0806.4972/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[10pt,preprint]{aastex}                                                                                                    
3: %\documentclass{emulateapj}
4: 
5: \shorttitle{Surface Detonations}
6: \shortauthors{Meakin et al.}
7: 
8: \newcommand{\etal} { et~al.\ }
9: \newcommand{\sol}{$M_\odot$}
10: \def\nuc#1#2{\relax\ifmmode{}^{#1}{\protect\text{#2}}\else${}^{#1}$#2\fi}
11: \def\solrat#1#2{$[$#1/#2$]$}
12: \def\msol#1{\relax$#1\,M_\odot\/$ }
13: \def\mcol{\multicolumn}
14: \def\mult{$\times$}
15: \def\msun{M$_{\odot}$}
16: \def\lang{\langle}
17: \def\rang{\rangle}
18: \def\tnm{\tablenotemark}
19: \def\tnt{\tablenotetext}
20: \def\iso#1#2{$^{#2}${#1}}
21: \def\betp{$\beta^+$}
22: \def\betm{$\beta^-$}
23: 
24: \begin{document}
25: \title{STUDY OF THE DETONATION PHASE IN THE GRAVITATIONALLY CONFINED DETONATION MODEL
26: OF TYPE Ia SUPERNOVAE}
27: \author{Casey A. Meakin\altaffilmark{1,2,3,5}, Ivo Seitenzahl\altaffilmark{3,4}, 
28:   Dean Townsley\altaffilmark{1,2,3}, George C. Jordan IV\altaffilmark{1,2},
29:   James Truran\altaffilmark{1,2,3}, Don Lamb\altaffilmark{1,2,4}}
30: \altaffiltext{1}{Center for Astrophysical Thermonuclear Flashes, University of Chicago, Chicago, IL}    
31: \altaffiltext{2}{Department of Astronomy and Astrophysics, University of Chicago, Chicago, IL}          
32: \altaffiltext{3}{Joint Institute for Nuclear Astrophysics, University of Chicago, Chicago, IL}          
33: \altaffiltext{4}{Enrico Fermi Institute, University of Chicago, Chicago, IL}                            
34: \altaffiltext{5}{Steward Observatory, University of Arizona, Tucson, AZ}                                
35: \email{casey.meakin@gmail.com}
36: 
37: \begin{abstract}
38:   We study the gravitationally confined detonation (GCD) model of Type
39:   Ia supernovae through the detonation phase and into homologous
40:   expansion.  In the GCD model, a detonation is triggered by the surface
41:   flow due to single point, off-center flame ignition in carbon-oxygen
42:   white dwarfs.
43:   The simulations are unique in terms of the degree to which non-idealized physics is
44:   used to treat the reactive flow, including weak reaction rates and a time dependent
45:   treatment of material in nuclear statistical equilibrium (NSE).
46:   Careful attention is paid to accurately calculating the final composition of material
47:   which is burned to NSE and frozen out in the rapid expansion following the passage of a 
48:   detonation wave over the high density core of the white dwarf; and an efficient method
49:   for nucleosynthesis post-processing is developed which obviates the need for costly network 
50:   calculations along tracer particle thermodynamic trajectories.  Observational diagnostics are 
51:   presented for the explosion models, including abundance stratifications and integrated yields.
52:   We find that for all of the ignition conditions studied here, a self regulating process comprised 
53:   of neutronization and stellar expansion results in final \iso{Ni}{56} masses of $\sim$1.1\msun.
54:   But, more energetic models result in larger total NSE and stable Fe peak yields.
55:   The total yield of intermediate mass elements is $\sim0.1$\msun and the explosion energies are 
56:   all around 1.5$\times$10$^{51}$ ergs. The explosion models are briefly compared 
57:   to the inferred properties of recent Type Ia supernova observations.
58:   The potential for surface detonation models to produce lower 
59:   luminosity (lower \iso{Ni}{56} mass) supernovae is discussed. 
60: \end{abstract}
61: 
62: \keywords{stars: evolution - stars: nucleosynthesis - supernovae -
63:   hydrodynamics }
64: 
65: 
66: \section{INTRODUCTION}
67: 
68: \par The currently favored model for Type Ia supernovae (SNe Ia) is the 
69: thermonuclear incineration of a white dwarf (WD) which has accreted mass
70: to near the Chandrasekhar limit from a binary companion 
71: \citep[e.g.,][]{branch1995,hillebrandt2000}. The enormous luminosity
72: and homogeneity in the properties of the light curves of SNe Ia make
73: them exceptionally good standard candles and as such have shown that the
74: expansion rate of the universe is accelerating and provided intriguing evidence
75: for a cosmological constant \citep{riess1998}.
76: 
77: \par Despite the success in using SNe Ia as cosmological probes and identifying
78: a plausible astrophysical progenitor site for the explosions, a detailed
79: understanding of the explosion mechanisms itself remains elusive. 
80: Several uncertainties stand in the way of a definitive solution to the
81: SNe Ia problem. On the one hand, the conditions under which the 
82: thermonuclear runaway commences remains poorly understood so that the
83: initial number and distribution of flamelets that seed the runaway
84: is still a free parameter. On the other hand, although significant progress has been
85: made in simulating flame fronts in multi-dimensional stellar models
86: \citep{gamezo2005,schmidt2006b,roepke2007a,townsley2007,jordan2007}, the challenge
87: associated with modeling an unresolved turbulent deflagration 
88: \citep[e.g.][]{schmidt2006a} with limited computational resources injects
89: an additional degree of uncertainty into the outcome of a model for any 
90: given choice of initial conditions.
91: 
92: %\par A flamelet which ignites off center will rise buoyantly towards the
93: %stellar surface until burning is quenched at low densities.
94: %In order of decreasing energy release, or equivalent flame bubble
95: %ignition further from the stellar center, the following evolutionary paths
96: %can occur  \citep[e.g.][]{livne2005}: (1) The nuclear energy released unbinds
97: %the star.  (2) The nuclear energy release falls short of unbinding the star
98: %but leads to significant expansion of the star which quenches the nuclear
99: %burning and prevents a significant surface flow. (3) The nuclear energy
100: %released during the deflagration phase is not enough to significantly expand
101: %the star and a large fraction of the hot ash is confined to the star in the
102: %form of a surface flow which eventually converges at the pole opposite to
103: %breakout.
104: 
105: %\par All three of these cases have been considered as possible scenarios for 
106: %Type Ia supernovae. The first case is known as a {\em deflagration} or 
107: %{\em pure deflagration} model and is distinct in the sense that no detonation 
108: %forms in this model. The second case, with a marginally bound
109: %core following deflagration, may detonate upon recollapse \citep{khoklov} 
110: %giving rise to a healthy Type Ia explosion.  Most recently, it has been proposed 
111: %that the surface flow in the third case initiates a detonation that consumes the 
112: %unburnt portion of the C/O core \citep{plewa2004,plewa2007} and producing a 
113: %luminous Ia explosion.  
114: 
115: %\par A caveate to the above scenarios is the possibility that deflagration to 
116: %detonation transitions (DDT) occur at locations where the nuclear flame makes
117: %a transition to the distributed burning regime (REF) and the remainder of the 
118: %unburnt stellar core is consumed by one or more detonation waves.  The possibility 
119: %of a DDT exists in any of the above scenarios and the evolution therefore depends 
120: %strongly upon the initial flamelet distribution.
121: 
122: %\par The overall problem requires mastery of at least three 
123: %reactive-hydrodynamic 
124: %subproblems, including: (1) The nature of the initial conditions and the birth 
125: %of a thermonuclear flame, including the spatial distribution of that initial flame, 
126: %its location within the star and whether or not it is composed of a single initial 
127: %spark or a distribution of flame kernels.  (2) The propagation of the nuclear flame 
128: %through the star, which is best described as a Rayleigh-Taylor unstable deflagration. 
129: %(3) The mode by which the (subsonic) deflagration makes a transition to a detonation. 
130: %Each of these problems continues to be an active field of research and have only 
131: %recently started to be investigated with multi-dimensional, multi-physics simulation 
132: %codes with an interesting degree of realism.
133: 
134: 
135: %\par The GCD model can be conveniently broken down into four distinct stages of 
136: %evolution, including: (1) The initial growth of the flame bubble and the subsequent 
137: %erruption of hot ash  onto the surface of the star. (2) The confinement of the hot 
138: %ash to the stellar surface by gravity, which drives a flow around the outside of 
139: %the star towards the anti-pode.   (3) The collision of the surface flow at the 
140: %anti-pode which leads to the formation of an inward and outwardly directed jet.  
141: %(4) The inwardly directed jet eventually leads to conditions  under which a 
142: %detonation arises and sweeps over, the remaining unburned material in the stellar core 
143: %producing a large fraction of a solar mass worth of iron peak and intermediate mass 
144: %elements, disrupting the star and imparting $\sim$10$^{51}$ ergs of kinetic energy to 
145: %the ejecta.
146: 
147: \par In this paper we describe progress on our ongoing effort to improve
148: the simulation of SNe Ia in multi-dimensions, including methods
149: to perform detailed nucleosynthesis post-processing in a computationally efficient manner. 
150: We extend the study of the GCD
151: model for a single ignition point slightly offset a range of distances from the center of the
152: star, as described in \citet{townsley2007}, through 
153: the detonation phase and into homologous expansion.
154: In \S\ref{sec:numerics} we describe the treatment of the reactive-hydrodynamics 
155: problem used in our simulation code. In \S\ref{sec:def} we review the relevant properties of 
156: the deflagration phase for single point flame ignition.  In \S\ref{sec:det} we examine 
157: in some detail the initiation of the detonation, the properties of the detonation 
158: wave which disrupts the star, and the resultant remnant morphology. 
159: In \S\ref{sec:yields} we discuss in detail the nucleosynthetic yields for the explosions
160: studied and decribe the methodology used to efficiently calculate iron peak yields
161: from the simulation data. We conclude with a summary of the salient features of the 
162: explosion models in light of observed Type Ia supernvoae.
163: 
164: 
165: \section{NUMERICAL METHODS: HYDRODYNAMICS AND NUCLEAR BURNING}
166: \label{sec:numerics}
167: \par In this section we review the computational tools used to model the hydrodynamic 
168: and nuclear evolution of the stellar plasma, including the treatment of subsonic 
169: (deflagration) and supersonic (detonation) burning fronts.  The basic code framework is 
170: FLASH \citep{fryxell2000}, a modular, block-structured adaptive mesh refinement (AMR), 
171: Eulerian, reactive-hydrodynamics code. We use a directionally split PPM solver \citep{colella1984} 
172: generalized to treat non-ideal gasses \citep{colella1985} to handle the hydrodynamic 
173: evolution.
174: 
175: \par The energetics scheme used to treat flames and detonation waves
176: in our simulations uses 3 progress variables to track carbon burning, 
177: NSQE relaxation, and NSE relaxation.  The rates connecting these burning 
178: stages are calibrated using a large (200 nuclide) nuclear reaction network 
179: for the conditions relevant to the Type Ia problem. Additionally, energy losses 
180: (through neutrino emission) and changes in the electron mole fraction
181: $Y_e$ due to weak interactions taking place in material which has burned to NSE are
182: incorporated.  Details can be found in \citet{calder2007,townsley2007,seitenzahl2008a}.
183: %\par Nuclear burning is coupled to the hydrodynamic evolution using an operator splitting 
184: %formulation whereby hydrodynamic and nuclear burning modules are called succesively over 
185: %the course of a single timestep \citep{fryxell2000}.
186: 
187: \par Both detonation waves and flames are impossible to resolve in full star simulations
188: because they are characterized by length scales that are more than ten orders of magnitude 
189: smaller than the radius of the white dwarf to be modeled, $R_{\rm wd}\sim10^8$ cm.  
190: Therefore, these reaction fronts must be treated in a special manner. Subsonic burning fronts 
191: (deflagrations) are advanced using an advection-diffusion-reaction (ADR) equation.
192: In short, a thickened flame front  ($\sim$4 grid zones wide) is advanced at a 
193: speed $v_f = \max(v_l, v_t)$, where $v_l$ is the laminar flame speed calculated 
194: by \citet{timmes1992} and  $v_t$ is a Rayleigh-Taylor driven 
195: turbulent flame speed. Details concerning the implementation, calibration and noise 
196: properties of the flame treatment can be found in \citet{townsley2007} and 
197: Asida et al. (2008, in preparation) and references therein.
198: 
199: \par Detonations are handled naturally by the reactive hydrodynamics solver
200: in FLASH without the need for a front tracker. This approach is possible because 
201: unresolved Chapman-Jouguet (CJ) detonations retain the correct jump conditions and propagation speeds. 
202: Numerical stability is maintained by preventing nuclear burning within the
203: shock.  This is necessary because shocks are artificially spread out over a few zones 
204: by the PPM hydrodynamics solver, which can lead to unphysical burning within
205: shocks that can destabilize the burning front \citep{fryxell1989}.
206: The energetics in the detonation differ from that in the deflagration front only
207: in how carbon burning proceeds, as represented by the first progress variable $\phi_1$
208: and an explicit carbon burning rate is used \citep{caughlan1988}. The additional burning stages
209: (NSQE and NSE relaxation) are tracked by $\phi_2$ and $\phi_3$ and are evolved in the same 
210: manner as in the post flame ash \citep{calder2007,townsley2007}.
211: At densities above $\sim 10^7$ g cm$^{-3}$ 
212: detonations propagating through a mixture that is equal parts \nuc{12}{C} and \nuc{16}{O} have 
213: Mach numbers that are larger than the CJ value, but by only a few percent 
214: \citep{gamezo1999,sharpe2001}.
215: Cellular structure smaller than the grid scale will be suppressed in our simulations 
216: but is free to form on resolved scales. The impact of cellular structure on the global evolution
217: of the model is still uncertain.
218: However, since cellular structure alters the detonation wave 
219: speed by only a few percent for the conditions being modeled \citep{timmes2000b} the effect is likely to be small. 
220: Additional details related to the treatment of detonation waves are discussed in \S\ref{sec:det}.
221: 
222: %In order to treat detonations in the explicit manner described, a carbon 
223: %burning reaction rate is needed \citep{caughlan1988} to advance the first
224: %progress variable in place of the flame model. The additional burning stages 
225: %($\phi_2$, $\phi_3$) are advanced using the relaxation timescales in the same 
226: %manner as in the flame.
227: 
228: %\par This approximate treatment of the nuclear burning has the following shortcoming: 
229: %For densities above $10^7$ g/cm$^3$ detonations propagating through a mixture that 
230: %is equal parts \nuc{12}{C} and \nuc{16}{O} are in the ``pathological'' regime 
231: %\citep{gamezo1999,sharpe2001}. [Basic features: detonation speeds vary less 
232: %than X\% over the region of interest.]  
233: %Detonation waves are unstable to perturbations transverse to the direction 
234: %of propagation which give rise to a so-called ``cellular structures'' on length 
235: %scales comparable to the induction length \citep{gamezo1999,timmes2000b}.  
236: %Suppresed on smaller scales by resolution -- what might be the subgrid-scale 
237: %effect of these instabilities? clumpiness, propagation speed differences. 
238: %(Summarize. Conclude by reciting the magnitude of the errors expected. $\sim$ a few \%)
239: 
240: \par Self gravity is calculated using a multi-pole solver with a maximum spherical harmonic
241: index l$_{max}$=10. The Helmholtz equation of state of \citet{timmes2000a} is used to 
242: describe the thermodynamic properties of the stellar plasma including contributions from 
243: blackbody radiation, ions, and electrons of an arbitrary degree of degeneracy.
244: 
245: \par Passive tracer particles are included in our simulations which record
246: the time history of the flow properties along Lagrangian trajectories. These records
247: can be used to calculate detailed nucleosynthetic yields as well as to provide
248: additional diagnostic for complex flows.  We use 10$^5$ tracer particles
249: for 2D models and 10$^6$ for 3D models.  The particles used in this study are
250: initialized at the beginning of each simulation with a mass weighted distribution.
251: In \S\ref{subsec:freezeout-method} we present a novel method to calculate post-explosion
252: yields which does not require the prohibitively expensive post-processing of a large
253: number of tracer particle with a nuclear reaction network, but rather uses information 
254: readily extracted from the tracers to calibrate an efficient table look-up scheme.
255: 
256: \section{FLAME IGNITION AND DEFLAGRATION}
257: \label{sec:def}
258: \par In this paper we extend the study of the GCD model for the single point flame ignition models of
259: \citet{townsley2007} through the detonation phase and into homologous
260: expansion. The general simulation
261: setup is the same, and we review it here briefly, along with a description of
262: the basic progression of the evolution preceding detonation.
263: After carbon burning ignites at the center of the white dwarf, a
264: convective core is formed which expands as it heats.  Our simulations begin
265: when the nuclear burning timescale becomes shorter than the eddy turnover
266: time, so that the first flamelet is ignited near (within a few hundred km of)
267: the center of the white dwarf.  As discussed by \citet{townsley2007}, there is
268: still significant uncertainty
269: in the form which the nuclear flame will take at birth (also see
270: e.g.~\citealt{woosley2004}), relating to the number and location of what are
271: generally assumed to be relatively small ($<1$ km) ignition regions.  For
272: reasons related to simplicity of setup and limitations of the imposed
273: cylindrical symmetry, we restrict our study to off-center, single point
274: ignitions in a quiescent background star.
275: The initial WD used in these simulations has a uniform temperature $4\times
276: 10^7$ K, a mass of \msol{1.365}\footnote[1]{This was erroneously given as
277: \msol{1.38} in \citet{townsley2007}, none of these parameters have
278: changed from that work to this.}, a central density of $2.2\times 10^9$ g
279: cm$^{-3}$, and is composed of equal mass fractions of \nuc{12}{C} and
280: \nuc{16}{O}.  This progenitor is much colder than reality, but it is
281: expected, and we have confirmed by comparison, that this does not have a
282: significant effect on the structure of the white dwarf or the dynamics 
283: of the explosion.  At the beginning of the
284: simulation a spherical region of radius $r_{\rm bub}$=16 km placed on
285: the polar axis at a range of distances between 
286: $r_{\rm off}=$20 and 100 km from the center of the star is converted to NSE 
287: ash in pressure equilibrium with the remainder of the star.  A summary of the 
288: initial flame bubble parameters studied in this paper is given in 
289: Table \ref{tab:models}.
290: 
291: \par The basic stages of single bubble flame evolution can be described in
292: terms of two key length scales, the grid resolution, $\Delta$, which sets the
293: limit to which we can resolve flame structure, and the critical wavelength,
294: sometimes called the fire polishing length, $\lambda_c\equiv 6\pi s^2/Ag$ 
295: \citep{khoklov1995}, where $s$ is the front propagation speed (flame speed)
296: and $A$ is the Atwood number 
297: $A=(\rho_{\rm fuel}-\rho_{\rm ash})/(\rho_{\rm fuel}+\rho_{\rm ash})$.  
298: Perturbations in a
299: Rayleigh-Taylor (R-T) unstable flame front smaller than $\lambda_c$ are
300: ``polished out'' by the propagation of the flame,  while those of larger
301: scale are enhanced by R-T growth, wrinkling the flame bubble.
302: \cite{townsley2007} distinguished three phases of flame evolution that occur
303: successively as the buoyant flame bubble grows in size and rises toward the
304: stellar surface: laminar bubble growth when $r_{\rm bub} \lesssim\lambda_c$,
305: resolved R-T unstable growth, and finally R-T unstable growth on the sub-grid
306: scale when $\lambda_c < \Delta$.  One immediate consequence of this progression 
307: and the increase of $g$ with radius, is that the resolution limits the largest
308: ignition offset position $r_{\rm off}$ which can be used and still start in 
309: the laminar growth phase. This limit is roughly 100 km for $\Delta=4$ km, the 
310: resolution of all the simulations here. At this resolution the critical wavelength
311: is less than the grid scale, $\lambda_c < \Delta$, outside roughly 400 km from the 
312: center of the star.
313: 
314: For even modest offsets, the hot ash is buoyant enough that it erupts from
315: the surface of the star before more than a few percent of the star is burned.
316: This creates a vigorous flow over the surface of the WD, which is still
317: relatively compact due to the small amount of burning.  The progress of this
318: eruption and flow is shown in Figure \ref{fig:breakout} for the case with
319: $r_{\rm off} = 40$ km.
320: As mentioned previously, there is indication from comparisons of recent work
321: \citep{roepke2007a,townsley2007,jordan2007} that the eruption pattern arising
322: from a given offset is dependent on the choice of burning model, and therefore is
323: currently uncertain.  This has important consequences at the collision
324: region because it sets the velocity and density structure of the
325: incoming flows as well as the surface gravity (via the degree of stellar
326: expansion) under which the collision occurs.
327: 
328: We study the cases from \citet{townsley2007}, which demonstrated collisions
329: that created detonation conditions, along with several supplementary cases near the
330: minimum offset distance that led to detonation conditions.  This gives the range $r_{\rm
331: off}=$20 to 100 km.  The expansion which occurs during the
332: deflagration and surface flow stages is very nearly homologous and has only a
333: small degree of asymmetry, such that most of the asymmetry is created later
334: during the detonation phase (see below).
335: Figure \ref{fig:homologous-deflagration} compares the scaled density profile
336: for the initial model and the 25 km and 100 km offset cases at the time the
337: detonation initiates.  Density is scaled by the central value and radius is
338: scaled by the distance from the center at which the density drops to $1/e$
339: times the central value.  Profiles along the equator and along the symmetry
340: axis are both shown, demonstrating that the star remains very symmetric out
341: to approximately 2 density scale heights away from the center.  This region
342: contains approximately 90\% of the stellar mass, and even the asymmetry
343: beyond this is fairly modest, but is likely to lead to some asymmetry in the
344: highest-velocity spectral features.
345: 
346: Based on the variation in expansion found in the resolution study performed
347: by \citet{townsley2007}, and other cases generally, it appears that the
348: conditions at the collision region and the density structure at detonation
349: are not a single parameter family in the mass burned, or equivalently nuclear
350: energy release, in the deflagration, $E_{\rm n,def}$, prior to the collision.
351: It does seem that $E_{\rm n, def}$ is the primary parameter, but other
352: contributing factors include the time dependence of the energy release,
353: morphology of the burning region in the flame plume, and the possibility of
354: secondary or tertiary ignition sites.  The sudden input of energy in the
355: deflagration puts the star in an oscillation, and the timing of the
356: detonation initiation with respect to this oscillation is important for
357: setting the density structure at detonation, and thereby the burning
358: products (see \S5).  The timing and magnitude of the
359: nuclear energy release will change the magnitude and phase of this full-star
360: oscillation, but further investigation, including studies of 3-dimensional
361: deflagration morphologies, is needed to characterize these relationships.
362: 
363: As an aside we note the impact of neutronization due to electron captures
364: during the deflagration.  Neutronization influences the dynamics of a rising
365: flame bubble by changing the average binding energy of the final NSE state
366: obtained as well as the electron pressure available per gram of stellar plasma. 
367: We assessed the impact of neutronization by recalculating several models through 
368: bubble rise with weak rates suppressed, by enforcing $\dot Y_e = 0$. Flame bubbles 
369: ignited closer to the stellar center, and hence at higher densities,
370: are more strongly affected. Models ignited at $r_{\rm off}$=40 km and 30 km 
371: {\em burned $\sim$10\% and $\sim$40\% more mass, respectively, with weak rates 
372: suppressed} while the model ignited at $r_{\rm off}=$80 km 
373: was negligibly affected by the weak reactions during bubble rise and breakout.
374: The effect that the weak reactions have on the burned mass depends on the
375: developement of turbulence which is not well represented in the 2D simulations
376: presented here.  Therefore, while we have demonstrated that weak reactions play
377: a non-negligible role in the present suite of models the impact that they have
378: on more realistic 3D flows remains an open question.
379: 
380: %(Based on a comparison with Dean's earlier runs to my more recent ones
381: % w/ and w/out weak rates it appears that the sensitivity of the calculations
382: % to the development of turbulence is very strong and a simple explanation of the
383: % differences remains lacking.)
384: %
385: % It appears 
386: % that the change in the binding energy of the final state dominates such that flames with 
387: % weak reactions suppressed release less energy in burning to NSE thereby producing less 
388: % buoyant ash which rises more slowly, burns to a larger lateral extent, and produces 
389: % more total burned mass prior to breakout.
390: %
391: % \par (Dean to Casey:  This is not consistent with the studies I did on this, in
392: % which the bubble rise was directly observed to be faster with neutronization
393: % turned off.  I didn't check the burned mass dependence.  It may be that the
394: % faster rise generates more turbulence...)
395: 
396: 
397: \section{DETONATION}
398: \label{sec:det}
399: 
400: %\par Even slightly off-center flame ignition in the core of white dwarf appears
401: %to strongly diminish the chances that a deflagration by itself can unbind a
402: %white dwarf and lead to an explosion comparable to those observed as Ia SNe. 
403: 
404: %\par While the conditions of flame ignition, including the spatial distributions and formation
405: %rate, remains unknown an increasing  number of studies have begun to explore the mapping 
406: %between proposed distributions and final outcomes, incorporating 
407: %ever more realistic physics and relying increasingly on 3D full star simulations
408: %\citep{gamezo2005,jordan2007,roepke2007a}.
409: %Although the correct treatment of a turbulent flame remains an active area of research,
410: %recent simulations which use state of the art turbulent flame models 
411: %cannot reproduce the observed features of normal luminosity Ia SNe when burning takes 
412: %place solely in a deflagration \citep{roepke2007b}. These results have renewed interest 
413: %in scenarios which undergo a transition to detonation after a phase of subsonic burning.
414: 
415: \par Single point off-center ignition results in a buoyant plume of hot ash which is brought 
416: to the surface of the star before more than a few percent of the stellar core is consumed 
417: by the flame (see M$_{\rm def}$ in Table~\ref{tab:models}). As the hot ash rises to the surface, 
418: the nuclear energy that is released excites a stellar pulsation which initially expands the star.  
419: Against this background pulsation, 
420: the hot ash from the burning is expelled from the stellar interior. A large fraction of this ash 
421: is confined to the star's surface by gravity.  This ash sweeps over the surface of the
422: star together with a flow of unburned stellar material which is pushed ahead of it.
423: In all but the most expanded model in our parameter study (i.e., those with ignition points
424: r$_{\rm off} \ge$ 25 km), the resulting surface flows converge at a point opposite to breakout 
425: which we refer to as the {\em collision region}  (Fig.~\ref{fig:breakout}). 
426: These converging surface flows result in a bi-directional, collimated jet-like 
427: flow which both expels material away from the star's surface and drives a flow of high 
428: temperature material into the stellar core.
429: The inward directed component of the collimated flow 
430: reaches high enough densities and temperatures that a ``surface detonation'' inititiates which 
431: sweeps over the core and completely disrupts the white dwarf, giving rise 
432: to a luminous supernova explosion. 
433: 
434: \par In the following subsections we describe the
435: characteristics of the bi-directional jet which forms in the colliding surface 
436: flow and initiates the detonation (\S\ref{subsec:jet}), 
437: we discuss the characteristics of the ensuing detonation phase of burning (\S\ref{subsec:det}), 
438: and we describe the final state of free expansion which results (\S\ref{subsec:remnant}).
439: A detailed analysis of the nucleosynthetic yields is presented in the next section, \S5.
440: 
441: %Only the model with an initial flame bubble 
442: %ignited 20 km off center fails to detonate within the time simulated. In this model the large 
443: %amplitude of the stellar pulsation produced by the energy release in the deflagration 
444: %stalls the surface flow and lowers the densities in the collision region to values 
445: %well below the canonical value of $\rho\sim$10$^7$ g cm$^{-3}$ necessary to trigger a 
446: %detonation \citep{niemeyer1997,seitenzahl2008b}.
447: 
448: 
449: 
450: 
451: \subsection{Jet Formation and Detonation Initiation}
452: \label{subsec:jet}
453: 
454: %% FORMATION AND INIITAL DEVELOPMENT OF THE JET
455: 
456: \par {\em Jet Formation and Characteristics.---} As the surface flow produced by the deflagration 
457: converges, material accumulates in a small region on the hemisphere opposite to the 
458: breakout location.  The material which initially piles up, consists of unburned
459: carbon and oxygen rich surface material which is pushed ahead of the ash as it flows around
460: the stellar surface.  As material accumulates in this region it is heated by compression until it reaches
461: temperatures sufficient to initiate carbon burning, which further heats the compressed material
462: and raises its pressure.  Shortly after the initial collision, a conical shock forms which separates the 
463: compressed material along the axis from the inflowing surface flow. The surface flow material
464: burns as it passes through this shock and ``accretes'' into the collision region.  
465: The resultant pressure in the collision region roughly
466: balances the ram pressure of the accreting surface flow, $p_{\rm coll} \sim [\rho v^2]_{\rm surf}$ (see Fig.~\ref{fig:jet-slice}).
467: The pressure achieved in the compressed region soon exceeds the (nearly hydrostatic) background pressure 
468: sufficiently that it redirects the accreting material and drives a bidirectional jet-like 
469: flow which has components aligned along the polar axis.  A closeup of the collision region 
470: thus formed is shown in Figure~\ref{fig:jet} (which corresponds to the region outlined
471: by the dashed box in Figure~\ref{fig:breakout}). The velocity vectors reveal the 
472: bidirectional nature of the flow. The ash from the deflagration is just approaching 
473: the collision region at the time shown, well after the collimated jet has formed.
474: The width of the jet increases with time as material continues to accrete into
475: the region, but retains structure on scales $<$50 km, which are well resolved in our
476: simulations which have a grid resolution of 4 km.
477: 
478: 
479: %% JET PROFILES
480: 
481: \par In Figure~\ref{fig:jet-slice} the flow properties along the jet axis are shown just prior to 
482: the onset of detonation for two 2D models and one 3D model.
483: The 2D models shown span the conditions studied in this paper, including the model with the most
484: expanded core (left-panel) and the least expanded core (middle-panel) which
485: detonate in our study.
486: All of the collimated flows share the same overall structure with
487: the more expanded stars having shallower density gradients in the collision region.
488: The velocity profiles are roughly linear, decreasing from a maximum inwardly
489: directed velocity of $\sim$ (1 to 2)$\times 10^9$ cm/s to a comparable
490: velocity directed away from the stellar surface.
491: While the inward flow is attended by a great deal of small scale internal 
492: substructure and turbulence (Fig.~\ref{fig:jet}), three distinct ``fronts'' are readily identifiable 
493: along the axis of the jet: a leading subsonic compression wave, followed by a fuel-ash boundary layer,
494: and finally an internal shock.  The fuel-ash boundary layer is marked in Figure~\ref{fig:jet-slice}
495: by the dashed vertical line. The material ahead of this line has not yet been compressed
496: to high enough densities that carbon burning can proceed.
497: The compression wave(s) which eminates from the head of the jet as it moves
498: into the star can be seen as perturbations preceeding the fuel-ash boundary in all of the variables 
499: plotted in Figure~\ref{fig:jet-slice} and can also be seen as the pressure waves extending into
500: the star ahead of the jet in Figure~\ref{fig:jet} (left-panel).  The compression 
501: wave moves into the star at the sound speed, which is 
502: $c_s\sim 3.5\times 10^8$ cm/s at this location. The head of the jet, as marked by 
503: the location of the fuel-ash boundary, moves inward at a fraction of the sound speed so that 
504: the size of the compressed region grows with time. 
505: Trailing behind the fuel-ash boundary is a shock wave which
506: separates the low density, high velocity flow produced in the collision region from the compression
507: wave which moves ahead of it. It is the ram pressure of this high velocity flow which drives the 
508: compression wave into the star.  The ram pressure of this high velocity flow is balanced by the 
509: gas pressure of the compressed, overlying material, as shown by the red line in the bottom panels of 
510: Figure~\ref{fig:jet-slice}.  In all of the models studied, the inward directed jet continues to
511: compress material, heating it to carbon burning conditions until a detonation arises
512: and distrupt the star.
513: 
514: \par An important question in the context of the present study concerns the extent to which the jet-like 
515: flows which develop depend upon the 2D geometry used.
516: Simulations of off-center ignition using 3D grids have been made \citep[e.g.][]{roepke2007a,jordan2007} 
517: with the general conclusions that focusing of surface flow also occurs in 3D and is not strongly 
518: diminished compared to 2D. As a point of direct comparison, we have simulated a 3D model from flame
519: ignition through detonation using the same methods as used in the 2D models presented here.  
520: This 3D model used a finest resolution of $\Delta$=8 km and was ignited
521: by a 16 km flame bubble displaced 80 km from the stellar center.  The development of the collision 
522: region and the subsequent detonation in the 3D model is remarkably similar to that found in the 2D models.  
523: For comparison, the profile of the jet formed in the 3D model just prior to detonation is included in 
524: the right panel of Figure~\ref{fig:jet-slice}.  
525: 
526: 
527: 
528: %% SHAPED CHARGE JETS
529: 
530: \par Jet formation within a converging flow and jet penetration are well studied phenomena.  
531: For instance, engineers have designed shaped charge explosives which create jets by explosively 
532: collapsing a convex ``liner'' material, most often cone-shaped, onto itself \citep[e.g.][]{birkhoff1948}.\footnote[2]{Mining engineers have employed similar methods as early as 1792.}
533: The jets thus formed have notoriously strong penetrative power and can
534: slice and penetrate sheets of steel which are several times thicker than
535: the shaped charge diameter and are applied in both military and industrial 
536: capacities including metal perforation, armor penetration, and oil well drilling.
537: While there are many differences between shaped charge jet formation and 
538: the collision region flows present in our calculations, the phenomena bear interesting 
539: similarities which may provide insight into the depth to which a converging surface flow may
540: penetrate the underlying carbon-oxygen rich layers of a white dwarf.  
541: The jet models which have been 
542: made to interpret experimental results estimate penetration depth by balancing 
543: the ram pressure of the jet material with that of the target material in a frame of reference 
544: that is moving with the jet-target interface. The penetration depth under the simplifying 
545: circumstances of constant density jet and target materials depend on only the density ratio and
546: the jet length.  This picture is greatly complicated in the stellar surface flow
547: case where compressibility plays a central role and the pressure balance at the jet-star interface 
548: is between the dynamical pressure of the jet and the gas pressure of the core. 
549: While it is
550: beyond the scope of the present paper to fully analyze the problem of compressible jet formation and 
551: penetration, we conclude by noting that the strong penetrative power observed in shaped charge jets
552: provides support for the deep penetration
553: seen in all of the simulations in our study which develop collision regions.  In all of the cases
554: which we study, the jets which have formed penetrate into denser layers of the white dwarf until a 
555: detonation occurs.
556: 
557: 
558: 
559: 
560: %% THE INITIATION OF THE DETONATION
561: 
562: \par {\em Detonation Initiation.---} Once the density of the material undergoing carbon burning
563: in the jet exceeds $\rho_{\rm det}\sim 10^7$ g/cm$^3$ a detonation initiates 
564: which then propagates away from the head of the jet at the Chapman-Jouguet speed,
565: $D_{CJ} \sim 1.2\times 10^9$ cm/s with Mach number $M = D_{CJ}/c_s \sim 3.4$.
566: The time sequence shown in Figure~\ref{fig:jet} captures the moment when the 
567: detonation initiates at the head of the jet and begins to spread outward. Because
568: of the weak dependence of the detonation wave speed on the upstream density, the detonation 
569: front radiates from its point of initiation nearly spherically. 
570: 
571: \par The initiation of the detonation, which takes place at the fuel-ash boundary, when 
572: $\rho\sim 10^7$ g/cm$^{3}$ and T$\sim 3\times10^9$ K., resembles 
573: a Zel'dovich gradient mechanism \citep{zeldovich1970,khokhlov1997}.  Detonation initiation through this process
574: involves a complicated interplay between burning and hydrodynamic flow
575: that requires a coherent build up of acoustic energy by the nuclear energy release.  
576: An often cited criteria for the initiation 
577: of a detonation in the context of degenerate carbon-oxygen material is that a ``critical'' 
578: mass of material needs to be heated and compressed above a certain temperature and density 
579: threshold \citep{arnett1994,niemeyer1997,roepke2007a}.  While these studies indicate the general
580: conditions under which detonations might readily arise, thermodynamic
581: conditions and heated masses alone represent a gross oversimplification of the underlying initiation process which
582: depends sensitively on the gradients of thermodynamic variables within the heated region.  
583: Since gradients play a central role, the resolution and the geometry of the flows being simulated,
584: such as those presented here, are important considerations when investigating the potential 
585: for detonation. The suite of simulations studied in this paper use a finest zone size which is 4 km 
586: and limits the steepness of temperature gradients which can be represented in our models.
587: And although detonations do arise in our simulations, drawing conclusions from the results of
588: simulations alone concerning the success or failure of detonation will require investigations at 
589: significantly higher resolution than has been possible to date.
590: 
591: \par We have made some efforts to address the robustness of initiation with a suite of simulation 
592: models which employ a patch of mesh refinement over the collision region having zones as fine as 
593: 125 m. One of the principal findings of this study, which is being prepared for publication
594: elsewhere (C. Meakin et al. in prep), is that the gradients at the head of the inward directed
595: jet component become steeper at higher resolution which at first appears to inhibit detonation. 
596: However, the higher resolution flows develop turbulent structures within the shear layers that 
597: form at the interface between the head of the jet and the background stellar material, 
598: such as through the Kelvin-Helmholtz instability, which thicken the fuel-ash boundary to an extent
599: that induction time gradients conducive to the spontaneous initiation of a detonation may develop after all.
600: 
601: %Although the initiation process is unresolved in the calculations presented here, 
602: %this mechanism is being studied with a suite of models which employ a patch of mesh 
603: %refinement over the jet region having zones as fine as $\sim 0.1$ km, a scale on which 
604: %the gradient mechanism (critical radius) is resolved \citep{niemeyer1997,seitenzahl2008b}.
605: %One of the conclusions of this study, which is being prepared for publication
606: %\citep{seitenzahl2008b}, is that the initiation of the detonation is robust within the 
607: %jet and forms through a gradient mechanism at the head of the jet.
608: 
609: 
610: 
611: %
612: %CONDITIONS FOR FAILURE OF A COLLISION REGION TO FORM
613: %
614: %\par Only one model in our suite fails to detonate within the simulated time. In this model
615: %the flame was ignited in a 16 km radius bubble offset 20 km from the stellar center.  
616: %The energy released in the deflagration leads to a large expansion of the
617: %stellar core which stalls the surface flow before collision and jet formation can
618: %take place. The total energy released in this model is not enough to unbind the
619: %star, however, and it will eventually collapse again, and remains a candidate for
620: %a pulsational delayed detonation model \citep{khoklov1991,arnett1994}.
621: %The boundary between failure and success in detonation does not depend on only one 
622: %parameter, the burned mass, but also depends on the both the spatial and temporal release 
623: %of the energy. For instance multi-point ignition, mixing caused by a turbulent background, 
624: %and initial gradients in the composition of the core might all lead to different
625: %degrees of expansion and surface flow strengths for the same amount of net burned mass
626: %in the deflagration phase.
627: 
628: %The high degree of nonlinearity and feedback inherent in this problem
629: %requires large scale, 3D simulations to study the mapping between
630: %ignition conditions and possible surface detonations in the context of colliding
631: %surface flows. In addition to the uncertainties inherent in turbulent flame models,
632: %viscosity dependent processes such as vortext shedding and advective instabilities 
633: %during the early phases of flame bubble evolution, which take place at Reynolds 
634: %numbers that are presently impossible to simulate, remain unclear.
635: 
636: 
637: 
638: \subsection{Propagation of the Detonation Wave over the Stellar Core}
639: \label{subsec:det}
640: 
641: \par Once the detonation wave forms it propogates outward from the spot 
642: of initiation nearly spherically, and consumes the unburned carbon and oxygen remaining 
643: in the core. The time sequence in Figure~\ref{fig:det-temp} shows the geometry of the detonation 
644: wave as it propagates over the stellar core. 
645: The detonation wave speed is a weak function of the upstream plasma density and varies by 
646: only $\pm$5\% for the conditions present in the uburned core, where $10^7<\rho<10^9$ g cm$^{-3}$ 
647: \citep[see Figure 2 of][]{gamezo1999}. The detonation wave traverses the expanded white dwarf 
648: in $t_{\rm cross}\sim 2 r_{\rm det}/{D_{CJ}}\sim 0.4$ s where the core size is roughly 
649: $r_{\rm det}\sim 2\times10^8$ cm and the detonation wave speed is $D_{CJ}\sim10^9$ cm s$^{-1}$.
650: 
651: 
652: \par As the detonation wave propagates it compresses upstream material prior
653: to burning.  Upstream material with a density greater than $\sim 10^7$ g cm$^{-3}$
654: is compressed and heated strongly enough by the shock that complete relaxation to nuclear 
655: statistical equilibrium (NSE) occurs before the rarefaction wave behind the detonation expands
656: the material and it freezes-out (see \S5). At lower upstream densities
657: relaxation to NSE is incomplete and the ash is composed of intermediate
658: mass elements (IMEs) such as Si, S, Ca, and Ar, i.e., the products of incomplete
659: silicon burning \citep[e.g.][]{woosley1973,arnett1996}.
660: 
661: \par Material which is compressed to densities exceeding $\sim10^8$ g cm$^{-3}$ in
662: the detonation wave develops a non-negligible neutron excess through electron capture 
663: reactions. The strong density dependence of the weak reaction rates limit this
664: neutronization to the central-most regions of the star as evident in 
665: Figure~\ref{fig:det-ye} which shows the spatial distribution of electron mole
666: fraction $Y_e$ as the detonation wave sweeps over the stellar core.  As discussed in \S5, the
667: final composition of the material burned to NSE, including the fraction which is \nuc{56}{Ni}, 
668: depends on the degree of  neutronization.
669: 
670: 
671: \par Detonation waves are subject to transverse instabilities which influence
672: the structure of the reaction zone and the reaction products and introduce
673: inhomogeneities in the downstream flow \citep[e.g.,][]{gamezo1999,timmes2000b,sharpe2001}.
674: Therefore, in order to faithfully capture in entirety the properties of the burning in a detonation 
675: wave the reaction length scale must be resolved.  An additional complication arises
676: in modeling detonations when the density scale height in the medium through which the detonation
677: propagates is comparable to or smaller than the reaction length.  Under these
678: conditions steady detonation wave theory cannot be applied and the resulting reactive-
679: hydrodynamic flow remains an active field of research \citep{sharpe2001}.  
680: In the context of a carbon-oxygen, near Chandrasekhar-mass white
681: dwarf (M$_{\rm Ch}$), such conditions arise when the upstream density is $\sim10^7$ g cm$^{-3}$.
682: Significant deviations from a Chapman-Jouguet detonation
683: may arise and influence the resulting intermediate mass element (IME) yield.  
684: Since IMEs, such as Si and Ca, are 
685: primary observational diagnostics of the explosion mechanism underlying SNe Ia 
686: \citep[e.g.][]{wang2003,wang2007}, these uncertainties have important implications
687: for modeling all delayed detonation scenarios.
688: 
689: \par In the models presented here, the stellar cores undergo only modest expansion
690: during the deflagration and detonation phases.  Between 90\% and 97\% of the unburned mass in the core 
691: has a density which exceeds $\sim10^7$ g cm$^{-3}$ at the time detonation initiates and 
692: all of this material undergoes complete relaxation to NSE, 
693: resulting in primarily \nuc{56}{Ni} and and a small fraction of stable Fe-peak elements 
694: (\S5 and Table~\ref{tab:models}).  Therefore, only a small amount of mass, which is confined to a thin 
695: shell in the outer part of the core, is burned to IMEs by the detonation wave.
696: Within this narrow shell the length scales associated with transverse
697: instabilities exceed the grid scale used ($\Delta$ = 4 km) \citep{gamezo1999} and our
698: numerical methods are sufficient to capture them.
699: However, material in this narrow region undergoes rapid expansion 
700: after the detonation wave passes and it quickly mixes with the turbulent layer of deflagration 
701: ash which lies immediately above it so that it is difficult to discern the presence of cellular 
702: structure if it did indeed arise.  Significantly higher fidelity simulations are required
703: in order to study the impact that transverse instabilities have under these conditions.
704: While these affects are negligible in the present suite of models, more expanded, lower 
705: density cores are likely to be much more strongly impacted by this uncertain physics.
706: 
707: 
708: \par Upon encountering the deflagration ash which enshrouds the star, 
709: the detonation wave transitions into a shock wave which accelerates
710: the hot ash.  After the detonation wave and the ensuing shock have propagated
711: off of the computational grid, what is left behind is a rapidly expanding
712: remnant consisting of a smoothly layered core of material burned to NSE with 
713: a thin shell of IMEs outside of that, surrounded by a turbulent
714: layer of ash from the deflagration composed of both NSE and
715: IME material.
716: 
717: 
718: 
719: 
720: 
721: \subsection{Transition to Free Expansion and Final Remnant Shape}
722: \label{subsec:remnant}
723: 
724: \par As the detonation wave traverses the stellar core it shifts the density
725: distribution so that the peak in density is initially moved in the positive 
726: $y$-direction. This can be seen in Figure~\ref{fig:det-propagation} which
727: presents a time series of velocity and density profiles spanning the time interval
728: over which the detonation wave traverses the stellar core.  The initial shift in the
729: density peak towards positive $y$ is due to the strong rarefaction which follows the detonation wave
730: and expands the material behind it on a very short timescale ($\tau_{\rm expand}\sim 0.4$ s).
731: Within $\sim$1 s following the passage of the detonation over the core, the density
732: peak moves back in the negative $y$-direction and ends up south of the equator (negative $y$).
733: The binding energy released in the detonation wave is converted into the kinetic energy 
734: associated with expansion within $\sim$1 s after the detonation wave completes its passage over 
735: the star.  The total energy budget is shown in Figure~\ref{fig:energy} for the model ignited
736: 25 km off center.  By $t\sim4$ s the remnant is transitioning into a state of free expansion 
737: and assumes a self-similar shape which is no longer changing with time and the radial velocity
738: is well described by a linear dependence on the distance from the center of the remnant.
739: Axial and equatorial profiles of density and radial velocity in the remnant are presented in 
740: Figures~\ref{fig:det-profiles} for two models which span the explosion outcomes in our study.
741: 
742: 
743: \par Shown in Figure~\ref{fig:dens-contours} is the late time ($t>4$ s) remant shape presented as 
744: a series of logarithmically spaced density contours for the same two models in Figure~\ref{fig:det-profiles}.  
745: What can be seen in this figure is that the asymmetry imparted by the off-center ignition and surface detonation
746: manifests as a shift in the center of density contours even though each individual contour
747: is well described by a circle. The overall shape of the density distribution, therefore, can 
748: be characterized by the radius $R_c$ and center $y_c$ of the circles which best describe each 
749: density contour. This information is shown in Figure~\ref{fig:shape}.  The degree to which these 
750: two curves ($R_c$ and $y_c$) approximate the remnant is shown by the thin line in 
751: Figure~\ref{fig:det-profiles}, which is the function $r(\rho) = y_c(\rho)\pm R_c(\rho)$.  
752: 
753: \par Superimposed over the relatively smooth overall shape of the final remnant are smaller scale density 
754: inhomogeneities due to the turbulent flow associated with the deflagration and the surface flow which 
755: preceeded detonation. These perturbations are quantified in the bottom panel of Figure~\ref{fig:shape} 
756: as root mean square (rms) deviations in density taken along the best fit circle at each density.   
757: The general trend is that more expanded stellar cores have larger
758: density perturbations in their surface layers at the time
759: of detonation.  This can be accounted for partly by the fact that more expansion results from
760: a larger amount of energy liberated in the deflagration which goes into powering the surface flow.  
761: Additionally, more expanded cores are less stably stratified at their surfaces (lower gravity and 
762: shallower pressure gradients) and so are more easily perturbed by the surface flow which passes over 
763: the star before detonation.
764: 
765: 
766: 
767: 
768: \section{NUCLEOSYNTHESIS}
769: \label{sec:yields}
770: 
771: 
772: \par The nucleosynthetic yield for the type of explosion model studied in this paper
773: consist of a mixture of ash due to both a deflagration and a detonation.  
774: The total amount of the star burned in the deflagration amounts to less than $\sim$0.1 \msun
775: with nearly the entire remaining mass of the star consumed by the detonation wave which follows.
776: The progress variables described in \S\ref{sec:numerics} which are used to parameterize
777: the compositional evolution and energy release due to nuclear burning allow us to calculate the 
778: bulk yield of IMEs and NSE material directly from the multi-dimensional
779: simulation data by performing simple sums \citep{calder2007,townsley2007}. In Table~\ref{tab:models} 
780: the total mass burned in the deflagration M$_{\rm def}$ and the detonation M$_{\rm det}$ is summarized,
781: including the budget of IME and NSE material.
782: The deflagration, propagated with the ADR flame model (\S\ref{sec:numerics}), produces a total yield which 
783: is approximately one third IMEs and two thirds NSE material, while more than 90\% of the material burned 
784: in the detonation is completely relaxed to NSE.  
785: 
786: \par A general feature of these explosion models is that higher density cores at the time of detonation 
787: produce a larger yield of NSE material and a smaller yield of IMEs.  This trend is summarized in 
788: Figure~\ref{fig:mnse-rhoc} which relates the final NSE yield to the central density of the white dwarf
789: at the time of detonation.  The total mass of material having a density exceeding $\rho=10^7$ g/cm$^3$
790: is also shown as a function of central density, and provides a good measure of the mass of material that will 
791: burn to NSE in the detonation. The relationship between the amount of mass above a certain density and 
792: the central density is a property of the initial white dwarf density structure 
793: and the wave form of the pulsation which is excited in the star 
794: by the deflagration. The dashed line in Figure~\ref{fig:mnse-rhoc} shows the relationship expected if the 
795: pulsation is described by the fundamental mode of the linear wave equation.  This mode fits the simulation 
796: data remarkably well considering how large (and non-linear) the pulsation amplitude is for the low central density end of the
797: figure.  But this can be understood by the fact that the fundamental mode is close to homologous, i.e., the 
798: displacement is nearly directly proportional to the radius of the white dwarf.
799: 
800: 
801: \par The data for the 3D model described in \S\ref{subsec:jet} has been included in 
802: Figure~\ref{fig:mnse-rhoc} for comparison, and shows that the overall character of the 
803: expansion driven by the deflagration is not dimensionality dependent, nor is the nucleosynthesis 
804: that takes place in the detonation.  Data from an additional 3D simulation which burned
805: significantly more mass in the deflagration is shown and labeled ``3D Multi''. This simulation 
806: data, which is part of an extended suite of 3D models investigating the mapping of ignition conditions to 
807: final explosion energies and nucleosynthetic yields (G. C. Jordan IV et al., in prep), was ignited
808: by a uniform distribution of 30 flame bubbles enclosed in an 80 km radius sphere having its center displaced 100 km
809: from the stellar center.  
810: This distribution of ignition points is intended to be more representative
811:  than single-point ignition of the non-axisymmetric conditions created by
812:  the interaction of the growing bubble with the pre-ignition convective
813:  core.
814: This model demonstrates that fundamental mode radial pulsation is a good description of
815: the core dynamics preceeding detonation even for significant degrees of expansion.
816: This model also demonstrates that off-center deflagration models are capable of producing a broad range of 
817: NSE and \iso{Ni}{56} masses, and not just the most luminous SNe Ia events.
818: 
819: 
820: \subsection{Iron Peak Freeze-Out Yields: Method}
821: \label{subsec:freezeout-method}
822: 
823: \par As NSE material expands and cools in the rarefaction that follows the detonation 
824: wave, nuclear reactions eventually cease, the composition no longer changes and the 
825: material is said to have gone through {\em freeze-out}. 
826: In this section we describe a methodology to efficiently and accurately calculate 
827: iron peak yields for material which burns to NSE and then freezes out in the expansion following the 
828: detonation wave.  In our hydrodynamic simulations $\sim10^5$ to $\sim10^6$ Lagrangian tracer 
829: particles are passively advected through the computational domain with an initial distribution that
830: evenly samples the underlying mass distribution. Nucleosynthetic yields can then be calcualted by 
831: integrating nuclear reaction networks along each of these particle trajectories and then summing the 
832: yields. However, when the large number of particle trajectories required for accurate yield estimates is
833: multiplied by the number of simulation models desired for study, the computational cost of this
834: brute force method of post-processing becomes prohibitively expensive. 
835: Therefore, we have developed an alternative approach to calculate the final composition of
836: material processed by the detonation, which takes advantage of the fact that the final nucleosynthetic 
837: yield $X_{i,f}$ of material burned to NSE depends only on the final entropy 
838: $S_f$, expansion timescale $\tau$, and degree of neutronization $\eta_f$ of the detonated material to a high 
839: degree of precision with $X_{i,f} = X_{i,f}(S_f,\eta_f,\tau)$.
840: 
841: 
842: 
843: % I. INDIVIDUAL TRAJECTORY
844: \subsubsection{Individual Trajectories}
845: 
846: \par The temperature and density of a generic tracer particle processed by the detonation
847: is presented in Figure~\ref{fig:trajectory}. The evolution of $Y_e$ and the abundances of 
848: nuclei along this trajectory have been calculated using a nuclear reaction network 
849: initialized with the initial composition of the white dwarf material in the simulation,
850: equal mass fractions of \nuc{12}{C} and \nuc{16}{O}. 
851: The network code used for the  integrations is a version of the network used in 
852: \citet{calder2007} expanded to 443 nuclear species (see Table~\ref{tab:network}). The 
853: thermonuclear reaction rates are taken from an expanded version 
854: ~\citetext{Schatz 2005, private communication} of the rate compilation 
855: REACLIB~\citep{thielemann1986,rauscher2000}.  We have also included the 
856: temperature-dependent nuclear partition functions provided by~\citet{rauscher2000}, 
857: both in the determination of the rates of inverse reactions and in our determination 
858: of NSE abundance patterns. Electron screening of thermonuclear 
859: reaction rates is incorporated, adopting the relations for weak screening and strong 
860: screening provided by~\citet{wallace1982} \citep[for additional details see the appendix of][]{calder2007}. 
861: Contributions from weak reactions are included using the rates provided by \citet{langanke2001}.
862: 
863: \par The time evolution during the rarefaction stage of several abundant
864: species along this trajectory, parameterized by plasma temperature, is shown in
865: Figure~\ref{fig:trajectory-burn} (solid lines).  
866: The NSE composition corresponding to the same density and neutron excess for each temperature along the rarefaction part of the trajectory is also shown for comparison (dashed lines). 
867: The NSE mass fractions were determined with the NSE solver described in \citet{calder2007} and \cite{seitenzahl2008a}, which uses the same nuclear physics as the reaction network code. 
868: Figure~\ref{fig:trajectory-burn} illustrates the degree to which adopting an NSE state at a particular ``freeze out temperature'' 
869: is a poor approximation to the final freeze-out abundances. This is true because nuclei freeze-out 
870: over a fairly large range in temperature and there is non-trivial evolution in a nuclide's abundance 
871: after it falls out of NSE but before it reaches its asymptotic freeze-out value.
872: 
873: \par The thermodynamic trajectories for Lagragian elements (i.e., tracer particles) processed 
874: by the detonation wave are well characterized by an exponential temperature evolution
875: 
876: \begin{equation}
877:   \label{eqn:temp}
878:   T(t) = T_0 \exp(-t/\tau)
879: \end{equation}
880: 
881: \noindent and a corresponding density evolution found by assuming adiabaticity
882: 
883: \begin{equation}
884:   \label{eqn:entropy}
885:   S(t) = S(T,\rho, \bar{A}, \bar{Z}) = S_f = (\hbox{constant})
886: \end{equation}
887: 
888: \noindent where $S_f$ is the final entropy in the post-detonation state and $\bar{A}$ and $\bar{Z}$ are 
889: the average atomic weight and charge of the plasma during the burn with $Y_e = \bar{Z}/\bar{A}$. 
890: The density in equation \ref{eqn:entropy} is found using the same Helmholtz equation of state 
891: \citep{timmes1999,timmes2000a,fryxell2000} used in the hydrodynamic 
892: simulations. A parameterized trajectory, is shown in Figure~\ref{fig:trajectory}
893: for comparison to the particle trajectory.  
894: 
895: \par The abundance evolution for this parameterized trajectory is presented in 
896: Figure~\ref{fig:trajectory-burn} (dotted line) and shows agreement to a high degree 
897: of precision with the tracer particle trajectory.
898: Because the composition is well described by a NSE distribution at temperatures above $T_9 \sim 5.5$
899: the final yields are not dependent on the peak temperature reached by the particle trajectory
900: and depend only on the entropy, the total amount of neutronization, and the 
901: expansion timescale. Final asympotic freeze-out yields are safely adopted when the plasma temperature 
902: drops below $T\sim 10^9$ K with no evolution in the abundances taking place at lower temperatures.  
903: Because the electron capture rates are a strong function of density, neutronization occurs in the short lived
904: high density region formed immediately behind the detonation front, while the material is still in NSE and 
905: well before freeze-out begins.
906: 
907: %\par The stability of the network integrations require that the abundances are initialized with 
908: %an appropriate NSE distribution, and this is accomplished using the NSE solver described in 
909: %\citet{calder2007} and \cite{seitenzahl2008a} which uses the same nuclear physics as the 
910: %reaction network code.
911: %\par Electron capture rates are a strongly increasing function of density so that most 
912: %neutronization in the plasma occurs in a narrow region of high density trailing the detonation 
913: %front while the material is still in NSE, well before freeze out.  This allows us to use a 
914: %constant electron fraction for our network integrations chosen as the final value in the 
915: %freeze out.
916: 
917: % II. ENTROPY-NEUTRONIZATION CORRELATION AND LOOKUP TABLE
918: 
919: \subsubsection{Systematic Properties of the Post-Detonation State and Generating a Lookup Table}
920: \par A tight correlation between the degree of neutronization $\eta_f$ and the final entropy 
921:  $S_f$ of the detonated material further simplifies the procedure and allows us to calculate 
922: the final composition for each grid zone in our simulations using a one parameter
923: freeze-out abundance lookup table $X_{i,f} = X_{i,f}(\eta_f)$.
924: The asymptotic values of the degree of neutronization $\eta_f = (1-2 Y_{e,f})$
925: and the entropy $S_f$ for all of the stellar matter burned into NSE in the detonation wave
926: is found to lie along a narrow ridge in the $S_f$-$\eta_f$ plane as shown
927: in Figure \ref{fig:ye-s-table}.
928: This correlation arises from the monotonic dependence of the entropy deposition
929: and the neutronization rate $\dot{Y_e}(\rho)$ on the post-shock plasma density 
930: in the detonation wave (which is itself a monotonic function of the pre-shocked, 
931: upstream density). (Note, this tight correlation doesn't exist for the material which is
932: burned in the deflagration so that the method described here is not applied to that phase of burning.)
933: 
934: A lookup table is constructed along the black line shown in 
935: Figure \ref{fig:ye-s-table} and has been sampled at 50 locations logarithmically spaced in neutron excess $\eta$.
936: For every value of $\eta$ in the table, the freeze-out abundances are calculated by integrating the nuclear reaction network, 
937: with weak interactions turned off, over an adiabatic analytic trajectories as described above for the corresponding value of 
938: $S_f$.  The network is initialized with NSE composition at $T_9 = 6.0$ and the integration constinued until $T_9<1.0$, at which 
939: time freeze-out has occured.  A summary of the stable iron peak yields for material along this locus of points is presented in 
940: Figure~\ref{fig:yields-table}.  The freeze-out abundances for a computational zone are then be found by using the table entry with 
941: the corresponding value of $\eta$. The table lookup is computationally fast, and once the table is created no additional network 
942: calculations are necessary.
943: 
944: As described above, the final freezeout yield depends on the expansion timescale $\tau$.
945: The expansion timescale, defined by eq.[\ref{eqn:temp}] and found by fitting an exponential to the temperature histories
946: of the tracer particles, varies smoothly across the face of the white 
947: dwarf in the narrow range $0.2 < \tau < 0.6$ s for all of the models simulated 
948: (the range is narrower for an individual explosion model). If one wanted  
949: to incorporate this information into the processing of the final yields, the range in expansion timescales would need 
950: to be reflected in the network calculations. 
951: In the work presented here, we adopt a central value of $\tau = 0.4$ s and 
952: we discuss the sensitivity of the final yields to variations in this value in \S\ref{subsec:final-yields} below.
953: 
954: % III. PRESENTATION OF FINAL YIELDS
955: \subsection{Iron Peak Freeze-Out Yields: Results}
956: \label{subsec:final-yields}
957: 
958: \par We calculate the final yield of stable iron peak isotopes for all of our explosion models
959: using the procedure outlined in \S\ref{subsec:freezeout-method} above. In Figure~\ref{fig:yields}
960: we present yields for isotopes in the mass range A = 45 to 68 (\nuc{45}{Ti} to \nuc{68}{Zn}), 
961: accounting for the decay of radioactive isotopes.  
962: The yields ($X_i$) are scaled to the \iso{Fe}{56} abundance ($X_{Fe}$, from the decay chain
963: \nuc{56}{Ni}$\rightarrow$\nuc{56}{Co}$\rightarrow$\nuc{56}{Fe}) and the corresponding relative 
964: solar system ratio ($X_{i,\odot}/X_{Fe,\odot}$) based on the abundances of \citet{lodders2003}.
965: In Table~\ref{tab:scaled-yields} we present the elemental abundances for the iron peak elements 
966: from Ti to Zn scaled to Fe and solar system ratios. In all cases we highlight results for three 
967: explosion models which bracket the range of initial conditions and final outcomes found in our 
968: simulation suite. In addition, we provide a detailed list of the final integrated iron peak yields in units
969: of solar mass in Table~\ref{tab:yields}, including the abundances of radioactive isotopes 
970: and their half lives.  This table can be used to determine the absolute yield of a particular
971: isotope, or to examine the isotopic ratios of specific elements of interest.
972: 
973: \par The iron peak yields are similar to pure deflagration
974: models such as the one-dimensional model of Nomoto \citep[W7 yields in][]{brachwitz2000} and the
975: three dimensional model presented in \citet{travaglio2004}. 
976: The iron peak yield for the pure deflagration models are more neutron rich than our 
977: models, however, because of the higher densities under which the deflagration burns material
978: to NSE. The highest density core to detonate in our model suite (with $r_{\rm off}$=100 km)
979: neutronized the most in the detonation and bears the most similarity to a pure deflagration 
980: model in terms of integrated yields.
981: The most neutron-rich isotopes of each element (e.g. \iso{Ti}{50}, \iso{Cr}{54},
982: \iso{Fe}{58}) have no appreciable contribution from the NSE material created by the detonation.  There
983: is likely some of these species in the small amount of deflagration material not included in our
984: post-processing.  Notably, none of our models produce untoward overabundances ($\gtrsim 2$ times solar,
985: indicated by the dotted lines) of either \iso{Fe}{54} or
986: \iso{Ni}{58}. Overproduction of these nuclides continues to be a serious shortcoming of deflagration
987: models, both spherical and multi-dimensional, which process much of the stellar interior to NSE before
988: expansion can occur.
989: The spatial distribution of the material
990: in our detonation models, however, remains layered in space and velocity while the burning products
991: in the pure deflagration model are strongly mixed due to the turbulent nature under which burning proceeds
992: in a deflagration \citep[e.g.][]{roepke2007b,gamezo2003}. 
993: 
994: \par Interestingly, the total yield of \iso{Ni}{56} in all of the explosion models presented here is
995: $\sim$1.1\msun  independent of the degree of expansion which takes place prior to detonation.
996: This is due to a self regulating process comprised of pre-expansion and neutronization which
997: counteract each other. While the total yield of material burned to NSE is larger for stars
998: which detonate at higher central densities, more neutronization takes place which shifts 
999: the iron peak yield to more neutron rich isotopes and away from \iso{Ni}{56} \citep[see e.g.][]{timmes2003}.  
1000: The highest density core at detonation produces overall more stable iron peak isotopes 
1001: but approximately the same \iso{Ni}{56} yield as the core with the lowest density at detonation.
1002: It can be seen from Figure~\ref{fig:mnse-rhoc}, however, that this trend cannot hold for 
1003: significantly more expanded cores since the total mass of high density material drops off 
1004: precipitously as lower central densities are reached and will therefore result in SNe Ia 
1005: explosions which have smaller \iso{Ni}{56} yields.
1006: 
1007: \par {\em Dependence on progenitor neutronization. ---}
1008: The progenitor white dwarf model used in our explosion calculations 
1009: is composed of equal parts \nuc{12}{C} and \nuc{16}{O} so that $Y_{e} = 0.5$ everywhere in the 
1010: unburned fuel prior to detonation.
1011: However, the progenitor is expected to develop a neutron excess before flame ignition
1012: both during the CNO and He burning cycles and during a $\sim$1000 yr epoch of hydrostatic 
1013: carbon burning which is sometimes referred to as ``simmering''. 
1014: Recent studies of the ``simmering'' epoch indicate that when carbon burning runs away locally
1015: and a flame is born, $\eta^{\rm sim} \approx 10^{-3}$
1016: \citep{piro2008,chamulak2008}, while stars with an initial metallicity comparable to solar 
1017: will develop a neutron excess of $\eta_{\odot} \approx 1.5\times 10^{-3}$  by the time core He burning 
1018: commences. 
1019: 
1020: \par The neutronization which takes place during the detonation is restricted to the densest, 
1021: central-most regions of the stellar core because of the strong density dependence of the electron 
1022: capture rates (see Figure~\ref{fig:det-ye}). The resulting distribution of neutronization
1023: is presented in Figure~\ref{fig:ye-s-cdf}. This distribution 
1024: extends to values lower than the minimum  $\eta_{\rm min}^{\rm sim}$ expected in the progenitor 
1025: prior to explosive burning.  We explore the impact that such a neutronization 
1026: floor will have on the yields by enforcing $\eta = \max(\eta_{\rm min},\eta)$ prior to 
1027: calculating the iron peak yields using the lookup table method described in \S\ref{subsec:freezeout-method}.
1028: The results are presented in Figure~\ref{fig:yields} (right panel) which shows yields calculated after 
1029: applying neutronization floors of $\eta_{\rm min} = 0$, $10^{-3}$, and $2\times 10^{-3}$ for the model 
1030: which was least neutronized during the detonation (with $r_{\rm off} = $25 km),
1031: and therefore has the most mass affected by a floor in $\eta$.
1032: The iron peak elements  which are primarily produced at low $\eta$ and are therefore most 
1033: strongly affected by a neutronization floor are V, Cr, Mn, and Zn, although the isotopic 
1034: ratios across the entire iron peak are affected.
1035: 
1036: \par {\em Sensitivity to scatter in $S_f$ and $\tau$. ---} 
1037: It is possible to construct a higher dimensional lookup table for calculating yields 
1038: which accounts for the scatter about the $\eta$-$S_f$ curve used to generate the table 
1039: (Figure~\ref{fig:ye-s-table}), but the total error associated with neglecting this scatter 
1040: in final entropy $S_f$ is small. In Figure~\ref{fig:yields-texp-entropy} (right) we present the total 
1041: variation in the yields due to shifting the $\eta$-$S_f$ curve in final entropy by $\pm$5\%, which is the range of the scatter. 
1042: Similarly, the freezeout timescale that takes place in the wake of the detonation wave
1043: has some scatter about the fiducial value of $\tau=0.4$ s that has been used for the results
1044: presented above, spanning the range $0.2 <\tau < 0.6$ s. 
1045: The total spread in yields adopting the extreme values for $\tau$ is shown in Figure~\ref{fig:yields-texp-entropy} (left). 
1046: The variation in the yield will be significantly smaller than illustrated by these figures since there 
1047: exists a smooth distribution between the extreme values of $S_f$ and $\tau$ 
1048: with the majority of the mass peaked about the central value.  
1049: %While the effects of scatter in 
1050: %$S_f$ and $\tau$ have a small impact on the final yields, it is important to investigate 
1051: %the range of conditions present in any model before applying the freezeout technique described in 
1052: %\S\ref{subsec:freezeout-method} above.
1053: 
1054: 
1055: \subsubsection{The velocity distribution of the yield.}
1056: \par {\em The yield of NSE material. ---}
1057: Inferring the abundance stratification in SNe Ia is possible by studying
1058: high fidelity, multi-epoch spectra; a technique which is proving to be a powerful new tool for
1059: constraining explosions models \citep{stehle2005,roepke2007b,mazzali2008}.  The total
1060: number of objects which have had detailed internal abundance stratifications reconstructed
1061: to date is small (only 2 at the time of writing, including 2002bo and 2004eo). These two low 
1062: luminosity SNe have inferred \iso{Ni}{56} masses in the range M[\iso{Ni}{56}]$\sim$0.43 - 0.52 \msun.  
1063: Despite the very different \iso{Ni}{56} masses between these two observed SNe and the explosion models 
1064: presented in this paper, it is interesting to compare the qualitative and quantitative properties of 
1065: the abundance stratifications in an attempt to understand the nature of and the diversity inherent 
1066: in the explosion mechanism.
1067: 
1068: \par For the case of SN 2004eo, \citet{stehle2005} find M[\iso{Ni}{56}]$\sim$0.43 \msun. In their
1069: reconstruction, the \iso{Ni}{56} mass fraction drops below 0.5 at $v_{\rm exp}\sim$ 7,000 km/s,
1070: and drops below 0.1 at 12,000 km/s.
1071: For the case of SN 2002bo, \citet{mazzali2008} find M[\iso{Ni}{56}]$\sim$0.52 \msun. The \iso{Ni}{56}
1072: mass fraction drops below 0.5 at $v_{\rm exp}\sim$ 10,000 km/s and drops below 0.1 at 15,000 km/s.
1073: In both of these SNe, a high mass fraction of stable Fe (X$_{Fe}\sim 1$) is inferred at low velocities 
1074: $v_{\rm exp} < $ 3,000 km/s.
1075: 
1076: \par The distribution of the elemental abundances as a function of the 
1077: radial velocity for our explosion models is presented in Figure~\ref{fig:yields-velocity}.
1078: In this figure, the elemental abundances are calculated by summing over isotopes and 
1079: taking into account radioactive decays with half lives less than 1 day.
1080: Two models are shown which bracket the final outcome of all the explosions modeled in this paper. 
1081: Both models produce $\sim$1.1\msun  of \iso{Ni}{56}.
1082: The model ignited with $r_{\rm off}$=25 km is the most expanded at the time of detonation, 
1083: neutronizes the least amount in the detonation, and has the lowest explosion energy, 
1084: E$_{\rm tot} = 1.45\times 10^{51}$ erg.  
1085: The contribution of stable Fe is the smallest in this model, having a mass fraction
1086: X$_{Fe}\sim$10$^{-3}$ out to  $v_{\rm exp}\sim$ 2,000 km/s.  The mass fraction of
1087: \iso{Ni}{56} drops below 0.5 at $v_{\rm exp}\sim$ 14,000 km/s, and drops
1088: below 0.1 at $\sim$ 16,000 km/s.  The model ignited with $r_{\rm off}$=100 km is the least 
1089: expanded at the time of detonation,  is neutronized the most by the detonation wave, and has 
1090: the largest explosion energy, E$_{\rm tot}=1.52\times10^{51}$ erg.
1091: Although the ejecta at low velocities is still dominated by \iso{Ni}{56},
1092: stable Fe with a mass fraction exceeding X$_{Fe}\sim$0.1 is present out to $v_{\rm exp}\sim$6,000 km/s.
1093: The mass fraction of \iso{Ni}{56} drops below 0.5 at a velocity of $v_{\rm exp}\sim$ 16,000 km/s and
1094: drops below 0.1 at 18,500 km/s.
1095: 
1096: \par While the total yield of \iso{Ni}{56} is significantly larger in the explosion models presented
1097: here, the qualitative layered structure of the remnant and the near absence of unburned
1098: carbon and oxygen are in good agreement between the models and the observations.  As discussed 
1099: in \S\ref{sec:yields} and summarized in Figure~\ref{fig:mnse-rhoc}, lower \iso{Ni}{56}
1100: masses can be produced in surface detonation models which release more energy during the
1101: deflagration phase.  However, detonations which take place at lower central densities 
1102: and produce smaller \iso{Ni}{56} masses, undergo signifiantly less neutronization and will 
1103: therefore fail to reproduce the stable Fe core which has been inferred in the two models 
1104: discussed above.  On the other hand, it is possible that the neutron rich region seen
1105: at low velocities in SNe Ia remnants are a vestige of the progenitor conditions at ignition. 
1106: The nature of the progenitor at flame ignition, including the central density and $Y_e$ 
1107: distribution, are uncertain and will remain so until the evolution leading up to ignition
1108: is better understood, including the much debated and poorly understood Urca process 
1109: \citep[see e.g.,][]{lesaffre2005,arnett1996}.
1110: 
1111: \par {\em The yield of non-NSE material. ---}
1112: The detailed yield for material which has not completely relaxed to NSE prior to freezeout 
1113: and is composed primarily of IMEs such as Si, S, and Ca is not presented here. In addition to 
1114: producing IMEs, material which has begun 
1115: silicon burning but has not yet reached an NSE state will contribute to the iron peak with isotopic
1116: ratios that are very different from that which reaches an NSE state.
1117: While the impact of this burning process is small for the suite of explosion models presented in this 
1118: paper, which produce primarily NSE material, it is essential to accurately calculate the composition
1119: and distribution of this material for lower luminosity (lower \iso{Ni}{56} mass) SNe Ia explosion models
1120: which will have a significantly larger contribution of non-NSE material.  
1121: Additionaly, although the total contribution of IMEs to the mass of the remnant is small in all of the 
1122: explosions presented here this material plays a central role in modeling the observational 
1123: signatures of these explosion models
1124: and is therefore crucial for comparing our calculations to observational data.
1125: Therefore, a procedure for determining incomplete silicon burning yields which is similar to the 
1126: method described in \S\ref{subsec:freezeout-method} is being developed (C. Meakin et al., in preperation).
1127: 
1128: %\section{DISCUSSION}
1129: %\par (ala Don Lamb: some words connecting the results of the models presented to the
1130: %observational data. in particular, how the inferred morphology and asymmetries detected
1131: %in spectra and light curves, and the distribution of composition throughout the remantn
1132: %compare with the models presented here.)
1133: 
1134: 
1135: 
1136: \section{CONCLUSIONS}
1137: 
1138: \par We have studied the final outcomes for a range of single point flame ignition models of 
1139: thermonuclear supernovae within the computational framework developed at the FLASH center
1140: (\S\ref{sec:numerics} and \citet{fryxell2000,calder2007,townsley2007,seitenzahl2008a}).
1141: For the first time in this work, we have extended the 3-stage reactive ash model for nuclear burning
1142: described in \citet{townsley2007}
1143: to study the ignition and propagation of the detonation mode of burning.
1144: As a result, our explosion models
1145: are unique in terms of the degree to which non-idealized nuclear physics are employed, including
1146: a non-static, tabularized treatment of the nuclear statistical equilibrium (NSE) state and the
1147: inclusion of contemporary weak reaction rates \citep{seitenzahl2008a}.
1148: In addition, we have demonstrated here, by reaction network post-processing of 
1149: recorded Lagrangian histories, that the 3-stage reactive ash model
1150: provides a suitable reproduction of fluid density and temperature histories to
1151: allow detailed nucleosynthesis, including self-consistent neutronization.
1152: Using these techniques, we have followed the progression of a thermonuclear flame (deflagration)
1153: from a single ignition point which is varied to successively larger distances from the center of a 
1154: carbon oxygen white dwarf, and we have described in detail the resulting surface flows and 
1155: detonation which ensue.
1156: 
1157: \par Detonations arise within a colliding surface flow for all models which are 
1158: ignited at a radial location which exceeds $\sim$ 20 km in our 2D simulations.
1159: Flames ignited closer to the stellar center release enough nuclear energy to signifianctly 
1160: expand the stellar core to a degree that it stalls the surface 
1161: flow, thus preventing a strong collision region and detonation.  The nuclear binding energy released 
1162: in these stalled surface flow models, however, is not enough to gravitationally unbind the star and 
1163: they remain viable candidates for a pulsational detonation upon recollapse \citep{khoklov1991,arnett1994}.
1164: Models which detonate release $\sim 2\times10^{51}$ erg in nuclear binding energy, 
1165: resulting in a supernova-like explosion with total energy $E_{\rm tot}\sim1.5\times 10^{51}$ erg.  
1166: 
1167: \par  In all of the models in our parameter study which produce supernova-like explosions, detonation 
1168: initiates within a jet-like flow which forms in the converging surface flow. This is in agreement with
1169: the results presented in \citet{kasen2007}. However, we do not find that the
1170: detonation initiates through a shock to detonation transition (SDT) as suggested by these authors, but
1171: instead find that the detonation occurs through a gradient mechanism. The initiation of the detonation
1172: takes place within the compressed gas which lies ahead of the high velocity jet, and ahead of the
1173: internal shock which forms within the jet (see \S\ref{sec:det} and Figure~\ref{fig:jet-slice}).  
1174: The focusing of the surface flow and the formation of the jet is also present in 3D simulations
1175: (\S\ref{sec:det}, and \citealt{jordan2007,roepke2007a}),
1176: and is therefore not an artifact of the 2D axisymmetric geometry used.
1177: 
1178: \par Within a few seconds after the detonation wave disrupts the stellar core, homologous expansion is
1179: beginning to be established. By t$\sim$ 4 s from flame ignition more than 90\% of the total energy 
1180: is in the kinetic energy of expansion, and the expansion velocity has acquired a linear dependence on radius. 
1181: The final remnant posesses both global and small scale asymmetries which will influence the observational signature.
1182: When the remnant enters the homologous expansion phase it is characterized by a smooth, layered 
1183: inner core surrounded by a low density, flocculent layer of deflagration ash which was dumped onto
1184: the surface of the star prior to detonation. The smooth inner core of the remnant has a global north-south 
1185: asymmetry due to off-center ignition and surface detonation, which is well characterized by circular isodensity
1186: contours which are progressively off-center at higher densities (see \S~\ref{subsec:remnant} and Figure~\ref{fig:shape}).
1187: 
1188: 
1189: \par We have analyzed in detail the nucleosynthesis of material burned to NSE in the detonation.
1190: These results have been generated from the multi-dimensional simulation data using a newly 
1191: developed post-processing method which takes advantage of the uniqueness of the NSE state and 
1192: systematic properties of detonation waves. The
1193: method, presented in \S\ref{sec:yields}, obviates the need for computationally prohibitive network calculations 
1194: along each of the millions of particle trajectories which are necessary for good mass resolution in 3D explosion models.
1195: This work addresses only material which has relaxed to NSE, which forms, by mass, nearly all of the yield from the
1196: 2-dimensional explosion models of this study.
1197: Extending the method to include detailed isotopic yields for material incompletely relaxed to NSE (incomplete carbon,
1198: oxygen, and silicon burning) is being developed and will be described in a forthcoming publication (C. Meakin, in prep).
1199: Nucleosynthesis of material processed in a deflagration instead of detonation burning mode can be processed with a
1200: similarly parameterized method, though requiring more parameters, if it reaches NSE.  This leaves only the relative
1201: minority of tracks in partially burned deflagration material to be processed directly (only a few percent of
1202: all the trajectories).
1203: 
1204: \par Larger offsets of the ignition point lead to less stellar expansion prior
1205: to detonation and therefore the production of more NSE material.  However, we find that
1206: the amount of \iso{Ni}{56} produced stays roughly fixed at $\sim 1.1 M_\odot$ for all
1207: of our 2-dimensional explosion models which extend down to a
1208: central density of $\rho_c\sim4\times 10^{8}$ g/cm$^{3}$ at the time of detonation.
1209: This regulation is due to the enhanced neutronization at the higher densities characteristic
1210: of the less-expanded cases.
1211: Higher density cores at the time of detonation result in more neutronization, and therefore
1212: a larger fractional yield of stable Fe-peak isotopes (e.g., \iso{Fe}{54} and \iso{Fe}{58}).
1213: The isotopic distribution we find in the Fe-peak is very similar to that found for pure deflagration 
1214: models, but is characterized by a lower degree of neutronization.  Less neutronization is a result of the 
1215: lower densities under which the burning proceeds in our surface detonation models compared to pure
1216: deflagrations, due to the pre-detonation expansion.
1217: Between 0.06 and 
1218: 0.14\msun\ of intermediate mass elements are produced at high velocities.
1219: Regions in which more than half of the mass is in the form of IMEs lie
1220: above an expansion velocity of 14,000 km/s for all of the 2-dimensional 
1221: detonation models calculated.
1222: 
1223: % probably should say more about IMEs compared with observation
1224: 
1225: \par We successfully reproduced the relationship between the central density and mass-density distribution
1226: in the pre-detonation expanded star by superposing on the hydrostatic star the lowest order radial mode
1227: calculated in a linear approximation.
1228: We find that much smaller \iso{Ni}{56} yields are expected in cores which undergo more expansion prior
1229: to detonation (see Figure~\ref{fig:mnse-rhoc}).  This degree of expansion appears to be achievable in 3-dimensional
1230: simulations which relax the constraints on axisymmetry of the ignition conditions necessary for 2-dimensional
1231: simulations.  Thus it is expected that more realistic simulations, which include the pre-ignition convection
1232: field and its effect on the growing flame bubble, will be characterized by such larger expansions.
1233: However, further analysis of such simulations, which will be the subject of future papers, is required.
1234: 
1235: % -- I think this is going a bit too far in speculation for the conclusions -- it is fine in the main text -Dean
1236: %In these models, almost no stable Fe will be synthesized
1237: %because detonation will occur at such low densities that the iron peak will be completely dominated by
1238: %\iso{Ni}{56}.  However, the progenitor is thought to undergo
1239: %some degree of neutronization during the simmering phase prior to flame ignition which will result in
1240: %a contribution of stable iron.
1241: 
1242: \par Future work on elucidating the SNe Ia explosion mechanism which is being pursued at the FLASH center
1243: involves the following. (1) We are extending our survey of the mapping between flame ignition conditions and final outcomes
1244: within the computational framework developed at the FLASH center, including multi-point ignition conditions and 3D models. 
1245: (2) A simulation pipeline is being constructed to generate synthetic observational diagnostics 
1246: for the explosion models, including light curves and spectra, which will allow a more direct comparison between 
1247: the systematic properties of the single degenerate Type Ia model and observational data.
1248: 
1249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% AKNOWLEDGEMENTS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1250: 
1251: \begin{acknowledgements}
1252:   We thank Snezhana Abarzhi for bringing to our attention the literature on
1253:   cummulative jets and shaped charges. We also thank Fang Peng for making her nuclear reaction network code available to us for this work.
1254:   This work is supported in part at the University of Chicago by the Depart of Energy under Grant B523820 to the ASC/Alliances
1255:   Center for Astrophysical Thermonuclear Flashes, and the National Science Foundation under Grant PHY 02-16783 for the 
1256:   Frontier Center ``Joint Institute for Nuclear Astrophysics'' (JINA), and at the Argonne National Laboratory by
1257:   the U.S. Department of Energy, Office of Nuclear Physics, under contract DE-AC02-06CH11357. 
1258: \end{acknowledgements}
1259: 
1260: 
1261: 
1262: 
1263: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% APPENDIX %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1264: 
1265: \begin{appendix}
1266: 
1267: \section{FREEZE OUT ABUNDANCES: DETAILED IRON PEAK YIELDS}
1268: 
1269: \par The iron peak nucleosynthetic yields for three models spanning the range
1270: of ignition conditions simulated are summarized in Table~\ref{tab:yields}.  
1271: The isotopes presented have been selected based on a limiting abundance 
1272: ($M_{i;0,f} > 10^{-20}$ \msun). Two columns are shown for each model including the 
1273: initial yield and the final yield after radioactive decays have been taken into 
1274: account.  The half lives and decay modes are presented for unstable isotopes.
1275: 
1276: \end{appendix}
1277: 
1278: 
1279: %%%%%%%%%%%%%%%%%%%%%%%%%% REFERENCES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1280: %\input{biblio.tex}
1281: 
1282: \begin{thebibliography}
1283: 
1284: \bibitem[Arnett(1996)]{arnett1996} Arnett, D.\ 1996, Supernovae 
1285: and Mucleosynthesis: An Investigation of the History of Matter, from the 
1286: Big Bang to the Present, by D.~Arnett.~Princeton: Princeton University 
1287: Press, 1996.,  
1288: 
1289: \bibitem[Arnett \& Livne(1994)]{arnett1994} Arnett, D., \& Livne, 
1290: E.\ 1994, \apj, 427, 315 
1291: 
1292: %\bibitem[Asida et al.(2007)]{asida2007} Asida, S. et al, 2007, in preperation.
1293: 
1294: \bibitem[Birkhoff et al.(1948)]{birkhoff1948} Birkhoff, G., 
1295:   MacDougall, D.~P., Pugh, E.~M., 
1296:   \& Taylor, G.\ 1948, Journal of Applied Physics, 19, 563 
1297:   
1298: 
1299: \bibitem[Brachwitz et al.(2000)]{brachwitz2000} Brachwitz, F., et al., 2000, \apj, 536, 934
1300: 
1301: \bibitem[Branch et al.(1995)]{branch1995} Branch, D., Livio, M., 
1302: Yungelson, L.~R., Boffi, F.~R., \& Baron, E.\ 1995, \pasp, 107, 1019 
1303: 
1304: \bibitem[Calder et al.(2007)]{calder2007} Calder, A.C.,  et al, 2007 \apj 656, 313C
1305: 
1306: \bibitem[Caughlan \& Fowler(1988)]{caughlan1988} Caughlan, G.~R., \& 
1307: Fowler, W.~A.\ 1988, Atomic Data and Nuclear Data Tables, 40, 283 
1308: 
1309: \bibitem[Chabrier \& Potekhin (1998)]{chabrier1998} Chabrier, G., Potekhin, A., 1998, \pre, 58, 4941
1310: 
1311: \bibitem[Chamulak et al.(2008)]{chamulak2008} Chamulak, D.~A., Brown, E.~F., Timmes, F.~X., 
1312: \& Dupczak, K.\ 2008, ArXiv e-prints, 801, arXiv:0801.1643 
1313: 
1314: 
1315: \bibitem[Colella \& Glaz(1985)]{colella1985} Colella, P., \& Glaz, 
1316: H.~M.\ 1985, Journal of Computational Physics, 59, 264 
1317: 
1318: \bibitem[Colella \& Woodward(1984)]{colella1984} Colella, P., \& 
1319: Woodward, P.~R.\ 1984, Journal of Computational Physics, 54, 174 
1320: 
1321: %\bibitem[Courant \& Friedrichs (1948)] {courant1948} Courant, R., \& Friedrichs, K.O., 1948, 
1322: %\textit{Supersonic Flow and Shock Waves}, Interscience, New York
1323: 
1324: \bibitem[Dursi \& Timmes(2006)]{dursi2006} Dursi, L.~J., \& 
1325: Timmes, F.~X.\ 2006, \apj, 641, 1071 
1326: 
1327: 
1328: \bibitem[Fryxell, Mueller \& Arnett(1989)]{fryxell1989} Fryxell, B., Mueller, W., \& Arnett, D., 
1329: MPIA technical report
1330: 
1331: \bibitem[Fryxell et al.(2000)]{fryxell2000} Fryxell, B., et al.\ 
1332: 2000, \apjs, 131, 273 
1333: 
1334: \bibitem[Gamezo et al.(1999)]{gamezo1999} Gamezo, V.~N., Wheeler, 
1335: J.~C., Khokhlov, A.~M., \& Oran, E.~S.\ 1999, \apj, 512, 827 
1336: 
1337: \bibitem[Gamezo et al.(2003)]{gamezo2003} Gamezo, V.~N., Khokhlov, 
1338: A.~M., Oran, E.~S., Chtchelkanova, A.~Y., 
1339: \& Rosenberg, R.~O.\ 2003, Science, 299, 77 
1340: 
1341: 
1342: \bibitem[Gamezo et al.(2005)]{gamezo2005} Gamezo, V.~N., Khokhlov, 
1343: A.~M., \& Oran, E.~S.\ 2005, \apj, 623, 337 
1344: 
1345: \bibitem[Hillebrandt \& Niemeyer(2000)]{hillebrandt2000} Hillebrandt, 
1346: W., \& Niemeyer, J.~C.\ 2000, \araa, 38, 191 
1347: 
1348: 
1349: \bibitem[Jordan et al.(2008)]{jordan2007} Jordan, G.~I., Fisher, 
1350:   R., Townsley, D., Calder, A., Graziani, C, Asida, S, Lamb, D., \& Truran, 
1351:   J.\ 2007, \apj, 681, 1448
1352: 
1353: 
1354: \bibitem[Kasen \& Plewa(2007)]{kasen2007} Kasen, D., \& Plewa, T.\ 2007, \apj, 662, 459 
1355: 
1356: 
1357: %\bibitem[Khokhlov(1989)] {khoklov1989} Khokhlov, A.M., 1989, \mnras, 239, 785
1358:   
1359: \bibitem[Khokhlov(1991)]{khoklov1991} Khokhlov, A.~M.\ 1991, \aap, 245, 114 
1360: 
1361: \bibitem[Khokhlov(1995)]{khoklov1995} Khokhlov, A.~M.\ 1995, \apj, 
1362: 449, 695 
1363: 
1364: 
1365: 
1366: \bibitem[Khokhlov et al.(1997)]{khokhlov1997} Khokhlov, A.~M., Oran, 
1367: E.~S., \& Wheeler, J.~C.\ 1997, \apj, 478, 678 
1368: 
1369: 
1370: \bibitem[Langanke \& Mart\'inez-Pinedo(2001)] {langanke2001} Langanke, K., Mart\'inez-Pinedo, G., 2001, 
1371: ADNDT, 79, 1
1372: 
1373: \bibitem[Lee \& Higgins(1999)]{lee1999} Lee,  J.H.S., \& Higgins, A.J. 1999. 
1374:   Phil. Trans. R. Soc. Lond. A 357, 3503
1375:   
1376: 
1377: \bibitem[Lesaffre et al.(2005)]{lesaffre2005} Lesaffre, P., 
1378: Podsiadlowski, P., \& Tout, C.~A.\ 2005, Nuclear Physics A, 758, 463 
1379: 
1380: \bibitem[Lodders(2003)]{lodders2003} Lodders, K.\ 2003, \apj, 591, 1220 
1381: 
1382: 
1383: \bibitem[Mazzali et al.(2008)]{mazzali2008} Mazzali, P.~A., Sauer, 
1384: D.~N., Pastorello, A., Benetti, S., \& Hillebrandt, W.\ 2008, \mnras, 456 
1385: 
1386: 
1387: 
1388: 
1389: 
1390: \bibitem[Niemeyer \& Woosley(1997)]{niemeyer1997} Niemeyer, J.~C., 
1391:   \& Woosley, S.~E.\ 1997, \apj, 475, 740 
1392: 
1393: \bibitem[Nomoto et al.(1984)]{nomoto1984} Nomoto, K., Thielemann, F.K., Yokoi, K, 1984, \apj, 286, 644
1394: 
1395: \bibitem[Piro \& Bildsten(2008)]{piro2008} Piro, A.~L., \& Bildsten, L.\ 2008, \apj, 673, 1009 
1396: 
1397: \bibitem[Plewa(2007)]{plewa2007} Plewa, T.\ 2007, \apj, 657, 942 
1398: 
1399: 
1400: \bibitem[Rauscher \& Thielemann(2000)]{rauscher2000} Rauscher, T., Thielmann, F.-K., 2000, ADNDT, 75, 1R
1401: 
1402: \bibitem[Riess et al.(1998)]{riess1998} Riess, A.~G., et al.\ 
1403: 1998, \aj, 116, 1009
1404: 
1405: \bibitem[R{\"o}pke et al.(2007a)]{roepke2007a} R{\"o}pke, F.~K., 
1406: Woosley, S.~E., \& Hillebrandt, W.\ 2007, \apj, 660, 1344
1407: 
1408: \bibitem[R{\"o}pke et al.(2007b)]{roepke2007b} R{\"o}pke, F.~K., 
1409: Hillebrandt, W., Schmidt, W., Niemeyer, J.~C., Blinnikov, S.~I., \& 
1410: Mazzali, P.~A.\ 2007, \apj, 668, 1132
1411: 
1412: \bibitem[Schmidt et al.(2006a)]{schmidt2006a} Schmidt, W., Niemeyer, 
1413:   J.~C., \& Hillebrandt, W.\ 2006, \aap, 450, 265
1414:   
1415: \bibitem[Schmidt et al.(2006b)]{schmidt2006b} Schmidt, W., Niemeyer, 
1416:   J.~C., Hillebrandt, W.,R{\"o}pke, F.~K.\ 2006, \aap, 450, 283
1417:   
1418: \bibitem[Seitenzahl et al.(2008a)]{seitenzahl2008a} Seitenzahl, I., Townsley, D., Peng, F., Truran, J., 
1419:   ADNDT, in press. %NSE table paper
1420:   
1421: \bibitem[Seitenzahl et al.(2008b)]{seitenzahl2008b} Seitenzahl, I., et al., in preperation. 
1422:   %detonation initiation paper
1423:   
1424: \bibitem[Sharpe(2001)]{sharpe2001} Sharpe, Gary J., 2001, \mnras, 322, 614
1425:   
1426:   
1427: \bibitem[Stehle et al.(2005)]{stehle2005} Stehle, M., Mazzali, 
1428: P.~A., Benetti, S., \& Hillebrandt, W.\ 2005, \mnras, 360, 1231 
1429: 
1430: 
1431: \bibitem[Thielemann et al.(1986)] {thielemann1986} Thielemann, F.-K., Truran, J., Arnould, M., 1986,
1432: Advances in Nuclear Astrophysics, ed. E. Vangioni-Flam et al. (Gif-sur-Yvette: Edidtions Fronti\`eres), 525
1433: 
1434: \bibitem[Timmes \& Arnett (1999)] {timmes1999} Timmes, F.~X.,Arnett, D., 1999, \apjs, 125, 277
1435: 
1436: \bibitem[Timmes et al.(2003)]{timmes2003} Timmes, F.~X., Brown, 
1437: E.~F., \& Truran, J.~W.\ 2003, \apjl, 590, L83 
1438: 
1439: \bibitem[Timmes \& Swesty(2000a)]{timmes2000a} Timmes, F.~X., \& 
1440: Swesty, F.~D.\ 2000, \apjs, 126, 501 
1441: 
1442: \bibitem[Timmes et al.(2000b)]{timmes2000b} Timmes, F.~X., et al.\ 
1443: 2000, \apj, 543, 938 
1444: 
1445: \bibitem[Timmes \& Woosley(1992)]{timmes1992} Timmes, F.~X., \& 
1446: Woosley, S.~E.\ 1992, \apj, 396, 649 
1447: 
1448: 
1449: \bibitem[Travaglio et al.(2004)] {travaglio2004} Travaglio, C., et al., 2004, \aap, 425, 1029
1450: 
1451: \bibitem[Townsley et al.(2007)]{townsley2007} Townsley, D., et al., 2007, \apj, submitted
1452: 
1453: \bibitem[Wallace et al.(1982)]{wallace1982} Wallace, R., Woosley, S., Weaver, T., 1982, \apj, 225, 1021
1454: 
1455: \bibitem[Wang et al.(2003)]{wang2003} Wang, L., et al.\ 2003, 
1456: \apj, 591, 1110 
1457: 
1458: \bibitem[Wang et al.(2007)]{wang2007} Wang, L., Baade, D., 
1459: \& Patat, F.\ 2007, Science, 315, 212 
1460: 
1461: 
1462: \bibitem[Woosley et al.(1973)]{woosley1973} Woosley, S.~E., Arnett, 
1463: W.~D., \& Clayton, D.~D.\ 1973, \apjs, 26, 231 
1464: 
1465: \bibitem[Woosley et al.(2004)]{woosley2004} Woosley, S.~E., Wunsch, 
1466: S., \& Kuhlen, M.\ 2004, \apj, 607, 921 
1467: 
1468: 
1469: \bibitem[Zel'dovich et al.(1970)]{zeldovich1970} Zel'dovich, Ya. B., Librovich, V. B., 
1470: Makhviladze, G. M., \& Sivashinsky, G. I. 1970, Acta Astron., 15, 313
1471: 
1472: \end{thebibliography}
1473: 
1474: 
1475: 
1476: %%%%%%%%%%%%%%%%%%%%%%%%%%%% FIGURES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1477: \clearpage
1478: %\input{figures.tex}
1479: 
1480: \begin{figure}
1481: %  \includegraphics{figs/rpv1_16o40.jpg}
1482:   \includegraphics[scale=0.75]{f1.jpg}
1483:   \caption{\label{fig:breakout}
1484:   This time sequence of ash abundance (represented by the $\phi_1$
1485:   progress variable) shows how a bubble ignited near the stellar core rises
1486:   buoyantly, erupts from the star's surface and drives a flow which is largely
1487:   confined to the surface of the star by gravity.  The black contour line indicates 
1488:   a density of 10$^7$ g cm$^{-3}$.  Eventually the surface flow converges at the 
1489:   opposite pole from the breakout location, compressing material in that region
1490:   until it begins to burn carbon. The dashed box in the right panel indicates
1491:   the region detailed in Figure \ref{fig:jet} below, where the converging
1492:   flow produces a jet that initiates a detonation wave.}
1493: \end{figure}
1494: 
1495: 
1496: \begin{figure}
1497: %  \includegraphics{figs/homologous_expansion.25_and_100.ps}
1498:   \includegraphics{f2.jpg}
1499:   \caption{The radial density profile scaled to the central density and the
1500:   density e-folding height for the initial white dwarf (thick grey
1501:   line), and at the time when a surface detonation initiates ($t_{\rm det}$,
1502:   see Table \ref{tab:models}) for flame bubbles ignited  at 25 (thin black lines)
1503:   and 100 km (thick black lines) off-center.
1504:   Equatorial (solid lines) and polar (dashed lines) profiles are shown
1505:   for both of the pre-detonation models and are well described by homologous
1506:   expansion with minimal asymmetry.
1507:     \label{fig:homologous-deflagration}}
1508: \end{figure}
1509: 
1510: 
1511: \begin{figure}
1512: %  \includegraphics{figs/jet_fig_16o40_press.jpg}%{figs/jet_initiation_final.ps}
1513:   \includegraphics[scale=0.6]{f3.jpg}
1514:   \caption{A time sequence showing the ``collision region'' and the bidirectional
1515:     jet-like flow for the model ignited 40 km off-center.
1516:     The pressure field and the velocity field are shown as the detonation wave initiates 
1517:     and begins to break away from the end of the inwardly moving jet component.
1518:     The red and black contour lines indicate carbon depletion at the 1\% and 99\% levels, 
1519:     respectively. In the region where the detonation initiates the
1520:     density and temperature are 10$^7$ g/cm$^3$ and 4$\times$10$^9$ K. The longest
1521:     velocity vector indicates a flow speed of $v_{\rm vec} = 10^9$ cm/s, while other
1522:     vectors have lengths linearly proportional to the flow speed.
1523:     \label{fig:jet}}
1524: \end{figure}
1525: 
1526: 
1527: \begin{figure}
1528:   \epsscale{1.0}
1529:   %\includegraphics{figs/jet_slices_w3d_wpram.jpg}
1530:   \includegraphics[scale=0.62]{f4.jpg}
1531:   \caption{Flow properties along the jet axis prior to detonation for 2D and 3D models, including
1532:     density, radial velocity, temperature, and gas pressure.  The radially directed ram pressure, 
1533:     $p_{\rm ram} = \rho v_r^2$, is shown in the pressure figure for each model by the thin red line.
1534:     The 2D models shown were ignited by 16 km radius flame bubbles with offset (left) $r_{\rm off} = 25$ km and 
1535:     (middle) $r_{\rm off} = 100$ km.  A low resolution ($\Delta = 8$ km) 3D model is shown (right) 
1536:     which was ignited by a 16 km radius flame bubble with offset $r_{\rm off} = 80$ km for comparison.
1537:     The dashed vertical lines mark the locations of the burning front for each model, taken to be
1538:     where $\phi_1 = 0.5$. In the top panel, the horizontal dot-dashed line marks a density of 
1539:     $10^7$ g/cm$^3$, a value above which detonation readily arises in the simulations once it reaches
1540:     a temperature of $T\sim 2\times 10^9$ K. The velocity zero point is marked by a dot-dashed horizontal line.
1541:     \label{fig:jet-slice}}
1542: \end{figure}
1543: 
1544: 
1545: \begin{figure}
1546:   %\includegraphics{figs/det_fig_16o40_temp.jpg}
1547:   \includegraphics[scale=0.6]{f5.jpg}
1548:   \caption{In this time sequence, the detonation wave breaks away from the jet in which it 
1549:     formed and sweeps across the stellar core.  The black line is the 10$^7$ g/cm$^3$ iso-density 
1550:     contour which is roughly coincident with the stellar surface.  
1551:     \label{fig:det-temp}}
1552: \end{figure}
1553: 
1554: 
1555: 
1556: 
1557: \begin{figure}
1558:   %\includegraphics{figs/ye_fig_16o40.jpg}
1559:   \includegraphics[scale=0.6]{f6.jpg}
1560:   \caption{The time evolution of the electron mole fraction ($Y_e$) is shown as the detonation 
1561:     wave passes over the stellar center.  The dip in $Y_e$ reveals that neutronization is taking 
1562:     place in the high density core where material has burned to NSE.  The expansion 
1563:     which follows the detonation freezes the $Y_e$ distribution as the material evolves into a 
1564:     supernova remnant. The neutron rich material (low $Y_e$) which surrounds the stellar
1565:     core is the ash from the deflagration which had burned at high densities before it
1566:     erupted from the star and spread out over the surface.
1567:     The black line is the 10$^7$ g/cm$^3$ iso-density contour and the light 
1568:     blue line indicates the contour of carbon depletion at the 99\% level.\label{fig:det-ye}}
1569: \end{figure}
1570: 
1571: 
1572: \begin{figure}
1573:   %\includegraphics{figs/vyslices.460_to_510.ps}{figs/ddslices.460_to_510.ps}
1574:   \includegraphics[scale=0.45]{f7a.jpg}
1575:   \includegraphics[scale=0.45]{f7b.jpg}
1576:   \caption{The profile of the (left) velocity and (right) density along the
1577:     polar axis is shown for several moments evenly spaced in time ($\delta t=$0.025 s)  
1578:     as the detonation wave passes across the stellar core for the model ignited
1579:     with a flame bubble 40 km from the stellar center.
1580:     \label{fig:det-propagation}}
1581: \end{figure}
1582: 
1583: 
1584: 
1585: 
1586: \begin{figure}
1587:   \epsscale{1.0}
1588:   %\includegraphics{figs/energy_25.ps}
1589:   \includegraphics{f8.jpg}
1590:   \epsscale{1.0}
1591:   \caption{The time evolution of the kinetic, internal, and  gravitational potential 
1592:     energy for a model with a flame bubble ignited 25 km from the stellar center.
1593:     The nuclear energy released by burning in the deflagration ($t < t_{det}$, with $t_{det} = 2.45$s) 
1594:     and the detonation ($t > t_{det}$) can be seen as a change in the sum of 
1595:     these three energy components (blue).\label{fig:energy}}
1596: \end{figure}
1597: 
1598: 
1599: 
1600: 
1601: \begin{figure}
1602:   %\includegraphics{figs/detprofiles_25and100.jpg}
1603:   \includegraphics[scale=0.7]{f9.jpg}
1604:   \caption{Late time ($t>4$ s) density and velocity profiles for post detonation state models
1605:     having ignition:
1606:     (left) 25 km and (right) 100 km off-center. The density
1607:     is scaled by the peak value and the position is scaled by the density e-folding
1608:     distance in the equatorial direction.
1609: %    The spatial scales for the 25 km and 100 km offset models
1610: %    are  $r_{\rho} = 1.41\times 10^9$ cm and  $r_{\rho} = 1.62\times 10^9$ cm, 
1611: %    respectively.
1612:     The thick gray line shows the scaled density profile 
1613:     of the initial white dwarf model, while the post detonation state model is shown 
1614:     by thick black lines for profiles along the: (solid) equatorial, and (dashed) polar axes.
1615:     The thin black line shows the reconstructed density profile along the polar
1616:     axis based on the contour fits presented in Figure~\ref{fig:shape} (see text for more details).
1617:     \label{fig:det-profiles}}
1618: \end{figure}
1619: 
1620: 
1621: 
1622: %\begin{figure}
1623: %  \epsscale{1.0}
1624: %  \includegraphics{figs/line-density-25and100.ps}
1625: %  \caption{Scaled line density, $\lambda/\lambda_{\rm max}$ (see eq.[\ref{eq:line-density}]), along
1626: %    the symmetry axis at the following times: (black) just prior to detonation, and 
1627: %    (gray) at late times ($t>4$ s).
1628: %    Shown are the models with flame bubbles ignited: (thick) 25 km, and 
1629: %    (thin) 100 km off-center.
1630: %    The spatial scales are the same density e-folding lengths as in Figure~\ref{fig:det-profiles}.
1631: %    \label{fig:line-density}}
1632: %\end{figure}
1633: 
1634: \begin{figure}
1635:   \epsscale{1.0}
1636:   %\includegraphics{figs/dens_contours_25and100.jpg}
1637:   \includegraphics[scale=0.7]{f10.jpg}
1638:   \caption{Late time ($t>4$ s) density contours for the models ignited: (left) 25 km, and
1639:     (right) 100 km off-center.  The contours mark the locations at which
1640:     $\ln (\rho/\rho_c) = -0.5, -1, -1.5, -2, -2.5$ where $\rho_c$ is the peak density.
1641:     The magnitude of the largest velocity vectors are (left) $v = 2.4\times 10^9$ cm/s, and 
1642:     (right) $v = 2.5\times 10^9$ cm/s.
1643:     The spatial scales are the same density e-folding lengths as in Figure~\ref{fig:det-profiles}.
1644:     \label{fig:dens-contours}}
1645: \end{figure}
1646: 
1647: 
1648: \begin{figure}
1649:   \epsscale{0.8}
1650:   %\includegraphics{figs/shape_25_and_100.ps}
1651:   \includegraphics{f11.jpg}
1652:   \caption{The iso-density contours of the remnant during late times 
1653:     ($t>4$ s, Figure~\ref{fig:dens-contours}) are well described by 
1654:     circles of radius $R_c$ which have centers that are offset from the origin
1655:     by an amount $y_c$ along the symmetry axis. The best fit radii and offsets are
1656:     shown here as a function of density contour value scaled to the central density 
1657:     for the models ignited 25 km (thin-line), and 100 km (thick-line) off-center
1658:     in the panels above: (top) radius, $R_c$; (middle) circle center, $y_c$. 
1659:     The spatial dimensions are scaled in terms of the e-folding density scale
1660:     height in the equatorial direction of the remnant.
1661:     (bottom) The degree of clumpiness is characterized by the ratio of the r.m.s. 
1662:     deviation in density along the best fit circle to the density contour value,
1663:     denoted $\delta\rho/\rho$.
1664:     The density perturbations at high density ($\ln\rho/\rho_c$>1.5) are due primarily
1665:     to the narrow trail of ash left behind as the flame bubble rises our of the
1666:     stellar core.
1667:     \label{fig:shape}}
1668: \end{figure}
1669: 
1670: 
1671: \clearpage
1672: 
1673: 
1674: \begin{figure}
1675:   \epsscale{1.0}
1676:   %\includegraphics{figs/mnse_rhoc.ps}
1677:   \includegraphics{f12.jpg}
1678:   \caption{The total mass of NSE material and \nuc{56}{Ni} created in the detonation and the 
1679:     total mass of high density ($\rho>10^7$ g/cm$^3$) matter at detonation is plotted against 
1680:     the central density at detonation for all of the 2D models studied. 
1681:     The total mass of material having a density which exceeds 
1682:     $\rho=5\times 10^6$, $7.5\times 10^6$, and $10^7$ g/cm$^3$ during a 0.25 s time period 
1683:     preceeding detonation is shown by the curves which terminate at the detonation
1684:     density for the 25 km and 100 km off-center ignition models.
1685:     Data points for two 3D models are shown for comparison: the data points labeled 
1686:     ``3D Single'' show the \nuc{56}{Ni}, NSE, and high density material masses for a 
1687:     3D model ignited  80 km  off-center.  The data point labeled ``3D Multi'' shows 
1688:     the central density and the high density material mass for a 3D multi-point ignition 
1689:     model which is described in the text (see \S\ref{sec:yields}).  The dashed line shows 
1690:     the relationship between central density and high density material for the initial
1691:     white dwarf model expanded by the (linear) fundamental pulsation mode.
1692:     \label{fig:mnse-rhoc}}
1693: \end{figure}
1694: 
1695: 
1696: 
1697: \begin{figure}
1698:   %\includegraphics{figs/traject_exp_solid.jpg}
1699:   \includegraphics[scale=0.6]{f13.jpg}
1700:   \caption{The thermodynamic trajectory of a Lagrangian tracer particle having
1701:     an expansion timescale $\tau = 0.42$s, and final entropy $s$ = 2.273 N$_A k$ and final
1702:     electron mole fraction $Y_e$ = 0.49873.  The dotted line shows the analytic 
1703:     adiabatic fit  to this trajectory, parameterized by $\tau$, $Y_e$, and $s$.
1704:     \label{fig:trajectory}}
1705: \end{figure}
1706: 
1707: 
1708: 
1709: 
1710: \begin{figure}
1711:   \epsscale{0.9}
1712:   %\includegraphics{figs/tj_exp_nse_many_highT.jpg}
1713:   \includegraphics[scale=0.6]{f14.jpg}
1714:   \caption{The time evolution is shown for various high abundance iron peak isotopes 
1715:     during the expansion which is here parameterized by the plasma temperature.
1716:     For each species the abundances have been calculated with a network using the thermodynamic 
1717:     trajectory of the tracer particle (solid) and the analytic fit (dotted). For comparison, the
1718:     NSE values are shown (dashed) using the thermodynamic conditions at each point along 
1719:     the particle trajectory.  All three are in good agreement until $T\sim5.5\times10^9$ K, 
1720:     below which the NSE distribution begins to deviate from the network calculation at various 
1721:     temperatures. The tracer particle and the analytic fit agree to a high level of precision 
1722:     through freeze out. \label{fig:trajectory-burn}}
1723: \end{figure}
1724: 
1725: 
1726: 
1727: 
1728: \begin{figure}
1729:   %\includegraphics{figs/ye_s.4_16_100.ps}
1730:   \includegraphics[scale=0.6]{f15.jpg}
1731:   \caption{The distribution of NSE mass in entropy and degree of neutronization 
1732:     for the  model with 100 km off-center bubble ignition.  The black line indicates the 
1733:     curve on which the freeze out yield table was calculated (\S\ref{subsec:freezeout-method}).
1734:     %(right) Cumulative distribution function showing the fraction of NSE material below 
1735:     %a certain degree of neutonization.
1736:     \label{fig:ye-s-table}}
1737: \end{figure}
1738: 
1739: 
1740: 
1741: \begin{figure}
1742:   %\includegraphics{figs/ye_s_cdf.ps}
1743:   \includegraphics{f16.jpg}
1744:   \caption{Cumulative distribution functions showing the fraction of NSE material below 
1745:     a certain degree of neutonization for the models ignited: 25 km, 40km, and 100 km 
1746:     off-center (from darkest to lightest, respectively).
1747:     \label{fig:ye-s-cdf}}
1748: \end{figure}
1749: 
1750: 
1751: \begin{figure}
1752:   %\includegraphics{figs/freezeout_table_unscaled.ps}{figs/freezeout_table_scaled.ps}
1753:   \includegraphics[scale=0.45]{f17a.jpg}
1754:   \includegraphics[scale=0.45]{f17b.jpg}
1755:   \caption{Iron peak freezeout yields, accounting for 
1756:     radioactive decay, as a function of the degree of neutronization.
1757:     The corresponding entropy of the material is related to the degree of neutronization, 
1758:     $S_f = S_f(\eta_f)$, by the black line shown in Figure~\ref{fig:ye-s-table}
1759:     and an expansion timescale of $\tau = 0.4$ s.
1760:     (left) Isotope mass fractions, X(\nuc{A}{Z}) for nuclide with atomic number A and 
1761:     proton number Z.
1762:     (right) Isotope mass fractions scaled to X(\nuc{56}{Fe}) and normalized to the 
1763:     corresponding solar system abundance ratios of \citet{lodders2003} where
1764:     [\nuc{A}{Z}/\nuc{56}{Fe}] =$\log_{10}$ 
1765:     (X(\nuc{A}{Z})/X(\nuc{56}{Fe})) - $\log_{10}$(X(\nuc{A}{Z})/X(\nuc{56}{Fe}))$_{\odot}$.
1766:     The dashed horizontal lines indicate where the scaled abundance ratio is equal to 0.5, 
1767:     1.0, and 2.0 times the solar system value.
1768:     \label{fig:yields-table}}
1769: \end{figure}
1770: 
1771: 
1772: \begin{figure}
1773:   %\includegraphics{figs/final_iron_peak_yields.ps}{figs/final_iron_peak_yields_25km_etamin.ps}
1774:   \includegraphics[scale=0.45]{f18a.jpg}
1775:   \includegraphics[scale=0.45]{f18b.jpg}
1776:   \caption{(left) The nucleosynthetic yields normalized by the solar system abundances are 
1777:     shown for three models which span the degree of pre-expansion and neutronization 
1778:     in our simulation suite.
1779:     (right) The variation in the yields due to imposing a neutronization floor of 
1780:     $\eta_{\rm min}$ = 0, $10^{-3}$, and $2\times 10^{-3}$ for the model ignited 
1781:     25 km off-center. Same notation as Figure~\ref{fig:yields-table} \label{fig:yields}}
1782: \end{figure}
1783: 
1784: 
1785: \begin{figure}
1786:   %\includegraphics{figs/final_iron_peak_yields_100km_texp.ps}
1787:   %{figs/final_iron_peak_yields_100km_entropy.ps}
1788:   \includegraphics[scale=0.45]{f19a.jpg}
1789:   \includegraphics[scale=0.45]{f19b.jpg}
1790:   \caption{The variation in nucleosynthetic yields due to:
1791:     (left) variations in expansion timescale, with $\tau_{exp} = $0.2, 0.4, and 0.6 s; 
1792:     and (right) entropy changes, where $\pm$5\% variations about the fiducial 
1793:     entropy-neutronization curve, $S_f = S_f(\eta_f)$, shown in Figure~\ref{fig:ye-s-table} 
1794:     have been used. Same notation
1795:     as Figure~\ref{fig:yields-table}.
1796:     \label{fig:yields-texp-entropy}}
1797: \end{figure}
1798: 
1799: 
1800: 
1801: \begin{figure}
1802:   %\includegraphics{figs/vyields.4_16_25.ps}{figs/vyields.4_16_100.ps}
1803:   \includegraphics[scale=0.45]{f20a.jpg}
1804:   \includegraphics[scale=0.45]{f20b.jpg}
1805:   \caption{Distribution of elemental abundances are shown as a function of 
1806:     expansion velocity for models having initial flame bubbles ignited (left) 
1807:     25 km and (right) 100 km from the stellar center. The elemental yields are 
1808:     calculated by taking into account radioactive decays with half lives less than 
1809:     1 day.  The dotted vertical line indicates the velocity above which less than 
1810:     95\% of the material has been burned to NSE and therefore our nucleosynthesis 
1811:     post-processing method (\S\ref{subsec:freezeout-method}) is no longer reliable. 
1812:     At velocities where less than 95\% of the material burns to NSE we show only 
1813:     the total fraction of NSE and intermediate mass elements (IMEs).  The red 
1814:     curve shows the fraction of the total stellar mass interior to the 
1815:     velocity. \label{fig:yields-velocity}}
1816: \end{figure}
1817: 
1818: 
1819: 
1820: %%%%%%%%%%%%%%%%%%%%%%%%% TABLES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1821: \clearpage
1822: \input{tab1.tex}
1823: \input{tab2.tex}
1824: \input{tab3.tex}
1825: \input{tab4.tex}
1826: 
1827: %\input{tables.tex}
1828: %\input{tables.appendix.tex}
1829: 
1830: 
1831: 
1832: 
1833: 
1834: \end{document}
1835: