to be submitted to AJWater ice in the Kuiper beltM.E. BrownDivision of Geological and Planetary Sciences, California Institute of Technology, Pasadena, [email protected]. SchallerNASA Dryden Aircraft Operations Facility, Palmdale, CA 93550 and National SuborbitalEducation and Research Center, University of North Dakota, Grand Forks, ND, 85202W.C. FraserDivision of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA91125ABSTRACTWe examine a large collection of low resolution near-infrared spectra of Kuiper beltobjects and centaurs in an attempt to understand the presence of water ice in theKuiper belt. We find that water ice on the surface of these objects occurs in threeseparate manners: (1) Haumea family members uniquely show surfaces of nearly purewater ice, presumably a consequence of the fragmentation of the icy mantle of a largerdifferentiated proto-Haumea; (2) large objects with absolute magnitudes of H < 3 (anda limited number to H = 4.5) have surface coverings of water ice – perhaps mixedwith ammonia – that appears to be related to possibly ancient cryovolcanism on theselarge objects; and (3) smaller KBOs and centaurs which are neither Haumea familymembers nor cold-classical KBOs appear to divide into two families (which we referto as “neutral” and “red”), each of which is a mixture of a common nearly-neutralcomponent and either a slightly red or very red component that also includes waterice. A model suggesting that the difference between neutral and red objects s due toformation in an early compact solar system either inside or outside, respectively, of the∼20 AU methanol evaporation line is supported by the observation that methanol isonly detected on the reddest objects, which are those which would be expected to havethe most of the methanol containing mixture.Subject headings: solar system: Kuiper belt — solar system: formation — astrochem-istry– 2 –1. IntroductionThe Kuiper belt is composed of low-temperature remnants of the outer regions of the proto-planetary disk which never became incorporated into planets. By analogy to short-period comets,which are derived from the Kuiper belt, and from cosmochemical considerations, it is expected thatwater ice is a major constituent of the composition of Kuiper belt objects (KBOs). That water icewas the first clearly identified consituent on the surfaces of small KBOs and of centaurs (formerKBOs which are currently on short-lived planet-crossing orbits) was thus not a surprise (Brownet al. 1998, 1999).For a large majority of KBOs which have been studied, water ice remains the only identifiablesurface constituent, even though the visible absorption features are so small that, in most cases, itis clear that water ice is a relatively minor component of the surface (Barkume et al. 2006; Guilbertet al. 2009). One major exception is Haumea and its satellites and collisional family, which appearto have surfaces composed of nearly pure water ice, thought to be exposed when the differentiatedicy mantle of the proto-Haumea was removed in a giant impact (Brown et al. 2007). The othermajor exception is the largest Kuiper belt objects, which are cold and massive enough to maintainvolatile atmospheres and frosts over the age of the solar system Schaller & Brown (2007), and whosebedrock is covered and thus unobservable.for the majority of KBOs with water ice present in the spectrum, little connection has beenmade between the water ice visible on the surface and any other properties of the KBOs. No simplecorrelation appears between dynamical or color properties and water ice absorption (Barkume et al.2006; Guilbert et al. 2009), so emphasis has instead mostly been on detailed modeling to determinesurface constituents (i.e. Barucci et al. 2011) and quantify the absence or presence of ice.Here we examine the moderate-sized and smaller KBOs and examine water and other ices indetail. The goal is to examine these objects as a class, rather than perform detailed modeling ofindividual objects, in the hope of statistically understanding the causes and states of ices on theseobjects.2. Observations and analysisIn order to examine the properties of ices in the Kuiper belt, we assemble a nearly uniformset of low-resolution (λ/∆λ ∼ 160) 1.5-2.4 mum reflectance spectra, almost all obtained usingNIRC – the first-generation near-infrared spectrograph at the Keck 1 telescope (Matthews & Soifer1994). Spectra were collected from the surveys of Brown (2000) and Barkume et al. (2008). Wealso include the NIRC spectrum of Charon (Brown & Calvin 2000) obtained with the identicalinstrument, and the only available spectrum of 2007 OR10, obtained with the FIRE spectrographat the Magellan telescope (Brown et al. 2011a).To augment this existing sample, we obtained new spectra of 15 KBOs and centaurs with– 3 –the NIRC spectrograph on the Keck telescope until the spectrograph was retired in 2009. Theobservations and data reduction were performed identically to the original Keck surveys, and detailsare given in Table 1. The KBO 19521 Chaos (1998 WH24) appeared unusual in the previous survey,so it was reobserved to determine if it has a unique spectral type or if the observations were faulty.The new spectrum closely resembles other KBO spectra, so we assume the previous spectrum wasin error and only retain the new spectrum. Figure 1 shows all of the newly obtained spectra.The full spectral sample (see Table 2) includes 64 objects. To examine ices on the smallerobjects, we remove from consideration objects known to be part of Haumea’s collisional family andalso the largest objects for which methane dominates the spectrum (Eris, Pluto, Makemake, andSedna). A total of 57 objects remain in the sample at this point.We are interested in characterizing the amount of water ice absorption present in each ofthe spectra, but spectral models give non-unique results for the fraction of water ice present ona surface. Differences in assumed grain size, type of mixing, and non-water ice components canlead to order-of-magnitude or greater variations in the derived water ice abundance. Frequently, ameasurement of the depth of the 2µm water absorption feature is used as a model-independent proxyfor the amount of water ice present (Brown 2000; Jewitt & Luu 2001; Barucci et al. 2011), but thismethod ignores much of the information contained in a broader spectrum. Barkume et al. (2008),therefore, developed a