1999). The eruption caused brightening at a wide wavelength range from 2.2 μm to 100 μm (Kóspál et al. 2007). After peak brightness the object started a gradual fading. The ISOPHOT-S spectra plotted in Figure 2 cover a period of 20 months, and the first one was obtained 4 months after the maximum brightness occurred. The gradual fading, which is discussed in detail in Kóspál et al. (2007), is well visible in the figure. The source is surrounded by a small elongated nebula, approximately 15'' in diameter, best seen in the K band, but also discernible at 3.6, 4.5, and 5.8 μm (Hodapp et al. 1996; Kóspál et al. 2007).

CK 1 is an embedded protostellar binary with a separation of 15, and P.A. of 10° (Eiroa & Leinert 1987). The southern component is brighter than the northern one, and their flux ratio decreases from 4.2 at 0.9 μm to 3.3 at 3.5 μm to 1.1–2.8 between 8 and 12.9 μm (Eiroa et al. 1987; Eiroa & Leinert 1987; Ciardi et al. 2005). Both the ISO and the Spitzer spectra represent the sum of the spectra of CK 1 north and CK 1 south (the Spitzer slit is oriented along the separation of the binary), thus they can be compared. Ciardi et al. (2005) present resolved 8–13 μm spectra of the two components, and the sum of the two spectra as well. This sum agrees very well with our ISOPHOT-S and Spitzer/IRS data.

[SVS76] Ser 4 is a small, dense cluster of deeply embedded low- and intermediate-mass YSOs (Pontoppidan et al. 2004). Judging from the JHK photometry of Eiroa & Casali (1989) and the Spitzer 3.6–24.0 μm fluxes from Harvey et al. (2007), most of the flux in the Spitzer slit comes from SVS4/9 and SVS4/10, while in addition to these two sources, SVS4/4, SVS4/8, SVS4/11, and SVS4/5, also contribute to the ISO flux. SVS4/5 is especially bright above 5 μm (very red source). Source confusion explains the differences between the ISO and Spitzer spectra. We note that in the case of the ISOPHOT-S spectrum, predicted background was subtracted, making the absolute level of the measured flux probably more uncertain than 10%.

CK 2 is an embedded YSO in the Serpens star-forming region. We found no indication of extended mid-IR emission in the literature. The source is close to the sensitivity limit of ISOPHOT-S, resulting in a very low signal-to-noise ratio, especially above 6 μm. The ISOPHOT-S spectrum also suffers from memory effect, and a predicted background is subtracted, which, considering how faint the source is, introduces additional uncertainties. Thus, we do not analyze the variability of CK 2. Nevertheless, the ISOCAM-CVF spectrum of Alexander et al. (2003) is significantly lower than our Spitzer/IRS spectrum, indicating possible variability.

S CrA is a PMS binary with a separation of 13–15, P.A. of 147°–160°, K-band flux ratio of 1.9–2.7, and L-band flux ratio of 2.5 (Baier et al. 1985; Zinnecker et al. 1991; Chelli et al. 1995; Prato et al. 2003). Carmona et al. (2007) obtained spatially resolved optical spectra for the two components and determined a spectral type of G5 Ve for the primary (S CrA NW) and K5Ve for the secondary (S CrA SE). S CrA is a well-known optical variable, with V-band variations of up to ≈2 mag (e.g., Kardopolov & Filipev 1981). Graham (1992) interpreted the photometric and spectroscopic changes of the object as variable accretion onto the star. Both the ISOPHOT-S aperture and the Spitzer/IRS slit included the whole flux of both components, thus, they can be compared.

R CrA, T CrA, and HH 100 IRS are part of Coronet, a small cluster of YSOs. The two brightest sources, R CrA and T CrA, are Herbig stars, while the remaining dozen (HH 100 IRS among others) are deeply embedded low-mass protostars (Forbrich et al. 2007, and references therein). R CrA illuminates the reflection nebula NGC 6729. Both R CrA and the nebula are highly variable at optical wavelengths (Graham & Phillips 1987). HH 100 IRS is also highly variable, with amplitudes up to 2.5 mag in the K band (Axon et al. 1982; Reipurth & Wamsteker 1983). Both R CrA and T CrA are possible spectro-astrometric binaries (Takami et al. 2003). Using speckle interferometry, Dewarf & Dyck (1993) found that R CrA is unresolved at 2.2 and 3.8 μm. Prusti et al. (1994) found no evidence for extended mid-IR emission larger than ≈5'' around R CrA and T CrA. Interestingly, R CrA displays the 4.27 μm feature of solid CO₂ ice in absorption and the 10 μm silicate feature in emission at the same time (Nummelin et al. 2001; Acke & van den Ancker 2004). No Spitzer spectra are available for R CrA and T CrA, but the comparison of our ISOPHOT-S spectra in Figure 2 and the ISO-SWS spectra published in Acke & van den Ancker (2004) indicates very similar flux levels. For HH 100 IRS, variability can be studied by comparing the two ISOPHOT-S spectra.

VV CrA is a T Tauri-type binary with a separation of 210, P.A. of 44°, and K-band flux ratio of 8.6. The optical secondary becomes brighter around the J band, and dominates the flux at longer wavelengths (Chelli et al. 1995). Ratzka et al. (2008, p. 519) report that the secondary faded significantly and "it is now fainter than the primary even in the N band." They also show 10 μm MIDI spectra for each component: VV CrA NE shows deep silicate absorption, while VV CrA SW is featureless. Our ISO spectrum contains flux from both components, while the Spitzer/IRS spectrum is centered on the NE component, but contains significant contribution from the SW as well. For this reason, we do not analyze variability for this source.

WW Vul is an isolated Herbig Ae star, which shows UXor-like fadings at optical wavelengths, although its disk probably has a larger inclination than for other UXors (Pugach 2004; Kozlova et al. 2006). WW Vul was observed twice with ISOPHOT-S and once with Spitzer. Both ISOPHOT-S spectra are offset corrected, but the large correction applied for the earlier one (TDT: 17600465) makes this more uncertain than the 10% assigned as a final uncertainty of the ISOPHOT-S spectra. For this reason, this spectrum in not used in our variability analysis, and we base our conclusions on the comparison of the other ISOPHOT-S spectrum (TDT: 51300108) and the Spitzer measurement.

BD +40 4124 and LkHa 224 are members of a cluster of young stars (Henning et al. 1998, and references therein). The brightest members of the cluster are the B2-type BD +40 4124 and the B5-type LkHa 224, but several fainter FGKM stars can be found within a few arcminutes; some of these are included in the ISOPHOT-S aperture. No Spitzer spectrum is obtained for these sources, however, an ISO-SWS measurement is available. Comparison of our ISOPHOT-S spectrum with the ISO-SWS reveals no significant differences. BD +40 4124 displays small-scale (<0.4 mag) quasi-periodic variability at optical wavelengths (Herbst & Shevchenko 1999; Strom et al. 1972).

PV Cep is an embedded YSO surrounded by the cometary reflection nebula RNO 125 (Thé et al. 1994). Both the star and the nebula are strongly variable (e.g., Scarrott et al. 1991; Lorenzetti et al. 2011). Kun et al. (2011) analyzed the brightness changes of PV Cep in the whole optical–infrared regime, and found significant flux variations below 10 μm. They conclude that the reason for the variability is a combination of changing accretion rate and changing circumstellar extinction. We found no indication in the literature that the object is extended at mid-IR wavelengths. The fact that the Spitzer spectrum has a higher flux than the ISOPHOT-S points to real variability.

SV Cep is a spectro-astrometric binary Herbig star (Wheelwright et al. 2010). Its optical light curve is characteristic of UXor-type variables with irregular, quasi-periodic variations due to inhomogeneities in a nearly edge-on circumstellar disk (e.g., Rostopchina et al. 2000). Juhász et al. (2007) investigated the long-term 0.55–100 μm variability of SV Cep and found a correlation between the optical and far-IR light curves. In the mid-IR regime, they found lower variability amplitudes (a peak-to-peak variability of ≈40% at 10 μm) and a hint for anti-correlation with the optical changes. They interpret the variability using a self-shadowed disk and a puffed-up inner rim with variable inner rim height. Their model also contained an optically thin envelope to account for the constant 25 μm flux. Our ISOPHOT-S spectrum plotted in Figure 2 and the ISO-SWS spectrum obtained 9 months earlier (Acke & van den Ancker 2004) are consistent with each other.