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: %\documentclass[manuscript]{aastex}
7: %\usepackage{apjfonts}
8: %\usepackage{natbib}
9: %\usepackage{graphicx}
10: %\usepackage{emulateapj}
11:
12: \shorttitle{Gas-grain chemistry in molecular clouds}
13: \shortauthors{Vasyunin et al.}
14:
15: \begin{document}
16:
17: \title{A unified Monte Carlo treatment of gas-grain chemistry for
18: large reaction networks. I. Testing validity of rate equations in molecular clouds}
19:
20: \author{A.I. Vasyunin}
21: \affil{Max Planck Institute for Astronomy, K\"onigstuhl 17, D-69117
22: Heidelberg, Germany} \email{vasyunin@mpia.de}
23:
24: \author{D.A. Semenov}
25: \affil{Max Planck Institute for Astronomy, K\"onigstuhl 17, D-69117
26: Heidelberg, Germany} \email{semenov@mpia.de}
27:
28: \author{D.S. Wiebe}
29: \affil{Institute of Astronomy of the Russian Academy of Sciences,
30: Pyatnitskaya str. 48, 119017 Moscow, Russia}
31: \email{dwiebe@inasan.ru}
32:
33: \and
34:
35: \author{Th. Henning}
36: \affil{Max Planck Institute for Astronomy, K\"onigstuhl 17, D-69117
37: Heidelberg, Germany} \email{henning@mpia.de}
38:
39: \begin{abstract}
40: In this study we demonstrate for the first time that the unified Monte Carlo
41: approach can be applied to model gas-grain chemistry in large reaction networks.
42: Specifically, we build a time-dependent gas-grain chemical model of the
43: interstellar medium, involving about 6000 gas-phase and 200 grain
44: surface reactions. This model is used to test the validity of the standard and
45: modified rate equation methods in models of dense
46: and translucent molecular clouds
47: and to specify under which
48: conditions the use of the stochastic approach is desirable.
49:
50: Two cases are considered: (1) the surface mobility of
51: all species is due to thermal hopping, (2) in addition to thermal
52: hopping, temperature-independent quantum tunneling for H and H$_2$
53: is allowed. The physical conditions characteristic for the core and
54: the outer region of TMC1 cloud are adopted. The gas-phase rate file
55: RATE\,06 together with an extended set of gas-grain and surface
56: reactions is utilized.
57:
58: We found that
59: at temperatures 25--30~K gas-phase abundances of H$_2$O, NH$_3$, CO
60: and many other gas-phase and surface species in
61: the stochastic model differ from those in the deterministic models by
62: more than an order of magnitude, at least, when
63: tunneling is accounted for and/or diffusion energies are 3x lower than
64: the binding energies. In this case, surface reactions, involving light species,
65: proceed faster than accretion of the same species.
66: In contrast, in the model without tunneling
67: and with high binding energies, when the typical timescale of a surface
68: recombination is greater than the timescale of accretion onto the
69: grain, we obtain almost perfect agreement between results of
70: Monte Carlo and deterministic calculations in the same temperature range.
71: At lower temperatures ($\sim10$~K) gaseous
72: and, in particular, surface abundances of most important molecules
73: are not much affected by stochastic processes.
74: \end{abstract}
75:
76: \keywords{astrochemistry ISM: abundances, clouds, molecules ---
77: molecular processes -- radio lines: ISM --- stars: formation}
78:
79: \section{Introduction}
80: Chemical processes are discrete in nature, and it has been realized
81: long time ago that their microphysically correct theoretical
82: treatment should rest upon stochastic methods \citep{Gillespie76}.
83: While in cellular biology Monte Carlo simulations of chemical
84: processes in cells are widespread, most models in astrochemistry are
85: based on deterministic rate equation (RE) approach, which has been
86: proved to be inaccurate for chemical processes on ``cell-like"
87: objects like dust grains and PAHs
88: \citep{TielensHagen82,Caselli_ea98,
89: hershem03,LipshtatBiham03,StantchevaHerbst04}. The concept of
90: mass-action kinetics, which lies behind rate equations, being
91: adequate for gas-phase reactions, is not appropriate in the case of
92: small surface population (number of particles $\leq1$ per
93: grain). This situation is often the case in grain surface catalysis.
94: Rates of surface reactions obtained through this approach may be
95: strongly incorrect, leading to an improper estimation of timescales
96: of fundamental processes, like H$_{2}$ formation, and significant
97: errors in abundances of other species.
98:
99: Up to now, a number of attempts have been made to assess the
100: significance of stochastic effects for astrochemical simulations and
101: to develop suitable numerical methods.
102: In the context of astrochemistry first attempt to treat grain
103: surface chemistry stochastically has been made by
104: \cite{TielensHagen82}. In this study, equilibrium abundances of
105: gas-phase species were used to obtain accretion rates and to
106: calculate time-dependent populations of surface species through
107: Monte Carlo approach. All surface species but H$_{2}$ were assumed
108: to remain on the grain surfaces.
109:
110: Subsequent attempts to account for the discrete nature of
111: surface reactions can be divided into two categories. The first
112: category encompasses different modifications to the rate equations
113: for grain surface chemistry \citep{Caselli_ea98, Caselli_ea02,
114: Stantcheva_ea01}. The main idea behind these modifications is to
115: restrict rates of diffusive surface reactions by accretion or
116: desorption rates of reactants. Thus, the so--called
117: accretion-limited regime is taken into account, when the rate of a
118: particular surface reaction is determined by the flux of reactants
119: accreting on a grain and not by their diffusion over the surface.
120: For simple reacting systems, results obtained with this
121: semi-empirical approach were found, at least, to be in closer
122: agreement with Monte Carlo (MC) computations than the treatment by
123: rate equations alone. However, the applicability of the modified
124: rate equations (MRE) to large chemical networks has not been tested
125: so far. Moreover, it is impossible to investigate purely stochastic
126: effects, like bistability, with this technique, still purely
127: deterministic in its nature \citep{LeBourlot_ea93,
128: ShalabieaGreenberg95, Lee_ea98, BogerSternberg06}.
129:
130: In the second category of studies the master equation is solved with
131: different methods. Because the chemical master equation describes
132: probabilities of a chemically reacting system to be in all possible
133: states of its phase space and the number of these states grows
134: exponentially with the number of species, their direct integration
135: is only possible for extremely simple systems, consisting of a very
136: few different components.
137: \cite{Biham_ea01}, \cite{Green_ea01} and \cite{Lipshtat_ea04} have
138: performed a direct integration of the master equation to investigate
139: H$_{2}$ formation. These studies demonstrated that in some cases
140: this fundamental process cannot be adequately described with the
141: rate equation approach. In \cite{Stantcheva_ea02},
142: \cite{Caselli_ea02} and \cite{StantchevaHerbst03} direct integration
143: of the master equation was applied to study the evolution of a
144: simple H---O---CO surface chemical network and deuterium
145: fractionation. Direct integration of the master equation in more
146: complex networks of grain surface reactions is hampered by the lack
147: of appropriate computing power. To the best of our knowledge, the
148: only successful simulation of a more extended surface chemical
149: network in astrochemistry, consisting of 19 reactions, with direct
150: integration of the master equation was accomplished by
151: \cite{StantchevaHerbst04}.
152:
153: A Monte Carlo approach based on continuous-time random-walk method
154: has been developed by \cite{Chang_ea05} to model recombination of
155: hydrogen atoms on interstellar grains. The advantage of this method is
156: that it allows to model formation of H$_2$ and other molecules on surfaces of arbitrary
157: roughness, which has later been demonstrated by \cite{CuppenHerbst07}.
158:
159: To relax computational requirements needed to solve the master
160: equation, the moment equation approximation was suggested by
161: \cite{LipshtatBiham03} and \cite{BarzelBiham07}. Currently, this is
162: the most promising approach to efficient simulations of gas-grain
163: systems with complex grain surface chemistry, because the moment
164: equations can be easily combined with rate equations used to
165: simulate gas-phase chemical processes. But, strictly speaking, the
166: moment equations are an approximation and their validity should be
167: verified by a comparison with exact methods.
168:
169: The only feasible technique allowing us to obtain an exact solution
170: to the master equation for complex chemically reacting systems is
171: Monte Carlo method developed by \cite{Gillespie76}. This approach is
172: widely used in molecular biology for simulation of chemical
173: processes in living cells and can also be applied to astrochemical
174: problems. Gillespie's Stochastic Simulation Algorithm (SSA) was
175: first used in complex astrochemical networks by \cite{Charnley98},
176: where SSA was applied to gas-phase chemistry only. Later a simple
177: grain surface H---O---CO chemical model, similar to the one
178: considered in \cite{Caselli_ea98}, was developed by
179: \cite{Charnley01}. In this model, gas-phase abundances of species
180: were assumed to be constant and desorption was neglected. No coupled
181: gas-grain chemical model was studied.
182:
183: The chemical master equation is usually
184: considered as a method to model surface reactions. However, due to
185: its universality, it can be used to model the gas-phase chemistry as
186: well. The reason why the rate equations are so popular as a tool to
187: study gas-phase chemistry in astronomical environments is the fact
188: that they are easier to implement and much more computationally
189: effective. However, if one uses the rate equations to model
190: gas-phase chemistry and some stochastic approach to model surface
191: chemistry one has to find a way to match the two very different
192: computational techniques. The first study in which a Monte Carlo treatment of grain surface
193: chemistry was coupled to time-dependent gas-phase chemistry is
194: \cite{Chang_ea07}. In this paper iterative technique is utilized to
195: combine very different Monte Carlo and rate equations approaches.
196: Calculations are performed for the time interval of $2\cdot10^{5}$
197: years and a relatively simple H---O---CO surface network consisting
198: of 12 reactions is used.
199:
200: Even though remarkable progress has been made to develop stochastic
201: approaches to astrochemistry, up to now there are no studies in
202: which a stochastic treatment of a complex grain surface chemical
203: network is fully coupled to time-dependent gas-phase chemistry.
204: The situation gets less complicated when
205: the same method is used for both gas-phase and surface chemistry. In
206: this paper, for the first time, we present a ``complete''
207: time-dependent chemical gas-grain model, calculated with Monte Carlo
208: approach. We solve the master equation with a Monte Carlo
209: technique, using it as a single method to treat gas-phase reactions,
210: surface reactions, and gas-grain exchange processes simultaneously.
211: The gas-phase chemical network includes more than 6000 reactions while
212: surface network consists of more than 200 reactions.This Monte Carlo model is used to test the validity of rate equations and modified
213: rate equations for a range of physical conditions, typical of
214: diffuse and dense clouds.
215: So far, no comparison of RE and MRE techniques against rigorous
216: stochastic approach over a wide range of
217: physical conditions, typical for astrophysical objects, has been
218: made.
219:
220: There is clear observational evidence that interstellar grains have
221: complex mineral composition and may have either amorphous or
222: crystalline structure. In general, it is believed that interstellar
223: dust particles are made of olivine-like silicates and some form of
224: amorphous carbon or even graphite \citep{DraineLee78}. The location
225: on a surface which may be occupied by an accreted molecule is
226: commonly referred to as a surface site. The density of sites and
227: binding energies for specific species strongly depend on the dust
228: material and its structure. Binding energies $E_{\rm b}$, that is,
229: potential barriers between adjacent sites, have been assumed by
230: \cite{HHL92} to be roughly 0.3 of corresponding desorption energies
231: $E_{\rm D}$.
232: In \cite{Pirronello_ea97a, Pirronello_ea97b, Pirronello_ea99},
233: \cite{Katz_ea99} and \cite{Perets_ea05} experimental results on the
234: H$_{2}$ formation, the ratio of diffusion energy to desorption
235: energy, $E_{\rm b}/E_{\rm D}$, was found to be close to 0.77 for H
236: atoms. This ratio determines the surface mobility of species and is
237: of utter importance for estimating surface reaction rates.
238:
239: Another factor defining rates of reactions involving H and H$_2$
240: (and their isotopologues) is related to the possibility of their
241: tunneling. Non-thermal quantum tunneling of these species allows
242: them to scan grain surface and find a reacting partner quickly. This
243: mechanism has been suggested as an explanation of efficient H$_2$
244: formation in the ISM. Later analysis of the cited experimental
245: results has lead \cite{Katz_ea99} to the conclusion that tunneling
246: does not happen. The absence of tunneling implies lower mobility, so
247: that surface processes are essentially reduced to accretion and
248: desorption. However, this conclusion turned out to be not the final
249: one. In other studies \citep[e.g.][]{CazauxTielens04} alternative
250: explanations of these experiments have been proposed and some
251: shortcomings of the theoretical analysis performed in
252: \cite{Katz_ea99} have been found. So, up to now the question of
253: presence or absence of tunneling effects on grain surfaces is not
254: settled, and the $E_{\rm b}/E_{\rm D}$ ratio is not well
255: constrained.
256: With this in mind, two different
257: models of grain surface reactions are considered: (1) the surface
258: mobility of all species is caused by thermal hopping only with high $E_{\rm b}/E_{\rm D}$ ratio equal to 0.77, (2)
259: temperature-independent quantum tunneling for H and H$_2$ is
260: included in addition to thermal hopping with low $E_{\rm b}/E_{\rm D}$ ratio equal to 0.3.
261:
262: The organization of the paper is as follows. In Section~2 we
263: describe physical conditions adopted in simulations and chemical
264: model used. Section~3 contains basics of stochastic reaction
265: kinetics and detailed description of the Monte Carlo code used for
266: simulations. In Section~4 the validity of rate equations and
267: modified rate equations is checked against Monte Carlo method.
268: First, global agreement between methods is investigated. Then, some
269: interesting species are discussed separately. In Section~5 a
270: discussion is presented. Section~6 contains the conclusions.
271:
272:
273: \section{Modeling}
274:
275: \subsection{Physical conditions}
276: \label{sec:cloud_models} In the present study, we consider grain
277: surface reactions under physical conditions typical of irradiated
278: translucent clouds, cold dark cores and infrared dark clouds. We consider temperatures $T$
279: between 10 and 50~K, densities $n(\rm H)$ between 200 and
280: $2\cdot10^4$~cm$^{-3}$ (three values of $n(\rm H)$ and five values of $T$), and visual extinctions of 0.2~mag at $n({\rm
281: H})=200$~cm$^{-3}$, 2.0~mag at $n({\rm H})=2\cdot10^3$~cm$^{-3}$ and
282: 15~mag at $n({\rm H})=2\cdot10^4$~cm$^{-3}$
283: \citep[e.g.,][]{SnowMcCall06,Hassel_ea08}. This $A_{\rm v}$ roughly corresponds
284: to the distance of order of 0.4--0.5 pc from the cloud boundary (at constant
285: density). The cloud is illuminated by the mean interstellar diffuse UV field. The dust temperature is
286: assumed to be equal to gas temperature. We do not study the earliest
287: stage of the molecular cloud evolution, which is essentially the
288: stage when atomic hydrogen is converted into hydrogen molecules. The
289: chemical evolution is simulated assuming static physical conditions.
290: While it is still a matter of debate how long does it take for a
291: typical isolated cloud core to become gravitationally unstable and
292: to start collapsing, we assume a short timescale of 1~Myr which
293: seems to be appropriate for the TMC1 cloud \citep[see
294: e.g.][]{Roberts_ea04, Semenov_ea04}.
295:
296: \subsection{Chemical model}
297: The chemical model is the same as described in
298: \citet{Vasyunin_ea08}. The gas-phase reactions and their rates are
299: taken from the RATE\,06 database, in which the effects of
300: dipole-enhanced ion-neutral rates are taken into account
301: \citep{Woodall_ea07}. All reactions with large negative activation
302: barriers are excluded. The rates of photodissociation and
303: photoionization of molecular species by interstellar UV photons are
304: taken from \citet{vDea_06}. The self- and mutual shielding of CO and
305: H$_2$ against UV photodissociation are computed as described in
306: \citet{van_Zadelhoff_ea03} using pre-calculated factors from
307: Tables~10 and 11 from \citet[][]{Lee_ea96}. Ionization and
308: dissociation by cosmic ray particles are also considered, with a
309: cosmic-ray ionization rate of $1.3\cdot10^{-17}$~s$^{-1}$
310: \citep{SpitzerTomasko68}.
311:
312: The gas-grain interactions include accretion of neutral species onto
313: dust grains, thermal and photodesorption of mantle materials,
314: dissociative recombination of ions on charged grains and grain
315: re-charging processes. The dust grains are uniform $0.1\,\mu$m
316: spherical particles made of amorphous silicates of olivine
317: stoichiometry \citep{Semenov_ea03}, with a dust-to-gas mass ratio of
318: $1\%$. The sticking probability is 100\%. Desorption energies
319: $E_{\rm D}$ and surface reaction list are taken from
320: \citet{GarrodHerbst06}.
321:
322: The grain surface is assumed to be compact, with surface density of
323: $2\cdot10^{14}$~sites\,cm$^{-2}$, which gives $\approx 3\cdot10^5$
324: sites per grain \citep{Biham_ea01}. We employ two models to
325: calculate the rates of surface reactions
326: (Table~\ref{tab:surf_mods}). In Model~T (T for tunneling) the tunneling timescale for
327: a light atom to overcome the potential barrier and migrate to
328: another potential well is computed using Eq.~(10) from
329: \citet{HHL92}, with the barrier thickness of 1\AA. In Model H (H for hopping) we do
330: not allow H and H$_2$ to scan surface sites by tunneling. The
331: diffusion timescale for a molecule is calculated as the timescale of
332: thermal hopping multiplied by the total number of surface site and
333: is given by Eq.~(2) and Eq. (4) from \citet{HHL92}. The hopping rates are
334: sensitive to the adopted values of diffusion energy, which are not
335: well constrained. Thus, we consider low and high diffusion energies
336: which are calculated from adopted desorption energies by multiplying
337: them by factors of 0.3 \citep[like in][]{HHL92} (Model T) and 0.77
338: \citep[][]{Katz_ea99} (Model H), respectively. For all considered
339: models, the total rate of a surface reaction is calculated as a sum
340: of diffusion or tunneling rates divided by the grain number density
341: and multiplied by the probability of reactions (100\% for processes without activation energy).
342:
343: Overall, our network consists of 422 gas-phase and 157 surface
344: species made of 13 elements, and 6002 reactions including 216
345: surface reactions. As initial abundances, we utilize the ``low
346: metallicity'' set of \citet{Lee_ea98}, where abundances of heavy
347: elements in the gas are assumed to be severely depleted. All
348: hydrogen is molecular initially. The chemical evolution for 1~Myr in
349: the classical deterministic approaches is computed with the fast
350: ``ALCHEMIC'' code\footnote{Available upon request: semenov@mpia.de}
351: in which the modified rate equations are implemented according to
352: \cite{Caselli_ea98}. No further modification for reactions with activation energy barriers is used \citep[see][]{Caselli_ea02}.
353:
354: \section{Stochastic reaction kinetics}
355:
356: \subsection{Theoretical foundations}
357: We consider a chemically reacting system which consists of $N$
358: different types of species \{$S_1$ ... $S_N$\} and $M$ chemical
359: reactions \{$R_1$ ... $R_M$\} \citep{Gillespie76}. All these species
360: are contained in a constant volume $\Omega$, in which local thermal
361: equilibrium is reached, so that the system is well mixed. We denote
362: the number of species of type $i$ at time $t$ as $X_i(t)$. The
363: ultimate goal is to determine the state vector $\vec{X}(t)=\{X_1(t)
364: ... X_N(t)\}$ of the system at any given time $t>t_0$, assuming
365: certain initial conditions $\vec{X}(t_0)=\vec{X}_0$.
366:
367: Let us assume that each reaction $R_j$ retains properties of a
368: Markov chain and can be considered as a set of independent
369: instantaneous events. Each chemical reaction is described by two
370: quantities, the discrete state vector $\vec{\nu}=\{\nu_{1j}, ...
371: ,\nu_{Nj}\}$ and the propensity function $a_j$. The components of
372: the state vector $\nu_{ij}$ represent the net change in populations
373: of the $i$th species due to the $j$th reaction. By definition, the
374: propensity function
375: \begin{equation}\label{rrate}
376: a_j(\vec{X})dt
377: \end{equation}
378: is the probability for a reaction $R_j$ to occur in the volume
379: $\Omega$ over the time interval $dt$. In analogy to the reaction
380: rate in the rate equation approach, for a one-body
381: %(first-order)
382: reaction the propensity function $a_j$ can be expressed through rate
383: constants and numbers of reactants as
384: \begin{equation}
385: a_j(\vec{X})=c_jx_1.
386: \end{equation}
387: Here, $c_j$ is the rate constant of the $j$th reaction and $x_1$ is
388: the absolute population of reagent. For a two-body
389: %(second-order)
390: heterogeneous reaction this expression changes as follows:
391: \begin{equation}
392: a_j(\vec{X})=c_jx_1x_2,
393: \end{equation}
394: where $c_j$ is the reaction rate constant, $x_1$ and $x_2$ are the
395: abundances of first and second reactants. If we deal with a two-body
396: homogeneous reaction (like H~+~H~$\rightarrow$~H$_{2}$), the
397: expression for its propensity function becomes somewhat more
398: complicated:
399: \begin{equation}
400: a_j(\vec{X})=\frac{c_j}{2}x_1(x_1-1)
401: \end{equation}
402: This expression reflects the fact that the rate of a homogeneous
403: two-body reaction is proportional to the number of all possible
404: pairs of its reactants. The term ($x_1-1$) is important for a proper
405: calculation of the reaction rate in the stochastic regime when the
406: average reagent population is close to unity and cannot be properly
407: reproduced by the rate equation approach. Note that the abundances
408: of species are always integer numbers in these expressions.
409:
410: These definitions allow to characterize the microphysical nature of
411: chemical processes and to establish a basis for stochastic chemical
412: kinetics. A vector equation that describes the temporal evolution of
413: a chemically reacting system by stochastic chemical kinetics is the
414: chemical master equation. To derive this equation one has to
415: introduce the probability of a system to be in the state
416: $\vec{X}(t)$ at a time $t>t_0$:
417: \begin{equation}\label{pdef}
418: P(\vec{X},t\mid\vec{X}_0,t_0)\equiv
419: Prob(\vec{X}(t)=\vec{X}\mid\vec{X}(t_0)=\vec{X}_0).
420: \end{equation}
421: Here $P(\vec{X},t\mid\vec{X}_0,t_0)$ is the conditional probability
422: density function of the time-dependent value $\vec{X}$. Temporal
423: changes of $\vec{X}$ are caused by chemical reactions with rates
424: defined by Eq.~(\ref{rrate}). Therefore, the change of
425: $P(\vec{X},t\mid\vec{X}_0,t_0)$ over the time interval $dt$ is the
426: sum of probabilities of all possible transitions from the state
427: $\vec{X_0}$ to the state $\vec{X}(t)$:
428: \begin{eqnarray*}
429: P\left(\vec{X},t+dt\mid\vec{X}_0,t_0\right)= \\
430: P\left(\vec{X},t\mid\vec{X}_0,t_0\right)\cdot\left\lbrack1-
431: \sum_{j=1}^M\left(a_j(\vec{X})dt\right)\right\rbrack+
432: \\
433: \sum_{j=1}^M P(\vec{X}-\vec{\nu_j},t\mid\vec{X}_0,t_0)\cdot
434: \left(a_j(\vec{X}-\vec{\nu}_j)dt\right)
435: \end{eqnarray*}
436: The first term in this equation is derived from the fact that the
437: system is already in the state $\vec{X}$ and describes the
438: probability for the system to leave the state during the time
439: interval $dt$. The second term defines the total probability for the
440: system to reach the state $\vec{X}$ from states
441: $\vec{X}-\vec{\nu_j}$ during the time interval $dt$ due to a
442: reaction $R_j$. Finally, we formulate the chemical master equation:
443: \begin{eqnarray}\label{mastereq}
444: \frac{\partial P(\vec{X},t\mid\vec{X_0},t_0)}{\partial t}=
445: \sum_{j=1}^M\lbrack
446: a_j(\vec{X}-\vec{\nu_j})P(\vec{X}-\vec{\nu_j},t\mid\vec{X_0},t_0)-
447: %\\
448: a_j(\vec{X})P(\vec{X},t\mid\vec{X_0},t_0)\rbrack
449: \end{eqnarray}
450: This vector equation looks similar to the conventional set of
451: balance equations. However, it properly takes into account the
452: discrete nature of chemical processes and, thus, has a much wider
453: range of applicability. Unfortunately, an analytic integration of
454: the chemical master equation is only possible in a very limited
455: number of cases \citep[several species and reactions, see
456: e.g.][]{StantchevaHerbst04}. Thus, one has to rely on various
457: numerical techniques to solve this equation.
458:
459: \subsection{Implementation of the Monte Carlo algorithm}\label{phys_model}
460:
461: In principle, any normalization of abundances can be used in
462: Eq.~(\ref{mastereq}). In particular, if we take $\Omega$ to be a
463: unit volume, we end up with the usual number densities, which are
464: widely used in astrochemical models. When dealing with mixed
465: gas-dust chemistry, one has to be more careful with the
466: normalization. As surface reactions can only proceed when both
467: reactants reside on a surface of the same grain, we define $\Omega$
468: as the volume of the interstellar medium that contains exactly one
469: dust grain,
470: \begin{equation}\label{gasvol_pergrain}
471: \Omega_{\rm MC} = \frac{\frac{4}{3}\pi\rho_{\rm
472: dust}r^{3}}{\rho_{\rm gas}\gamma},
473: \end{equation}
474: where $r$ is the grain radius, $\rho_{\rm dust}$ is the mass density
475: of grain material, $\rho_{\rm gas}$ is the mass gas density, and
476: $\gamma$ is the dust-to-gas (0.01) mass ratio. Here grains are
477: assumed to be spheres of equal size. As we also assume that the
478: interstellar medium is well-mixed, so that the volume $\Omega_{\rm
479: MC}$ is representative for any volume of the real interstellar
480: medium with the same physical conditions.
481:
482: This setup allows to construct the unified master equation for all
483: three kinds of processes listed above (gas phase reactions,
484: accretion/desorption processes and grain surface reactions). This
485: equation is solved with the stochastic Monte Carlo algorithm
486: described in \citet{Gillespie76}. We implemented this technique in a
487: FORTRAN77 code which allows to simulate all chemical reactions in
488: the network in a self-consistent manner. Such a ``brute-force''
489: approach requires substantial CPU power and cannot be utilized in
490: massive calculations. In the present study, it is used as a
491: benchmark method to simulate chemical evolution in astrophysical
492: objects. A typical run on a single Xeon 3.0GHz CPU takes between 10
493: hours and several days of computational time, and involves several
494: billions of time steps. If one is only interested in a single point then
495: the model can be used not only for benchmarking, but
496: also for practical purposes. Unfortunately, it becomes impractical
497: if abundances are computed for a number of spatial locations. High density also
498: slows down the computation significantly.
499:
500: Due to the fact than the Monte Carlo technique operates with
501: integers, the smallest abundance of a molecule is 1. Given the dust
502: properties in our model (radius $r=10^{-5}$ cm, density $\rho_{\rm
503: dust}=3$ g cm$^{-3}$), a dust-to-gas ratio of 1/100 means that
504: volume per one grain is about $10^{12}/n({\rm H})$ cm$^3$, where
505: $n({\rm H})$ is the number density of hydrogen nuclei. Therefore, an
506: absolute population of 1 corresponds to a relative abundance of
507: $10^{-12}$ with respect to the total number of hydrogen atoms (see
508: Eq. \ref{gasvol_pergrain}). This is the lowest abundance directly
509: resolvable in our model. However, the huge amount of tiny time steps
510: taken during the calculations allows to average stochastic
511: abundances over wider time spans of 10--1\,000 years and push the
512: smallest abundance resolvable by the Monte Carlo code below
513: $10^{-12}$:
514: \begin{equation}
515: \overline{X}=\frac{\sum_{i=1}^{N}{X_{i}\Delta t_{i}}}{\sum_{i=1}^{N}{\Delta t_{i}}}
516: \end{equation}
517: where $X_{i}$ is the abundance of species after the $i$th time step,
518: the $\Delta t_{i}$ is the time step in seconds and $N$ is the amount
519: of time steps over which the averaging is performed. In such an
520: averaging procedure, the noise in the random variable $X_{i}$
521: decreases with $N$ as $\sqrt{N}$ and enables us to resolve
522: abundances below $10^{-12}$. In our simulations we averaged
523: abundances over $10^{6}$ time steps. Therefore, the resolution of
524: our calculations is about 10$^{-15} - 10^{-16}$ cm$^{-3}$. The
525: average value of the Monte Carlo time step in Gillespie's algorithm
526: is equal to an inverse sum of rates of all chemical reactions in the
527: model. For example, in the dense cloud model many molecules form
528: rapidly at initial times and less actively toward the end of
529: evolution (Fig.~\ref{TimeStep_vs_Time}). Time intervals,
530: corresponding to $10^{6}$ time steps in a medium with density
531: $10^{4}$ cm$^{-3}$ vary from 10 to $7\cdot10^{2}$ years for early
532: and late time moments, correspondingly.
533:
534: \section{Comparison of the Monte Carlo method with rate equations}
535:
536: While the Monte Carlo (MC) technique seems to be the most adequate
537: method to solve the chemical master equation, it is computationally
538: extremely demanding. Rate equations (RE) do not account for the
539: stochastic nature of surface processes and thus may produce spurious
540: results in some circumstances. On the other hand, they do require
541: much less computer power and are easier to handle. Therefore rate
542: equations will remain an important tool in theoretical
543: astrochemistry.
544:
545: Given the importance of surface processes, it is thus necessary to
546: isolate regions in the parameter space where rate equations should
547: not be used at all or, at least, should be used with caution.
548: Similar comparisons, having been made so far, are based on a very
549: limited number of species and reactions \citep[e.g.][]{Charnley98,
550: Charnley01, RuffleHerbst00, Green_ea01, Caselli_ea02,
551: Stantcheva_ea02, StantchevaHerbst03, StantchevaHerbst04,
552: LipshtatBiham03, BarzelBiham07, Chang_ea07}. As we implemented a
553: unified Monte Carlo approach to the gas-phase and surface chemistry,
554: which is capable to treat a ``regular size'' chemical network, we
555: are able to perform a comprehensive comparison between predictions
556: of the MC and the (M)RE approaches.
557:
558: For both the Model~T and
559: the Model~H, the rate equations (RE), the modified rate approach
560: (MRE) (for surface chemistry) and the Monte Carlo method are used to
561: simulate gas-phase and surface chemistry. The differences in
562: abundances discussed are related to surface chemistry and/or
563: dust-gas interactions. For the gas-phase chemistry, all models
564: produce identical results. Each run represents a combination of the tunneling treatment (T or H) and the surface chemistry treatment (RE, MRE, or MC). In total, we considered six models denoted as T-RE, T-MRE, etc.
565:
566: \subsection{Global Agreement}
567:
568: To perform an overall comparison, in Figures~\ref{hasg_rate} and%
569: ~\ref{obih_rate} we show diagrams of global agreement between MC,
570: RE, and MRE calculations as a function of $n$ and $T$ for
571: evolutionary times $10^4$, $10^5$, and $10^6$ years. For a given
572: species, the two methods, being compared, are assumed to agree at
573: time $t$ if they produce abundances, that differ by no more than an
574: order of magnitude. As we have mentioned above, in the MC method we
575: are able to resolve abundances down to $10^{-15}$, which would lead
576: to a formal disagreement between abundances, say, $10^{-15}$ in the
577: MC method and $10^{-16}$ (or smaller) in the RE method (all abundances are
578: relative to the total number of H atoms). This kind of
579: disagreement is not meaningful from the observational point of view
580: because such small relative abundances are extremely difficult to
581: observe with a satisfactory accuracy. In addition, even though lower
582: abundances can be calculated by the RE method, their actual accuracy
583: is limited by the numerical interpolation. In the following
584: discussion, species, for which both methods predict relative
585: abundances less than $10^{-12}$, are excluded from further
586: consideration. In Figures~\ref{hasg_rate} and ~\ref{obih_rate},
587: contours labeled with 0.9 mean that the two methods give abundances
588: that differ by less than an order of magnitude for 90\% of species
589: with abundances higher than $10^{-12}$ in both methods.
590:
591: An order-of-magnitude agreement criterion may seem to be too coarse.
592: However, when stochastic effects do not play an important role, all
593: three method produce results, which are nearly identical. To
594: illustrate this, we show in Figure~\ref{h2cccc} the evolution of the
595: H$_2$CCCC abundance at $n=2\cdot10^4$~cm$^{-3}$ and $T=10$\,K,
596: simulated with the MC and RE methods. The species is chosen because
597: its abundance strongly varies with time. It can be seen that the
598: difference between the two methods is very small.
599:
600: \subsubsection{Model~T}
601:
602: In Model~T the species are highly mobile---atomic and molecular
603: hydrogen because of tunneling, and all the other species because of
604: the low energy barrier for diffusion. Mobility drives rapid surface
605: reactions with rates which sometimes exceed the accretion/desorption
606: rate. Surface reactions, which are on average faster than accretion,
607: are not permitted in the MC approach when surface populations of
608: their reagents are low ($\sim$~1), but they are allowed in the RE
609: approach, where the surface abundance of a species can be much less
610: than one. Because of that, differences between Monte Carlo and rate
611: equation methods are quite noticeable in Model~T.
612:
613: At temperatures below 20\,K and above $\sim40$\,K the agreement
614: between the stochastic method and the rate equation method (both RE
615: and MRE) is about 85\% or better at all times and densities
616: (Figure~\ref{hasg_rate}). Around $T\sim10$\,K the residual
617: discrepancy is mainly caused by complex surface species, which have
618: zero abundances in Model~T-MC and are overproduced in Model~T-RE and
619: Model~T-MRE, so that their abundances are just above the adopted
620: cutoff of $10^{-12}$. If we would raise the cutoff to $10^{-11}$,
621: the agreement would be almost 100\%. At $T\sim50$\,K the agreement
622: is also nearly perfect, as the chemistry is almost a purely
623: gas-phase chemistry under these conditions (dust temperature being
624: equal to gas temperature).
625:
626: All the major discrepancies are concentrated in the temperature
627: range between 25--30\,K. In this range the accretion rate (which
628: depends on gas temperature) is high enough to allow accumulation and
629: some processing of surface species, while the correspondingly high
630: desorption rate (which depends on dust temperature and $A_{\rm V}$)
631: precludes surface production of complex molecules. The latter
632: process is adequately described in the MC runs only.
633:
634: At earlier times ($t\la10^5$ years) the largest differences in this
635: temperature range are observed at lower densities. They are caused by
636: an overproduction of `terminal' surface species, like water,
637: ammonia, hydrogen peroxide, carbon dioxide, etc., in the RE and MRE
638: calculations. At later times ($t\sim10^6$~years) situation changes
639: drastically. The discrepancies shift toward higher density, and the
640: overall agreement falls off to about 60\% for both RE and MRE
641: calculations. Even though the stochastic chemistry is only important
642: on dust surfaces, by this time it also influences many gas-phase
643: abundances due to effective gas-grain interactions.
644:
645: At $t=10^6$~years, in the RE calculation at $T=30$\,K, among 234 species with
646: abundances higher than $10^{-12}$, 95 species (including 81
647: gas-phase species) have an order of magnitude disagreement with the
648: MC run. In the MRE run at $T=25$\,K, the number of species with
649: abundances above $10^{-12}$ is 221. Among them 83 species disagree
650: with the MC results by more than an order of magnitude. In both
651: calculations these are mostly carbon-bearing species, like CO,
652: CO$_2$, CS, to name a few, and, in particular, carbon chains (like
653: some cyanopolyynes). It is noteworthy that not only the abundances
654: of neutral species disagree, but also the abundances of some ions,
655: including C$^+$ and S$^+$.
656:
657: \subsubsection{Model~H}
658:
659: In Model~H the mobility of all species on the surface (including H
660: and H$_2$) is caused by thermal hopping only, which is slow because
661: of the high $E_{\rm b}/E_{\rm D}$ ratio. One of the consequences of
662: these slow surface reaction rates is that results produced by the
663: modified rate equations almost do not differ from results of
664: conventional rate equations. The essence of the modification is to
665: artificially slow down surface reactions to make their rates
666: consistent with accretion/desorption rates. In Model~H all surface
667: reactions are slow anyway, so that modifications never occur. Thus,
668: we only compare the results of the MC and RE calculations.
669:
670: At $t=10^4$ years the percentage agreement between the two runs
671: never falls below 80\% and most often is actually much better than
672: the adopted order-of-magnitude criterion.
673: At later times, the discrepancy appears in the same temperature
674: range as in Model~T, i.e., between 25 and 30\,K. Unlike Model~T,
675: both at $t=10^5$~years and $t=10^6$~years the disagreement shows up
676: at low density. The set of discrepant species is nearly identical at
677: both times and consist mostly of ices, overproduced by rate
678: equations. These ices include some key observed species. For
679: example, NH$_{3}$, H$_{2}$O, CO$_2$ ices are all abundant in the
680: Model~H-RE despite the low density, with a surface H$_{2}$O
681: abundance reaching $3\cdot10^{-7}$ by $10^6$ years. On the other
682: hand, the Model~H-MC results in very small or zero abundances for
683: the same species. More chemically rich ices in Model~H-RE bring about
684: enhanced gas-phase abundances for the same molecules. The only
685: species which is {\em under}produced by the rate equation approach
686: is surface C$_2$O.
687:
688: \subsection{Selected Species}
689:
690: The diagrams presented above indicate areas in the parameter space,
691: relevant to molecular clouds, where deterministic methods fail to
692: describe stochastic surface processes. However, a global agreement
693: does not necessarily imply that the abundances of key species are
694: also correctly calculated. In the following we discuss a few
695: important gas-phase species and ices, for which an order of
696: magnitude (or more) disagreement has been found.
697:
698: \subsubsection{Model~T}
699:
700: Because surface reaction rates are high in this model, there are
701: many species for which stochastic and deterministic methods give
702: quite different results. The most abundant trace molecule, CO, is
703: among these species. In Figure~\ref{COtwo}, we show agreement
704: diagrams for Models T-RE and T-MRE (top row) and the CO abundance
705: evolution for $n-T$ combinations corresponding to the worst
706: agreement. Both the RE and MRE methods fail to reproduce the late CO
707: evolution at high density and temperatures of 25--30\,K (where the
708: global agreement is also worst in this model), but each in a
709: different way. While in Model~T-RE CO is underabundant by an order
710: of magnitude, in Model~T-MRE it is overabundant by the similar
711: amount. This difference is related to the treatment of CO
712: $\longrightarrow$ CO$_2$ conversion on dust surfaces, where
713: underabundance of the CO ice in Model~T-RE and overabundance in
714: Model~T-MRE are observed at almost all times. At later time, these
715: surface abundances just start to propagate to gas-phase abundances.
716:
717: Overproduction of the CO$_2$ ice in Model~T-RE also consumes carbon
718: atoms which would otherwise be available not only for CO, but also
719: for more complex molecules, in particular, carbon chains, starting
720: from C$_2$, observed in diffuse clouds, and C$_2$S, used as a
721: diagnostics in prestellar cores. In the RE and MRE calculations
722: these molecules, like CO, show trends opposite to that of CO$_2$
723: ice: when surface CO$_2$ is overabundant with respect to the MC
724: model, carbon chains are underabundant and vice versa.
725:
726: When CO molecule sticks to a grain, it either desorbs back to the gas-phase,
727: where it may participate in further processing, or is converted into CO$_2$.
728: As the desorption energy is quite high for CO$_2$, it acts as a sink
729: for carbon atoms at these temperatures. In the accretion-limited regime,
730: rate equations overproduce CO$_2$ with respect to the MC method. Thus,
731: we see less CO and other carbon-bearing species in the gas-phase. The MRE
732: method helps, probably, too much to account for accretion-limited CO
733: processing and essentially quenches surface CO $\longrightarrow$ CO$_2$
734: conversion. It is worth noting that at high density in this particular temperature range
735: (and for adopted CO desorption energy) CO balances between complete
736: freeze-out (at $T < 20$\,K) and near absence on dust surface (at $T>30$\,K).
737: Thus, even relatively minor changes in treatment may lead to noticeable consequences.
738:
739: Another carbon-bearing molecule that shows an interesting behavior
740: is methanol in the gas phase. Diagrams, comparing its abundance
741: computed with the MC approach and with the rate equations, are
742: presented in Figure~\ref{CH3OHtwo}. The agreement between results of
743: the MC calculations and the RE calculations is almost perfect, at
744: least, within the scatter produced by the MC simulations. However,
745: the modified rate equations, which tend to improve agreement between
746: stochastic and deterministic methods for many other species, in this
747: particular case underpredict methanol abundance by more than two
748: orders of magnitude in comparison with the MC run. The same is true
749: for surface methanol as well. It is interesting to note that the region in the
750: parameter space, where MC and MRE methanol abundances disagree, extends
751: down to lowest temperatures considered.
752:
753: An interesting pattern is presented by the abundance of water,
754: ammonia and their ices (Figure~\ref{water}). These species are
755: severely overproduced by the RE methods at $t=10^5$~years, but after
756: this time the modified rate equations are able to restore the
757: agreement with the MC method, so that by $10^6$ years difference
758: between MC and MRE runs is not significant. This is an example of a
759: situation where the MRE help to produce realistic results, at least,
760: at later times.
761: We show plots only for gas-phase abundances
762: because these are what is really observed. However, these abundances
763: are only a reflection of surface processes, and surface water and
764: ammonia abundances behave in exactly the same way. It is
765: hard to name a single reaction which is responsible for the
766: difference between results of RE and MRE methods in this case.
767: Let's consider water as its chemistry is somewhat simpler.
768: The primary formation reaction for water is H~+~OH, with
769: only a minor contribution from H$_2$O$_2$ (see below).
770: However, hydroxyl is not produced in H~+~O reaction, as one
771: might have expected. Because of the paucity of H atoms
772: on the grain surface, hydroxyl formation is dominated by reaction
773: \[
774: {\rm O}+{\rm HNO}\longrightarrow{\rm NO}+{\rm OH}.
775: \]
776: Abundance of HNO is restored in reaction
777: \[
778: {\rm H}+{\rm NO}\longrightarrow {\rm HNO}.
779: \]
780: These two reactions form a semi-closed loop for which
781: the sole result is OH synthesis out of an H atom and an O atom.
782: Evolution of surface O abundance seems to be the key to the
783: difference between the RE and MRE models. In the RE model
784: the number of O atoms on a surface is determined by the balance
785: between their accretion from the gas phase and consumption
786: in O~+~HNO reaction. The same processes define O abundance
787: in the MRE model as well, but only at $t\la10^5$ years. After this
788: time O+HNO reaction slows down significantly due to modifying
789: correction, and O abundance is controlled purely by accretion and
790: desorption thereafter (as it is in the MC model at all times). As
791: a result, OH and H$_2$O abundances decrease, getting closer to the
792: MC model prediction. A similar mechanism is at work in the case
793: of ammonia as well.
794:
795: \subsubsection{Model~H}
796:
797: The only parameter region where a discrepancy occurs for Model~H is
798: located at $n=2\cdot10^2$~cm$^{-3}$ and $T=25$\,K. Because of the
799: low density, the degree of molecular complexity at these conditions
800: is also low, and by $10^6$ years only 73 species have abundances
801: higher than $10^{-12}$ at least in one of the calculations. Of these
802: 73 species, 25 show disagreement between the MC calculations and the
803: RE calculations.
804:
805: In the top row of Figure~\ref{model_h} we show the low density
806: evolution of gas-phase water and molecular oxygen abundances. Both
807: plots show a similar trend. In the MC calculations the abundance of
808: the molecule stays nearly constant. At earlier times this
809: steady-state behavior is reproduced in the RE run as well, but later
810: the abundance grows and exceeds the MC abundance by more than an
811: order of magnitude. It is interesting to note that the {\em
812: average\/} O$_2$ abundance in the MC run seems to decrease somewhat
813: at later times, and this trend is reproduced by the RE model. The behavior
814: of gas-phase abundances is a reflection of grain-surface abundances.
815: It must be kept in mind that surface water abundance is quite high in
816: the H-RE model, reaching almost $10^{-6}$. Even though evaporation rate
817: is not very high, it does increase gas-phase abundance up to a level of a
818: few time $10^{-10}$. Also, it is a low density model, so desorption is
819: primarily caused by photons.
820:
821: In Figure~\ref{misc} we show the evolution of H$_2$O$_2$ surface
822: abundance and O$_2$ gas-phase abundance in Model~T-RE at
823: $n=2\cdot10^4$~cm$^{-3}$ and the gas-phase ammonia abundance in
824: Model~H-RE at $n=2\cdot10^3$~cm$^{-3}$. These plots demonstrate that
825: disagreement does not always mean simply higher or lower abundances.
826: Different processes determine the H$_2$O$_2$ abundance at different
827: times, and while the RE method is able to capture some of these
828: processes, others are obviously missed. Our analysis shows that this behavior
829: is related to the treatment of reaction with H atoms. Specifically, surface
830: abundance of H$_2$O$_2$ is controlled by three reactions only.
831: It is produced in a reaction H + O$_2$H and destroyed by atomic hydrogen
832: in reactions producing either H$_2$O + OH or O$+2$H + H$_2$. (Note
833: that in the considered model water ice is almost irreversible sink
834: for O atoms as there are no surface reactions destroying water, and
835: desorption is negligible.) The formation reaction is fast as it does not have an
836: activation barrier. So, initially, as H and O atoms start to stick to a grain,
837: surface abundance of H$_2$O$_2$ steadily grows. At the same time,
838: there is not enough H atoms for either of destruction reactions. Then, in the MC model
839: at around 1300 years the first destruction reaction (H~+~H$_2$O$_2$
840: $\longrightarrow$ H$_2$O~+~OH), which has a slightly lower activation
841: barrier than the other one (H~+~H$_2$O$_2$ $\longrightarrow$ O$_2$H~+~H$_2$),
842: starts to ``steal'' some hydrogen atoms from the formation reaction, irreversibly
843: removing one O atom per reaction from the H$_2$O$_2$ (re)formation process.
844: This is the reason for the sharp fall-off of H$_2$O$_2$ abundance. On the
845: other hand, in the RE model both destruction and formation reactions may
846: occur simultaneously, which prevents the first destruction reaction from
847: having such a dramatic effect. Note that the second reaction restores
848: an O$_2$H molecule which may react again with an H atom to restore an
849: H$_2$O$_2$ molecule.
850:
851: H$_2$O$_2$ is not unique in this kind of behavior. There are some other
852: surface species which show more or less similar trends, that is, almost
853: perfect agreement at some times and quite noticeable disagreement
854: at other times. These are some carbon chains (C$_2$H$_2$, C$_2$N)
855: or simpler molecules (CS, NO). Another equally dramatic example
856: is represented by C$_2$H$_2$. The mechanism is similar, that is,
857: it is related to the sequence of hydrogen additions which is treated
858: differently in the MC model and the RE model.
859:
860: Evolution of the O$_2$
861: abundance computed with the RE method follows the evolution computed
862: with the MC model, but most of the time the RE curve is somewhat
863: higher than the MC curve. Finally, the plot for ammonia shows the
864: situation when the RE method correctly describes the average
865: abundance evolution. However, the abundance predicted by the MC
866: method fluctuates so wildly, that at each particular $t$ we can
867: detect disagreement with high probability, using our formal
868: order-of-magnitude criterion.
869:
870: \section{Discussion}\label{diss}
871:
872: In the previous sections we investigated the validity of the
873: (modified) rate equation approach to grain surface chemistry under
874: different physical conditions encompassing diffuse clouds, giant
875: molecular clouds, infrared dark clouds etc. Evolution of the medium
876: is studied with an extended gas-grain chemical model over a time
877: period of 10$^{6}$ years. We utilize the unified stochastic Monte
878: Carlo approach, applied simultaneously to gas-phase and
879: grain-surface reactions. The results are used to test the validity
880: of conventional deterministic approaches. In general, differences in
881: results obtained with deterministic and stochastic methods strongly
882: depend on the adopted microphysical model of surface chemistry. In
883: Model T, where tunneling for atomic and molecular hydrogen is
884: permitted and diffusion/desorption energies ratio is low ($E_{\rm
885: b}/E_{\rm D}=0.3$), discrepancy both for RE and MRE methods is very
886: significant. At low temperatures (10~-~20\,K) only abundances of
887: surface species are discrepant by more than an order of magnitude.
888: At moderate temperatures (25--40\,K), due to active gas-grain
889: interaction, incorrect treatment of grain surface chemistry becomes
890: important for gas-phase abundances, too, and leads to dramatic
891: decrease of overall agreement. At even higher temperatures agreement
892: becomes better again due to limited surface chemistry. In Model H
893: with no tunneling and high $E_{\rm b}/E_{\rm D}$ ratio, surface
894: species are much less mobile, and the agreement between
895: deterministic and stochastic methods is better. The only exception
896: is low density ($n({\rm H})=10^{2}$~cm$^{-3}$) with moderate
897: temperatures (25~--~30\,K), where average residence time of species
898: on grain surface is still lower than the average interval between
899: reaction events. In general, the stochastic chemistry severely
900: affects abundances and, thus, must be taken into account in chemical
901: models of a warm and moderately dense medium.
902:
903: Modifications in rate equations, aimed at taking into account the
904: accretion-limited regime, do not provide a significant improvement
905: over `canonical' rate equations in Model~T. Close inspection of
906: Figure~\ref{hasg_rate} shows that at earlier times
907: ($t\le10^5$~years) modified rate equations mostly produce results,
908: which are rather {\em less\/} consistent with the `exact' Monte
909: Carlo solution, than results of the RE method. Only at later times
910: ($t\sim10^6$~years) results of the MRE method become more consistent
911: with the `exact' solution than the RE results. In the previous
912: section, we have already shown some examples of species, for which
913: the MRE method actually worsens the agreement.
914:
915: Because at many of considered physical conditions results obtained
916: with Monte Carlo method deviate significantly from those obtained
917: with (modified) rate equations, the natural question to ask is
918: whether the inability of the RE method to treat these combinations
919: of parameters might have caused mis-interpretation of observational
920: data. To answer this question, one would need to consider an object,
921: for which a large volume of observational data is available, so that
922: simultaneous comparison for many molecules is possible. So far,
923: there seems to be just one molecular cloud, which is studied in
924: necessary details. This is a sub-region of the TMC1 cloud where
925: cyanopolyynes reach high concentrations, the so-called TMC1-CP.
926: The physical conditions in this object ($T \sim 10$~K, $n_{\rm H}
927: \sim 10^4$~cm$^{-3}$) do fall into the range adopted in our model
928: \citep{BM_89}. However, the TMC1-CP is probably too cold to show
929: significant dependence of gas-phase abundances from surface
930: processes.
931:
932: We may expect that the disagreement will be more significant
933: in hotter objects. The range of physical conditions studied in this
934: paper is quite narrow, as it has only been chosen to provide an
935: initial view on the importance of stochastic treatment for surface
936: chemistry. There are two directions which are worth to follow for
937: further study. One of them is related to diffuse clouds. In this
938: paper we most of the time assumed initial conditions in which all
939: the hydrogen is molecular even at low density. Another
940: simplification, which is, strictly speaking, not valid in low
941: density medium, is equality of dust and gas temperature. So, it is
942: interesting to check if stochastic effects work in predominantly
943: atomic gas with different gas and dust temperatures. Another
944: direction is chemistry at later stages of star formation, that is,
945: in hot cores, hyper- and ultracompact HII regions, protoplanetary
946: disks. In particular, in dark cold disk midplanes, which are poor in
947: atomic hydrogen, stochastic effects may severely influence surface
948: synthesis of hydrogenated species, like formaldehyde, which are
949: later transferred by turbulent mixing in warmer disk areas and
950: desorbed into gas phase \citep{aikawa_co_disk}. In these
951: circumstances one would definitely need to use more accurate methods
952: to model surface chemistry. Stochastic effects may be even more
953: pronounced in the dynamically evolving medium, like during slow
954: warm-up phase in hot cores or ``corinos''. Complex (organic)
955: molecules will be among the most affected species.
956:
957: While the Monte Carlo method is the most direct approach, it is by
958: no means practical. So, other methods are to be developed, in
959: particular, the method of moment equations, if we want to understand
960: deeper the chemical evolution in warm and dense astrophysical objects.
961:
962: \section{Conclusions}
963:
964: In this study, a gas-grain chemical model is presented, consisting
965: of about 600 species and 6000 reactions with stochastic description
966: of grain surface chemistry. For the first time both gas phase and
967: grain surface reactions are simulated with a unified Monte Carlo
968: approach. This unified model is used to test the validity of rate
969: equations and modified rate equations
970: for a set of physical conditions, relevant to translucent
971: clouds, dark dense cores, and infrared dark clouds.
972:
973: A comparison of results obtained with deterministic rate equations
974: (RE) and modified rate equations (MRE) with results of Monte Carlo
975: simulations was performed for this set of physical conditions and
976: two different models of surface chemistry. We found
977: that results obtained with both RE and MRE approaches sometimes significantly
978: deviate from those obtained with Monte Carlo calculations in Model~T,
979: where surface species are highly mobile and most of the grain
980: chemistry occurs in the accretion limited regime. While at low
981: temperatures (10\,K -- 20\,K) mainly RE and MRE abundances of
982: surface species show significant discrepancies, at moderately high
983: temperatures (25\,K -- 30\,K) even abundant gas-phase species like
984: CO considerably deviate from the Monte Carlo results. At these
985: temperatures grain surface chemistry is still very active and at the
986: same time processes of accretion and desorption are very efficient.
987: The gas-phase abundances of many species are influenced by surface
988: chemistry that is correctly described in Monte Carlo simulations
989: only. In Model H surface species are much less mobile than in Model
990: T and the agreement between the (modified) rate equations approach
991: and Monte Carlo simulations is almost perfect. The only parameter
992: region where this discrepancy is significant is at a temperature of
993: $T=25$\,K and at a low density $n({\rm H})=2\cdot10^{2}$ cm$^{-3}$.
994: On the whole, results of modified rate equations seem to be in
995: somewhat closer agreement with Monte Carlo simulations than results
996: of the conventional rate approach. But still, the agreement is far
997: from being perfect, so modified rates should be used with care. We
998: conclude that stochastic effects have a significant impact on
999: chemical evolution of moderately warm medium and must be properly taken
1000: into account in corresponding theoretical models. In the studied parameter range,
1001: stochastic effects are, in general, not important at low ($T\sim10$\,K)
1002: and high ($T\sim50$\,K) temperatures.
1003:
1004: \acknowledgements
1005: Part of this work was supported by the Russian
1006: \emph{Dynasty} foundation. DW is supported by the RFBR grant 07-02-01031. Authors are grateful to Sergey Koposov for
1007: valuable and fruitful discussions and assistance in data analysis
1008: software development. Authors are
1009: thankful to the anonymous referee for valuable comments and
1010: suggestions. This research has made use of NASA's
1011: Astrophysics Data System.
1012:
1013:
1014: %\bibliographystyle{apj}
1015: %\bibliography{apj-jour,references}
1016: \input{ms.bbl}
1017:
1018:
1019: %Types of surface microphysics
1020: \clearpage
1021: \input{tab1.tex}
1022:
1023: \clearpage
1024: \begin{figure}
1025: \plotone{f1.eps} \caption{The Monte Carlo time step as a function of
1026: time in the model of the chemical evolution of a dense cloud.}
1027: \label{TimeStep_vs_Time}
1028: \end{figure}
1029:
1030: \clearpage
1031: \begin{figure}
1032: \plotone{f2.eps} \caption{Global agreement diagrams for Model T at
1033: three time moments. Percentage of species is shown in contour
1034: labels, for which abundances computed with the rate equations and
1035: with the Monte Carlo method differ by no more than an order of
1036: magnitude. Grayscale map with arbitrary contours is added for
1037: clarity. Darker color corresponds to worse agreement. Results are
1038: presented for conventional rate equations (top row) and for modified
1039: rate equations (bottom row).} \label{hasg_rate}
1040: \end{figure}
1041:
1042: \clearpage
1043: \begin{figure}
1044: \plotone{f3.eps} \caption{Same as in Figure~\ref{hasg_rate}, but for
1045: Model H.} \label{obih_rate}
1046: \end{figure}
1047:
1048: \clearpage
1049: \begin{figure}
1050: \plotone{f4.eps} \caption{Evolution of H$_2$CCCC gas-phase abundance at
1051: $n=2\cdot10^4$~cm$^{-3}$ and $T=10$\,K, simulated with the MC and RE
1052: methods (Model~T). Dotted line indicates raw MC results, solid line
1053: shows MC results after averaging. Results of RE integration are
1054: shown with dashed line.} \label{h2cccc}
1055: \end{figure}
1056:
1057: \clearpage
1058: \begin{figure}
1059: \plotone{f5.eps} \caption{(Top row) Agreement diagrams for CO in Model~T.
1060: Contour labels correspond to the logarithm of the ratio of gas-phase CO
1061: abundance in the MC run to that in the RE (top left) and MRE (top
1062: right) runs. (Bottom row) Gas-phase CO abundance evolution in the RE run
1063: ($n=2\cdot10^4$~cm$^{-3}$ and $T=30$\,K; bottom left) and in the MRE
1064: run ($n=2\cdot10^4$~cm$^{-3}$ and $T=25$\,K; bottom right).}
1065: \label{COtwo}
1066: \end{figure}
1067:
1068: \clearpage
1069: \begin{figure}
1070: \plotone{f6.eps} \caption{(Top row) Agreement diagrams for methanol.
1071: Contour labels correspond to the logarithm of the ratio of gas-phase CH$_3$OH
1072: abundance in the MC run to that in the RE (top left) and MRE (top
1073: right) runs. (Bottom row) Gas-phase methanol abundance evolution in the RE run
1074: ($n=2\cdot10^3$~cm$^{-3}$ and $T=25$\,K; bottom left) and in the MRE
1075: run ($n=2\cdot10^3$~cm$^{-3}$ and $T=25$\,K; bottom right).}
1076: \label{CH3OHtwo}
1077: \end{figure}
1078:
1079: \clearpage
1080: \begin{figure}
1081: \plotone{f7.eps} \caption{Water and ammonia gas-phase abundance evolution in
1082: the RE run in Model~T ($n=2\cdot10^2$~cm$^{-3}$ and $T=25$\,K; left) and in the
1083: MRE run ($n=2\cdot10^2$~cm$^{-3}$ and $T=25$\,K; right).}
1084: \label{water}
1085: \end{figure}
1086:
1087: \clearpage
1088: \begin{figure}
1089: \plotone{f8.eps}
1090: \caption{Water and molecular oxygen abundances at $n=2\cdot10^2$~cm$^{-3}$
1091: and $T=25$\,K in Model~H.}
1092: \label{model_h}
1093: \end{figure}
1094:
1095: \clearpage
1096: \begin{figure}
1097: \plotone{f9.eps} \caption{Abundance evolution for selected species.
1098: Models and physical conditions are indicated on the diagrams.}
1099: \label{misc}
1100: \end{figure}
1101:
1102: \end{document}
1103: