Spitzer Observations of Interstellar Object 1I/'Oumuamua

Observations were obtained with Spitzer/IRAC (Fazio et al. 2004) as part of DDT program 13249. Seven Astronomical Observing Requests (AORs) were used, six of ∼5 hr duration with 166 × 100 s frames, and a final 2.9 hr (clock time) AOR with 94 × 100 s frames, for a total of 1090 frames and 30.2 hr on-source frame time (acquired over 33 hr of clock time). The observations were divided in this way because of limits in the number of commands and data allowed in a single AOR. The data were taken with the "Moving Single" target mode with Full Array readouts, using a small cycling dither pattern. Two frames were taken at each dither position, to reduce the overhead of moving after each frame. Images were obtained in both arrays, but only the 4.5 μm channel was nominally centered on the target position, since the object was expected to be brighter in that IRAC channel. With the information known at that time, we estimated that with this integration time we could achieve a 3σ or better detection of the object if it was at its expected maximum brightness during the time of observation.

'Oumuamua was discovered on 2017 October 19 (and announced as an interstellar body on 2017 October 25; Bacci et al. 2017), but because of the constraints of the Spitzer observability zone, the earliest that the Spitzer observations could begin was late on November 20 (Figure 1). The ephemeris used to develop the original Spitzer observation sequence was based on ground-based astrometric data through the end of October and had a prediction uncertainty larger than the Spitzer FOV. On November 9, the Magdalena Ridge Observatory collected additional ground-based observations, which we used together with preliminary high-precision astrometry from Micheli et al. (2018) to refine the orbit of 'Oumuamua. We found that the revised predicted positions could potentially put the object very close to or off the edge of the array for many frames in the AORs constructed with the original ephemeris. The Spitzer Science Center (SSC) was able to replan the observations with the latest orbit information. The first AOR began executing at 2017 November 21 10:13:26 UT, and the last AOR completed at 2017 November 22 18:52:06 UT; this is around 2.5 months after 'Oumuamua's perihelion passage. The average heliocentric distance of 'Oumuamua during the observations was 2.0 au and the average Spitzer-centric distance was 1.8 au; the average phase angle was around 31° (Figure 1). This geometry changed only very slightly during the 33 hr of clock time needed to carry out these observations. The rate of motion on-sky in these observations was around 68 arcsec/hr.

Zoom In Zoom Out Reset image size

The data reduction method used was similar to that described in Mommert et al. (2014b). A mosaic of the field was constructed from the data set itself and then subtracted from the individual basic calibrated data (BCD) frames. After subtraction of the background mosaic, residual background sources and bright cosmic ray artifacts were masked in the individual BCDs before being mosaicked in the reference frame of the moving object.

Micheli et al. (2018) later collected ground-based and Hubble Space Telescope astrometry of 'Oumuamua, eventually extending the observational arc through 2018 January 2. Based on this longer sampling of the trajectory, they reported a 30σ detection of a non-gravitational acceleration acting on the motion of the object, inferred from position measurements over time. This acceleration was not visible or expected when the Spitzer observation sequence was built in November and would have resulted in an ephemeris correction of about 100 arcseconds at the time of the Spitzer observations (Figure 2). This correction is along the Line of Variation (Milani et al. 2005), i.e., the direction corresponding to the semimajor axis of the uncertainty ellipse, which corresponds to a position angle (north to east) of 818. Though this correction is a statistically significant deviation (7.7σ) from the gravity-only ephemeris used to build the Spitzer observation sequence, the updated ephemeris still falls inside the Spitzer field of view, which is 5.2 arcmin on each side. The final mosaic presented below and our data analysis are based on the most recent solution for the position (i.e., Micheli et al. 2018), so the only impact on our observations between the pre-HST solution (used for planning our observations) and the post-HST solution would be the error in the rate over the individual 100 s integrations. The difference between the two solutions (i.e., the degree of trailing introduced) over that length of time is completely negligible.

Zoom In Zoom Out Reset image size

Our final mosaic is shown in Figure 3 along with the predicted location of 'Oumuamua. There are no bright coherent sources in this image, so we conclude that we did not confidently detect the source. The 1σ noise level in the final mosaic is ∼0.1 μJy per PSF. This noise level was determined by recovering synthetic sources of various brightnesses that were injected in the final mosaic. Sources as faint as 0.3 μJy could be reliably found and extracted with an error of 0.1 μJy. This noise floor is consistent with our expectations from Spitzer observations of other very faint moving objects (Mommert et al. 2014a, 2014b).

Zoom In Zoom Out Reset image size

There are several ∼2σ "blobs" (essentially, single pixels) in the image, and a "source" that is around 1σ that is located within the uncertainty ellipse. Since our final image is stacked in the moving frame of the target, the likelihood of any of these blobs corresponding to a different true astrophysical object, which would have to be moving at the same rate as "Oumuamua over 30 hr, is vanishingly small. Nevertheless, the presence of these blobs in the image at the 1σ or 2σ level implies that there is correlated noise somewhere in our data stream. The final image has residuals from background stars not fully removed from the mosaics, which cause some streaking across the image. There are likely also fainter cosmic rays and other low-level array effects that survive our filtering and enter into the final image. These residuals get smeared out because of the offsets and mapping between the instrument pixels and the final image pixels on a smaller scale that will lead to correlated "noise". The final image does not look like an image with only random pixel values in each pixel, but is consistent with what we would expect for the object of interest being too faint to detect. Alternately, this could be a tenuous detection of 'Oumuamua at 1σ, or around 0.1 μJy. In the analysis below, we use a 3σ upper limit for our calculations, which implies non-detection, or, at best, a weak detection.

There is no significant vignetting in the 4.5 μm IRAC channel near the edge of the field. However, there could be an impact on sensitivity from the object being off the edge of the detector for some frames due to the dithering. The 3σ position uncertainty ellipse shown in yellow in Figure 3 is fully covered by all exposures to within 2% of the median coverage of the central region (small variations are caused by rejection of bad pixels or pixels affected by background objects or cosmic rays during the exposures). To the right of the yellow ellipse, the coverage drops off roughly linearly along the red path until the end where it reaches a value of around 12% lower coverage than the central part of the image. Therefore, at that extreme end, the upper limit flux would then be 0.32 μJy for objects at this position in the mosaic. The coverage is similar along lines perpendicular to the major axes of the ellipses shown in Figure 3. Given the small area affected and the small (<10%) difference we simply use the global 3σ (0.3 μJy) upper limit for our analysis below.

Based on the discovery of non-gravitational accelerations acting upon the orbit of 'Oumuamua (Micheli et al. 2018), we investigate the possibility of dust and gas activity in this object during our observations. Our non-detection enables the placement of upper limits on the production rates of dust, as well as CO and CO₂ gas; we are unable to constrain the production of other gas species. We use the same formalism that we used in detecting cometary behavior in the NEO Don Quixote (Mommert et al. 2014c)—Bauer et al. (2015) used similar approaches with WISE data—and our measured 3σ 4.5 μm flux density limit (0.3 μJy). Within a 6 pixel (5.2 arcsec) radius aperture (our standard size) we derive Afρ ≤ 0.07 cm following the definition of A'hearn et al. (1984), where A is albedo, f is the filling factor, and ρ is the linear radius of the emission (here, an upper limit). Assuming a dust particle radius of 10 μm, a dust bulk density of 1 g cm⁻³, an albedo of 0.03 that is comparable with cometary dust and compatible with the range of possible albedos that we derived for 'Oumuamua, and a dust ejection velocity equal to the expansion velocity of gas at this distance from the Sun (Ootsubo et al. 2012), we find a 3σ upper limit on the dust production rate of 9 kg s⁻¹. Similarly, we calculate the 3σ upper limit on the CO₂ gas production rate as 9 × 10²² molecules s⁻¹. This can be scaled into a 3σ upper limit for the production of CO (∼9 × 10²¹ molecules s⁻¹) based on the ratio of the CO₂ and CO fluorescence efficiencies (Crovisier & Encrenaz 1983).

This CO upper limit is much lower than the Micheli et al. (2018) value of 4.5 × 10²⁵ molecules s⁻¹ (the most sensitive search in the literature) and implies that the outgassing from 'Oumuamua cannot have CO (or, presumably, CO₂) as a significant component, though the Micheli et al. (2018) CO production rate assumes a relative large body and albedo of 4%. If 'Oumuamua's size were 10–20 times smaller than the Micheli et al. (2018) diameter of 220 m, then the amount of CO outgassing at the upper limit would produce sufficient acceleration. However, an effective spherical diameter of 10–20 m would require an unacceptably high albedo and unacceptably low η, as described below, so this argument is rejected. Overall, we find these upper limit production rates and our upper limit of Afρ to be very low compared to the ensemble of comets (A'Hearn et al. 1995; Ootsubo et al. 2012), supporting the inactivity of 'Oumuamua during our observations.

Our analysis of 'Oumuamua's physical properties is based on a measured flux density upper limit and thermal modeling performed with the NEATM. This model has been specifically designed for use on near-Earth asteroid observations and has been shown to be reasonably accurate over a wide range of cases (Harris et al. 2011; Mommert et al. 2018). It is applicable to thermal emission from any airless body and has been used extensively for comet nuclei as well (Lisse et al. 2005, 2009; Fernandez et al. 2013). A more sophisticated thermophysical model (Mommert et al. 2014a, 2014b, 2018) is not appropriate here due to the lack of information on the target (e.g., spin pole orientation and complex rotation state; shape is somewhat known but not uniquely so) and the upper-limit nature of the flux density measurement.

'Oumuamua is known to have a high-amplitude light curve (e.g., Jewitt et al. 2017; Knight et al. 2017; Meech et al. 2017; Bolin et al. 2018; Micheli et al. 2018) with, most likely, a period of 6–8 hr. Since our observations spanned 33 hr (clock time), any light curve effects are smoothed out, and we observe only the average flux. Furthermore, even though the Spitzer viewing geometry of 'Oumuamua is very different from that seen by observatories on and near the Earth, because 'Oumuamua is in an excited rotation state (Belton et al. 2018; Drahus et al. 2018; Fraser et al. 2018), we likely observed the same time-averaged projected surface area that would have been seen from Earth. Furthermore, the light curves presented in Belton et al. (2018) are not sinusoidal but rather have broad maxima and narrow minima, so our 33 hr integration is likely not corrupted by faint epochs in the light curve. Even in the case of a 55 hr period, one possible solution suggested by Belton et al. (2018), our observations span a significant fraction of the entire rotation and therefore included something close to the time-averaged cross-sectional area, except in the case of a pathological orientation.

We do not include uncertainties on the ratio of infrared to optical reflectances, as the impact of this ratio barely affects the overall results of this study, especially in the light of the large uncertainties on the beaming parameter η and hence the geometric albedo p_V.

We investigate the applicability of NEATM for this study given the high aspect ratio of 'Oumuamua (Meech et al. 2017; McNeill et al. 2018) and the assumption of sphericity in NEATM. For this purpose, we use an asteroid thermophysical model (Mommert et al. 2014a, 2014b, 2018) to derive the thermal and reflected solar flux density of the body, assuming both a highly elongated shape and a highly oblate shape (following Belton et al. 2018). Based on McNeill et al. (2018), we assume a triaxial ellipsoidal shape with semimajor axes 6:1:1 for the highly elongated shape and 1:$\sqrt{6}$: $\sqrt{6}$ for the highly oblate shape, both in arbitrary units. Furthermore, we use the geometry during our Spitzer observations, the period (7.34 hr) derived by Meech et al. (2017), and assume a geometric albedo of 0.03 (in agreement with our NEATM-derived lower limit) and typical small-body values for thermal inertia and surface roughness. We simulate the flux density observed at Spitzer over one-quarter of the target's rotation and derive the average flux density, which is the quantity measured in our observations by combining all available data. Finally, we form the ratio of the average flux density derived for a spherical body (NEATM assumption) to the average flux density derived from the elongated shape and oblate shape models. Deriving this ratio minimizes the effects of the choice of the geometric albedo, surface roughness, and thermal inertia used in the simulation.

We find that this flux density ratio varies as a function of the target's spin axis latitude (as a proxy for the aspect angle of our observations). In the case of the elongated shape, a spin axis latitude of 90° (equator-on view), the ratio is 1, and rotational effects are averaged out during our observations. The ratio decreases to 0.5 for spin axis latitudes approaching 0 (pole-on view), which represents an extreme case. In the case of the oblate shape, we find ratios of around [0.5, 1, 2] for β = [0, 32.7, 90] degrees, respectively. As no information on the spin axis orientation of 'Oumuamua is available, we use the average¹⁵ latitude of 327, leading to a flux density ratio of 0.5 for the elongated shape model and 0.7 for the prolate shape. This mismatch between the flux densities of the different shapes is insignificant compared to the uncertainties introduced by the lack of knowledge of the surface properties (η) of 'Oumuamua. We therefore conclude that our use of NEATM is acceptable. We also note that the uncertainties in the results from the H_V uncertainties are small compared to the uncertainties from our lack of constraints on η.

4.3.1. Low-albedo Solution

4.3.2. Mid-range Albedo Solution

4.3.3. High-albedo Solution

There are several possible interpretations of our results, as follows. We note that in all cases, given our upper limit on CO and CO₂ production rates, some other gas species (e.g., water) must also have been emitted to explain the non-gravitational acceleration observed by Micheli et al. (2018). We can place no constraint on these other gas species.

(1) 'Oumuamua could have a high η, which would not be unusual for asteroids but would be very unusual for comets—although a body in an excited rotation state might have a higher than expected η value. In the high η case, the albedo is low, which means the diameter is large. However, a large body implies a large outgassing rate, which we do not see for CO and CO₂ and for dust. Conversely, (2) 'Oumuamua could have a low η, in accordance with expectations for cometary bodies. However, this requires an albedo that is much higher than that expected for comets. This high albedo corresponds to a small diameter, which is favored, considering our upper limits on gas and dust production. (3) Intermediate values of η, albedo, and diameter are also possible, though these are not really consistent with any expectations.

In conclusion, there is no simple asteroidal or cometary physical model that agrees with expectations and previous work (including non-gravitational acceleration) and our results for all of η, albedo, and diameter. One plausible explanation is that 'Oumuamua was a dormant comet nucleus reactivated, after millions of years in interstellar space, by heating during its close passage by the Sun. This reactivation either destroyed the thin dark mantle expected to be created by cosmic rays and galactic ultraviolet radiation (e.g., Lisse et al. 1998, 2004) and/or coated the surface with an optically thick layer of new, fresh ice. In the latter case, the bright coating could plausibly have come from CO, CO₂, or water, as follows.

We assume that 'Oumuamua is outgassing 9 × 10²² molecules of CO₂ per second (see above). In the high-albedo case, the effective spherical diameter of 'Oumuamua is around 98 m, and the surface area is therefore around 3 × 10⁴ m² (taking 49 m as the radius of the equivalent sphere). If we require a uniform surface layer that is 10 μm thick—so that the surface appears bright with CO₂ ice for observations made at 4.5 μm—then the volume of this surface layer is around 0.3 m³.

The density of CO₂ ice is approximately 1.5 g cm⁻³. The mass required to create a surface layer of 0.3 m³ is therefore 4.5 × 10⁵ g. We calculate the number of CO₂ molecules required as

$$\begin{eqnarray*}&&\displaystyle \frac{4.5\times {10}^{5}\,{\rm{g}}}{44\,{\rm{g}}/\mathrm{mole}}\times 6.02\times {10}^{23}\,\mathrm{molecules}/\mathrm{mole},\end{eqnarray*}$$

which is around 6 × 10²⁷ molecules of CO₂. Given the CO₂ upper limit of 9 × 10²² molecules s⁻¹ derived above, depositing 6 × 10²⁷ molecules of CO₂ requires only around 67000 s, or around 0.8 days—far less than the few weeks of 'Oumuamua's perihelion passage time. Thus, even if the efficiency of this process is small, it is still quite plausible that a low level of activity—induced by solar heating of a near-subsurface CO₂ reservoir—could produce enough material to coat the surface with bright, fresh CO₂ and increase the albedo to the relatively high value required in our high-albedo case.

Alternately, heating of water ice into gas could present a plausible scenario. Cometary dust to gas ratios are typically around 5:1, so our dust emission upper limit of 9 kg s⁻¹ corresponds to 1.8 kg s⁻¹ as an upper limit for gas emission. If we assume that all of this gas emission is in H₂O, then we find

$$\begin{eqnarray*}&&\displaystyle \frac{1.8\times {10}^{3}\,{\rm{g}}\,{{\rm{s}}}^{-1}}{18\,{\rm{g}}/\mathrm{mole}}\times 6.02\times {10}^{23}\,\mathrm{molecules}/\mathrm{mole},\end{eqnarray*}$$

which is around 6 × 10²⁵ molecules s⁻¹, enough to produce the non-gravitational accelerations reported by Micheli et al. (2018).

CO+CO₂ ice in typical solar system comets is around 15% of the water abundance. Here our limits imply around 0.15% for this ratio—a factor of 100 times smaller. This could imply that 'Oumuamua was heated to ∼100 K prior to our observations—either by our Sun, or before entering our solar system. 'Oumuamua, if propelled by water ice sublimation at 6 × 10²⁵ molecules s⁻¹ while producing only ≤9 × 10²² molecules s⁻¹ of CO+CO₂, must therefore have been previously devolatilized of these more volatile ices.

Unfortunately, we do not have pre-perihelion observations to compare to these post-perihelion observations to test the hypothesis that 'Oumuamua brightened during its perihelion passage. Further modeling of 'Oumuamua's outgassing—whether CO, CO₂, or some other species—would be very beneficial.

A plausible analogy for such activity-produced resurfacing is the well-studied comet 67P, the target of the Rosetta mission. Keller et al. (2017) and Liao et al. (2018) showed that the nucleus of 67P was partially resurfaced through re-condensation of volatiles released from the nucleus; Liao et al. (2018) found that the deposition rate of water ice could be up to several microns in an hour near perihelion. While this does not correspond directly to low levels of activity and an albedo increase, as proposed here for 'Oumuamua, it is nevertheless evidence that activity can resurface small-body surfaces after perihelion passage, at the order of magnitude required for 'Oumuamua (1–10 μm deposited in days or weeks). Bolin et al. (2018) saw no color changes as a function of rotation, which could imply a relatively uniform resurfacing process. This suggested emission must be too small to create a measurable change in velocity after the first 'Oumuamua observations (i.e., the beginning of the observational arc), or else occurred after perihelion but before the discovery observation of 'Oumuamua (see Figure 1), in order to be consistent with the results reported in Micheli et al. (2018).

Another possible analogy is the well-studied comet Shoemaker-Levy 9. Sekanina (1995) shows that after the breakup of this body from tidal forces exerted by Jupiter, many fragments appeared intrinsically brighter—as if fresh ice had just been revealed or deposited onto their surfaces. It is possible that the shape and/or rotation state of 'Oumuamua were affected by its passage near the Sun, and interior volatiles may also have been liberated onto the surface at the same time.