THE ATOMIC AND MOLECULAR CONTENT OF DISKS AROUND VERY LOW-MASS STARS AND BROWN DWARFS

The strongest atomic/ionic lines detected in Spitzer spectra of T Tauri disks are the H i (7–6) at 12.37 μm and the [Ne ii] at 12.81 μm (e.g., Pascucci et al. 2007; Lahuis et al. 2007). We have searched for these emission lines in our sample and report the first [Ne ii] detections toward BD disks (see Figure 3).

Zoom In Zoom Out Reset image size

To estimate line fluxes and upper limits, we followed the procedure outlined in Pascucci et al. (2007). In brief, when a [Ne ii] line is detected, we fit a Gaussian plus a first-order polynomial to the spectral region within ±0.05 μm of the line (10 pixels). In the case of nondetections, we fit the same wavelength region with a first-order polynomial and compute a 3σ upper limit from the rms of the continuum-subtracted spectrum. The 1σ uncertainties on detected lines are also computed from the rms of continuum-subtracted spectra in a region just outside the line. Our results are summarized in Table 4. The brightest [Ne ii] emission is from BD #4 with a line flux of ∼23 × 10⁻¹⁶ erg s⁻¹ cm⁻². Line fluxes from BD #1 and #2 are ∼3 and five times weaker, respectively. We also note that the emission toward source #2 is only a tentative detection (slightly above 2σ) and should be confirmed via future observations with James Webb Space Telescope (JWST). For the other five targets, we estimate 3σ upper limits in the [Ne ii] line that range between (4–9) × 10⁻¹⁶ erg s⁻¹ cm⁻². Similar upper limits are found for the H i(7–6) line, which is never detected in our sample. We will discuss in Section 5.1 the likely origin of the [Ne ii] emission toward very low mass stars and BDs.

The Spitzer IRS SH module covers many rotational transitions from H₂O and OH, as well as rovibrational bands from simple molecules such as C₂H₂, HCN, and CO₂ (Q-branches at 13.7, 14.0, and 14.97 μm, respectively). These molecules are frequently detected toward T Tauri stars with disks and likely trace the warm disk atmosphere (e.g., Pontoppidan et al. 2010). The H₂ infrared lines have been also detected toward T Tauri disks (Lahuis et al. 2007; Carr & Najita 2011), though some detections may be tracing outflows or cloud emission rather than the disk itself (Lahuis et al. 2007).

Here, we report the first detections of H₂ pure rotational lines toward BD disks (see Figure 4). The procedure to measure line fluxes and upper limits is identical to that used for the [Ne ii] lines (Section 3.1 and Table 4). Firm H₂ detections (⩾3σ) are only available for sources #3, 4, and 6 in the H₂ S(2) at 12.28 μm and for source #4 in the H₂ S(1) at 17.03 μm. Tentative detections (above 2σ) are also reported for #1 and 3 in the H₂ S(1) transition (see Figure 4 and Table 4). Object #4 is the only one where both H₂ transitions are firmly detected and point to similar line fluxes. When lines are not detected, 3σ upper limits are found to be ∼(4–10) × 10⁻¹⁶ erg s⁻¹ cm⁻². We will further discuss these detections and upper limits in Section 5.2.

Zoom In Zoom Out Reset image size

We also report detections of the C₂H₂ and HCN rovibrational bands toward objects #2, 3, 6, and 7 (see later the continuum-subtracted spectra in Figure 8—the latter HCN detection is very weak). In all objects, the C₂H₂ band emission appears stronger than the HCN emission. This pattern was already recognized in low-resolution Spitzer spectra from 19 very low mass stars and BDs belonging to the Cha I star-forming region (Pascucci et al. 2009) and is opposite to what is seen in T Tauri disks (Pascucci et al. 2009; Carr & Najita 2011). The five-times-higher resolution of the spectra presented here enables us to detect fainter emission and model the shape of the band emission (see Section 4). We also present the first CO₂ detections toward BD disks (objects #2, 3, and 7) and tentative detections of H₂O lines in the spectra of objects #1 and #4 (see Figure 5). We do not have any detection of OH lines in our spectra.

Zoom In Zoom Out Reset image size

To quantify the strength of the molecular emission, we measure fluxes in the H₂O 17.22 and 18.17 μm lines⁶ and in the organic molecular bands using a procedure similar to that outlined in Salyk et al. (2011). First, we fit a linear continuum using pixels shortward and longward of the emission. Then, we subtract the continuum emission and integrate the flux within wavelength regions that cover the full emission (see Note to Table 5). Line flux upper limits are calculated from the rms in the continuum-subtracted spectra using line widths appropriate to each transition: an unresolved line width for the H₂O rotational transitions, three times the resolution for the C₂H₂ and HCN bands, and two times the resolution for the CO₂ band. The 1σ uncertainties on detected lines are also calculated from the rms in the continuum-subtracted spectra. Table 5 reports our line flux measurements and 3σ upper limits when lines are not detected. The strongest band emission is typically from C₂H₂, followed by HCN and by CO₂. Water lines are the weakest and tentatively detected at the 2σ–3σ level only toward the two sources with no emission from organic molecules (objects #1 and #4). These tentative detections should be prime candidates for followup observations with JWST. Figure 6 compares molecular line flux ratios from our sample with those from T Tauri disks (Pontoppidan et al. 2010; Salyk et al. 2011). While there is some overlap between the two samples, very low mass star and BD disks tend to occupy the upper right corner of the plot. The data in Figure 6 confirm the result from Pascucci et al. (2009) that BD disks have larger C₂H₂/HCN flux ratios than the average T Tauri disk. In addition, it shows that the HCN/H₂O (and C₂H₂/H₂O) flux ratios tend to be larger in BD than in T Tauri disks. We will further discuss the significance of this result in Section 5.3.

Zoom In Zoom Out Reset image size

Ground-based observations have spectrally resolved the brightest [Ne ii] lines detected with Spitzer toward T Tauri stars (Herczeg et al. 2007; van Boekel et al. 2009; Najita et al. 2009; Pascucci & Sterzik 2009; Pascucci et al. 2011; Sacco et al. 2012; Baldovin-Saavedra et al. 2012). These data have demonstrated that [Ne ii] emission can trace either (1) jets and outflows far from the star (typically toward high-accreting stars) or (2) the hot (∼10,000–5000 K) disk surface ionized by stellar EUV and/or X-ray photons (in moderate- to low-accreting stars, M_acc ≲ 10⁻⁸ M_☉ yr⁻¹). When disk emission is detected, most [Ne ii] lines are blueshifted by a few km s⁻¹ with respect to the stellar velocity, pointing to unbound gas in a photoevaporative thermal wind (e.g., Pascucci & Sterzik 2009; Sacco et al. 2012). It is debated whether [Ne ii] emission traces either the fully ionized EUV or the partially ionized X-ray disk layer, hence mass loss rates via photoevaporation are yet unconstrained (e.g., the recent review by Alexander et al. 2013).

To decide whether our [Ne ii] detections trace jets/outflows or the BD disk surface, we plot the [Ne ii] luminosity ($L_{{\rm Ne\,\scriptsize{II}}}$) as a function of the stellar mass accretion rate (M_acc). Together with our sample, we include all other Spitzer detections and upper limits reported in the literature for T Tauri stars (see Figure 9). The complete [Ne ii] sample spans almost six orders of magnitude in M_acc and three orders of magnitude in $L_{{\rm Ne\,\scriptsize{II}}}$. When we extrapolate the $L_{{\rm Ne\,\scriptsize{II}}}$–M_acc relation found for T Tauri stars with known jets (Güdel et al. 2010) to the BD regime, we see that only the strongest of our [Ne ii] detections (from source #4) can be attributed to jets or outflows. Indeed, Bouy et al. (2008) detected strong forbidden optical lines (O i, S ii, and N ii) in the high-resolution spectrum of source #4 (2MASS J04442713+2512164), suggesting that this object is undergoing significant mass loss via a jet or outflow. We conclude that most of the [Ne ii] emission from source #4 is not associated with the disk but produced in circumstellar material shocked by jets or outflows.

Zoom In Zoom Out Reset image size

BDs #1 and 2 have lower $L_{{\rm Ne\,\scriptsize{II}}}$ than source #4, by factors of ∼3 and 5, respectively (the detection toward #2 is only tentative). It is interesting to speculate what X-ray or EUV fluxes are needed to reproduce these lower luminosities, assuming that the [Ne ii] emission originates entirely in the ionized surface of BD disks. In the simplest scenario in which neon atoms are ionized solely by stellar EUV photons, we can convert the observed [Ne ii] luminosities into the rate of ionizing photons reaching the disk surface (Φ_EUV, Hollenbach & Gorti 2009, HG09). The [Ne ii] detections from #1 and 2 correspond to Φ_EUV of 0.6–1.2 × 10⁴⁰ s⁻¹. These values are ∼5–10 times lower than the Φ_EUV inferred for T Tauri stars (Alexander et al. 2005; Pascucci et al. 2012) and represent upper limits to the EUV luminosities impinging on the disk surface. This is because part, or most of, the [Ne ii] emission could arise from a disk layer ionized by stellar X-rays (see below). It is unclear whether EUV photons are from the stellar chromosphere and/or accretion processes (see, e.g., Alexander et al. 2005). If the latter dominate, then Φ_EUV should scale with the accretion luminosity or, equivalently, with the stellar FUV luminosity (L_FUV), since the two quantities are tightly correlated (Yang et al. 2012). L_FUV for T Tauri stars cluster between 0.001–0.01 L_☉ (see Figure 12 in Yang et al. 2012) while the L_FUV we estimate for our sample are ∼10–1000 times lower (see Tables 1 and 3). X-ray luminosities (L_X) decrease with stellar mass less steeply than L_FUV. The L_X of our sources (most of which are estimated from the source mass—see Table 3) are ∼10–100 times lower than the average L_X of T Tauri stars (∼10³⁰ erg s⁻¹; Güdel et al. 2007). These trends suggest that X-rays, rather than EUV photons, are the main ionizing source of neon atoms in BD disk atmospheres. According to the models of Schisano et al. (2010), unscreened X-ray luminosities of ∼5 × 10²⁸ erg s⁻¹ are necessary to produce $L_{{\rm Ne\,\scriptsize{II}}}\sim 10^{27}$ erg s⁻¹, which is close to our [Ne ii] fluxes and 3σ upper limits. Therefore, the fact that the majority of our disks remain undetected in the [Ne ii] line is consistent with their low X-ray luminosities. In summary, a disk origin is plausible for the [Ne ii] emission reported toward BDs #1 and 2, and X-ray ionization is likely the major contributor to neon ions in their disk atmosphere.

The detection of [Ne ii] emission from BD disk atmospheres shows that even young very-low-mass objects have enough high-energy photons to create a fully or partially ionized hot surface. Whether this surface is unbound and part of a photoevaporative flow, as in moderate- to low-accreting T Tauri stars, cannot be tested with the IRS low resolving power. A resolution of ∼10 km s⁻¹ is needed to spectrally resolve lines tracing the flow and detect the few km s⁻¹ blueshifts in peak centroid with respect to photospheric lines. Due to the faintness of our sample, and BD disks in general, this test will have to wait for high-resolution spectrographs mounted on 40-m-class telescopes.

H₂ pure rotational lines are found to be relatively weak toward T Tauri stars with respect to emission from other molecules and from atoms or ions. The Spitzer "Cores to Disks" legacy program reports that only 6 out of 76 young (∼Myr-old) disks present the H₂ S(2) line in their mid-infrared spectra (Lahuis et al. 2007). The H₂ S(1) line is detected in only 1 out of 76 spectra. Because of the large beam of Spitzer and the lack of "sky" observations for this program, extended (envelope, local cloud, jet) emission can potentially contaminate some of the reported line fluxes (Lahuis et al. 2007). Deeper observations and a better sky subtraction increase the H₂ detection rate. Carr & Najita (2011) identify the H₂ S(1) transition toward 6 out of their 11 Taurus T Tauri stars but do not provide H₂ S(2) fluxes because of contamination from much stronger H₂O and OH lines around 12.28 μm. Our observing technique and data reduction are similar to theirs. In addition, our sample exhibits weaker water and OH emission than T Tauri disks (see Section 3.2) resulting in an H₂ detection rate similar to that reported by Carr & Najita (2011), even toward sources that are about an order of magnitude fainter in the continuum.

Similarly to the continuum emission, our H₂ fluxes are about an order of magnitude lower than those observed toward T Tauri disks (Carr & Najita 2011) and predicted from disks irradiated by FUV and X-ray photons from T Tauri stars (Nomura et al. 2007). This is in line with the lower FUV and X-ray luminosity of very low mass stars and BDs, and suggests that our H₂ lines are indeed tracing disk gas. In the assumption that this gas is in LTE and the lines are optically thin,⁷ we can use the H₂ S(2) and H₂ S(1) detections (or upper limits in one of the two) to compute a gas excitation temperature, T_ex (see, e.g., Thi et al. 2001 for the equations). Our measurements are consistent with $F_{\rm H_2 S(1)} \sim F_{\rm H_2 S(2)}$, which implies T_ex = 645 K (with the assumptions outlined above). Including the 1σ uncertainties on the measured fluxes for objects #4 (Table 4) expands the range of possible T_ex to 450–1200 K. If we then use these temperatures to compute a gas mass, we find that the H₂ rotational lines trace only ∼2–7 × 10⁻⁴ M_J (with the largest value for gas at the lowest T_ex in the range reported above). Gas masses of BD disks (M_disk) have not been measured. However, the few submillimeter and millimeter detections of the continuum dust emission suggest that the M_disk/M_BD ratio is a few percent, if the gas-to-dust-ratio is equal to the interstellar medium value of 100 (Klein et al. 2003; Scholz et al. 2006; Ricci et al. 2012). This result points to total BD disk masses of ∼M_J, more than three orders of magnitude larger than what we find from the H₂ pure rotational lines. This discrepancy is not surprising since models of disks irradiated by stellar FUV and X-rays predict that H₂ pure rotational lines trace mainly the warm surface of protoplanetary disks and are thus insensitive to the more abundant and cooler disk gas (e.g., Nomura et al. 2007).

The combined H₂ and [Ne ii] results show that BD disks have an overall structure similar to T Tauri disks: a hot ionized layer tracing the uppermost disk atmosphere and a warm (∼600 K) molecular layer at lower vertical heights.

We use the molecular column densities inferred from our LTE modeling (Section 4) to compute relative abundances of simple organic molecules and water in disks around very low mass stars and BDs. This approach complements the discussion of line flux ratios in Section 3.2 and has been previously taken for T Tauri disks (Carr & Najita 2011; Salyk et al. 2011). There are a number of caveats in using this simple approach (e.g., Carr & Najita 2011). Perhaps the most important one is that the radial and vertical regions probed by these molecules may be different, whereas they are often assumed to be the same due to the lack of spatial information with Spitzer observations. In addition, these infrared emission lines trace the warm disk surface, which may not be representative of the overall disk composition.

With these caveats in mind, our modeling results suggest that C₂H₂ is typically more abundant than HCN in BD disks and both C₂H₂ and HCN are more abundant than CO₂. The column density of water is very sensitive to the assumed gas temperature and useful limits on its abundance can be placed only if gas is at T ⩾ 600 K (even in objects #2 and 3). Figure 10 compares column density ratios of C₂H₂/HCN versus HCN/H₂O for our BD disks (red symbols) and T Tauri disks from the literature (black symbols). Horizontal red dashed lines connect the two water column densities reported in Table 6. The most obvious difference between BD and T Tauri disks is in their C₂H₂/HCN column density ratios. BD disks appear to have a systematically higher C₂H₂/HCN column density ratio: the median ratio for BD disks is ∼4 while that for the complete T Tauri sample is ∼0.6, not considering the seven C₂H₂ upper limits in the T Tauri sample. Thus, on average, C₂H₂ is about seven times more abundant than HCN in BD disks than in T Tauri disks. This difference was first pointed out from the analysis of low-resolution Spitzer spectra (Pascucci et al. 2009) and is corroborated by our higher resolution spectra on a different sample of BD disks. This finding is also in line with the typically higher C₂H₂/HCN strength in BD than in T Tauri disks (Section 3.2 and Figure 6). If water vapor is warm (T ⩾ 600 K), then Figure 10 also hints at a reduced abundance of H₂O in BD disks with respect to T Tauri disks. This possible trend is also apparent in the model-independent plot of line flux ratios, especially for the U Sco sources (Figure 6).

Zoom In Zoom Out Reset image size

We now turn to theoretical models of disks irradiated by stellar FUV and X-rays (Agúndez et al. 2008; Najita et al. 2011; Walsh et al. 2012) to seek explanations for the trends in molecular abundances discussed above. We note that models report vertical column densities integrated over the height of the disk whereas observations measure line-of-sight column densities and infrared observations like ours only trace the warm disk atmosphere. So any comparison (in this as well as in other works) relies on the assumption that vertically integrated column densities are representative of the column densities in the warm disk atmosphere.

The models of Agúndez et al. (2008) are for typical T Tauri disks around Sun-like stars and include FUV heating but not X-ray heating. They find that the steady-state vertical column density of H₂O is relatively constant out to ∼3 AU from the star. On the contrary, the organic molecules HCN, C₂H₂, and CO₂ have a stronger radial dependence, with C₂H₂ (and even more CO₂) becoming less abundant in the very inner disk (inside ∼0.5 AU). At ∼1 AU, the radius likely probed with Spitzer around T Tauri stars, HCN is more abundant than C₂H₂ by a factor of ∼10, in agreement with inferred column density ratios for several T Tauri disks (Carr & Najita 2011; Salyk et al. 2011). Only beyond ∼2.5 AU does C₂H₂ become more abundant than HCN in the Agúndez et al. (2008) model. The models of Walsh et al. (2012) include both FUV and X-ray heating. As in Agúndez et al. (2008), they predict that H₂O is the most abundant molecule in the gas phase (after H₂) inside the snow line. However, unlike Agúndez et al. (2008), they find that HCN remains more abundant than C₂H₂ throughout the radial extent of the disk. The differences among these model results are likely due to differences in the computed gas disk temperature. Our higher C₂H₂/HCN in BD disks cannot be explained as the result of these two molecules tracing different radii in the Walsh et al. (2012) models. In the Agúndez et al. (2008) models, our result would suggest that C₂H₂ traces cooler gas, at larger radii, than HCN. This does not seem to be the case, since our T estimates from fitting the C₂H₂ band do a good job in reproducing the profile of the HCN emission band. If the predicted vertical column densities reflect atmospheric column densities, a depletion of water in BD disks (as hinted at by our line flux ratios and LTE modeling) cannot be explained as the result of radial dependencies in molecular column densities. Detailed thermochemical models of BD disks should be carried out to determine whether the different properties of BDs and their disks could explain the observed trends.

Along these lines, Najita et al. (2011) have explored how changing the properties T Tauri disks and stellar heating might affect molecular abundances. As a cautionary note, the models of Najita et al. (2011) include accretional and stellar X-ray heating but neglect FUV heating and UV-induced processes, which generally are found to be more important than X-ray processes (Walsh et al. 2012). Najita et al. (2011) find that low accretional heating (lower disk temperature) reduces the warm molecular columns of most species but further decreases the C₂H₂/HCN ratio while keeping large column densities of H₂O. Both trends contrast with our results, suggesting that the lower accretional heating of BD disks is not responsible for the observed trends in molecular abundances. Increasing the grain size from 0.7 to 7 μm in the disk atmosphere increases the warm molecular columns of most species as well as the C₂H₂/HCN ratio. While this trend is in qualitative agreement with the observed one, 7 μm grains in the disk atmosphere enhance the C₂H₂/HCN ratio to only 0.07—still too low in comparison with the observed one. At the same time, it is very unlikely that BD disks have even larger grains in their atmosphere because many of them show 10 μm silicate emission features in their low-resolution Spitzer spectra (Pascucci et al. 2009). Such emission features point to the presence of dust grains with sizes ⩽5 μm. Another way of increasing the C₂H₂/HCN ratio in the models of Najita et al. (2011) is by decreasing L_X (green symbols in Figure 10). However, the C₂H₂/HCN ratios we measure for BDs are reproduced for an L_X of 10²⁶ erg s⁻¹, almost two orders of magnitude lower than that in our sample (see Tables 1 and 3). A decrease in L_X should also result in a very modest change (just a factor of a few) in the HCN/H₂O ratio, with water being more abundant than HCN by a factor of ∼100 like in T Tauri disks. This latter trend is not consistent with our BD sample if the water-emitting layer is warm (T ⩾ 600 K). Even if the water nondetections alone cannot exclude a lower gas temperature, the detection of H₂ and other organic molecules show that a warm T ∼ 600 K molecular layer is present in BD disk atmospheres. In T Tauri disk atmospheres, this layer has also a large H₂O column density inside ∼1 AU and produces the observed water emission spectrum (Najita et al. 2011). Hence, it is plausible that there is some warm (T ⩾ 600 K) water in BD disk atmospheres but perhaps not enough to be detectable in our spectra. Finally, lowering the O/C ratio from 1.5 to 1 (blue symbols in Figure 10) enhances the C₂H₂/HCN and HCN/H₂O molecular ratios, which is in line with our trends. In the framework of these models, BD disk atmospheres appear to have a lower O/C ratio (close to 1) than T Tauri disk atmospheres (see also Mandell et al. 2012 who find an O/C = 1.5 from the analysis of near-infrared C₂H₂ and HCN rovibrational bands in T Tauri disks). When oxygen is depleted relative to carbon, water abundances decrease while the abundance of hydrocarbons, especially C₂H₂ and CH₄, increase (Najita et al. 2011). Unfortunately, the high-resolution Spitzer spectrograph does not cover the rovibrational band of CH₄, and testing the inference of an enhanced carbon chemistry in BD disks will require additional observations. We also note that the CO₂ abundance decreases only slightly (by about a factor of two) when going from an O/C = 1.5 to 1, meaning that C₂H₂ and HCN remain more abundant than CO₂, in agreement with our finding.

Finally, we wish to discuss our results in the context of the proposed correlation between the strength of HCN/H₂O and the disk mass of T Tauri stars. Najita et al. (2013) attribute this correlation to higher disk masses producing more nonmigrating planetesimals that lock up water ice beyond the snow line, thereby decreasing the O/C ratio in the inner disk. Of our very-low-mass-star and BD disks, only three (#1, 3, and 4) have measured masses (Scholz et al. 2006). Their average disk mass is ∼10⁻³ M_☉, which agrees with the disk mass over the stellar mass being about 1% throughout the M spectral type and down to the BD regime (Andrews et al. 2013; Mohanty et al. 2013). If BDs follow the trend proposed by Najita et al. (2013), we should have measured HCN/H₂O flux ratios as low as ∼0.6 for the water line at 17.22 μm,⁸ indicating the prevalence of water vapor (high O/C ratio) in the inner region of low-mass disks. On the contrary, we find HCN/H₂O flux ratios all larger than 3 (see Figure 6), pointing to a depletion of water vapor in the inner regions of BD disks. Clearly, disks around very low mass stars and BDs do not follow the proposed trend between the HCN/H₂O flux ratio and the disk mass.

Could our observations still be explained with an analogous process, i.e., could low-mass BD disks have formed more nonmigrating icy planetesimals than T Tauri disks? As we mention in the Introduction, our BD sample may be slightly biased toward older objects with no strong jets or outflows. These disks may have had more time to produce large nonmigrating icy planetesimals than the perhaps younger sample of T Tauri disks discussed in the literature so far. Even for disks in the same star-forming region, there is growing observational evidence that the first steps of planet formation proceed faster in BD than in T Tauri disks at radial distances tracing the same disk temperature. First, 10 and 20 μm silicate emission features suggest that the warm (100–500 K, Juhász et al. 2009) atmospheres of BD disks contain, on average, larger grains than those of T Tauri disks (Pascucci et al. 2009; Riaz et al. 2012). Second, detailed SED modeling of "coeval" disks finds that disks around very low mass stars and BDs are more settled than disks around T Tauri stars, which also points to faster grain growth followed by dust settling (Szűcs et al. 2010). Finally, the detection of millimeter-sized grains in two young BD disks (Ricci et al. 2013) demonstrates that, even in the cold outer regions of these low-mass disks, large grains can form in a Myr timescale and be retained. The recent theoretical work of Pinilla et al. (2013) highlights some of the main physical processes that could contribute to the observed differences between BD and T Tauri disks. Pinilla et al. (2013) show that the growth timescale of micron-size particles due to settling is directly proportional to the square root of the stellar luminosity. This implies that small grains in BD disks grow an order of magnitude faster than those in T Tauri disks that trace regions at similar temperatures. At the same time, because the radial drift velocity is inversely proportional to the square root of the stellar mass, dust particles in BD disks are depleted faster than in T Tauri disks. While a mechanism to slow down the depletion of millimeter- or meter-size grains is needed in T Tauri disks (Pinilla et al. 2012), that mechanism must be more efficient in BD disks to explain the recent detections of millimeter dust particles. The models of Pinilla et al. (2013) support the view that the first steps of planet formation are faster in BD than in T Tauri disks. Unfortunately, these models do not follow the growth of dust up to planetesimals (∼km in size), which is when gas drag does not significantly affect their orbital dynamics. Our inference of a lower O/C in BD disks than in T Tauri disks perhaps suggests that the faster dust growth in BD disks extends to the stage of nonmigrating planetesimals.