1: %% manuscript produces a one-column, double-spaced document:
2: %\documentclass[manuscript]{aastex}
3:
4: %% preprint2 produces a double-column, single-spaced document:
5: \documentclass[12pt,preprint]{aastex}
6: %\usepackage{apjfonts}
7: %\usepackage{natbib}
8: \usepackage{graphicx}
9: %\usepackage{emulateapj5}
10: %\citestyle{aa}
11:
12: \shorttitle{Chemistry in Disks}
13: \shortauthors{Vasyunin et al.}
14:
15: \begin{document}
16:
17: \title{Chemistry in Protoplanetary Disks: A Sensitivity Analysis}
18:
19: \author{A.I. Vasyunin}
20: \affil{Max Planck Institute for Astronomy, K\"onigstuhl 17,
21: D-69117 Heidelberg, Germany}
22: \affil{Ural State University, ul. Lenina 51, Yekatirinburg, 620083
23: Russia}
24: \email{vasyunin@mpia.de}
25:
26: \author{D. Semenov, Th. Henning}
27: \affil{Max Planck Institute for Astronomy, K\"onigstuhl 17,
28: D-69117 Heidelberg, Germany}
29: \email{semenov,henning@mpia.de}
30:
31: \author{V. Wakelam}
32: \affil{Universit\'e Bordeaux 1, CNRS, Laboratoire Astrophysique de Bordeaux , BP89 33270 Floirac, France}
33: \email{Valentine.Wakelam@obs.u-bordeaux1.fr}
34:
35: \author{Eric Herbst}
36: \affil{Ohio State University, Department of Physics, 174 West 18th
37: Avenue, Columbus, OH 43210-1106 USA}
38: \email{herbst@mps.ohio-state.edu}
39:
40: \and
41:
42: \author{A.M. Sobolev}
43: \affil{Ural State University, ul. Lenina 51, Yekatirinburg, 620083
44: Russia}
45: \email{andrej.sobolev@usu.ru}
46:
47: \begin{abstract}
48: We study how uncertainties in the rate coefficients of chemical
49: reactions in the RATE\,06 database affect
50: abundances and column densities of key molecules in protoplanetary
51: disks. We randomly varied the gas-phase reaction rates within their
52: uncertainty limits and calculated the time-dependent
53: abundances and column densities using a gas-grain chemical model and a flaring steady-state
54: disk model. We find that key species can be separated into two distinct groups according
55: to the sensitivity of their column densities to the rate uncertainties. The first group
56: includes CO, C$^+$, H$_3^+$, H$_2$O, NH$_3$, N$_2$H$^+$, and HCNH$^+$.
57: For these species, the column densities are not very
58: sensitive to the rate uncertainties but the abundances in specific regions are.
59: The second group includes CS, CO$_2$, HCO$^+$, H$_2$CO, C$_2$H, CN, HCN, HNC
60: and other, more complex species, for which high abundances and abundance uncertainties co-exist in the same disk
61: region, leading to larger scatters in the column densities.
62: However, even for complex and heavy molecules, the dispersion in their column densities is not more than a factor
63: of $\sim 4$. We perform a sensitivity analysis of the computed abundances to rate uncertainties and identify
64: those reactions with the most problematic rate coefficients. We conclude that the rate coefficients of
65: about a hundred of chemical reactions need to be determined more accurately
66: in order to greatly improve the reliability of modern astrochemical models.
67: This improvement should be an ultimate goal for future laboratory studies and theoretical
68: investigations.
69: \end{abstract}
70:
71: \keywords{accretion disks -- astrochemistry --- methods: statistical ---
72: molecular processes --- stars: planetary systems: protoplanetary disks}
73:
74:
75: \section{Introduction}
76: %23.04.07: R.I.P. EBN!
77: More than 140 organic and inorganic molecular species consisting of up to 13 atoms
78: have been identified in space so far\footnote{\url{http://astrochemistry.net}},
79: including such complex species as dimethyl ether \citep[CH$_3$OCH$_3$;][]{Snyder_ea74}
80: and acetamide \citep[CH$_3$CONH$_2$;][]{Hollis_ea06}. The rich variety of the observed
81: molecules implies that many more relevant yet undiscovered species
82: must be involved in the processes of their formation and destruction. Still, only a tiny
83: fraction of these species have been firmly detected in several protoplanetary
84: disks, including CO (and its isotopes), CN, HCN, HNC, H$_2$CO, C$_2$H, CS,
85: HCO$^+$, H$^{13}$CO$^+$, DCO$^+$, and N$_2$H$^+$
86: \citep{Dutrey_ea97,Kastner_etal1997,Zadelhoff_etal2001,Qi_etal2003,Thi_etal2004}.
87: Multi-molecule, multi-transition observations of emission lines with (sub-)millimeter
88: interferometers and single-dish antennas provide a wealth of information about planet-forming disks resembling
89: the young Solar Nebula \citep{Simon_ea00,Aikawa_ea03,Qi_ea05,Pety_ea06,Dutrey_ea07}.
90:
91: An ultimate goal of disk studies is the reconstruction of the
92: evolutionary history and spatial abundance distribution of various
93: species, which requires sophisticated chemical models
94: \citep{Willacy_Langer00,Markwick_ea02,van_Zadelhoff_ea03,Ilgner_ea04,Semenov_ea05,Aikawa_ea06,TG07}.
95: The analysis and modeling of observational data acquired with
96: limited spatial and spectral resolution is not complete without
97: taking all possible uncertainties into account such as instrumental
98: errors, uncertainties in the distance to source, orientation, etc.
99: Whereas these uncertainties can in general be reduced either by
100: using better observational data provided by current interferometers
101: (PdBI, SMA) equipped with a new generation of receivers or in the
102: future with the advent of more powerful instruments like ALMA,
103: EVLA, and Herschel, there is an \emph{intrinsic} source of
104: ambiguity in all the chemical models -- our limited knowledge of the
105: reaction rate coefficients.
106:
107: Modern astrochemical databases include up to about 4\,500 gas-phase reactions and $\sim 450$
108: species \citep{umist95,umist99,OSU03,rate06}, but many of these reactions have
109: poorly estimated rate coefficients with uncertainties of about a factor of 2 and larger.
110: For example, radiative association reactions lead to the formation of new complex species out of
111: smaller ones through the photon relaxation of an excited collisional molecular complex.
112: Under low-density space conditions where three-body processes are unlikely,
113: these reactions may have rate coefficients as high as $\sim 10^{-9}$~cm$^3$\,s$^{-1}$ and as
114: low as $\sim 10^{-17}$~cm$^3$\,s$^{-1}$, depending on the density of vibrational states of the complex.
115: The radiative association rates are usually difficult to measure in the laboratory as well as to predict
116: theoretically, especially for bigger species \citep[][]{Bates_51,Williams72,Herbst_80}.
117:
118: In contrast, dissociative recombination of molecular ions almost always proceeds very rapidly, in particular
119: at low temperature, with rate coefficients that can be accurately
120: obtained \citep{Florescu_Mitchell06}. However, the products and their relative branching fractions branching
121: channels of many
122: dissociative recombination reactions are not known precisely, particularly at low temperatures
123: \citep[see though][]{Semaniak_ea01,Ishii_ea06,Ojekull_ea06}, which may spoil theoretical predictions
124: \citep[see, e.g.][]{Millar_ea88,Geppert_ea05a}.
125:
126: Recently, \citet[Paper~I hereafter]{Vasyunin_ea04} investigated the influence of uncertainties in rate coefficients
127: on molecular abundances in dense and diffuse clouds, using the gas-phase UMIST\,95
128: database. They found that the abundance uncertainties of simple species are limited to within about an order of
129: magnitude and increase substantially with the number of atoms in the molecule, though the uncertainties still
130: do not exceed the observational errors for simple molecules. They proposed a
131: sensitivity analysis to identify those reactions that introduce the largest errors
132: in the computed concentrations. \citet[hereafter Paper~II]{Wakelam_ea05}
133: studied uncertainties in the gas-phase chemistry of hot cores and demonstrated that at late times, $\ga 10^4$~yr,
134: modeled abundances of important molecules can suffer from large uncertainties due to poorly known rate coefficients.
135: \citet{Wakelam_ea06} focused on dense cloud chemistry and took uncertainties in the physical
136: parameters into account, as well as comparing the osu.2003 and UMIST\,99 chemistries. These authors came to
137: conclusions that grain-surface reactions should be considered in order to achieve better consistency with the
138: observational data, even though there is a large degree of ambiguity in simple cloud models based on static
139: physical and chemical structures. Finally, \citet{Wakelam_ea06b} expanded this study by using their uncertainty analysis to show that there is a bistability in the abundances of
140: many species that are hyper-sensitive to the adopted value of the ratio of the cosmic ray ionization rate of helium to that of hydrogen
141: \citep[see also][]{Pineau_des_Forets_ea92,Shalabiea_Greenberg95,Boger_Sternberg06}.
142: Furthermore, \citet{Izzard_ea07} recently studied the effect of proton-capture reaction rate uncertainties
143: in the NeNa and MgAl chains on the abundances of the Ne, Na, Mg and Al isotopes produced in intermediate-mass
144: AGB stars.
145:
146: Despite previous work in rate uncertainties, to the best of our knowledge, there have been no attempts in the
147: literature to study how these reaction
148: uncertainties, which are unlikely to be eliminated in the near future,
149: affect the results of astrochemical modeling in the wide range of
150: physical conditions typical of protoplanetary disks. In this paper,
151: we extend the previous analyses by \citet[Paper~I]{Vasyunin_ea04} and
152: \citet[Paper~II]{Wakelam_ea05} to the conditions of a low-mass protoplanetary
153: disk surrounding a young T Tauri star. The aims of our study are
154: several. First, we analyze how large the abundance and column
155: density scatter for key species are due to uncertainties in rate coefficients.
156: Second, we investigate
157: how these uncertainties vary with disk location.
158: Third, we isolate the reactions whose rate
159: uncertainties contribute most to the abundance scatter and which are therefore worth studying in detail. Last,
160: we predict how the overall consistency of theoretical models will be improved
161: after these reaction rates are better constrained.
162: %A message for observers is given for which
163: %species are to look and which column densities are to get in order
164: %to put constrains on the yet problematic rate values.
165:
166: The organization of our paper is the following. The adopted disk
167: physical structure and time-dependent gas-grain chemical model are
168: presented in Sect.~\ref{phys_par}. The Monte Carlo method used to introduce the
169: uncertainties in the RATE\,06 reaction rates is described in Sect.~\ref{method}.
170: In Section~\ref{res} we report computed distributions of the mean molecular
171: abundances and column densities and their errors in the whole disk, and analyze
172: the influence of the rate coefficient uncertainties on these quantities. A correlation
173: method that allows determination of the role of various reactions in the abundance uncertainties
174: as a function of time and disk location is outlined, and results for several key species
175: are presented in Sect.~\ref{corr_method}. Particular attention is paid to the identification
176: of the most uncertain reactions for the chemical evolution of key species in the entire disk.
177: In Sect.~\ref{cons_impr} we predict to what extent the abundance scatter due to the rate
178: uncertainties will be lowered when accurate rate values of several dozens most problematic reactions
179: are available. The remainder of Section \ref{diss} is concerned with problems in different classes of reactions.
180: Final conclusions are drawn in Sect~\ref{concl}.
181:
182:
183: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
184: % Model:
185: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
186: \section{Disk model and uncertainty approach}
187: \label{mod}
188:
189: \subsection{Physical structure and chemical model of the disk}
190: \label{phys_par}
191: In our simulations, we adopted the 1+1D steady-state irradiated disk model with vertical
192: temperature gradient that represents
193: the low-mass Class~II protoplanetary disk surrounding the young T Tauri star DM Tau \citep{DAlessio_ea99}.
194: The disk has a radius of 800~AU, an accretion rate $\dot{M}=10^{-8}\,M_\sun$\,yr$^{-1}$, a viscosity
195: parameter $\alpha = 0.01$, and a mass $M\simeq0.07\,M_\sun$ \citep{Dutrey_ea97,Pietu_ea07}.
196: The thermal and density structure of the disk is shown in Fig.~\ref{disk}.
197: As a disk age we used a value of $\sim 5$~Myr, which has been derived by \citet[][]{Simon_ea00}
198: based on evolutionary track modeling of the central star. % of \citet{DM97,BCAH,PS99,SDF}.
199: In chemical simulations an outer disk region beyond the distance of 50~AU from the central star
200: is considered. This is the only routinely accessible disk region with existing (sub-)millimeter interferometers.
201:
202: We assumed that the disk is illuminated by UV radiation from the central star with
203: an intensity $G=410\,G_0$ at $r=100$~AU ($G(r) \propto r^{-0.5}$) and by interstellar UV radiation with
204: intensity $G_0$ in plane-parallel geometry \citep[][]{Draine_78,van_Dishoeck88,Bea03,Dutrey_ea07}.
205: The dust grains are assumed to be uniform $0.1\,\mu$m spherical particles made of amorphous silicates with olivine
206: stoichiometry \citep{Semenov_ea03}, with a dust-to-gas mass ratio of $1\%$.
207: The self-and mutual-shielding of CO and H$_2$ against UV photodissociation is computed using the pre-calculated
208: factors from Tables~10 and 11 in \citet[][]{Lee_ea96}.
209:
210: Three other high-energy sources in the model that drive chemistry
211: either through dissociation, ionization, or desorption are cosmic
212: rays, decay of short-lived radionuclides, and stellar X-rays. The
213: X-ray ionization rate in a given disk region is computed according
214: to the results of
215: \citet{Glassgold_ea97a,Glassgold_ea97b} with parameters for their
216: high-metal depletion case. In this model, the
217: thermal $\sim 5$~keV X-ray photons are
218: generated at $\sim 0.1$~AU above the central star, with a total
219: X-ray luminosity of $\approx 10^{30}$~erg\,cm$^{-2}$\,s$^{-1}$
220: \citep{Glassgold_ea05}. The cosmic-ray ionization rate is assumed to
221: be 1.3$\times$10$^{-17}$~s$^{-1}$. The ionization rate caused by
222: radionuclide decay (primarily $^{26}$Al and $^{60}$Fe) is $6.5
223: \times 10^{-19}$~s$^{-1}$ \citep{Finocchi_Gail97}. We adding the
224: X-ray and radionuclide ionization rates to those involving cosmic
225: ray particles (CRP) and CRP-induced UV photons.
226:
227: A modified static gas-grain chemical model that includes gas-grain
228: interactions without surface reactions and turbulent mixing has been
229: utilized \citep[][]{Semenov_ea05}. We allow only surface formation
230: of molecular hydrogen using the approach of
231: \citet{Hollenbach_McKee79}. This modification significantly
232: decreases the numerical demands of our chemical code and makes it
233: computationally tractable while allows us to include all necessary
234: physical processes. The gas-phase reaction rates and their
235: uncertainties are taken from the recent RATE\,06 database, in which
236: the effects of dipole-enhanced ion-neutral rates are included
237: \citep{rate06}. To calculate photoreaction rates through the disk,
238: we adopt pre-computed fits of \citet{van_Dishoeck88} instead of
239: integrating the wavelength-dependent cross sections over the local
240: UV spectrum \citep[see][]{van_Zadelhoff_ea03}.
241: %, with some rates updated according to the recent measurements (see Table~\ref{umist_update}).
242:
243: Gas-grain interactions include the accretion of neutral molecules onto dust surfaces with a sticking
244: efficiency of 100$\%$, dissociative recombination of ions on charged dust grains, and grain re-charging, as well as
245: UV-, CRP-induced and thermal desorption of surface species.
246: Desorption energies $E_{\rm des}$ are mostly taken from the recent
247: osu.2007\footnote{\url{https://www.physics.ohio-state.edu/~eric/research.html}}
248: database \citep{GH_06} or roughly estimated by analogy for about 20 molecules.
249: % Desorption energies $E_{\rm des}$ are mostly taken from
250: % \citet{Hasegawa_Herbst93} or roughly estimated using simple expression $E_{\rm des} \approx 50 \times A$, where $A$ is
251: % the atomic weight of the species \citep[see Eq.~5 in][]{Wiebe_ea03}.
252:
253: Overall, our network consists of positively and
254: negatively as well as neutral dust grains, 420 gas-phase and 157
255: surface species made of 13 elements, and 5773 reactions. Among these
256: 5773 reactions are 4517 gas-phase reactions, 2 charge exchange
257: reactions for dust grains, 940 dissociative recombination reactions
258: of ions on charged and neutral grains, and 314 accretion/desorption
259: processes.
260:
261: As initial abundances, we have adopted the so-called ``low
262: metal'' set of \citet{Lee_ea98} as listed in Table~\ref{init_abund} and assumed that all hydrogen is mostly locked
263: in its molecular form. In these initial abundances, the standard solar elemental composition is depleted
264: in heavy elements by, e.g., 200 times for S up to more than $10^4$ for Fe, Cl, P, and F.
265: The mostly atomic initial abundances are chosen instead of those from a molecular cloud model because this choice allows
266: the abundance uncertainties to accumulate in a long sequence of chemical reactions with imprecise rates, starting
267: from basic processes that reach steady-state at early times during the disk evolution. \citet{Willacy_ea98}
268: have shown that most of the molecular abundances are hardly affected by the choice of input abundances.
269: Note also that the resulting abundance uncertainties can be sensitive to the initial elemental concentrations
270: \citep{Wakelam_ea06b}. However, we do not allow the elemental concentrations to vary in order to keep the computations
271: in reasonable limits.
272:
273: \subsection{Method to model rate uncertainties}
274: \label{method}
275: Each reaction rate coefficient in the RATE\,06
276: database is given in the standard Arrhenius form and thus relies on
277: 3 parameters: $\alpha$ (the absolute value at room temperature),
278: $\beta$ (the index for the power law dependence of the rate on
279: temperature), and $\gamma$ (the activation energy barrier in K). A
280: specific expression for the reaction rate coefficient depends on the
281: type of chemical process: bimolecular reaction, direct cosmic-ray
282: ionization reaction, photodissociation, etc. Note that in all cases
283: the rate coefficient scales linearly with the parameter $\alpha$.
284:
285: In addition to these three parameters, the RATE\,06 reaction rates are characterized by an
286: accuracy estimate (A, B, C, and D), where the uncertainties are smaller than $25\%$, $50\%$,
287: within a factor of 2, and within an order of magnitude.
288: % parameter $F$ that fall into 5 distinct groups:
289: % $F=1.25$, $F=1.5$, $F=2$, $F=10$, and highly uncertain (though none of the
290: % reactions in RATE\,06 is assigned to this class).
291: Almost all reactions with measured rates belong to the first group
292: ($\sim 1\,400$ reactions), while those reactions with rates that
293: were ``guessed'' or derived by analogy are assigned to the third
294: group ($\sim 2\,800$ reactions), which contains most of the
295: ion-neutral reaction rates. The second group includes about 300
296: mostly neutral-neutral and photodissociation reactions, whereas the
297: fourth group consists of only 4 photodissociation reactions
298: \citep{rate06}.
299:
300: In accord with the latest osu.2007 network, we adopted a rate uncertainty of one order of magnitude
301: for radiative association reactions.
302: Such a large uncertainty factor is justified by the difficulty
303: in calculating or measuring these rate coefficients \citep[e.g.,][]{Herbst_85}.
304: The uncertainty factors vary widely for the photoreaction rates since the corresponding
305: frequency-dependent cross-sections have limited accuracy and have been
306: obtained only for a fraction of molecules in the UMIST database
307: \citep[][]{van_Dishoeck88,van_Dishoeck_ea06}.
308:
309: Our method to model abundance uncertainties is based on the computation of a large set
310: of chemical models, using identical physical and initial conditions and the same
311: chemical network, but with randomly varied rate coefficients within their uncertainty
312: limits.
313: %Here we do not take into account additional uncertainties in the modeling
314: %results that may arise due to the poorly known
315: %temperature dependence of the rates, branching ratios and products, or energy barriers.
316: We utilized the same method to introduce uncertainties in the rate
317: coefficient values as described in
318: \citet{Dobrijevic_Parisot98}, \citet{Dobrijevic_ea03}, and \citet{Wakelam_ea05}. The rate
319: coefficient for each gas-phase reaction $i$ is randomly
320: chosen over a log-normal distribution with median $\alpha_i$ and
321: dispersion $F_i$. Consequently, the rate coefficient of the $i$-th
322: reaction spans an interval between $\alpha_i/F_i$ and $\alpha_i\times
323: F_i$ with a probability of 68$\%$ ($1\sigma$). We generated a sequence
324: of these log-normally distributed rates for all gas-phase reactions
325: in our chemical network using the following expression:
326: \begin{equation}
327: \alpha_i^l=\alpha_i\times(F_i)^{\epsilon},
328: \end{equation}
329: where $\alpha_i^l$ is the $l$-th realization of the $i$-th reaction
330: rate, $\alpha_i$ is the standard RATE\,06 rate value for the $i$-th
331: reaction, $F_i$ is the dispersion, or the uncertainty factor, for
332: this rate, and $\epsilon$ is randomly distributed from -1 to 1
333: with an uniform distribution law.
334:
335:
336: %Despite the fact that our chemical model also includes gas-grain interactions,
337: Using this approach, we allowed the rates of only 4517 gas-phase reactions
338: to vary in our network. The rates of all other processes (gas-grain interactions,
339: dissociative recombination of molecular ions on grains, and recharging of dust grains)
340: were kept constant.
341: % We also did not take into account uncertainties in the values of sticking probabilities, the
342: % desorption energies, etc.
343: Not did we account for possible variations among the parameters of the disk
344: physical model. Such an idealization allowed us to focus solely on an investigation of how uncertainties in
345: gas-phase reactions affect the results of the disk chemical modeling.
346:
347: Using 8\,000 realizations of the RATE\,06 network, we simulated 5~Myr of chemical evolution in the outer,
348: $r\ge 50$~AU, disk of DM Tau.
349: The disk grid consists of 5 radial (50, 100, 200, 380, and 760~AU) and 10 equidistant vertical points
350: (with step sizes of 3.2, 7.6, 18, 41.7, and 96 AU for the considered radii).
351: Such a huge number of utilized chemical models with randomly varied rates leads to low values of
352: statistical noise in the distributions of the abundance scatters at any particular disk location,
353: assuring the correctness of the sensitivity analysis. Moreover, a large number of varied reaction rates allows
354: an analysis of all reactions in the chemical network, in contrast to models with limited chemistry
355: included in previous works \citep[e.g.,][]{Wakelam_ea05}. Still, even with the modest $5\times10$ spatial
356: resolution of the adopted disk model and without surface reactions, the overall computational time needed
357: to calculate the chemistry with these 8\,000 networks was about 3 days on a 4-CPU PC machine
358: (3.0~GHz Xeon, 8~Gb RAM).
359:
360:
361: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
362: % Results:
363: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
364: \section{Results}
365: \label{res}
366:
367:
368: \subsection{Abundance distribution profiles}
369: \label{db_abunds}
370: Before we perform a detailed analysis of the influence of the rate uncertainties
371: on the modeled molecular abundances, we investigate the abundance distribution profiles.
372: Because we utilized a log-normal distribution
373: for the rate coefficients, one might expect that for each molecule the abundance
374: distribution should show a normal (Gaussian) profile in logarithmic scale.
375: If this hypothesis is correct, one can use the dispersion -- a measure of the scatter
376: of randomly varied values that have a Gaussian distribution -- as a convenient
377: quantity characterizing the abundance and column density uncertainties.
378:
379: As shown in Paper~II, this assumption may not be fullfiled for certain species and certain time steps.
380: For example, when molecular abundances of a particular species show a steep decline or increase
381: with time then at that moment the corresponding histogram of
382: the abundance distribution can have several peaks and thus be far from a Gaussian shape because of sampling effects.
383: Bimodal distributions can also be obtained if the system is highly sensitive to a small variation of specific parameter.
384: \citet{Wakelam_ea06b} showed that dense cloud chemistry hypersensitive, and even bistable, to
385: the ratio between He and H$_2$ cosmic-ray ionization rates in particular conditions.
386: In this case our uncertainty method would tend to overestimate the dispersion of the modeled abundances.
387: The disk chemistry is very rich in a sense that it proceeds vastly differently in various disk
388: regions and at different times, so it is natural to expect that the abundance distributions
389: of some species can sometime deviate from a normal distribution.
390:
391: We carefully studied this problem and found that such a situation happens rarely at late times,
392: $\ga 1$~Myr, for which our analysis will be performed. In Fig.~\ref{hists}
393: we show several representative histograms of the abundance distributions at 5~Myr, using
394: HCO$^+$ and CO as an example. For an inner disk region, at $r\approx 97.5$~AU, $z \approx 30$~AU,
395: a strong deviation from the Gaussian shape is clearly visible, with 2 distinct peaks and a gap
396: in between (Fig.~\ref{hists}, top panels). Note that this is one of the few exceptional cases
397: that is discussed in detail
398: below. Everywhere else through the disk the HCO$^+$ and CO abundances have nearly Gaussian-like profiles
399: (Fig.~\ref{hists}, bottom panels). This condition holds for other key species as well, implying that the
400: averaged abundances in the disk should not differ from the values computed with the standard RATE\,06 network.
401: Moreover, in Paper~I we found that the Gaussian shape of the abundance distribution
402: is also preserved for the uniform distribution of the rate uncertainties in linear scale.
403:
404: In the rest of this paper, we will refer to one standard deviation ($1\sigma$) in
405: the modeled abundance distribution as the ``abundance uncertainty''. In our notation ``abundance scatter''
406: represents twice the abundance uncertainty in logarithmic scale ($2 \log \sigma$).
407: It is the abundance uncertainties that have to be taken into account
408: when interpreting the observational data with a chemical model.
409:
410:
411: \subsection{Distribution of the mean abundances in the disk}
412: \label{mean_abunds}
413: Although the evolution of all species was calculated in 8\,000 realizations of the chemical model, in what
414: follows we focus on several key molecules that are used as tracers of disk physical structure and
415: chemical composition.
416: Our list of important species includes molecules that are widely used for kinematic studies and
417: determination of the density and temperature (CO, CS, H$_2$CO, NH$_3$, and HCN)
418: as well as for probing such major parameters as ionization degree (HCO$^+$),
419: radiation fields (CN, HCN), and the yet-to-be detected C$^+$ as a dominant ion in the disk surface
420: \citep{Dutrey_ea97,Kastner_etal1997,Aikawa_ea03,Dutrey_ea07}.
421:
422: We use time-dependent abundances to analyze the chemical evolution of these species in detail.
423: The analysis is based on the calculation of the relative importance of each reaction in the chemical network
424: for the abundance change of a molecule under investigation at a given time \citep{Wiebe_ea03}.
425: This allows us to weight all reactions in the model and isolate a few most important pathways
426: for chemical evolution of one or several species under specific physical conditions.
427: In the case of protoplanetary disks, such an analysis has to be performed for several tens of representative cells.
428:
429: In Fig.~\ref{Abu2D} we show the mean abundances of these molecules
430: in the disk at 5~Myr, which are obtained by averaging the results of
431: the chemical network. The vertical extension of the disk is scaled
432: by one hydrostatic scale height, $H(r) = \sqrt{2}C_s/\Omega$, where
433: $C_s$ is the midplane sound speed and $\Omega$ is the Keplerian
434: frequency \citep{Dartois_ea03}. The unity pressure scale height is
435: about 25 and 320~AU at distances of 100 and 800~AU, respectively.
436:
437: The ``sandwich''-like chemical structure of the disk is clearly seen in Fig.~\ref{Abu2D}.
438: In the ``dark'', cold, and dense midplane at $\la 0.5-1$ pressure scale height, the molecules
439: freeze out onto dust grains within $100-100\,000$~yr and low steady-state fractional abundances of
440: $\sim 10^{-14}-10^{-10}$are reached. The abundances of the gas-phase species
441: in the midplane at late times are sustained by the CRP-induced desorption of the mantle materials.
442: The midplane abundance of CO $\sim 10^{-8}$, is higher because this molecule is very abundant
443: ($\approx 10^{-4}$) in the gas phase initially, and since CO ice is rather volatile because of its low
444: CO desorption barrier of $\sim 930$~K \citep{Bisschop_etal2006}.
445: Overall, the gas-phase abundances in the midplane tend to slightly increase with radius as the disk
446: surface density and thus the density in midplane decrease and gas-grain interactions become less intense,
447: which slows down the freeze-out of gas-phase molecules onto dust grain surfaces.
448: % , whereas the penetration
449: % of the interstellar UV radiation and thus photoevaporation becomes more efficient at
450: % larger distances from the star, $r>300$~AU.
451:
452: The surface layer of the disk ($Z/H(r)\ga 2$) is deficient in molecules due to
453: strong irradiation by high-energy stellar and interstellar photons. Even CO, which is self-shielded and
454: mutually-shielded by H$_2$ against UV photodissociation cannot survive. Among the considered species, only C$^+$
455: reaches its maximal concentration and becomes the most abundant ion in the surface layer, with an abundance of
456: $\sim 10^{-4}$. The thickness of this highly ionized layer ($\sim 1$ pressure scale
457: height down from the disk surface) stays nearly constant with radius for all key species but highly reactive
458: radicals CS and CN, whose abundances are already high ($\sim 10^{-9}$) at scale pressure heights as high as
459: $Z/H(r) \approx 2$ at all radii (Fig.~\ref{Abu2D}).
460:
461: In contrast, polyatomic species tend to concentrate in the warm, chemically rich intermediate layer
462: ($\la 1-2$ pressure scale height, depending on the radius). The model-averaged
463: abundances for the considered species, $\sim 10^{-10}-10^{-4}$, are close to the values computed
464: with the standard RATE\,06 network and those found in previous work
465: \citep{Aikawa_Herbst99,Willacy_Langer00,Aikawa_ea02,Willacy_ea06,Dutrey_ea07}.
466:
467: Now let us consider the chemical processes that form and destroy the considered species in the intermediate
468: layer. In contrast to dense cloud models, for many species in this disk zone steady-state is
469: reached by $\sim 10^3-10^5$ yr. This is due to the higher densities encountered in the warm layer
470: ($\ga 10^6-10^8$~cm$^{-3}$)
471: than in the cloud cores ($\sim 10^4-10^6$~cm$^{-3}$) and a non-negligible flux of the UV radiation. Impinging interstellar
472: UV photons allow many surface species to desorb back into the gas phase \citep{Hasegawa_Herbst93,GH_06},
473: thus reducing their surface abundances and overall importance of gas-grain interactions for
474: disk chemical evolution at late times, $t\ga 10\,000$ yr.
475:
476: The evolution of CO and HCO$^+$ in the intermediate layer is
477: governed by a small set of reactions at all radii. The CO molecules
478: are formed via reactions of atomic oxygen with CH, CH$_2$, and C$_2$
479: at $t\la 1\,000$~yr, and partly through the dissociative
480: recombination of HCO$^+$ afterwards (at this stage steady-state has
481: been reached and there is a loop between CO and its protonated
482: form). The light hydrocarbons that are precursors of CO are
483: themselves rapidly created within a few years via the radiative
484: association of C with H and H$_2$, and CH with H$_2$. The CO is
485: mainly removed from the gas phase by accretion onto the dust grain
486: surfaces in outer, cold regions and by reactive collisions with
487: helium ions in the inner, X-ray ionized part of the intermediate
488: layer. In the upper part of this layer, CO is also photodissociated.
489: The ion HCO$^+$ initially forms via reactions of O with CH and
490: ionized light hydrocarbons, CH$^+_{\rm n}$ (n=1--3), and the
491: ion-neutral reaction between CO and CH$_5^+$ (also producing
492: methane). At a later time of $\ga 1\,000$~yr, HCO$^+$ becomes the
493: most abundant ion in the intermediate layer and the dominant molecular
494: ion in the disk, with a fractional abundance of
495: $\sim 10^{-9}$. At this steady-state stage, HCO$^+$ is mainly
496: produced via reactive collisions between CO and H$_3^+$ and
497: destroyed by dissociative recombination with electrons and
498: negatively charged grains.
499:
500: The chemically related H$_2$CO molecule is produced through reaction
501: of CH$_3$ with oxygen atoms. At a later time of $\sim10^5$~yr
502: H$_2$CO is involved in a simple formation-destruction cycle. It
503: starts with protonation of formaldehyde by HCO$^+$, H$_3$O$^+$, and
504: H$_3^+$, which is followed by dissociative recombination into
505: H$_2$CO (33\%), CO (33\%), or HCO (33\%). Apart from
506: photodissociation and freeze out, another direct destruction channel
507: for H$_2$CO in the upper, more ionized part of the intermediate
508: layer is the ion-neutral reaction with C$^+$, which
509: produces either CO and CH$_2^+$, HCO$^+$ and CH, or H$_2$CO$^+$ and C.
510: The chemical evolution of formaldehyde typically reaches steady-state at about $10^5$~yr
511: with a gas-phase abundance of $\la 3\times10^{-10}$.
512: Note that in the same disk chemical model as discussed above but with a set of surface
513: reactions taken from \citet{GH_06} the overall (gas-phase and surface) abundance
514: of formaldehyde is decreased by a factor of several as it is converted to methanol by hydrogenation
515: on grain surfaces.
516:
517: The photostable radical CN is abundant ($10^{-8}$) at an upper, less molecularly rich disk layer
518: ($Z/H(r) \approx 1.8$, Fig.~\ref{Abu2D}). The initial formation channel for the cyanogen radical is the radiative
519: association between C and N, followed by the neutral-neutral reaction of N and CH, and dissociative recombination
520: of HCNH$^+$ at later times ($t\ga 10^4$~yr).
521: Other, less important formation pathways for CN at this stage are neutral-neutral reactions between
522: C and NO producing CN and O as well as between C and OCN leading to CN and CO.
523: The destruction of CN mostly proceeds through photodissociation and neutral-neutral reaction
524: with N forming highly photostable nitrogen molecules and C.
525:
526: In turn, HCNH$^+$ is produced by ion-neutral reactions
527: between H$_2$ and HCN$^+$ and between N and C$_n$H$_2^+$ ($n=3,4$)
528: at early times ($t\la 10^3$~yr). Later the steady-state recycling of
529: HCNH$^+$ is reached, which involves dissociative recombination of
530: HCNH$^+$ into either HCN (33\%), HNC (33\%), or CN (33\%), followed
531: by re-production of HCNH$^+$ via protonation of HCN and HNC by
532: either H$_3^+$, H$_3$O$^+$, or HCO$^+$ as well as via the
533: ion-neutral reaction of CN with H$_3^+$.
534:
535: The chemistry of HCN is tightly related to the evolution of CN and
536: HCNH$^+$. HCN is mainly produced by neutral-neutral reactions
537: involving nitrogen and methylene or methyl at early evolutionary
538: stages, $t\la 10^4$~yr. Later the evolution of HCN proceeds as a
539: part of the HCNH$^+$ formation-destruction loop. The major
540: destruction routes for gas-phase HCN include freeze-out onto dust
541: grains and the ion-neutral reaction with C$^+$, which forms
542: CNC$^+$ and H. Another important destruction channel for HCN in the
543: upper disk layers at $\sim 1.5-2$ scale heights is
544: photodissociation. Steady-state for HCN is typically reached within
545: $\ga 10^4$~yr, with an abundance of about a few $\times 10^{-10}$.
546:
547: The layer where abundant ammonia exists is located somewhat deeper
548: toward the disk midplane compared with HCN and especially CN because
549: NH$_3$ can be photodissociated by UV photons shortward of $\approx
550: 1\,950$~\AA \citep{van_Dishoeck88}. The main formation route for
551: ammonia is dissociative recombination of NH$_4^+$, which itself is
552: formed through a sequence of hydrogen insertion reactions starting
553: with the production of the NH$^+$ ions from N$^+$ reacting with
554: H$_2$. Upon formation, NH$^+$ repeatedly reacts with molecular
555: hydrogen and gains an addition hydrogen atom until NH$_4^+$ is
556: formed. The destruction of ammonia proceeds mainly via
557: ion-neutral reactions with C$^+$ and S$^+$ in the upper
558: intermediate layer as well as H$_3^+$, H$_3$O$^+$, and HCO$^+$ in a
559: deeper region, which leads again to NH$_4^+$ and either H$_2$,
560: H$_2$O, or CO. For all radii, the abundance of ammonia reaches a
561: steady-state value of $\la 3\times 10^{-10}-10^{-9}$ at very late
562: times of $\ga 10^5-10^6$~yr.
563:
564: The chemical evolution of CS is governed by neutral-neutral
565: reactions between sulphur atoms and either CH, CH$_2$, or C$_2$
566: initially, and later, at $t\ga 10^3$~yr, by an extended
567: formation-destruction cycle of H$_{\rm n}$CS$^+$ (n=1--3) and
568: HC$_2$S$^+$. At this evolutionary stage these complex ions are
569: produced by ion-neutral reactions of protonated water,
570: H$_3^+$, and HCO$^+$ with H$_{\rm n}$CS (n=0--2). The initial routes
571: to form the HCS$^+$ ion are the reactions between CH$_3^+$ and S or
572: CS$^+$ and H$_2$, whereas HC$_2$S$^+$ forms predominantly via
573: reaction of CH$_3$ and S$^+$. The H$_3$CS$^+$ ion is mostly produced
574: in reactive collisions of methane with sulphur ions. Upon formation,
575: HCS$^+$ dissociatively recombines into CS ($\sim 10\%$) or CH ($\sim
576: 90\%$), while HC$_2$S$^+$ breaks up equally into C$_2$S or CS. The
577: H$_2$CS$^+$ and H$_3$CS$^+$ ions recombine with equal probabilities
578: into CS and either HCS or H$_2$CS. Finally, the major destruction
579: channels for CS are photodissociation and the ion-neutral
580: reaction with C$^+$. The steady-state abundance for CS, $\sim
581: 10^{-9}$, is usually attained at about $10^3-10^5$~yr in our model.
582:
583:
584: \subsection{Distribution of the abundance and column density uncertainties}
585: \label{err_dist}
586: For each key molecule, we derived the distribution of abundance uncertainties (dispersion) in the disk.
587: The results are shown in Fig.~\ref{UncertDistrib}. These uncertainties are quite different in different disk regions.
588: Note that the overall abundance uncertainties are not much larger for the more complex (heavy) species
589: among considered, in comparison to those with fewer atoms (e.g., compare C$^+$, CO, HCO$^+$, and H$_2$CO).
590: However, if we would focus on even more complex species (e.g., consisting of $>7$ atoms), the abundance uncertainties
591: will in general increase with the number of atoms in the molecule, as shown by \citet{Vasyunin_ea04} and \citet{Wakelam_ea06}.
592: % This is a direct indication that the abundance uncertainties are caused by uncertain rate values
593: % of a restricted number of key reactions. This is in contrast to the conclusions of Paper~I based on
594: % the outdated UMIST\,95 ratefile and a cold cloud model. Since we adopted the recent RATE\,06 database, which contains
595: % more reactions with accurately derived rates and allows the chemistry to evolve under a wide range of temperatures,
596: % densities, and high-energy radiation fields, the overall abundance uncertainties through the disks do not
597: % increase significantly with the number of atoms in a species.
598:
599: Among the key reactions, the rates of photodissociation and
600: photoionization processes can be uncertain by factors of 2 and
601: larger. Moreover, their absolute values depend on the details of the UV radiative transfer
602: modeling \citep{van_Zadelhoff_ea03}.
603: Thus, it is not surprising that in the disk
604: atmosphere at $Z/H(r)\ga 1.5-2$ large abundance uncertainties with a
605: factor of $\la 10-30$ are reached for most of the species, in
606: particular across the interface between the intermediate layer and
607: the disk atmosphere. A notable exception is ionized carbon, because
608: it contains essentially all of the element carbon in the disk
609: atmosphere and thus maintains a high steady-state abundance that is
610: not affected by chemistry. This effect leads to relatively low
611: values of the C$^+$ abundance uncertainties, in particular a factor
612: of $\la 3$. The same is true for CO, which locks up all carbon in
613: the disk midplane and intermediate layer at $\la 2$ pressure scale
614: height and here has its lowest abundance uncertainty, a factor of
615: $\la 2$ (Fig.~\ref{UncertDistrib}).
616:
617: There is a disk region adjacent to the midplane at $r\sim 100$~AU, $Z/H(r) \sim 1$, where
618: the abundance uncertainties of the considered carbon-bearing species including CO show an increase of up to one order
619: of magnitude (see HCO$^+$, Fig.~\ref{UncertDistrib}).
620: There, the chemistry of CO and hence all chemically related species
621: (e.g., HCO$^+$, H$_2$CO, etc.) are prone to the hypersensitivity at $\ga 10^6$~yr
622: caused by the rate uncertainty of the X-ray ionization of helium atoms, as
623: discussed in \citet{Wakelam_ea06b}.
624: This region is moderately warm, with a temperature $T\sim 25-40$~K, and well shielded from the UV radiation
625: from the central star and the interstellar
626: UV radiation ($A_{\mathrm V} \la 2-5$~mag). The thermal bremsstrahlung X-ray photons ($\sim 5$~keV)
627: generated at
628: $\sim 0.1$~AU above the star, however, are able to penetrate into this region \citep{Glassgold_ea97a},
629: with a total ionization rate of $\sim 10^{-16}$~s$^{-1}$. These X-ray photons produce helium
630: ions that rapidly react and destroy the CO molecules (forming O and C$^+$). In turn, these C$^+$ ions are used
631: to reform CO but also to slowly produce carbon chain molecules (e.g., C$_2$H$_2$) and cyanopolyynes (e.g., HC$_3$N),
632: which are removed from the gas phase by freeze out onto the dust surfaces. These heavy surface molecules lock up most of
633: the elemental carbon at $t\ga 1$~Myr such that the gas-phase abundance of CO is reduced to $\sim 10^{-9}-10^{-7}$.
634: The overall efficiency of the carbon chain formation and the decline of the CO abundance at late times
635: sensitively depend on the He ionization rate. Consequently, the CO histogram of the abundance distribution at 5~Myr
636: has a double-peaked, non-Gaussian shape, as shown in Fig.~\ref{hists}.
637:
638: The abundances of nitrogen-bearing species such as CN, HCN, and NH$_3$ in the region around the midplane at
639: $r>200$~AU (Fig.~\ref{UncertDistrib}) possesses a high sensitivity to the rate of the CRP ionization of He.
640: The histograms of CN, HCN, and NH$_3$ are not fully Gaussian at 5~Myr in this region, and our method tends
641: to overestimate the resulting abundance uncertainties.
642: %The overall effect is not as severe as that for the carbon-bearing species because the outer midplane is well shielded
643: %from the stellar X-ray photons, and cosmic rays are the only ionizing source.
644: This region is so cold that at late times, $t\ga 3\times 10^5$~yr,
645: most molecules are removed from the gas phase by freeze-out. At this
646: evolutionary stage, H$_3^+$ and later H$^+$ become the most abundant
647: charged species. The primal formation pathway for H$^+$ is the slow
648: ion-neutral reaction between He$^+$ and H$_2$. The rise of the H$^+$
649: abundance at $\sim 5$~Myr($ \la 10^{-8}$) occurs in part because of
650: the slowness of the radiative recombination of ionized hydrogen
651: atoms in the molecular-deficient gas. The total electron
652: concentration increases in tandem with H$^+$, and leads to more
653: rapid dissociative recombination of HCNH$^+$ and NH$_4^+$ (precursor
654: ions for CN, HCN, and NH$_3$). Consequently, the previously low
655: gas-phase abundances of these nitrogen-bearing molecules increase at
656: $t\sim5$~Myr by factors of several. The steepness and evolutionary
657: pattern for these abundance gradients depend sensitively on the
658: abundance of helium ions.
659:
660: In the rest of the intermediate layer, the chemical evolution of
661: the gas-phase species is initiated by radiative association
662: reactions with high uncertainties but governed later by
663: ion-neutral and neutral-neutral reactions with many accurately
664: estimated rates. The situation in the disk midplane is similar,
665: though at later times the chemical evolution of many gas-phase
666: species is governed by the steady-state accretion-desorption
667: life-cycle, which has no uncertainties in our model. This effect
668: leads to lower abundance uncertainties of a factor of $\sim 3$ than
669: occur in the upper disk region (see Fig.~\ref{UncertDistrib}).
670:
671: It is interesting to note that for CO, C$^+$, and NH$_3$ the uncertainty peak does not coincide
672: with the maximal relative abundance in the disk. One might expect that the scatter in their column
673: densities could be smaller than for other considered species (CS, CN, HCN, HCO$^+$, and H$_2$CO),
674: for which uncertainty peaks and maximum abundances tend to coincide.
675:
676: In Fig.~\ref{col_dens}, the radial distributions of the column
677: densities at 5~Myr are plotted for the considered species (dotted
678: lines). In addition, in Table~\ref{abs_uncer} we compile typical
679: intrinsic uncertainties of the column densities of a larger set of
680: key molecules in protoplanetary disks at 100 AU.
681: Indeed, the overall uncertainty in the column densities of CO,
682: C$^+$, and NH$_3$ as well as H$_3^+$, H$_2$O, N$_2$H$^+$, and
683: HCNH$^+$ is less than a factor of about 2. Using this uncertainty
684: value as a criterion, we assign these species to the so-called first
685: uncertainty group. On the other hand, the column densities of CS,
686: CN, HCN, HCO$^+$, and H$_2$CO as well as CO$_2$, C$_2$H, and HNC
687: have larger error bars up to a factor of 4. These species belong to
688: the second uncertainty group. It is likely that the species from the
689: first sensitivity group are more reliable observational tracers than
690: the species from the second one.
691:
692: Note that the column density uncertainties do not exceed a factor of about 4 even for formaldehyde
693: and are comparable with observational uncertainties. However, for heavier species (e.g., HC$_9$N),
694: which are not included in our study, the uncertainties are likely larger, as shown in Paper~I and Paper~II.
695: In our model, these uncertainties slowly
696: increase with time and typically reach steady-state at $\ga 10^5$~yr.
697: The column density uncertainties can be used as error bars of the theoretically
698: predicted quantities when comparing with observational data.
699: Moreover, one can envisage a situation when, for an object with a well-studied structure, high-resolution
700: observations followed by advanced modeling will allow putting such tight constraints on molecular abundances or
701: column densities that it may help to determine the rate coefficients of some key reactions \emph{observationally}.
702:
703:
704: \section{Discussion}
705: \label{diss}
706: In the previous section, we demonstrated how and to what extent the uncertainties of the rate
707: coefficients affect the computed abundances and column densities
708: of several key species in the disk. In this section, we discuss the relative
709: contributions of individual reactions to the uncertainties of these key molecules.
710:
711:
712: \subsection{Sensitivity analysis method}
713: \label{corr_method}
714: The chemical evolution of even a simple molecule usually involves a large number of reactions in the adopted
715: chemical network. The non-linearity of the equations of chemical
716: kinetics often makes it a challenge to find a direct link
717: between the rate value of a certain reaction and the molecular
718: abundances. Thus an efficient numerical method is required \citep[see, e.g.,][]{Turanyi_97}.
719:
720: Using a sensitivity analysis based on a simple correlation method, we have shown in Paper~I
721: that under the physical conditions of dark and diffuse clouds the rate uncertainties of only a handful of
722: chemical reactions affect the accuracy of the resulting abundances to the highest extent.
723: For the disk chemistry studied in this paper a modified, two-stage sensitivity analysis was performed.
724:
725: First, for each selected molecule, we calculated time-dependent
726: linear correlation coefficients between the abundances and rate
727: coefficients for all 8\,000 realizations of the RATE\,06 network,
728: and 19 logarithmically taken time steps over the
729: 5~Myr of disk evolution at each disk cell. The coefficients
730: $R_{ij}(t)$ are given by the expression
731: \begin{equation}
732: R_{ij}(r,z,t)=\frac{\sum_{l}{(x_i^l(r,z,t)-\overline{x_i(r,z,t)})(\alpha_j^l-\overline{\alpha_j})}}{
733: \sqrt{\sum_{l}{(x_i^l(r,z,t)-\overline{x_i(r,z,t)})}^2\sum_{l}{(\alpha_j^l-\overline{\alpha_j})}^2}},
734: \end{equation}
735: where $x_i^l(r,z,t)$ is the molecular abundance of the $i$-th species
736: computed with the $l$-th realization of the UMIST network,
737: $\overline{x_i(r,z,t)}$ is the standard (mean) abundance of this species,
738: $\alpha_j^l$ is the $j$-th reaction rate coefficient for the $l$-th
739: realization of the RATE\,06 network, and $\overline{\alpha^j}$ is
740: the standard rate coefficient of the $j$-th reaction.
741:
742: This is in contrast to Paper~I where the
743: analysis of statistically significant reactions was performed for only the final time step.
744: Since at various evolutionary stages the chemistry of a molecule is typically governed by a restricted number
745: of chemical processes, at any particular time step the abundance correlation will be high only for some of the
746: key reactions, as shown in Fig.~\ref{CorrInTime} for the case of ammonia.
747: The obvious drawback of such a ``single-time'' approach is that it does not
748: account for the cumulative character of the abundance uncertainties.
749: %, which is shown to be important in the above section.
750: Therefore, we integrated the absolute values of
751: time-dependent correlation coefficients over the entire evolutionary
752: time of 5~Myr. These quantities are called ``cumulative correlation
753: coefficients''.
754:
755: With this cumulative criterion, one can identify those reactions whose rate uncertainties are the most
756: important for the abundance scatter of one molecule in one specific disk region. As a general criterion for
757: the entire disk, we utilized the cumulative correlation summed up over all 50 disk grid cells, followed
758: by summation over the 8 considered species (CO, C$^+$, CS, HCO$^+$, H$_2$CO, CN, HCN, NH$_3$).
759: One should bear in mind that the results of our analysis
760: rely on the inherent assumption that the absolute values of the rate coefficients and their uncertainties
761: in RATE\,06 are accurately determined. This assumption may not be fullfiled for all important reactions in the
762: RATE\,06 database as we will show below. Such reactions will not be treated correctly by our sensitivity analysis.
763:
764:
765: \subsection{Identification of the most problematic reactions}
766: \label{cons_impr}
767: We calculated disk-averaged cumulative correlations for all 5773 reactions in our chemical network and for
768: 8 key species: CO, C$^+$, CS, HCO$^+$, H$_2$CO, CN, HCN, and NH$_3$. The corresponding cumulative correlations,
769: normalized in descending order, steadily decline with the number of reaction. The 56 most ``problematic''
770: reactions and their weights for these 8 molecules are listed in Table~\ref{imp_reac}. We term by ``weight''
771: the corresponding cumulative correlations summed over the all 50 disk cells and renormalized to the maximum value
772: of $1148.1$, which is achieved for the RA reaction between H$_2$ and C.
773:
774: These 56 reactions correspond to
775: $1\%$ of the total number of reactions in our chemical model. Moreover, such an apparently small number of reactions
776: is chosen to stimulate experimental and theoretical studies of the rate coefficients of the reactions with high
777: astrochemical importance. Note also that some of the most ``problematic'' reactions in Table~\ref{imp_reac} have
778: been carefully studied as their rates have small uncertainties of only factors of 1.25--1.5.
779:
780: In order to demonstrate that these 56 reactions are indeed important and their rate uncertainties
781: strongly affect the abundances, we re-calculated the disk chemical evolution but allowed the rates of these 56
782: reactions to vary by only a factor of 1.25 (and 2 for radiative association). Such idealization is thought to
783: mimic the situation when all of these 56 rates will be accurately measured in the laboratory or derived theoretically.
784: % YOU ONLY MAKE IT CLEAR AFTER THE FACT THAT THE 56 REACTIONS HAVE LARGE UNCERTAINTIES. PERHAPS A SENTENCE TO THIS EFFECT
785: % BELONGS EARLIER IN THE PARAGRAPH. I ALSO THINK THAT YOU SHOULD TEST YOUR ASSUMPTION BY REDUCING THE UNCERTAINTIES
786: % OF SOME OF THE MANY REACTIONS NOT IN YOUR LIST OF 56.
787: %
788: % I THINK THAT YOU SHOULD LIST THE CORRELATION COEFFICIENTS IN THE TABLE AND DISCUSS WHAT THE TOTAL RANGE IS FROM MOST
789: % TO LEAST INPORTANT. WHY DID YOU ARBITRARILY CHOOSE 56? IS THERE A SIGNIFICANT DIFFERENCE BETWEEN REACTION 56 AND
790: % REACTION 57 REGARDING THE CORRELATION COEFFICIENT?
791:
792: The resulting uncertainties in the molecular column densities are shown in Fig.~\ref{col_dens} (solid lines)
793: and reported in {Table~\ref{imp_reac} (last column). Compared with the initial column density uncertainties,
794: the refined chemical model leads to smaller error bars for most of the considered molecules, in particular
795: for CO, HCO$^+$, H$_2$CO, CN, and H$_2$O. However, the improvement is not so obvious for nitrogen-bearing species
796: (HCN, HNC, NH$_3$, HCNH$^+$) as well as CO$_2$, CS, and C$_2$H. Their chemical evolution is governed by a large set
797: of key reactions that are only partly included in the list of the 56 most problematic reactions
798: (Table~\ref{abs_uncer})\footnote{An extended version of this table is freely
799: available from the authors upon request.}.
800:
801: The fact that the uncertainties in column densities decrease not only for the 8 key species, but also for
802: some other molecules in Table~\ref{abs_uncer} clearly indicates that some of the reactions from Table~\ref{imp_reac}
803: are relevant for their chemical evolution. For H$_3^+$ these are the ionization of H$_2$ and reactions of H$_3^+$
804: with C, O, and CO. As we have shown in Sect.~\ref{db_abunds}, reactions with H$_3^+$ are an essential ingredient
805: of disk chemistry. The decrease in the abundance error bars for H$_3^+$ in the model with refined rate
806: uncertainties leads to smaller abundance uncertainties in the related species: HCO$^+$, H$_2$CO, H$_2$O,
807: HCN, HNC, HCNH$^+$, and CS.
808: %In the absence of surface reactions, the only effective formation pathway for CO$_2$
809: %is the neutral-neutral reaction between CO and OH \citep{Prasad_Huntress79}.
810: Table~\ref{imp_reac} contains a number of reactions important for the evolution of CO and CO$_2$;
811: e.g., formation of light hydrocarbons,
812: neutral-neutral reactions of these hydrocarbons with oxygen, and CO photodissociation. The former reactions
813: affect the evolution of C$_2$H, while photodissociation is an important process for CN.
814:
815: Below we discuss the contributions of each reaction type in detail.
816:
817:
818: \subsubsection{Radiative association}
819: \label{RA_imp} The most problematic reactions in the disk are
820: radiative association (RA) reactions. Radiative association leads to
821: the formation of a larger molecule from two smaller species upon
822: their collision, and requires the emission of excess energy in the
823: form of a photon. RA reactions allow the formation of new bonds and
824: more chemically ``advanced'' species \citep{van_Dishoeck_Blake98}.
825: The corresponding rate values are often so low for small species
826: that they are difficult to obtain with high accuracy
827: \citep{Williams72,Herbst_80,Bates_Herbst88a,Smith89}. Nonetheless,
828: assigned uncertainties for many of these reactions in RATE\,06 are
829: only a factor of 2 and less. Since we assume that the rate
830: coefficients of such reactions may vary by an order of magnitude, in
831: accord with the recent OSU database, it is not surprising that some
832: RA reactions are identified as among the most problematic reactions
833: for disk chemistry. Its importance is also rise due
834: to the fact that we start calculation from the atomic initial
835: abundances and use time-integrated correlation coefficients. It
836: means that at early evolutionary time chemistry is dominated by RA
837: reactions and, therefore, its time-integrated correlation
838: coefficients increase greatly at early times.
839:
840: Two RA reactions are particularly important for the chemical evolution and accumulation of the abundance uncertainties.
841: First is the reaction between H$_2$ and C, which produces CH$_2$ everywhere through the disk at all times. The rate coefficient of this reaction is not well known, with
842: with an estimated value of $10^{-17}$~cm$^3$\,s$^{-1}$ at $T = 10-300$~K \citep{Prasad_Huntress80,rate06}.
843: %STATE THE TEMPERATURE. BY THE WAY, THIS RATE IS PROBABLY A WILD GUESS.
844: This reaction is relevant for the evolution of CO, HCO$^+$, H$_2$CO, CS, and HCN. % (see Fig~\ref{all_spec_corr_map}).
845: Second, the reaction between H$_2$ and C$^+$ leads to the formation of
846: CH$_2^+$ (important for C$^+$, HCO$^+$, CS, NH$_3$, and CN),
847: with an estimated rate coefficient of $\approx 4\times10^{-17}$~cm$^3$\,s$^{-1}$ \citep{Herbst_85}. These
848: two reactions initiate processes of C and H addition in the chemical
849: model and thus should be relevant for nearly all species containing
850: carbon and hydrogen.
851:
852: The next most important RA reaction (8th place out of 56) is the formation of protonated methane from molecular hydrogen
853: and CH$_3^+$ \citep[$\alpha=1.3\times 10^{-14}$~cm$^3$\,s$^{-1}$, $\beta=-1$;][]{Smith89},
854: % YOUR REFERENCE IS PROBABLY NOT RIGHT; THE REACTION HAS ACTUALLY BEEN MEASURED BY TWO GROUPS HEADED BY GERLICH AND
855: % GORDON DUNN
856: which is an important molecular ion for CO destruction.
857: More recently the rate of this reaction has been measured at low temperature of 10~K
858: by \citet[]{G95}, $1.1\pm0.1 \times 10^{-13}$~cm$^3$\,s$^{-1}$ for para-H$_2$.
859:
860: Much less important (27th and 28th place) are two slow reactions
861: between C and H or N producing CH or CN radicals. It is interesting
862: to note that the latter reaction has different rate coefficients in the
863: OSU and RATE\,06 databases \citep{OSU03,rate06}. While in
864: osu.2007 this reaction rate has no temperature dependence and no activation
865: barrier \citep[$\alpha=10^{-17}$~cm$^3$\,s$^{-1}$;][]{Prasad_Huntress80}, in RATE\,06
866: the corresponding rate is low, $7.9\times10^{-19}$~cm$^3$\,s$^{-1}$, and possesses a small barrier of 96~K
867: \citep[relevant temperature range is between 300 and 2\,700~K;][]{SA00}.
868: Thus the RATE\,06 and osu.2007 rate coefficients for
869: the RA reaction between C and N differ by an order of magnitude in a wide temperature range.
870:
871: The situation is similar for the RA reaction between molecular hydrogen and CH
872: that forms CH$_3$ -- an important radical for the evolution of formaldehyde.
873: The corresponding osu.2007 rate at 10~K is $2.5\times 10^{-16}$~cm$^3$\,s$^{-1}$
874: ($\alpha=3.25\times 10^{-17}$~cm$^3$\,s$^{-1}$, $\beta=-0.6$). In contrast, in RATE\,06 this
875: reaction has a small barrier of 11.6~K and the 10~K rate value is lower by an order of magnitude,
876: $1.8\times 10^{-17}$~cm$^3$\,s$^{-1}$
877: \citep[$\alpha=5.09\times 10^{-18}$~cm$^3$\,s$^{-1}$, $\beta=-0.71$, and $\gamma = 11.6$~K;][]{BS97}.
878:
879: In addition to radiative association, we list radiative recombination
880: reactions of C$^+$ and CH$_3^+$ in the same group in Table~\ref{imp_reac}.
881: The radiative recombination of ionized carbon is a key
882: reaction for the evolution of C$^+$. % (Fig.~\ref{all_spec_corr_map}).
883: Its rate coefficient at room temperature is $4.67\times
884: 10^{-12}$~cm$^3$\,s$^{-1}$ with an inverse temperature exponent of
885: 0.6 \citep[][]{rate06}. The assumed uncertainty of this rate
886: (the factor of 10) in osu.2007 is in contrast to the results of
887: \citet{Nahar_Pradhan97}, who have theoretically predicted this rate
888: value and found that the corresponding uncertainty is only $\sim
889: 10\%$. In the RATE\,06 database the uncertainty of
890: this rate constant is set to 50\%.
891:
892: Finally, the radiative recombination of CH$_3^+$ is the
893: least probable channel in the reactive collisions between methyl
894: ions and electrons. The rate of this process is estimated to be much higher than for
895: other recombination reactions involving atomic species, particularly in cold
896: regions, and might be accurately determined in laboratory
897: \citep[$\alpha=1.1 \times 10^{-10}$~cm$^3$\,s$^{-1}$,
898: $\beta=-0.5$;][]{umist95,umist99,rate06}. In the osu.2007 database, this rate has a
899: steeper dependence on temperature, $\beta=-0.7$ \citep{OSU03}.
900:
901: We conclude that RA reactions require a particular attention
902: because their rate uncertainties are often large and not well known so that the modeling results can be heavily affected.
903:
904:
905: \subsubsection{Cosmic ray ionization}
906: \label{CRP_imp}
907: The next most important reactions that contribute strongly to the resulting abundance scatter
908: of many molecules are the cosmic ray ionization of molecular hydrogen and helium atoms.
909: The cosmic ray particles able to penetrate deeply into the disk are mostly high-energy protons
910: ($E\ga 100$~Mev) and heavy nuclei (e.g. iron, $E\sim 1$~Gev)
911: \citep{DalgarnoMacCray72,LJO85,DolginovStepinski94}.
912: Direct ionization initiate chemical evolution in dark
913: and cold regions where cosmic ray particles remain the main ionization source \citep{Herbst_Klemperer73}.
914: In the disk this is the outer midplane at $r\ga 30$~AU and the lower part of the intermediate
915: layer \citep{Gammie96,Semenov_ea04}. %, see Fig~\ref{all_spec_corr_map}.
916: Moreover, since we added the X-ray ionization rate to the CRP rate value in an attempt to crudely treat the effect
917: of high-energy stellar radiation, it is not surprising that high correlation coefficients for
918: some CRP-ionization reactions are reached in the disk atmosphere and intermediate layer at $r\la 100$~AU
919: for several key species such as CO (see Fig.~\ref{UncertDistrib}).
920: % (see NH$_3$ in Fig~\ref{all_spec_corr_map}).
921: A more accurate approach to model X-ray chemistry in disks is required
922: \citep{Neufeldea94,Maloneyea96,LeppDalgarno97,Staeuber_ea05}.
923:
924: It is natural that because the cosmic ray ionization of H$_2$ and He is the primal chemical process, it is
925: also the key factor for the accumulation of abundance scatters.
926: In the RATE\,06 database the corresponding rate coefficients are estimated to be accurate within a factor of 2
927: \citep{rate06}. However, this rate uncertainty has two origins.
928:
929: First, there is the \emph{physical} uncertainty that is related to our limited knowledge of
930: the absolute CRP flux and its penetrating ability, which may vary in different astrophysical objects
931: \citep{Spitzer_Tomasko68,DolginovStepinski94,Caselli_ea98}.
932: While the direct ionization process driven by CRPs is well studied and the corresponding cross sections and rates
933: are well known \citep{Solomon_Werner71}, the second \emph{chemical} uncertainty resides in the treatment of the
934: ionization by energetic secondary electrons \citep{Glassgold_Langer_73}. Here the elemental composition
935: of the gas plays a decisive role \citep{Ilgner_ea06a}.
936:
937: The CRP-induced ionization rate of H$_2$ currently used in all UMIST
938: and OSU databases was taken from the PhD thesis of Black (1975). It
939: was computed for a mixture of atomic and molecular hydrogen but new
940: calculations for a mixture of hydrogen and helium would be required
941: in order to have a more consistent rate value \citep[see discussion
942: in][]{Wakelam_ea06b}. Other CRP-ionization reactions listed in
943: Table~\ref{imp_reac} do not have a reference in the RATE\,06
944: database so it is hard to judge how accurate their rates are.
945:
946: Apparently, the rate uncertainty of the CRP-driven reactions can be partially eliminated
947: by more accurate modeling of the full cascade of ionizing events, but the physical uncertainty will
948: still remain, which needs more efforts from observers. \citet{Wakelam_ea06b} have demonstrated how crucial this
949: may be for the results of pure gas-phase chemical models of static molecular clouds.
950:
951:
952: \subsubsection{Photoreactions}
953: \label{PD_imp}
954: Another set of chemical processes with problematic rates are
955: photodissociation (PD) and photoionization (PI) reactions, as listed in Table~\ref{imp_reac}.
956: % and Fig.~\ref{all_spec_corr_map}.
957: These reactions are important in those regions where UV photons are either able to penetrate
958: (the disk atmosphere and intermediate layer) or are induced by cosmic ray particles
959: \citep[disk midplane;][]{PrasadTarafdar83}.
960: High-energy UV photons destroy or ionize gas-phase molecules and allow surface species to desorb back into the gas
961: phase.
962:
963: While both the PD and PI cross sections for many species can be
964: measured or calculated with a rather good accuracy of $\la 50\%$
965: \citep{van_Dishoeck88,van_Dishoeck_ea06}, the major difficulty in
966: obtaining accurate photorates is our limited knowledge of the UV
967: radiation field inside protoplanetary disks. \\citet{Bea03}
968: and \citet{van_Dishoeck_ea06} have shown that many T Tauri stars
969: emit strong, non-blackbody UV radiation. Their spectra resemble that
970: of the interstellar (IS) UV radiation
971: field, with a large fraction of the UV flux emitted in the
972: Ly$\alpha$ line. \citet{van_Zadelhoff_ea03} have demonstrated the
973: importance of scattering for the UV penetration deep into the disk
974: interior. Furthermore, for the two most abundant molecules, H$_2$
975: and CO, that effectively dissociate through partly overlapping lines
976: shortward of about $1\,120$~{\AA}, the optical thickness in these
977: lines can become so high that self- and mutual-shielding have to be
978: taken into account
979: \citep{DS70,van_Dishoeck88,Lee_ea96,Jonkheid_ea07}.
980:
981: In the UMIST and OSU databases, the photorates have mostly been
982: adopted from the compilation of \citet{van_Dishoeck88} and
983: \citet{RJLD91}, where they have been computed using a simple
984: plane-parallel approach and the IS (IS) UV radiation
985: field of \citet{Draine_78}. The IS UV flux is characteristic of a
986: diluted radiation field from an early B star ($T_{\rm eff}\sim
987: 30\,000$~K) located at a distance of $\sim10$~pc, with a cut-off at
988: 912~{\AA} due to absorption by interstellar atomic hydrogen
989: \citep{Spitzer_78,vanDishoeck94}. Obviously, it is not
990: representative of the UV radiation from much cooler Herbig Ae and T
991: Tauri stars surrounded by circumstellar disks. Recently,
992: \citet{van_Dishoeck_ea06} have recalculated the corresponding PD and
993: PI rates, using more appropriate UV spectral distributions typical
994: of Herbig Ae, T Tauri, and Sun-like stars.
995:
996: The 10 problematic photodissociation and photoionization reactions found by the sensitivity
997: analysis are involved in either the formation of key species (C$^+$ and CN), their destruction
998: (CO, CN, CS, NH$_3$, and HCN), or the destruction of parental molecules (OH, CH, and CH$_2$).
999: Their rates in the RATE\,06 database are in general inaccurate by a factor of 2 with some exceptions.
1000: The photodissociation of CO proceeds indirectly, through predissociation,
1001: and depends on the amount of molecular hydrogen and CO molecules in the line of sight to the source of UV
1002: radiation \citep{van_Dishoeck87}, and thus has a rate uncertainty of a factor of 10 (see Table~\ref{imp_reac}).
1003: In contrast, the rates of direct photodissociation of NH$_3$, OH, and HCN are inaccurate by the factor of 1.5
1004: only \citep{RJLD91}. Despite such small error bars, the latter three reactions
1005: were identified as problematic reactions for disk chemistry.
1006:
1007: % We conclude that most of the photorate uncertainties derive from inaccurate treatment of the UV penetration and
1008: % uncertainties in the measured UV flux and UV spectral energy distribution, unless their cross section were not derived
1009: % by analogy.
1010:
1011:
1012: \subsubsection{Ion-neutral and neutral-neutral reactions}
1013: \label{INNN_imp}
1014: Ion-neutral (IN) and neutral-neutral (NN)
1015: reactions form the largest fraction of the problematic reactions (32
1016: out of 56) for disk chemistry, but their individual importance is
1017: not as significant as for CRI and RA reactions
1018: (Table~\ref{imp_reac}).
1019: % The only reaction of that kind that appears in Fig.~\ref{all_spec_corr_map} is
1020: % reaction between N and CH$_2$ producing HNC and H.
1021:
1022: Ion-neutral reactions are usually exothermic and rapid, and
1023: lead to the formation of new species by bond rearrangement. Their
1024: rate coefficients can often be obtained by simple Langevin theory,
1025: which relates the isotropic polarizability of the species and its
1026: reduced mass \citep{Clary_88}. For molecules that possess a large
1027: dipole moment, the rate values can be significantly enhanced at low
1028: temperature by the long-distance Coulomb attraction between the
1029: positive ion and negatively charged side of the molecule
1030: \citep{ASC85,Clary85}.
1031: % THE PROPER REFERENCE SHOULD BE TO CLARY OR DAVID SMITH AND NIGEL ADAMS.
1032: Moreover, many neutral-neutral reactions with small barriers or without a barrier can also be fast
1033: under interstellar conditions. The rates of most radical-radical reactions and even some radical-stable neutral
1034: reactions are controlled by long-range attractive forces and thus do not get smaller at low temperatures
1035: \citep[see, e.g.,][]{Smith_88,SQTea_93,SSCea94,CSTea97}.
1036: % MANY REFERENCES TO THE WORK OF IAN SMITH AND IAN SINS FOR RADICAL-NEUTRAL REACTIONS. PLEASE PUT SOME IN.
1037: The typical rate coefficient for these reactions is about an order
1038: of magnitude lower than for ion-neutral processes. The rates
1039: of some of the fast neutral-neutral reactions at low temperatures
1040: have been measured in the laboratory and predicted theoretically
1041: \citep[e.g.,][]{Clary_ea94,SQDea94,ChLPSea01,OSU03}.
1042: % AGAIN, YOU ARE MISSING MANY EXPERINENTALLY STUDIED REACTIONS BY SINS AND SMITH.
1043:
1044: In RATE\,06, most of the IN and NN rate coefficients are uncertain
1045: by a factor of 2 and less, up to the accuracy of $25\%$
1046: (Table~\ref{imp_reac}). However, a major ambiguity that resides in
1047: the IN rate constants is the treatment of their temperature
1048: dependence. In two versions of the RATE\,06 database the rate value
1049: for the reaction involving a polar molecule and a molecular ion can
1050: have a negative dependence on temperature ($\beta = -0.5$; adopted
1051: in this work) or no temperature dependence ($\beta=0$), while its
1052: rate coefficient fits the value at room temperature. In contrast, in
1053: the OSU network the corresponding IN rates based on low
1054: temperature theoretical estimations to be most accurate at $\sim 10$~K
1055: even though the listed reaction rate coefficients are scaled with respect
1056: to 300 K with $\beta \sim -0.5$. Two problematic
1057: reactions involving a polar molecule and an ion with significantly
1058: different rates in RATE\,06 and osu.2007 are those between C$^+$ and
1059: NH, forming CN$^+$ and H ($\alpha_{\rm osu.2007} = 4.6 \times
1060: 10^{-9}$~cm$^3$\,s$^{-1}$ vs. $\alpha_{\rm RATE\,06} = 7.8 \times
1061: 10^{-10}$~cm$^3$\,s$^{-1}$) and between S$^+$ and CH, producing
1062: CS$^+$ and H ($\alpha_{\rm osu.2007} = 4.4 \times
1063: 10^{-9}$~cm$^3$\,s$^{-1}$ vs $\alpha_{\rm RATE\,06} = 6.2 \times
1064: 10^{-10}$~cm$^3$\,s$^{-1}$).
1065:
1066: Other IN reactions with different rates in the RATE\,06 and osu.2007
1067: databases are reactions involving an atom and a molecular ion.
1068: The break up of molecular hydrogen upon collision with helium ion
1069: involves a small barrier of 35~K and $\alpha = 3.7\times
1070: 10^{-14}$~cm$^3$\,s$^{-1}$ in RATE\,06
1071: \citep[the relevant temperature range is between 10K and 300K;][]{rate06}, while in
1072: osu.2007 the same rate coefficient is about 5 times smaller and has no barrier
1073: \citep{OSU03}. However, this results in difference of the osu.2007 and RATE\,06 rate values
1074: of an order of magnitude at most.
1075:
1076: Similarly, the OSU reaction between molecular
1077: hydrogen and ionized ammonia produces hydrogen atom and protonated
1078: ammonia with a rate coefficient of $1.5 \times
1079: 10^{-14}$~cm$^3$\,s$^{-1}$ and a steep temperature dependence
1080: ($\beta=-1.5$). In RATE\,06 this reaction has a non-Langevin rate with $\beta = 0$.
1081: At temperatures below 20~K its rate coefficient is twice as large, $\alpha = 3.36\times 10^{-14}$~cm$^3$\,s$^{-1}$,
1082: and a negative activation barrier of -35.7~K exists \citep{AS_84}. At temperatures between 20 and 300~K
1083: this reaction has no barrier and $\alpha = 2\times 10^{-13}$~cm$^3$\,s$^{-1}$. For temperatures above 300~K the
1084: rate coefficient is $1.7 \times 10^{-11}$~cm$^3$\,s$^{-1}$ and a barrier is about 1\,000~K
1085: \citep{Fehsenfeld_ea75}. Consequently, the osu.2007 and RATE\,06 rate values can differ by up to one order
1086: of magnitude.
1087:
1088: Finally, the IN
1089: reaction between H$_3^+$ and O produces only OH$^+$ and H$_2$ in
1090: OSU, but in RATE\,06 a second, less probable alternative channel
1091: that leads to the production of ionized water and atomic hydrogen is
1092: given \citep{MMcE_00}.
1093:
1094: Note that the 4 IN reactions with the smallest rate uncertainty of
1095: $25\%$ are still problematic for disk
1096: chemistry. Such high sensitivity of the disk modeling results to the rates of these
1097: reactions is caused by their importance for the
1098: evolution of CO, HCO$^+$, and HCN.
1099: The most important one is the IN reaction between helium ions and CO, which leads to the hypersensitivity of
1100: the final abundances of many carbon-bearing species in the inner part of the disk
1101: intermediate layer (see Fig.~\ref{hists}).
1102: The other reactions include the primal route to the formation of HCO$^+$ and destruction of CO:
1103: H$_3^+$ + CO $\rightarrow$ HCO$^+$ + H$_2$ \citep[$\alpha = 1.7 \times 10^{-9}$~cm$^3$\,s$^{-1}$;][]{Kim_ea75}
1104: as well as the destruction of molecular nitrogen by ionized helium
1105: \citep[$\alpha = 6.4 \times 10^{-10}$~cm$^3$\,s$^{-1}$;][]{rate06} and
1106: destruction of HCN by ionized carbon
1107: \citep[$\alpha = 3.1 \times 10^{-9}$~cm$^3$\,s$^{-1}$, $\beta=-0.5$;][]{Clary_ea85}.
1108:
1109: All of the 15 problematic neutral-neutral reactions for disk chemistry are reactions involving atoms
1110: (O, N, C, H, S) and either light hydrocarbons (C$_{\rm n}$H$_{\rm m}$, n=1,2, m=1--3) or CN, HCO, and H$_2$CO.
1111: In contrast to the IN reactions, the NN reactions are mainly important for the chemical evolution of the key
1112: species in warm disk regions, in particular reaction with atomic oxygen \citep{TG07}.
1113: Though their rates are rather accurate, with uncertainties that are typically not higher than $50\%$,
1114: some of the NN rates are still different in RATE\,06 and osu.2007 databases.
1115:
1116: The NN reaction between N and HCO produces only OCN and H in osu.2007, with a rate coefficient of
1117: $10^{-10}$~cm$^3$\,s$^{-1}$ \citep{OSU03}. In RATE\,06 this reaction has in addition 2 other channels
1118: that form either HCN and O ($\alpha = 1.7 \times 10^{-10}$~cm$^3$\,s$^{-1}$)
1119: or CO and NH \citep[$\alpha = 5.7 \times 10^{-12}$~cm$^3$\,s$^{-1}$, $\beta=0.5$, and $\gamma =1\,000$~K;][]{rate06}.
1120:
1121: Another reaction in the list is H + CH, which leads to C and H$_2$.
1122: In osu.2007, the rate coefficient has a weak dependence on temperature and no barrier
1123: \citep[$\alpha = 2.7 \times 10^{-11}$~cm$^3$\,s$^{-1}$, $\beta=0.38$;][]{umist99},
1124: while in RATE\,06 this rate has $\beta=0$ and an $80$~K barrier
1125: \citep[$\alpha = 1.3 \times 10^{-10}$~cm$^3$\,s$^{-1}$;][]{rate06}.
1126:
1127: Finally, the neutral-neutral reaction between H and CH$_2$,
1128: which forms CH and H$_2$, has the osu.2007 rate coefficient of
1129: $2.7 \times 10^{-10}$~cm$^3$\,s$^{-1}$ taken from the UMIST\,95 database,
1130: but this value is only $6.6 \times 10^{-11}$~cm$^3$\,s$^{-1}$ in RATE\,06.
1131:
1132: We conclude that the reaction rates of many important
1133: ion-neutral and neutral-neutral reactions are not known accurately
1134: enough, as a major controversy in their temperature dependence still
1135: persists. This is particularly true for the reactions involving an
1136: ion and a polar molecule, whose rates in RATE\,06 can differ by
1137: a factor of $\sim 5$ at 10~K for the dipole/non-dipole versions of this ratefile.
1138: The ion-polar rates in osu.2007 can be larger than those in the dipole
1139: version of RATE06 by another factor of 4-5, especially for linear neutral
1140: reactants, in which allowance is made for sub-thermal rotational populations
1141: via the so-called ``locked-dipole'' approach.
1142: % THIS SUMMARY IS FAR TOO OPTINISTIC: WHAT ABOUT THE ION-POLAR REACTION RATES WHERE YOU ACTUALLY HAVE CHOICES IN RATE06
1143: % THAT CAN LEAD TO GREATER THAN ORDER-OF-MAGNITUDE UNCERTAINTIES AT 10 K????
1144:
1145:
1146: \subsubsection{Dissociative recombination}
1147: \label{DR_imp}
1148: Molecular ions are efficiently converted into other, less complex molecules via dissociative
1149: recombination (DR) with electrons or charged grains.
1150: Similarly to IN reactions, these processes are especially fast at low temperatures
1151: with typical rate coefficients of about $10^{-6}$~cm$^{-3}$\,s$^{-1}$ at $10$~K due to the long-range Coulomb
1152: attraction. The DR rates can be measured rather accurately with afterglow and storage ring techniques
1153: \citep{Florescu_Mitchell06}. However, the sensitivity of their branching ratios with respect to the
1154: temperature is mostly unexplored and may not be described by a simple power law at low $T$
1155: \citep{Petrignani_ea05}.
1156:
1157: A strong influence of the DR rates and branching ratios on the results of chemical models of molecular clouds
1158: has been found \citep{Millar_ea88,Semaniak_ea01,Geppert_ea05a}. Our sensitivity method picked up two
1159: DR reactions that introduce significant uncertainties in computed abundances in protoplanetary disks.
1160: These include dissociative
1161: recombination of protonated ammonia into ammonia and hydrogen atoms (50th place out of 56), and a main
1162: destruction channel for HCO$^+$ (19th place, see Table~\ref{imp_reac}).
1163: Note that their rates are supposed to be known within an uncertainty of $25\%$.
1164:
1165: The DR of HCO$^+$ in RATE\,06 has one dissociative channel to CO and H with the total
1166: rate coefficient of $2.4 \times 10^{-7}$~cm$^3$\,s$^{-1}$ and $\beta=-0.69$ \citep{Mitchell_90}.
1167: In the osu.2007 database 2 additional channels are listed, which lead to the formation of either OH and C or CH and O but
1168: with a much smaller probability of $\sim 4\%$ each. The DR of NH$_4^+$ in both RATE\,06 and osu.2007
1169: involves 3 branching channels into NH$_2$ and either two hydrogen atoms ($21\%$) or one molecular hydrogen
1170: ($10\%$), or ammonia and atomic hydrogen ($69\%$) with the total rate of $1.5 \times 10^{-6}$~cm$^3$\,s$^{-1}$
1171: and $\beta=-0.5$ \citep{Vikor_ea99}.
1172:
1173: Still, more effort needs to be invested into the (re-)investigation of the DR branching ratios and their
1174: temperature-dependence, especially at low $\sim10$--20~K temperatures.
1175:
1176:
1177: \section{Summary and conclusions}
1178: \label{concl} The influence of gas-phase reaction rate uncertainties
1179: on the results of disk chemical modeling has been studied by a Monte
1180: Carlo method. The rate coefficients in the RATE\,06 network were
1181: varied 8\,000 times within their uncertainty limits using a
1182: log-normal distribution. Sets of abundances, column densities, and
1183: their error bars were computed for a number of key species in
1184: protoplanetary disks. We found that typical uncertainties of the
1185: molecular column densities do not exceed a factor of 3--4 even for
1186: the largest of the key molecules, which is comparable with
1187: observational uncertainties. The column densities of CO, C$^+$,
1188: H$_3^+$, H$_2$O, NH$_3$, N$_2$H$^+$, and HCNH$^+$ have particularly
1189: small error bars. A straightforward correlation analysis between
1190: molecular abundances and reaction rates was performed for the entire
1191: evolutionary time, and the most problematic reactions involving CO,
1192: C$^+$, CS, HCO$^+$, H$_2$CO, CN, HCN, and NH$_3$ were identified. We
1193: showed that the rate coefficients of about a hundred chemical
1194: reactions constituting only a few percent of the entire RATE\,06
1195: database need to be determined more accurately in order to
1196: significantly decrease uncertainties in the modeled abundances and
1197: column densities of the key observable molecules. We argue that it
1198: is worthwhile to (re-)investigate, either experimentally or
1199: theoretically, the rate coefficients of basic radiative association
1200: and cosmic ray ionization reactions as well as the
1201: temperature-dependence of key ion-neutral and neutral-neutral
1202: reactions, and the branching ratios and products of dissociative
1203: recombination processes. The rate uncertainties of cosmic ray
1204: ionization and photoreactions are partly due to uncertain physical
1205: parameters such as the CRP flux and UV penetration, and thus can be
1206: decreased with better physical models and observational data.
1207:
1208:
1209: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1210: % Acknowledgments:
1211: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1212: \begin{acknowledgements}
1213: A.V. appreciates financial support through a grant from the Dynasty
1214: Foundation and IMPRS fellowship. E. H. acknowledges the support of the National Science Foundation for
1215: his research program in astrochemistry. A. S. acknowledges the support of the Russian Foundation for Basic Research
1216: (grant 07-02-00628-a). Authors are thankful to the anonymous referee for valuable comments and suggestions.
1217: This research has made use of NASA's Astrophysics Data System.
1218: \end{acknowledgements}
1219:
1220:
1221: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1222: % References:
1223: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1224: %\bibliographystyle{apj}
1225: %\bibliography{apj-jour,references}
1226: \input{ms.bbl}
1227:
1228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1229: % Figures:
1230: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1231: \clearpage
1232: \begin{figure}
1233: \includegraphics[width=0.28\textwidth,clip=,angle=90]{fig1a.eps}
1234: \includegraphics[width=0.28\textwidth,clip=,angle=90]{fig1b.eps}
1235: \caption{Temperature (K) and density (cm$^{-3}$) structure of the adopted disk model.
1236: The total scale height of the disk $Z_{\rm max}$ increases with the radius such that at
1237: $r=100$~AU it has a value of 90~AU, whereas $Z_{\rm max} \approx 1\,300$~AU at $r=800$~AU.}
1238: \label{disk}
1239: \end{figure}
1240:
1241: \clearpage
1242: \begin{figure}
1243: \includegraphics[width=0.45\textwidth,clip=]{fig2.eps}
1244: \caption{Histograms of the abundance distributions show the number of chemical models
1245: in which a given molecular abundance is reached at 5~Myr. (Top) Non-Gaussian, double-peak
1246: distributions are shown for HCO$^+$ (left) and CO (right) in the inner disk region
1247: ($r\approx 97.5$~AU, $z \approx 30$~AU).
1248: (Bottom) Nearly Gaussian distributions are shown for the same species but located
1249: farther out from the central star ($r\approx 382$~AU, $z \approx 205$~AU).
1250: The point numbers in the boxes refer to the disk cells for which these calculations were
1251: performed.}
1252: \label{hists}
1253: \end{figure}
1254:
1255:
1256: \clearpage
1257: \begin{figure}
1258: %\includegraphics[angle=90]{orig_ab.eps}
1259: \includegraphics[width=0.65\textwidth,clip=,angle=90]{fig3.eps}
1260: \caption{Distribution of mean molecular abundances relative to the total amount of hydrogen nuclei
1261: for several observationally important species in the disk at 5~Myr. The distributions are obtained by
1262: averaging the 8\,000 model results calculated with randomly varied reaction rates. Results are
1263: shown for CO, C$^+$, CS, HCO$^+$ (top row, from left to right) and H$_2$CO, NH$_3$, HCN, and CN
1264: (bottom row) at $r\ge 50$~AU. The disk vertical axis is given in units of the pressure scale height.}
1265: \label{Abu2D}
1266: \end{figure}
1267:
1268:
1269: \clearpage
1270: \begin{figure}
1271: \includegraphics[width=0.65\textwidth,clip=,angle=90]{fig4.eps}
1272: \caption{The same as in Fig.~\ref{Abu2D} but for abundance uncertainties (logarithmic units).}
1273: \label{UncertDistrib}
1274: \end{figure}
1275:
1276:
1277: \clearpage
1278: \begin{figure}
1279: \includegraphics[width=0.65\textwidth,clip=,angle=90]{fig5.eps}
1280: \caption{The same as in Fig.~\ref{UncertDistrib} but for radial distributions of the column densities.
1281: The central solid line represents the model-averaged column densities.
1282: The corresponding column density scatters computed with the standard model are depicted with dotted lines.
1283: The column density scatters calculated with the chemical model where the rates of the 56 most problematic reactions
1284: are varied by small factors of 1.25--2 only are shown with solid lines.}
1285: \label{col_dens}
1286: \end{figure}
1287:
1288:
1289: \clearpage
1290: \begin{figure}
1291: \label{CorrInTime}
1292: \includegraphics[width=0.4\textwidth,clip=,angle=90]{fig6.eps}
1293: \caption{The largest correlation coefficients of
1294: relevant reactions for the evolution of NH$_3$ as a function of time
1295: in the disk intermediate layer ($r=97$~AU, $z=30$~AU).}
1296: % WE NORMALLY DON'T USE SUCH LONG DISCUSSION AS FIGURE CAPTIONS, SO I WOULD ELININATE THE SUBSEQUENT DISCUSSION... I ESPECIALLY DON'T UNDERSTAND THE ANTI-CORELATION BUSINESS.....
1297: % Note that the correlation of the NH$_3$
1298: % abundances with the uncertainty in the reaction of the CRP-ionization
1299: % of helium atoms becomes the largest ($\approx +0.35$) at $t\la
1300: % 10$~yr only. Similarly, cosmic ray ionization of H$_2$ and N have
1301: % significant impact on the NH$_3$ abundances only at early
1302: % evolutionary time, $t<10$~year. The correlation coefficient of
1303: % the radiative association reaction C$^+$ + H$_2$ producing
1304: % CH$_2^+$ shows more complicated, Gaussian-like behavior
1305: % with a peak value of $\sim 0.25$ at about 50~years. The largest
1306: % anticorrelation ($\approx -0.1$) corresponds to the photodissociation of
1307: % NH$_3$ into NH$_2$ and H, which is important during the entire evolutionary span.
1308: \end{figure}
1309:
1310:
1311: % \clearpage
1312: % \begin{figure}
1313: % \label{all_spec_corr_map}
1314: % \includegraphics[width=0.9\textwidth,clip=]{fig7.eps}
1315: % \caption{The distributions of the absolute correlations of the 3 most significant
1316: % reactions for the abundance uncertainties of CO, C$^+$, CS, HCO$^+$, NH$_3$, HCN,
1317: % CH$_3$OH, and CN (from top to bottom).}
1318: % \end{figure}
1319:
1320:
1321:
1322: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1323: % Tables:
1324: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1325: % Initial abundances:
1326: \clearpage
1327: \begin{deluxetable}{ll}
1328: \tablewidth{0pt} %\tabletypesize{\footnotesize}
1329: \tablecaption{Initial abundances\label{init_abund}}
1330: \tablehead{\colhead{Species} & \colhead{n(X)/n(H)}}
1331: \startdata
1332: He & 9.75(-2)$^{\rm a}$ \\
1333: H$_2$ & 4.99(-1) \\
1334: H & 2.00(-3) \\
1335: C & 7.86(-5) \\
1336: N & 2.47(-5) \\
1337: O & 1.80(-4) \\
1338: S & 9.14(-8) \\
1339: Si & 9.74(-9) \\
1340: Na & 2.25(-9) \\
1341: Mg & 1.09(-8) \\
1342: Fe & 2.74(-9) \\
1343: P & 2.16(-10)\\
1344: Cl & 1.00(-9) \\
1345: F & 1.00(-10) \\
1346: \enddata
1347: \tablenotemark{a}\tablenotetext{a}{A(-B) means A$\times $10$^{-B}$}
1348: \end{deluxetable}
1349:
1350:
1351: % Predicted column densities for some observationally important species:
1352: \clearpage
1353: \begin{deluxetable}{lll}
1354: \tablewidth{0pt} %\tabletypesize{\footnotesize} %\rotate
1355: \tablecaption{Typical column density uncertainties \label{abs_uncer}}
1356: \tablehead{\colhead{Species} & \colhead{Standard} & \colhead{Improved} \\
1357: \colhead{} & \colhead{model} & \colhead{model}
1358: }
1359: \startdata
1360: H$_3^+$ & factor $\sim$2 & $\sim$1.5 \\
1361: C$^+$ & factor $\sim$2 & $\sim$1.5 \\
1362: CO & factor $\sim$1.25 & 1.1 \\
1363: CO$_2$ & factor $\sim$3.5 & $\sim$2.5 \\
1364: CS & factor 3 & 2.5 \\
1365: HCO$^+$ & factor $\sim 3$ & $\sim$1.5 \\
1366: H$_2$CO & factor 3 & $\sim$1.5 \\
1367: CN & factor 4 & 2 \\
1368: HCN & factor $\sim 3$ & 2.5 \\
1369: HNC & factor 3 & $\sim2.5$ \\
1370: N$_2$H$^+$ & factor 2.5 & 2 \\
1371: NH$_3$ & factor 2.5 & 2.5 \\
1372: HCNH$^+$ & factor 2.5 & 2.5 \\
1373: H$_2$O & factor 1.7 & 1.4 \\
1374: C$_2$H & factor 4 & $\sim 3$\\
1375: \enddata
1376: %\tablecomments{}
1377: \end{deluxetable}
1378:
1379: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1380: % Most relevant reactions for the abundance uncertainties:
1381: \clearpage
1382: \begin{deluxetable}{llllll}
1383: \tablewidth{0pt}
1384: \tabletypesize{\footnotesize} %\rotate
1385: \tablecaption{Most problematic reactions\label{imp_reac}}
1386: \tablehead{\colhead{Reaction} & \colhead{} &\colhead{} & \colhead{Uncertainty} & \colhead{Weight} & \colhead{Type} }
1387: \startdata
1388: {\bf H$_2$ + C } & {\bf $\rightarrow$ } & {\bf CH$_2$ + h$\nu$ } & {\bf factor 10 } & 1.0 & {\bf RA } \\
1389: {\bf H$_2$ + C$^+$} & {\bf $\rightarrow$ } & {\bf CH$_2^+$ + h$\nu$ } & {\bf factor 10 } & 4.2(-1) & {\bf RA } \\
1390: H$_2$ + CH & $\rightarrow$ & CH$_3$ + $h\nu$ & factor 10 & 1.6(-2) & RA \\
1391: H$_2$ + CH$_3^+$ & $\rightarrow$ & CH$_5^+$ + $h\nu$ & factor 10 & 7.6(-2) & RA \\
1392: %O + e$^-$ & $\rightarrow$ & O$^-$ + $h\nu$ & factor 10 & RA \\
1393: %H + O & $\rightarrow$ & OH + $h\nu$ & factor 10 & RA \\
1394: C + H & $\rightarrow$ & CH + $h\nu$ & factor 10 & 2.3(-2) & RA \\
1395: C + N & $\rightarrow$ & CN + $h\nu$ & factor 10 & 2.2(-2) & RA \\
1396: C$^+$ + e$^-$ & $\rightarrow$ & C + $h\nu$ & factor 10 & 1.2(-2) & RR \\
1397: CH$_3^+$ + e$^-$ & $\rightarrow$ & CH$_3$ + $h\nu$ & factor 10 & 1.2(-2) &RR \\
1398: & & & \\
1399: {\bf H$_2$ + C.R.P.} & {\bf $\rightarrow$ } & {\bf H$_2^+$ + e$^-$ } & {\bf factor $2^{\rm *}$ } & 1.4(-1)& {\bf CRI } \\
1400: H$_2$ + C.R.P. & $\rightarrow$ & H$^+$ + H + e$^-$ & factor 2 & 1.0(-2) & CRI \\
1401: He + C.R.P. & $\rightarrow$ & He$^+$ + e$^-$ & factor 2 & 2.3(-1) & CRI \\
1402: N + C.R.P. & $\rightarrow$ & N$^+$ + e$^-$ & factor 2 & 9.7(-2) & CRI \\
1403: %H + C.R.P. & $\rightarrow$ & H$^+$ + e$^-$ & factor 2 & CRI \\
1404: & & & \\
1405: C + $h\nu$ & $\rightarrow$ & C$^+$ + e$^-$ & factor 2 & 3.8(-1) & PI \\
1406: C + C.R.P.$h\nu$ & $\rightarrow$ & C$^+$ + e$^-$ & factor 2 & 6.7(-2) & PI \\
1407: CO + $h\nu$ & $\rightarrow$ & O + C & factor 10 & 5.6(-2) & PD \\
1408: CH + $h\nu$ & $\rightarrow$ & C + H & factor 2 & 2.3(-2) & PD \\
1409: CH$_2$ + $h\nu$ & $\rightarrow$ & CH + H & factor 2 & 6.8(-3) & PD \\
1410: CN + $h\nu$ & $\rightarrow$ & N + C & factor 2 & 3.2(-2) & PD \\
1411: %CO$_2$ + $h\nu$ & $\rightarrow$ & CO + O & factor 2 & PD \\
1412: CS + $h\nu$ & $\rightarrow$ & S + C & factor 2 & 2.6(-2) & PD \\
1413: %HNC + $h\nu$ & $\rightarrow$ & CN + H & factor 2 & PD \\
1414: %H$_2$O + $h\nu$ & $\rightarrow$ & OH + H & factor 1.5 & PD \\
1415: %CH + $h\nu$ & $\rightarrow$ & CH$^+$ + e$^-$ & factor 2 & PI \\
1416: NH$_3$ + $h\nu$ & $\rightarrow$ & NH$_2$ + H & factor 1.5 & 8.2(-3) & PD \\
1417: HCN + $h\nu$ & $\rightarrow$ & CN + H & factor 1.5 & 1.5(-2) & PD \\
1418: OH + $h\nu$ & $\rightarrow$ & O + H & factor 1.5 & 7.8(-3) & PD \\
1419: & & & \\
1420: He$^+$ + CO & $\rightarrow$ & O + C$^+$ + He & factor 1.25 & 9.6(-3) & IN \\
1421: He$^+$ + N$_2$ & $\rightarrow$ & N + N$^+$ + He & factor 1.25 & 7.4(-3) & IN \\
1422: H$_3^+$ + C & $\rightarrow$ & CH$^+$ + H$_2$ & factor 2 & 3.3(-2) & IN \\
1423: H$_3^+$ + O & $\rightarrow$ & OH$^+$ + H$_2$ & factor 1.5 & 1.0(-2) & IN \\
1424: H$_3^+$ + CO & $\rightarrow$ & HCO$^+$ + H$_2$ & factor 1.25 & 7.3(-3) & IN \\
1425: H$_2^+$ + N & $\rightarrow$ & NH$^+$ + H & factor 2 & 4.3(-2) & IN \\
1426: H$_2$ + He$^+$ & $\rightarrow$ & He + H$^+$ + H & factor 2 & 2.0(-2) & IN \\
1427: H$_2$ + He$^+$ & $\rightarrow$ & He + H$_2^+$ & factor 2 & 2.1(-2) & IN \\
1428: H$_2$ + NH$_3^+$ & $\rightarrow$ & NH$_4^+$ + H & factor 1.5 & 4.4(-2) & IN \\
1429: H$^+$ + O & $\rightarrow$ & O$^+$ + H & factor 1.5 & 8.4(-3) & CT \\
1430: % N + NH$^+$ & $\rightarrow$ & N$_2^+$ + H & factor 2 & IN \\
1431: C$^+$ + CH & $\rightarrow$ & C$_2^+$ + H & factor 2 & 1.3(-2) & IN \\
1432: C$^+$ + OH & $\rightarrow$ & CO$^+$ + H & factor 2 & 3.1(-2) & IN \\
1433: C$^+$ + NH & $\rightarrow$ & CN$^+$ + H & factor 2 & 1.1(-2) & IN \\
1434: C$^+$ + HCN & $\rightarrow$ & CNC$^+$ + H & factor 1.25 & 7.9(-3) & IN \\
1435: C + HCO$^+$ & $\rightarrow$ & CO + CH$^+$ & factor 2 & 2.7(-2) & IN \\
1436: CH + S$^+$ & $\rightarrow$ & CS$^+$ + H & factor 2 & 2.6(-2) & IN \\
1437: CH$_2^+$ + O & $\rightarrow$ & HCO$^+$ + H & factor 2 & 7.2(-3) & IN \\
1438: NH$_3^+$ + Mg & $\rightarrow$ & NH$_3$ + Mg$^+$ & factor 2 & 1.1(-2) & CT \\
1439: % C$^+$ + HNC & $\rightarrow$ & C$_2$N$^+$ + H & factor 2 & IN \\
1440: %C + CH$_3^+$ & $\rightarrow$ & C$_2$H$^+$ + H$_2$ & factor 2 & IN \\
1441: & & & \\
1442: C + CH$_2$ & $\rightarrow$ & C$_2$H + H & factor 2 & 3.6(-2) & NN \\
1443: C + C$_2$H & $\rightarrow$ & C$_3$ + H & factor 2 & 8.3(-3) & NN \\
1444: %O + H$_2$CO & $\rightarrow$ & CO + H + OH & factor 1.25 & NN \\
1445: %N + CH$_2$ & $\rightarrow$ & HNC + H & factor 1.5 & NN \\
1446: N + CH$_2$ & $\rightarrow$ & HCN + H & factor 1.5 & 8.5(-2) & NN \\
1447: N + CN & $\rightarrow$ & N$_2$ + C & factor 2 & 4.6(-2) & NN \\
1448: N + HCO & $\rightarrow$ & HCN + O & factor 2 & 2.9(-2) & NN \\
1449: H + CH$_2$ & $\rightarrow$ & CH + H$_2$ & factor 1.25 & 1.9(-2) & NN \\
1450: H + CH & $\rightarrow$ & C + H$_2$ & factor 1.5 & 1.1(-2) & NN \\
1451: CH + O & $\rightarrow$ & HCO$^+$ + e$^-$ & factor 1.5 & 4.0(-2) & NN \\
1452: CH + O & $\rightarrow$ & CO + H & factor 1.25 & 8.3(-3) & NN \\
1453: CH + N & $\rightarrow$ & CN + H & factor 1.25 & 2.3(-2) & NN \\
1454: CH + S & $\rightarrow$ & CS + H & factor 2 & 1.6(-2) & NN \\
1455: CH$_2$ + S & $\rightarrow$ & CS + H$_2$ & factor 2 & 6.3(-2) & NN \\
1456: CH$_2$ + O & $\rightarrow$ & CO + H + H & factor 1.25 & 1.0(-2) & NN \\
1457: %CH$_2$ + O & $\rightarrow$ & HCO + H & factor 1.25 & NN \\
1458: CH$_3$ + O & $\rightarrow$ & H$_2$CO + H & factor 1.25 & 1.4(-2) & NN \\
1459: %O + HCO & $\rightarrow$ & CO$_2$ + H & factor 1.25 & NN \\
1460: & & & \\
1461: %H$_3$O$^+$ + e$^-$ & $\rightarrow$ & H$_2$O + H & factor 1.25 & DR \\
1462: HCO$^+$ + e$^-$ & $\rightarrow$ & CO + H & factor 1.25 & 3.1(-2) & DR \\
1463: NH$_4^+$ + e$^-$ & $\rightarrow$ & NH$_3$ + H & factor 1.25 & 8.1(-3) & DR \\
1464: %H$_3$O$^+$ e$^-$ & $\rightarrow$ & OH + H + H & factor 1.25 & DR \\
1465: \enddata
1466: \tablenotemark{*}\tablenotetext{*}{These rates are partly uncertain
1467: due to uncertainties in physical parameters.} \tablecomments{(CRI)
1468: Cosmic ray ionization; (CT) Charge transfer; (DR) Dissociative
1469: recombination; (IN) ion-neutral reaction; (NN)
1470: Neutral-neutral reaction; (PD) Photodissociation; (PI)
1471: Photoionization; (RA) Radiative association; (RR) Radiative
1472: recombination. Three most important reactions for the disk chemistry
1473: as identified by the sensitivity analysis are written in boldface.}
1474: \end{deluxetable}
1475:
1476:
1477:
1478: \end{document}
1479: