1: \documentclass[12pt,preprint]{aastex}
2: %\usepackage{epsfig}
3: \shorttitle{Simulations of White Dwarf Disk Boundary Layers}
4: \shortauthors{Balsara et al.}
5: \begin{document}
6: \bibliographystyle{apj}
7:
8: \title{Simulations of the Boundary Layer Between a White Dwarf
9: and its Accretion Disk}
10:
11: \author{Dinshaw S. Balsara \& Jacob Lund Fisker}
12: \affil{Department of Physics, University of Notre Dame, Notre Dame, IN 46556 \\
13: dbalsara@nd.edu; jfisker@nd.edu}
14:
15: \author{Patrick Godon\altaffilmark{1} \& Edward M. Sion}
16: \affil{Department of Astronomy and Astrophysics,
17: Villanova University, Villanova, PA 19085 \\
18: patrick.godon@villanova.edu; edward.sion@villanova.edu}
19:
20: \altaffiltext{1}
21: {Visiting at the Space Telescope Science Institute, Baltimore, MD 21218,
22: USA, godon@stsci.edu}
23:
24:
25: \begin{abstract}
26:
27: Using a 2.5D time-dependent numerical code we recently developed, we solve the full compressible Navier-Stokes equations to determine the structure of the boundary layer between the white dwarf and the accretion disk in non-magnetic cataclysmic variable systems. In this preliminary work, our numerical approach does not include radiation. In the energy equation, we either take the dissipation function ($\Phi$) into account or we assume that the energy dissipated by viscous processes is instantly radiated away ($\Phi=0$). For a slowly rotating non-magnetized accreting white dwarf, the accretion disk extends all the way to the stellar surface. There, the matter impacts and spreads towards the poles as new matter continuously piles up behind it. We carry out numerical simulations for different values of the alpha viscosity parameter ($\alpha$), corresponding to different mass accretion rates. In the high viscosity cases ($\alpha = 0.1$), the spreading boundary layer sets off a gravity wave in the surface matter. The accretion flow moves supersonically over the cusp making it susceptible to the rapid development of gravity wave and/or Kelvin-Helmholtz shearing instabilities. This BL is optically thick and extends more than 30 degrees to either side of the disk plane after only 3/4 of a Keplerian rotation period ($t_K$=19s). In the low viscosity cases ($\alpha =0.001$) , the spreading boundary layer does not set off gravity waves and it is optically thin.
28:
29: \end{abstract}
30:
31: \keywords{accretion, accretion disks -- binaries: close --- novae,
32: cataclysmic variables --- white dwarfs --- methods: numerical}
33:
34: \section{Introduction: Accreting White Dwarfs in Cataclysmic
35: Variables}\label{sec:introduction}
36:
37: Cataclysmic variables (CVs) form an interesting class of short-period
38: close binary systems comprising a hot white dwarf (WD) and a relatively
39: lower mass red dwarf star filling its Roche lobe \citep{cra56,kra62}.
40: In such systems hydrogen-rich matter from the red dwarf exits through
41: the inner Lagrange point (L1) and flows towards the white dwarf.
42: In the absence of strong magnetic fields, the
43: matter forms an accretion disk around the white dwarf due
44: to the excess angular momentum originating from the orbital motion of
45: the binary \citep{Prendergast68,fla74,lub75}. On-going accretion at a low rate
46: (quiescence) is interrupted every few weeks to months by intense
47: accretion (outburst)
48: of days to weeks - a dwarf nova (DN) accretion
49: event \citep{bat72}, thereby increasing the luminosity of the systems by several
50: magnitudes. A thermal instability in the accretion disk is believed to
51: trigger the increase in the mass transfer rate ($\dot{m}$)
52: through the disk and thus an increase in the rate of gravitational
53: energy release \citep{Cannizzo88}.
54:
55: CV systems are divided in sub-classes according to the duration,
56: occurrence and amplitude of their outburst \citep{hac93,Warner95,rit98}:
57: e.g. dwarf nova systems (DNs)
58: are non-magnetic and accrete through a disk, they spend most
59: of their time in the quiescent state; nova-like systems (NLs) are
60: disk-systems found mostly in the high outburst state; polars are
61: devoid of an accretion disk due to their strong magnetic fields (the
62: matter is funneled through the magnetic field lines onto the poles
63: of the WD); and intermediate polars (IPs) have truncated inner
64: disks due to their moderately strong magnetic fields.
65: Another CV subclass is the classical nova (or just ``nova'';),
66: characterized by an episode of unstable thermonuclear burning
67: (the thermonuclear runaway - TNR; \citet{ros68}).
68: All CV system are believed to undergo a classical nova explosion
69: every few thousand years or more, when enough hydrogen-rich
70: material accumulated in the envelope to reach the ignition point
71: at the electron-degenerate base of the envelope - where it is
72: compressed under the large gravity of the WD
73: \citep{sta71a,sta71b,Starrfield72}.
74: In the present work we concentrate on the study of the non (or
75: weakly) magnetized DN systems, where the accretion disk extends
76: all the way to the surface of the WD.
77:
78: Since the white dwarf is the most common end-product of stellar evolution
79: ($\approx 90$\% of all the stars in the Galaxy have evolved or will
80: evolve into white dwarfs) and the accretion disk is the most
81: common universal structure resulting from mass transfer with angular
82: momentum, and both can be directly observed in cataclysmic variable
83: systems (in the ultraviolet), an understanding of the accretion process
84: in cataclysmic variables is the first step in a global understanding
85: of accretion in other systems throughout the universe. These include
86: young stellar objects, accretion onto neutron stars and black holes,
87: and the most difficult to study, active galactic nuclei.
88: Therefore, accreting white dwarfs in cataclysmic variables are the best
89: {\it astronomical laboratories} to study the physics of accretion
90: disks. \\
91:
92: \subsection{The Accretion Disk \& Boundary Layer in One-Dimension}
93:
94: In the disk, magneto-hydrodynamic (MHD)
95: turbulence \citep{Shakura73} due to a magneto-rotational
96: instability \citep{Balbus94a} dissipates potential energy and transfers
97: angular momentum outward. As a result, the disk matter slowly spirals
98: inwards towards the white dwarf \citep{Pringle81}.
99: The total potential energy of accretion can be released at the
100: maximum rate
101: \begin{equation}
102: L_{acc}=\frac{GM_*\dot{m}}{r_*} = \dot{m} r_*^2 \Omega_K^2(r_*) ,
103: \end{equation}
104: where $G$ is the gravitational constant, $M_*$ is the mass of the
105: WD, $r_*$ its radius, $\dot{m}$ is the mass accretion rate, and
106: $\Omega_K(r_*)$ is the Keplerian angular velocity at one
107: stellar radius $r_*$.
108: This is the maximum amount of energy that can be extracted from the
109: accretion process per unit of time
110: and the actual luminosity can be smaller than this
111: (see below).
112: In the standard disk theory \citep{Shakura73,Lynden-Bell74},
113: the accretion disk is axisymmetric, geometrically thin in the
114: vertical dimension (it has a thickness $H$ such that $H/r<<1$),
115: and the energy dissipated by the (turbulent)
116: viscosity is instantly radiated locally in the $\pm z$ directions.
117: Only half of the accretion luminosity is emitted by the disk
118: ($L_{disk}=L_{acc}/2$), since the matter is still moving at a nearly
119: Keplerian velocity, $v_K\approx \sqrt{GM_*/r_*}$, before it is ultimately
120: accreted \citep{Lynden-Bell74}.
121:
122: The remaining accretion energy must, therefore, be dissipated
123: in order for the matter to be accreted onto the surface
124: of the more slowly rotating WD \citep{Pringle81}.
125: The disk matter is decelerated in that region where the inner disk
126: reaches the stellar surface: the boundary layer (BL).
127: The height of the BL can be comparable to
128: the scale height $H$ of the accretion disk.
129: The remaining rotational kinetic energy of the accreting matter
130: dissipated in the
131: boundary layer per unit of time ($L_{BL}$) is nearly equal to half of the
132: total luminosity of the accreting matter ($L_{acc}$),
133: namely:
134: \begin{equation}
135: L_{BL}=
136: \left( 1 - \beta^2 \right) L_{disk} =
137: \frac{\dot{m} r_*^2}{2} ( \Omega_K^2(r_*) - \Omega_*^2 ) ,
138: \end{equation}
139: where $\beta=\Omega_*/\Omega_K(r_*)$,
140: and $\Omega_*$ is the stellar angular velocity.
141: This is so because the material accreted onto
142: the WD surface corotates with it and keeps
143: a fraction of the rotational kinetic energy.
144: \citet{Kluzniak87}, however, pointed out that
145: part of the BL energy goes into
146: spinning up the accreting star through the shear
147: at the stellar equator, and hence, the fraction
148: that can be radiated away is smaller than the one given in eq.(2).
149: Namely, one has
150: \begin{equation}
151: L_{BL}=
152: ( 1- \beta )^2 L_{disk} =
153: \frac{\dot{m} r_*^2}{2} ( \Omega_K(r_*) - \Omega_* )^2.
154: \end{equation}
155: Therefore, for a non-rotating WD: $L_{BL}=L_{disk}=L_{acc}/2$.
156: However, for a rotating WD $L_{BL} < L_{disk}$,
157: and the total energy
158: radiated from the accretion process ($L_{BL}+L_{disk}$)
159: is actually smaller than $L_{acc}$ as defined by eq.(1).
160: The energy ``kept'' by the corotating accreted material
161: per unit of time is :
162: \begin{equation}
163: \beta^2 L_{disk} =
164: \frac{\dot{m} r_*^2}{2} \Omega_*^2,
165: \end{equation}
166: and the power invested to spin up the star through the shear at the equator
167: (eq.2-eq.3) is:
168: \begin{equation}
169: [ (1-\beta^2) - (1-\beta)^2 ] L_{disk} =
170: 2 \beta ( 1- \beta )
171: L_{disk} = \beta ( 1 - \beta ) L_{acc} .
172: \end{equation}
173: The above relations are correct to order of
174: $\eta = 1 - r_m/r_*$, where $r_m$ is the radius at which the gradient
175: of the angular velocity changes sign $\partial \Omega / \partial r =0$
176: (for further details of the one-dimensional analysis
177: see also \citet{Regev83,Kley91,Popham95}).
178: Usually $\eta$ is of the order of $H/r$ or smaller.
179: For a WD star rotating at 10\% of the Keplerian velocity
180: (i.e. several 100km/s)
181: the BL is expected to radiate $L_{BL}=0.81L_{disk}$, while a fraction
182: of $0.18L_{disk}$ is invested in spinning up the star.
183: For a WD star rotating at 30\% of the Keplerian velocity
184: (i.e. $\sim 1000$km/s)
185: the BL is expected to radiate $L_{BL} = 0.49L_{disk}$.
186: For a star rotating near break-up the angular velocity in the disk does not
187: have an extremum and decreases steadily outward, therefore
188: the disk is expected to spin-down the fast rotating WD
189: \citep{pop91}. In this case there is no boundary layer.
190:
191: Since the BL is much smaller than the disk itself and each radiate
192: almost equal amounts of energy, the disk emits mainly in the optical
193: and near UV, whereas the much hotter BL emits in the far UV and in X-rays.
194: During outburst, when $\dot{m}\approx 10^{-9}-10^{-8} M_{\odot}$/yr
195: \citep{war87,Cannizzo88}, observations (e.g. \citet{cor80,mau95,mau04})
196: reveal that the BL is optically thick
197: and emits mainly in the FUV and soft X-ray bands
198: with an effective temperature $T_{eff}$ of a few $10^5$K
199: as expected by the theory
200: (\citet{Pringle79,Regev83,Godon95,Popham95,Collins98b,Obach99}).
201: During quiescence, when $\dot{m}\approx 10^{-12}-10^{-10} M_{\odot}$/yr
202: \citep{war87}, observations (e.g. \citet{muk04,Pandel03,Pandel05})
203: reveal that the boundary layer is optically thin
204: and emits in the hard X-ray band with a temperature
205: of the order of $10^8$K, as expected by the theory
206: (e.g. \citet{Pringle79,Tylenda81,King84,Shaviv87,Narayan93,Popham99}).
207: However, in addition, in the outer region of the BL, where the BL
208: meets the disk, the optical thickness becomes larger again ($\sim 1$)
209: and that region emits in the FUV with a temperature $T_{eff}$ reaching
210: $\sim 10^5$K \citep{Popham99}.
211:
212: \subsection{Observational Background}
213:
214: Observationally, it was shown almost three decades ago
215: that
216: non-magnetic CVs do emit some fraction of their luminosity in the
217: X-ray bands (e.g. \citet{cor80,Cordova81,Becker81,Cordova83,Patterson85};
218: the ROSAT All-Sky Survey \citep{Beuermann93}).
219: As expected from the one-dimensional standard disk and boundary layer
220: theories, during quiescence hard X-rays (10keV and
221: higher) were observed from a small region close to the WD (e.g.
222: \citet{vanTeeseling96,Mukai97}), while during
223: outburst this emission is replaced by soft X-ray and EUV
224: (see e.g. the review of \citet{Mauche96} and the references therein).
225: The EUV region of the spectrum
226: is very difficult to observe because the absorption cross section
227: of (ISM) neutral hydrogen is very high. Because of this, only a
228: few systems have been successfully observed in the EUV, such as e.g. VW Hyi
229: (\citet{Mauche91}; for which $N(H)\sim 6\times 10^{17}$cm$^{-2}$,
230: \citet{Polidan90}).
231: Assuming that the X-ray and EUV emissions are from the BL, and
232: the optical and UV emissions are from the disk, many previous studies
233: found a very low ratio $L_{BL}/L_{disk}$ ($\approx 0.001-0.04$,
234: \citet{Mauche91,Hoare91}) during outburst, and a ratio of
235: $L_{BL}/L_{disk}\approx 0.25$ in quiescence (e.g. for VW Hyi,
236: \citet{Belloni91} assuming that the WD contributes 50 percent
237: of the UV flux).
238: The X-ray observations of {\it underluminous} boundary layers culminated
239: with the {\it ROSAT} observations of ten cataclysmic variables
240: (BA Cam, YZ Cnc, GP Com, VW Hyi, WX Hyi, TY PsA, V3885 Sgr,
241: CY UMa, CY Vel, IX Vel)
242: by \citet{vanTeeseling94}, in which 8 systems
243: were caught in quiescence
244: and 2 systems were caught in outburst. \citet{vanTeeseling94}
245: derived X-ray fluxes from their observations and UV fluxes from
246: existing {\it IUE} observations and found that ratio of the X-ray Luminosity
247: to the UV+Optical Luminosity is much smaller than one.
248:
249: Many processes were discussed to explain the "missing
250: boundary layer" (e.g. \citet{Ferland82,Shaviv87,King97,Meyer94,Ponman95}).
251: The main idea was that the kinetic energy of the BL could also be
252: converted into winds (e.g. \citet{King84}), WD or belt rotation
253: (e.g. \citet{Kippenhahn78}),
254: heating (e.g. \citet{Shavivand87,Regev89};
255: or maybe advected into the outer stellar envelop
256: (e.g.. \citet{Godon96,Godon97,Popham97}).
257: Some systems have been observed to have a WD rotating at a rather large
258: rotational velocity, of the order of $\sim 1,000$km/sec
259: (e.g. \citet{che97,Pandel05}), which implies (from eq.3)
260: $L_{BL}/L_{disk} = 1/2$.
261: However, the observed X-ray luminosities have been much smaller than
262: this and would imply a near-Keplerian rotation rate for many systems,
263: which is not the case.
264:
265: Recent XMM-Newton observations \citep{Pandel05},
266: taking into account the contribution
267: of the disk, WD and BL in the optical, UV and X-ray bands, found no
268: evidence of an underluminous BL in 8 quiescent dwarf novae
269: (OY Car, WW Cet, AB Dra, U Gem, VW Hyi, T Leo, TY PsA, SU UMa)
270: and the
271: data are consistent with $L_{BL}\approx L_{disk}$.
272: The main difference with previous studies was that the WD contribution
273: was taken into account with realistic temperatures taken from the literature
274: and could dominate over the disk in the UV.
275: \citet{Pandel05} basically assumed $L_{disk}=L_{UV}+L_{opt}-L_{WD}$
276: and $L_{BL}=L_X$, and found $L_{BL}/L_{disk} \sim 1$ for 6 objects
277: among 8. For VW Hyi and U Gem (with $L_{BL}/L_{disk} \sim 1/3, 1/2$
278: respectively) they suggest that the inner disk is
279: truncated at $r\sim 3R_{WD}$ to explain the discrepancy.
280:
281: \citet{Godon05} further computed the contribution of
282: all the emitting components (WD, disk, BL) in the optical, UV and
283: X-ray based on the existing multiwavelength
284: observations of VW Hyi, the standard
285: disk model \citep{Pringle81}
286: and \citet{Popham99}'s model of the boundary layer.
287: \citet{Popham99} has shown that (in quiescence) the BL emits part
288: of it energy in the UV band from that region where
289: the outer edge of the BL meets (and radiates energy into)
290: the optically thick inner edge of the disk.
291: \citet{Godon05} suggested that the
292: second component often observed in the quiescent UV spectra of DNe
293: (the so-called accretion belt also detected in the {\it FUSE} spectra of
294: VW Hyi --- \citep{god04}) is the UV emission from the outer BL.
295: Taking the UV contribution of this second component
296: ($L_{2nd}$) into account \citet{Godon05} showed that
297: the luminosity of the BL of VW Hyi in quiescence
298: ($L_{BL} = L_X + L_{2nd}$) is as expected
299: from the theory (namely $L_{BL}=L_{disk} = L_{opt} + L_{UV} - L_{WD}$),
300: therefore agreeing with \citet{Pandel05} that
301: {\it there is no missing boundary layer} even for VW Hyi and without
302: disk truncation.
303:
304: On the other hand, spectroscopic UV observations
305: of accreting WDs have also reached a rather advanced stage.
306: Observations have been carried out for a range of CVs,
307: \citep[to cite just a few]{Long93,Froning01,Szkody02,Welsh03,Sion04c,Godon04}
308: in quiescence and outburst. The
309: observations suggest that the spectrum is made up of several parts,
310: specifically, the underlying WD, the accretion disk, the accretion belt
311: that might form on the surface of the star, the hot spot
312: (where the matter inflowing from the L1 point hits the outer
313: rim of the accretion disk).
314: The temperature of each of these components proves to be a most important
315: diagnostic, though as better data come in, we can hope that other
316: hydrodynamical variables also become important diagnostics. The case of
317: VW Hydri is especially important because STIS measurements \citep{Sion04c}
318: have captured that object during its transition to outburst.
319: The VW Hyi transition to outburst caught by {\it STIS} was apparently
320: an outside-in outburst. The UV lagged the optical. The
321: {\it STIS} showed that longer FUV wavelengths changed much faster
322: and manifested the flux sooner than shorter wavelengths.
323: It has been found that the temperature of the accreting star can be
324: raised by as much as 50\% during outburst. The elevated WD temperature
325: returns to its quiescence value soon after the outburst.
326: The elevated surface temperature of the star may, therefore, be a good
327: diagnostic of the energy dissipated as the accretion stream impacts,
328: spreads out and slows down and reaches co-rotation with the star.
329: The rate at which the WD cools following the heating during the
330: outburst is not only diagnostic of the mass/depth of the heated surface
331: layers but also potentially a good diagnostic of the optical depth
332: of the accreted material.
333: A detailed understanding of BL physics may
334: also help one understand the so-called UV delay where the rise in the
335: UV emission lags the rise in the optical by several hours, indicating
336: that the UV emission might be more symptomatic of the physics of the
337: BL \citep{Livio92}.
338: It is important to note here that
339: our present numerical simulations are relevant
340: to the changes in the WD temperature observed on
341: a time scale of the order of outburst/quiescence cycle (e.g. VW Hyi),
342: and do not apply to compressional heating taking place
343: on a much longer evolutionary time scale
344: \citep{Sion95,Godon02,Godon03,Piro05}.
345: \\
346:
347: \subsection{The Two-Dimensional Boundary Layer}
348:
349: The BL has typically been solved in a model where the averaging of
350: the vertical structure which is employed in accretion disk studies
351: \citep{Shakura73} has been extrapolated to the surface. One also makes
352: the additional simplifying assumption that the vertical velocity in
353: the boundary layer is zero, just as it is in the outer parts of a
354: thin disk model. This reduces the description of the BL to a one
355: dimensional problem in the radial direction \citep{Pringle81,Meyer82,
356: Popham95,Collins98b}, where one actually solves the ``slim disk''
357: equations for the BL region (namely, the one-dimensional
358: disk equations + radiation in the radial direction; as first
359: proposed in the pioneering work of \citet{Regev83}).
360: However, this assumption is not necessarily correct since the BL may
361: spread out \citep{Ferland82}. The best one-dimensional models of
362: the BL
363: \citep{Popham95} predict a rise in the BL temperature during outburst
364: that is much larger than the observed one.
365:
366: Recognizing the multi-dimensionality of the problem several authors
367: attacked the problem directly using numerical methods
368: (e.g. \citet{Robertson86})
369: culminating in simulations which included flux-limited radiative
370: transport as well as different viscosities \citep{Kley87a,Kley89a,Kley91}.
371: However, while the efforts of \citet{Kley87a,Kley89a,Kley91} were
372: very bold and impressive, the high numerical resolution needed to
373: solve the problem has until now precluded following
374: the problem numerically
375: on a long evolution time scale and or resolving the fine structure of
376: the flow.
377: The importance of the meridional flow was demonstrated
378: in protostars and FU Ori stars already by \citet{Kley96,Kley99}, who
379: show how matter spread to the poles to completely engulf the accreting
380: star. However,
381: in accretion onto a compact star the problem has not been
382: solved because of the much smaller scales.
383:
384:
385: As the numerical two-dimensional simulations failed to follow the
386: evolution of the accretion onto the surface of the compact star,
387: an analytical approach was developed by
388: \citet{Inogamov99},
389: who implemented the one-dimensional BL equations with
390: an analytic treatment of the meridional direction (assuming a shallow-water
391: equation), where the vertical
392: velocity in the boundary layer is permitted to be non-zero.
393: This analytical treatment was carried out for an
394: accreting neutron-star (NS), and it was shown that the BL could
395: ``spread'' and cover a significant part of the NS.
396: Later on, \citep{Piro04} applied Inogamov \& Sunyaev's treatment
397: of the ``spread layer'' to accreting white dwarf in outburst (the optically
398: thick case).
399: The analytical work of \cite{Piro04} indicates that the
400: numerical setup must include sufficient resolution in the radial and
401: meridional planes to capture the dynamics and fine structure of the BL.
402: It is the purpose of the present simulations to actually provide such
403: a numerical study, by following the evolution of the accretion onto
404: (and into!)
405: the surface of the WD in the inner region of the disk and close to
406: the equatorial plane ($i<30$deg).
407:
408: In the one dimensional picture the energy kept by the corotating
409: accreted material per unit time is given by eq.4. In two dimension,
410: however, as the matter possibly spreads evenly onto the WD surface,
411: its moment of inertia is rather similar to that of a spherical shell
412: and the energy kept by the accreted material per unit of time becomes
413: \begin{equation}
414: \frac{1}{2}\Omega_*^2 \frac{2}{3} \dot{m} r_*^2 = \frac{2}{3} \beta^2 L_{disk},
415: \end{equation}
416: and the remaining energy (eq.4)-(eq.6)
417: \begin{equation}
418: \frac{1}{3} \beta^2 L_{disk}
419: \end{equation}
420: is available to further spin up the WD.
421: As the matter moves toward the poles, to spread evenly on the WD surface,
422: it has excess of angular velocity/momentum (due to the
423: differential velocity) which is added to the accreting WD.
424: We therefore see that even for the most simplistic two-dimensional
425: model of the boundary layer, the spreading of matter on the WD
426: surface involves a significant fraction of the accretion energy
427: (e.g. up to 10\% of the disk luminosity for a star rotating at
428: 1,000km/sec --- eq.7).
429: In the case of a star rotating near break-up velocity, accretion
430: at the equator {\it adds angular momentum to the star as the matter
431: spreads toward higher latitudes}. This contributes to a balancing
432: effect to the spin-down of the an accreting star rotating near
433: break-up as described by \citet{pop91}.
434:
435:
436: \subsection{The Importance of the Boundary Layer}
437:
438: The structure of the BL is important to help disentangle
439: and physically characterize
440: the different emitting components in an accreting WD system.
441: The details of the emitted spectrum
442: depends sensitively on the detailed structure of the boundary
443: layer and how it changes in response
444: to enhanced rates of accretion \citep{Fisker05c}. The boundary
445: layer is also important because it may play a role in the
446: still uncertain mechanism which drives the outflowing
447: bi-polar winds seen in dwarf novae during outburst and in nova-like
448: variables during their high optical brightness states.
449: Our boundary
450: layer simulations will eventually provide the theoretical framework
451: required to physically interpet the X-ray, EUV and FUV emission
452: observed in compact binaries containing accreting degenerate stars.
453:
454:
455: However, the structure of the BL is also important because it
456: determines how the accreted matter ultimately distributes itself in
457: the envelope of the WD, which is important for classical novae.
458: In classical novae,
459: the thermonuclear runaway (TNR) ejects part or all of
460: the envelope. The initial CNO composition of the burning material
461: should be strongly enhanced compared to the accreted material to
462: account for composition of the observed ejecta \citep{Starrfield72}.
463: Several mechanisms have been suggested for this CNO enhancement (see
464: \cite{Jose05} and references therein). They can be roughly divided
465: into pre-burst mixing between accreted material and the underlying CO
466: rich WD \citep{Kippenhahn78,MacDonald83,Rosner01,Alexakis04} by accretion
467: driving instabilities or mixing with the underlying material when the
468: convective zone of the thermonuclear runaway extends deep enough to
469: dredge up CO material \citep{Starrfield72,Glasner97}.
470:
471: However, before an exploration of the radiation emission characteristics
472: of the boundary layer can be carried out with full radiation hydrodynamics,
473: we must understand the dynamical processes that lead to the formation
474: of the boundary layer.
475: In this first stage our goal is to calculate the dynamical structure of the boundary layer with sufficient numerical resolution to capture the dynamical
476: evolution of the accretion flow and its interaction with the stellar
477: surface (see \cite{Fisker05c}), and we leave the
478: radiation-hydrodynamical study for later.
479:
480: Our model is presented in section~\ref{sec:model}, the results are given
481: and discussed in section~\ref{sec:results}. Section~\ref{sec:conclusions}
482: presents the conclusions. The equations we are solving are written
483: down explicitly in Appendix~\ref{appendix:model}
484: while the initial and boundary conditions are described in
485: Appendix~\ref{appendix:domainsetup}.
486:
487:
488: \section{Computational model}\label{sec:model}
489:
490: The source of the angular momentum transport in the BL probably
491: involves a combination of magnetic fields and turbulence. Since the
492: angular velocity in the BL is not a decreasing function of radial distance
493: from the star, it is not clear that the magnetorotational instability (MRI)
494: \citep{Balbus94a} provides angular momentum transport in the BL.
495: Actually, \citet{Brandenburg96} found that the effective alpha viscosity
496: parameter $\alpha$ would go to zero for a rotation law
497: $\Omega \sim r^{-q}$ when $q<0$. Nor is the
498: MRI essential at the BL because the accretion disk can directly exchange
499: angular momentum and mass with the outer layers of the star if an efficient
500: coupling mechanism is found between the disk and the star. The source of
501: viscosity at the BL is still debated in the literature (see
502: \cite{Popham95,Piro04,Godon05}).
503: \citet{Inogamov99} assumed that the viscosity is due to purely
504: hydrodynamical turbulence as in the deceleration of subsonic or supersonic
505: flow above a solid surface. The case for a purely hydrodynamical
506: turbulence in the boundary layer was already debated earlier
507: \citep{Zahn90,Dubrulle93},
508: however the physical process is likely to take place
509: in three dimensions \citep{Orszag80}, e.g. by means of
510: streamwise vortices (e.g. \citet{Hamilton94}).
511: The most common instability in a flow over a {\it curved surface},
512: is that of the boundary layer flow over a concave surface unstable to the
513: centrifugal instability (as it violates Rayleigh's criterion for stability
514: --- \citep{Rayleigh16}). This instability leads to turbulence through the
515: formation of the G\"ortler vortices
516: \citep{Saric94}, which tap energy from the laminar flow and poor it into
517: the turbulence (this is a non-linear instability leading to a subcritical
518: transition to turbulence).
519: However, the boundary layer flow on a convex surface is not subject to this
520: instability, rather the opposite, the centrifugal force in such a
521: flow is "restoring". Therefore, the instability that is the most likely
522: to take place in the star-disk boundary layer is the Kelvin-Helmhotz
523: shearing instability which will occur for a sufficiently large shear
524: (Richardson criteria, e.g. \citet{Drazin81}).
525:
526:
527:
528: Following
529: \citep{Shakura73} , we parametrize the efficiency of the angular momentum
530: transport with an $\alpha$ viscosity prescription. In a multidimensional
531: calculation $\alpha$ has to be spatially concentrated at the accretion disk
532: and the stellar surface and the formulations developed in \citep{Papaloizou86}
533: and \citep{Kley91} are used here. Specifically, we
534: used $\nu = \alpha c_s min ( H, h_p)$ where $H$ is the scale height of the disk
535: and $h_p$ is the local pressure scale height in the boundary layer.
536: For these calculations, the pressure scale height in the boundary layer
537: that develops on the surface of the star is always smaller than the
538: scale height of the disk. The compressible Navier-Stokes
539: equations themselves are written explicitly
540: in the Appendix \ref{appendix:model}.
541: Describing the angular momentum transport with a simple shear coefficient
542: means that the dynamics follows the Navier-Stokes equations.
543: Here the Navier-Stokes equations as given by \cite{Mihalas84} are solved
544: in spherical coordinates ($r$,$\theta$,$\phi$) on an axisymmetric mesh
545: with 384 ratioed zones in the radial direction and 128 ratioed zones
546: in the meridional range spanning 0 to 30 degrees from the disk plane
547: -- same as \cite{Fisker05c}.
548: The star was taken to be a non-rotating $0.6M_\odot$ WD with a radius
549: of $9\times 10^{8} cm$ . The radial extent of our computational domain
550: extended from $8.9\times 10^{8} cm$ to $1.1\times 10^{9} cm$. The
551: r-directional zones were concentrated towards the inner radial boundary
552: with each zone being 1\% larger than the previous one. The $\theta$-directional
553: zones were also ratioed with the smallest zones being closest to the
554: equator and each zone being 1.9\% larger than the previous one.
555: Such a ratioed zoning makes it possible for us to capture five scale
556: heights of the star's atmosphere as well as the vertical structure of
557: the disk. The zone ratioing also concentrates zones around the disk-star
558: interface. Unlike \citep{Kley91} , who used a mesh with 85
559: zones in each of the radial and meridional directions, our mesh has
560: substantially better resolution. The scale heights of WD atmospheres
561: are now known to be substantially smaller than estimated in
562: \citep{Kley91,Kley87a}, making the larger meshes used in this work more
563: essential. The higher resolutions, as well as the use of higher order
564: Godunov methods, enable us to substantially reduce the role of numerical
565: diffusion.
566:
567: In this paper, we wish to study not just the spin-down of the accretion
568: disk but also the spin-up of the star. For that reason, we retained five
569: scale heights of the WD's outer atmosphere. Care was taken to resolve
570: each stellar scale height in the radial direction with at least
571: twenty zones, which ensures a
572: numerically accurate, well-resolved and stable stellar atmosphere.
573: To ensure good resolution of the disk's structure, we retained at
574: least fifty zones across a disk scale height in the meridional direction.
575: The physical conditions
576: describing the structure of the disk are given in Appendix
577: \ref{appendix:domainsetup}.
578: Typical surface temperatures for hydrodynamically accreting WDs are
579: $\sim$ 30,000K and typical disk temperatures are usually taken to
580: be $\sim$ 100,000K. This choice of temperatures would make the scale heights
581: of the disk and WD atmosphere too small to be resolved with the above-mentioned
582: resolutions. For this reason, we systematically allow the temperatures
583: of both the disk and WD atmosphere to be one order of magnitude larger
584: than the physical values. As a result, the simulations were carried
585: out with a stellar temperature of T=300,000K
586: and a disk temperature of T=1,000,000K. While this might seem be a hot
587: temperature for a disk in a CV, it is still much less than the virial
588: temperature and the corresponding disk thickness is $H/r=0.03$
589: instead of $H/r=0.01$ for a $10^5$K disk. The sound speed used in the
590: formulae for the viscosity $\nu$ was derived for the temperatures used
591: in the simulations.
592: We assume a very tenuous and very hot halo which consists of the same material
593: as the disk (and the star). The mass in this halo is extremely small, making the
594: halo dynamically unimportant and its
595: sole use is to provide pressure balance at the upper
596: boundaries of the accretion disk
597: and the star. The fluids that make up the disk, halo and stellar atmosphere were
598: tagged with passive scalars and we are, therefore, in a position to assert that
599: the halo gas simply provides pressure support with minimal mixing into the
600: gas that makes up the disk or the stellar atmosphere.
601: Such a model with a disk and halo that are in dynamical equilibrium
602: with each other was briefly described in \cite{Balsara04}.
603: Since that previous work did
604: not include the stellar atmosphere, appendix
605: \ref{appendix:domainsetup} describes
606: the physical conditions that were assumed to set up
607: the WD atmosphere, the accretion disk and the halo.
608: Symmetry across the disk plane is assumed.
609:
610: Observations, and their interpretation in the context of the
611: \cite{Popham95} model have suggested that the interesting values
612: of $\alpha$ range from 0.1 during outburst to 0.001 during quiescence.
613: \citet{Godon05} derived $\alpha=0.004$ in the boundary layer from
614: the XMM-Newton observations of VW Hyi in quiescence \citep{Pandel03}.
615: For that reason, we performed
616: a parameter study with $\alpha=0.1$, $\alpha=0.03$, $\alpha=0.01$,
617: $\alpha=0.005$, and $\alpha=0.001$ . The same value of $\alpha$ was used
618: in the boundary layer and in the disk. This would seem to be a very reasonable
619: assumption if the same physical mechanism (such as a thermal instability,
620: \cite{Cannizzo98} or magnetic instability \cite{Livio98}) were to make
621: the disk fluid and the boundary layer fluid turbulent. Our simulations
622: do not include the effect
623: of radiative transfer in the boundary layer. It is worth noting that
624: in the optically thick limit, the BL retains most of the energy that is
625: generated by the viscous stresses. As a result, by retaining the viscous
626: energy generation terms in
627: eq.~[\ref{eq:heatequation}] and assuming that the alpha viscosity parameter
628: is large ($\alpha \sim 0.1$) we mimic the situation where the
629: boundary layer is optically thick during outburst.
630: Simulations were also carried out
631: by excluding the viscous energy generation terms in
632: eq.~[\ref{eq:heatequation}] and assuming $\alpha \sim 0.001$; in
633: that situation we mimic the limit where the boundary layer is
634: optically thin during quiescence, i.e. the
635: BL radiates away all the heat that is generated by the viscous stresses.
636: Our results, therefore, bracket
637: the two extreme cases. Should the high $\alpha$
638: simulations produce BLs with optical depths in excess of $50$ or $100$ , the
639: optically thick runs would find direct applicability to the astrophysical
640: problem. Likewise, should the low $\alpha$ simulations produce optically thin
641: boundary layers, the optically thin runs would find direct applicability to the astrophysical
642: problem. We show in the course of this
643: work that such trends are indeed observed in
644: the simulations. In future work, we will include a treatment of the radiative
645: transfer terms in the BL thus obtaining results that are free of current limitations.
646: The full range of simulations that we report on here with the various values of
647: $\alpha$ and the inclusion or exclusion of viscous energy generation terms are
648: listed in Table 1.
649:
650: Since the inner radial boundary extends into the star,
651: we enforced reflective boundary conditions at that boundary. While such
652: a boundary condition would reflect any waves that reached
653: the star's surface, we find that in practice the waves do not propagate
654: to this depth in the present simulations. The use of such a reflective
655: boundary condition represents a compromise. While it might reflect back
656: waves that propagate more than five scale heights into the star, our
657: experience has shown that surface waves almost never propagate to that
658: depth. The present boundary conditions do have the positive
659: consequence that they prevent any unexpected mass or momentum
660: inflow into the computational domain from the rest
661: of the star.
662: The outer radial (computational) boundary was designed to respond to
663: inflow or outflow of fluid at the outer open boundary, and it
664: was therefore treated by imposing the boundary conditions on the
665: characteristics of the flow as described in \citep{Godon96b}.
666: Thus outer boundary conditions were imposed directly
667: on the incoming characteristics, and computed values from the variables inside
668: the domain were imposed (propagated) on the outgoing characteristics.
669: The strategy works well, especially when combined with Riemann solvers
670: which also work on the same principle of resolving the inflowing and outflowing waves. The equatorial boundary condition in the $\theta$-direction
671: was reflective (symmetric) and the boundary condition at the other
672: $\theta$-directional boundary was specified as continue.
673:
674:
675:
676: \clearpage
677: \begin{table}
678: \begin{tabular}{ccc}
679: \hline
680: model & $\alpha$ & $\Phi$ \\
681: \hline
682: v1 & 0.1 & yes \\
683: v2 & 0.03 & yes \\
684: v3 & 0.01 & yes \\
685: v4 & 0.005 & yes \\
686: v5 & 0.001 & yes \\
687: nv1 & 0.1 &no \\
688: nv2 & 0.03 &no \\
689: nv3 & 0.01 &no \\
690: nv4 & 0.005 &no \\
691: nv5 & 0.001 &no \\
692: \hline
693: \end{tabular}
694: \caption{This table shows our ten models. The first column indicates
695: the model name. The second column indicates the $\alpha$-viscosity
696: employed in the model and the last column shows whether dissipative
697: heating was included in the model.}
698: \end{table}
699:
700: \clearpage
701:
702: The spatially and temporally second order algorithms
703: in \verb+RIEMANN+ have been described in
704: \cite{Roe96,Balsara98a,Balsara98b,Balsara99a,Balsara99b,Balsara04}
705: and use many ideas from higher order WENO schemes
706: (see \cite{Jiang96,Balsara00}) to reduce dissipation.
707: The matter in the model is subject to the central gravitational field
708: of the underlying WD.
709: The model uses an ideal gas ($\gamma=5/3$) of a fully ionized
710: solar composition ($\mu=0.62\,\textrm{g/mole}$) and assumes no
711: radiative transport.
712:
713: \section{Results and Discussions}\label{sec:results}
714: In the next subsections, we focus on several important aspects
715: of the boundary layer dynamics as follows.
716: In section~\ref{subsec:den} we focus on the density profile of the boundary
717: layer.
718: In section~\ref{subsec:prs} we consider the pressure profiles in
719: the boundary layer.
720: In section~\ref{subsec:ang_pol}, we focus mainly on the the evolution of
721: angular momentum in the boundary layer .
722: In section~\ref{subsec:ricnum}
723: we check the stability of the flow in the boundary layer.
724: In section~\ref{subsec:grav}, we discuss accretion based instabilities
725: and in section~\ref{subsec:optical} we relate
726: the computations to observations.
727:
728:
729:
730: \subsection{Density Structure of the Boundary Layer}\label{subsec:den}
731:
732: Figs 1a and 1b show the logarithm of the density in
733: the boundary layer with $\alpha=0.1$ and $\alpha=0.001$ respectively.
734: Figs 1c and 1d and do the same for the runs
735: with the same values of $\alpha$ but with
736: the viscous heating switched off in eq.~[\ref{eq:heatequation}].
737: The full extent of the
738: computational domain is shown in the $\theta$-direction. Only a
739: small portion of the inner radial direction is shown and the values on
740: the x-axis of the plot are in units of $10^{9} cm$, making it possible to
741: measure the radial coordinate . The same convention for labeling figures
742: applies to all other figures in this paper where flow variables are imaged.
743:
744: %\clearpage
745: \begin{figure}
746: %\plottwo{f1a.eps}{f1b.eps}
747: \caption{
748: (a) left panel.
749: The logarithmic density for model
750: v1 is shown after one Keplerian rotation.
751: (b) right panel.
752: The logarithmic density for model
753: v5 is shown after one Keplerian rotation.
754: (c) next page - left panel.
755: The logarithmic density for model
756: nv1 is shown after one Keplerian rotation.
757: (d) next page - right panel.
758: The logarithmic density for model
759: nv5 is shown after one Keplerian rotation. The full extent of the
760: computational domain is shown in the $\theta$-direction. Only a
761: portion of the radial direction is shown and the values on
762: the x-axis of the plot are in units of $10^{9} cm$ ,
763: making it possible to measure the radial coordinate.
764: }
765: \end{figure}
766: %\clearpage
767: %{\plottwo{f1c.eps}{f1d.eps}}
768: %\clearpage
769:
770: In all cases,
771: the poloidal velocity is overlaid as vectors, enabling us to trace
772: the flow of matter on the surface of the star. Thus Fig. 1a
773: pertains to the optically thick limit (outburst state) while
774: Fig. 1d corresponds to
775: the optically thin limit (quiescent state).
776: We see that a thick, dense boundary layer
777: forms in both Figs. 1a and 1c while that is not the case in Figs. 1b and 1d.
778: Thus, physically thick boundary layers form in outburst
779: and the result is independent of whether the viscous energy
780: generation is included in the energy equation. The boundary layers
781: that form in quiescence tend to be physically thin.
782: This shows that the viscosity is the primary discriminant in determining
783: the structure of the BL. The poloidal velocity vectors in Figs 1a
784: and 1c also show us that the velocity increases with increasing
785: distance from the star's surface. The velocity vectors in
786: Figs. 1b and 1d show a similar trend though it is harder to see because
787: of the smaller physical extent of the BL. Such a velocity structure is also
788: known to occur in terrestrial fluid dynamics when a viscous fluid
789: flows over a stationary solid surface, forming a boundary layer at
790: the solid's surface. This shows us that our simulations are producing
791: a valid result which is consistent with our intuition. It also shows
792: us that our decision to refer to these star-disk layers as boundary
793: layers is well-motivated.
794:
795: The accretion rate is high enough in the $\alpha=0.1$ cases that the
796: infalling matter depresses the stellar atmosphere close to the equator.
797: While this is not so evident in Fig. 1a--1d, we will show in
798: section~\ref{subsec:grav} that the infalling matter can excite gravity
799: waves on the surface of the star. Likewise, in section~\ref{subsec:ang_pol}
800: we will show that the high $\alpha$ case spins up a significant portion
801: of the star's atmosphere while itself being spun down. The net result
802: of this process is that the toroidal velocity of
803: the disk becomes sub-Keplerian at
804: larger radii from the star as $\alpha$ increases, resulting in broader boundary
805: layers. Fig. 2 shows the mass accreted as a function of time.
806: This figure was generated by integrating the mass from
807: the disk-star interface to the top of the BL where $dv_\phi/dr\equiv 0$.
808: Fig. 2 shows us that the
809: accretion rate is higher in the high-$\alpha$ cases, as expected.
810: Fig. 2 also shows us that for the high $\alpha$ cases the accretion rate
811: seems to drop off with time. This is entirely a result of the fact that
812: this round of simulations does not include radiative transfer. As a result,
813: the base of the boundary layer heats up, with a corresponding increase
814: in pressure. The radiative cooling times in a real accretion disk are short
815: enough (i.e. even smaller than an orbital time) that the boundary
816: layer would cool down quite rapidly. The consequent pressure reduction
817: would then permit accretion to proceed unimpeded.
818:
819: %\clearpage
820: \begin{figure}
821: %\plotone{f2.eps}
822: \caption{The mass accretion is shown as a function of
823: time for models v1--v5.}
824: \end{figure}
825: %\clearpage
826:
827: Because of resolution constraints, we had to use temperatures that
828: are somewhat larger than those that prevail on accreting white dwarfs.
829: In subsequent work, we intend to overcome this constraint.
830: It is, nevertheless, interesting to relate physical
831: depth to optical depth of the
832: boundary layer. In doing that, it is important to exclude the disk fluid
833: that is spinning close to the Keplerian velocity. We, therefore, define
834: the disk-boundary layer interface as the region where the gradient of the
835: angular velocity vanishes: $\partial \Omega / \partial r = 0$ at
836: $r=r_m$.
837: In the boundary layer ($r<r_m$) the matter is
838: spinning with a toroidal velocity that
839: is smaller than the Keplerian velocity.
840: For that sub-Keplerian accreted material, we plot
841: the optical depth due to Thompson scattering as a function of meridional
842: angle in Fig. 3.
843: Here we define the optical depth as $-\int_\infty^0 \kappa
844: \rho_{disk}dr$, where $\kappa=0.34\,\textrm{g}\,\textrm{cm}^{-3}$
845: is the Thomson scattering opacity of a fully ionized solar composition.
846: This constitutes the minimum amount of scattering and thus provides
847: a lower bound of the opaqueness of the matter.
848:
849:
850:
851: The boundary layer is technically defined by the radius where the radial
852: gradient of the accretion flow's angular velocity disappears. We use
853: that definition for the rest of this paper. Because we track the fluids that
854: make the disk, halo and star, we are in a position to isolate just the
855: disk material that is within the boundary layer. The optical depth of this
856: material (in the radial direction) is very important because it sets the
857: emission characteristics of the accretion belts that have been identified
858: in the observational literature. Fig. 3 shows us that physically thick boundary
859: layers in the high-$\alpha$ limit also tend to be optically thick while
860: physically thin boundary layers in the low-$\alpha$ limit tend to be
861: optically thin. The physical implication of that is that during outburst
862: the BL could suppress the emission from a fraction of the star's surface.
863: If the BL is optically thick with an optical depth $\gg 1$ then the
864: hottest part of the boundary layer, which prevails at the disk-star interface
865: would not be visible. However, during both the transitions, from outburst
866: to quiescence and vice-versa, it is possible that this hot layer might become
867: visible, giving one a direct view of boundary layer heating.
868: This might show up as an additional UV component such as the one
869: seen by \cite{Sion04c} in VW Hydri during its transition from
870: quiescence to outburst. A similar UV component has been detected in
871: U Gem by \cite{Long93,Froning01,Szkody02} during its transition from
872: outburst to quiescence. Obtaining matched measurements of the velocity
873: and UV excesses would allow one to further corroborate the scenario
874: presented here.
875: We see that the model in Fig. 1a produces an optically thick
876: boundary layer during outburst.
877: It is, therefore, consistent with our claim in Section 2 that inclusion
878: of the viscous dissipation term in eq.~[\ref{eq:heatequation}] provides
879: a rather realistic representation of the structure of the
880: boundary layer in outburst. Likewise, the model in
881: Fig. 1d, results in an optically thin boundary layer during quiescence.
882: It is thus consistent with our claim that excluding the viscous dissipation term in eq.~[\ref{eq:heatequation}] is similarly more representative of
883: the structure of the boundary layer in quiescence.
884:
885:
886: %\clearpage
887:
888: \begin{figure}
889: %\plotone{f3.eps}
890: \caption{The optical depth along the radial
891: direction is shown for models v1--v5 as a function of
892: the angle from the equatorial plane.}
893: \end{figure}
894:
895:
896: %\clearpage
897:
898:
899: \subsection{Pressure in the Boundary Layer}\label{subsec:prs}
900:
901: Figs 4a and 4b show the logarithm of the pressure in the boundary
902: layer with $\alpha=0.1$ and $\alpha=0.001$ respectively.
903: Figs 4c and 4d do the same for the runs
904: with the same values of $\alpha$ but with
905: the viscous heating switched off in eq.~[\ref{eq:heatequation}].
906:
907: %\clearpage
908: \begin{figure}
909: %\plottwo{f4a.eps}{f4b.eps}
910: \caption{
911: (a) left panel.
912: This figure shows the logarithmic pressure for model v1 after
913: one Keplerian rotation.
914: (b) right panel.
915: This figure shows the logarithmic pressure for model v5 after
916: one Keplerian rotation.
917: (c) next page - left panel.
918: This figure shows the logarithmic pressure for model nv1 after
919: one Keplerian rotation.
920: (d) next page - right panel.
921: This figure shows the logarithmic pressure for model nv5 after
922: one Keplerian rotation.
923: }
924: \end{figure}
925:
926: %\clearpage
927:
928: %{\plottwo{f4c.eps}{f4d.eps}}
929:
930: %\clearpage
931:
932: From Figs. 4a and 4b we see that the pressure of the accreted fluid
933: is highest in the equatorial
934: plane. This high pressure can be attributed to a combination of viscous
935: dissipation as well as the ram pressure due to infall of accreting material.
936: The pressure gradient in the meridional direction on the surface of the
937: star, therefore, drives the meridional flow that develops on the WD's surface.
938: The models that were used in Figs. 4c and 4d did not include viscous heating
939: in eq.~[\ref{eq:heatequation}]. They, nevertheless, show that the pressure
940: of the accreted fluid is highest in the equatorial plane and in this
941: instance, the increased pressure is entirely attributable to the
942: ram pressure due to infall of accreting material. We also see that the
943: equatorial pressure increase in Fig. 4b is less than
944: that in Fig. 4a and, similarly, for Figs. 4b and 4d.
945: The smaller pressures in Figs. 4c and
946: 4d relative to Figs. 4a and 4b can be attributed to the
947: additional contribution from viscous heating.
948: Fig. 4a--4d, therefore, serves
949: to show us that the formation of the pressure gradient in the meridional
950: direction, which then drives the meridional flow in the boundary layer,
951: is a very commonplace phenomenon. In other words, any flow that is put
952: in a similar geometry and is made to experience similar viscous stresses would
953: naturally form a similar boundary layer, showing the ubiquity of boundary
954: layer formation. (The only other requirement that such a BL flow has to satisfy
955: as the disk material migrates polewards on the star's surface is the ability to
956: lose angular momentum. The next sub-section will show that it can accomplish this
957: very efficiently by spinning up the underlying layers of the stellar atmosphere.)
958: Thus the decision by \cite{Inogamov99,Piro04} to include
959: a non-zero meridional velocity in their models was of central importance
960: in forming the inherently multi-dimensional boundary layers that we
961: see in our simulations. Such boundary layers
962: also form in accreting neutron stars, and TTauri stars that are
963: going through the FU Orionis phenomenon
964: \citet{Kley96,Kley99,Balsara05}.
965:
966: \subsection{Toroidal and Poloidal Velocities}\label{subsec:ang_pol}
967:
968: As the simulations start, the shear force transfers angular momentum
969: between differentially rotating disk annuli so that matter can move
970: inwards and accrete on the star at the footpoint of the disk.
971: Angular momentum transfer between the disk material and the stellar
972: surface is also necessary so matter can move to higher latitudes.
973: Otherwise, the centrifugal barrier prevents matter from leaving the
974: footpoint of the disk. It is this transfer of angular momentum which
975: spins up the existing surface layers of the star.
976: Angular momentum is also directly advected onto the star, because the
977: orbiting disk material eventually accretes and forms the new surface.
978:
979: Fig.5a shows the resulting specific angular momentum after 3/4 of
980: a Keplerian evolution for the $\alpha=0.1$ case, whereas
981: fig.5b shows it for the $\alpha=0.001$ case.
982: Fig.5a shows that the halo is spun up as is the underlying
983: star at the footpoint, where the disk connects with the star.
984: Moreover, we observe that by this
985: time in the simulation the disk material has spun
986: up all five atmospheric scale heights
987: of the star at equatorial latitudes. At higher
988: latitudes the spreading disk material
989: is less dense and thus takes longer amounts of time
990: to drag the star's surface around with it.
991: In fig.5b, which is at the same time as fig.5a and uses
992: the same color scale, we see that the disk has not spun up
993: the star's atmosphere even at the stellar equator.
994:
995: %\clearpage
996:
997: \begin{figure}
998: %\plottwo{f5a.eps}{f5b.eps}
999: \caption{
1000: (a) left panel.
1001: A color plot of the specific angular momentum,
1002: $\Omega=v_\phi/r$, for the $\alpha=0.1$ case after 3/4 of
1003: a Keplerian rotation period.
1004: (b) right panel.
1005: A color plot of the specific angular momentum for
1006: the $\alpha=0.001$ case after 3/4 of a Keplerian rotation
1007: period.
1008: }
1009: \end{figure}
1010:
1011: %\clearpage
1012:
1013: Fig.6 shows the toroidal velocity and sound speed in
1014: the disk's midplane as a function of radius for the simulated
1015: alpha-viscosities.
1016: The cases of $\alpha=0.01$, $\alpha=0.005$, and $\alpha=0.001$ do not
1017: show any significant spin-up of the star after 3/4 of a Keplerian
1018: rotation, whence there is not significant shear connection between the
1019: disk and the stellar surface. Therefore disk matter retains its Keplerian
1020: motion much closer to the star which means that even though the accretion
1021: rates are smaller, significant amounts of the dissipation can still be
1022: generated in the BL close to the star.
1023:
1024: %\clearpage
1025: \begin{figure}
1026: %\plotone{f6.eps}
1027: \caption{
1028: Toroidal ($\phi$) velocity in the disk's midplane
1029: shows the extent of the boundary layer as well
1030: as the sound speed after 3/4 of a Keplerian rotation period.
1031: Note that $v_\phi$ asymptotes to a higher value than
1032: $c_{s,\alpha}$.}
1033: \end{figure}
1034: %\clearpage
1035:
1036: The matter in the layers of the star that we simulate is indeed
1037: non-degenerate. Even so, the high temperatures cause it to have
1038: a rather low specific molecular viscosity. As a result, turbulence and/or
1039: threaded magnetic fields might nevertheless be the two most prominent ways to
1040: transfer a significant amount of angular momentum to the deeper
1041: layers of the WD \citep{Durisen73a}, where the relative motion
1042: drives the mixing between the accreted surface material and the
1043: deeper layers.
1044: A supersonic component in the toroidal direction over most of the
1045: BL is obtained for all values of $\alpha$ (see fig.6).
1046: The supersonic toroidal velocities mean
1047: that the flows are susceptible to the rapid
1048: development of gravity wave and/or Kelvin-Helmholtz
1049: instabilities in three dimensions.
1050: Such instabilities can be studied by taking a small slice of the interface
1051: from our simulations and extending it in the toroidal direction
1052: in a three dimensional
1053: simulation. Our present simulations could then provide
1054: the velocities which drive
1055: such instabilities. The calculation of the turbulent
1056: mixing which obtains from these
1057: instabilities must therefore be calculated using other
1058: models \citep{Kippenhahn78,MacDonald83,Rosner01,Alexakis04}.
1059:
1060:
1061:
1062: Fig. 7a shows the poloidal Mach number of the accreted material
1063: on the surface of the star for $\alpha=0.1$. Fig. 7b shows the same
1064: for $\alpha=0.001$. The same scale is used for both figures. We see that
1065: the Mach number is mildly transonic in Fig. 7a while it is subsonic
1066: in Fig. 7b.
1067: Only the $\alpha=0.1$ case has a transonic component in the poloidal
1068: directions, whereas the poloidal flows for $\alpha=0.03$, $\alpha=0.01$,
1069: $\alpha=0.005$, and $\alpha=0.001$ remain subsonic (see figs. 5a
1070: and 5b) \citep{Fisker05c}.
1071:
1072:
1073: %\clearpage
1074:
1075: \begin{figure}
1076: %\plottwo{f7a.eps}{f7b.eps}
1077: \caption{
1078: (a) Left panel.
1079: This figure shows the poloidal Mach number with poloidal
1080: velocity vectors overlaid for model v1 after one Keplerian rotation.
1081: (b) Right panel.
1082: This figure shows the poloidal Mach number with poloidal
1083: velocity vectors overlaid for model v5 after one Keplerian rotation. }
1084: \end{figure}
1085:
1086: %\clearpage
1087:
1088:
1089: \subsection{Hydrodynamic Instability in the Boundary Layer}
1090: \label{subsec:ricnum}
1091:
1092: The present simulations are based on an alpha-viscosity formulation,
1093: which implicitly assumes an underlying model for the sub-scale turbulence.
1094: However, the material that accretes on to a white dwarf has very
1095: low viscosity. Observations have not revealed the existence
1096: of magnetic fields at the WD surface. Besides, the radial gradient
1097: of the angular velocity has the wrong sign for the MRI to act.
1098: It is, therefore, interesting to explore the role of other hydrodynamical
1099: instabilities at the surface of the star. It is important to remember
1100: that the alpha-viscosity formulation does smear the velocity gradients.
1101: Yet, if these gradients are amenable to the development of a persistent
1102: hydrodynamical instability then the simulation should reveal its existence
1103: to us. It is in that spirit that we try to identify an instability
1104: that might give rise to a sub-scale hydrodynamical turbulence on
1105: an accreting white dwarf's surface.
1106:
1107: There are several instabilities
1108: that can possibly lead to turbulence in boundary layers, such as
1109: the centrifugal instability (Rayleigh's criterion),
1110: the shearing Kelvin-Helmholtz instability, or even the
1111: Rayleigh-Taylor instability (buoyancy forces). Depending on the
1112: exact angular velocity profile, pressure and density gradients
1113: each of the forces involved compete, some are stabilizing
1114: while other are destabilizing.
1115: The sufficient condition for linear stability of
1116: a rotating, radially stratified fluid under
1117: the influence of (a radial) gravity was given by \citet{Sung1974}
1118: \begin{equation}
1119: \frac{1}{\rho} g_{eff} S_{\varpi} +
1120: \frac{k_z^2}{M} \Phi - \frac{1}{4} \frac{m^2}{M}
1121: \left( \frac{d \Omega}{d \varpi } \right)^2 \ge 0 ,
1122: \end{equation}
1123: where, $m$ and $k_z$ are the angular and vertical mode numbers
1124: (in cylindrical coordinates $(\varpi, \phi, z)$),
1125: $M$ is defined as $M= k_z^2 + m^2/\varpi^2$,
1126: $g_{eff}$ is the effective gravity defined as
1127: $$
1128: g_{eff} = \frac{GM}{\varpi^2}- \varpi \Omega^2 ;
1129: $$
1130: $S_{\varpi}$ is the Schwarzschild discriminant \citep{Schwarzschild1906}
1131: $$
1132: S_{\varpi} = \left( \frac{d \rho}{d\varpi} \right)_{ad}
1133: - \left( \frac{d \rho}{d\varpi} \right) ,
1134: $$
1135: in which
1136: $$
1137: \left( \frac{ d \rho}{d\varpi} \right)_{ad} = \frac{\rho}{\Gamma_1 p}
1138: \frac{dp}{d\varpi}
1139: $$
1140: is the adiabatic density gradient;
1141: and $\Phi$ is the Rayleigh discriminant \citep{Rayleigh1880,Rayleigh16}
1142: $$
1143: \Phi = \frac{1}{\varpi^3} \frac{d}{d\varpi} \left(
1144: \varpi^2 \Omega \right)^2 .
1145: $$
1146: In the present case we do not consider vertical modes $k_z$ and
1147: only consider the equatorial flow for which $r=\varpi$.
1148: There are two distinct cases as follows.
1149:
1150: The axi-symmetric
1151: case (mode $m=0$) gives the Solberg-H\o iland criterion
1152: \citep{Sung1974,Sung1975}
1153: $$
1154: \frac{1}{\rho} g_{eff} S_{r} + \Phi \ge 0.
1155: $$
1156: The perturbations of an accretion disk to a convective
1157: instability has already been studied in this context
1158: \citep{Livio1977,Rudiger2002,Johnson2006}.
1159: In the limit of vanishing rotational velocity, the Solberg-H\o iland
1160: criterion simply leads to the classical Schwarzschild condition
1161: $S_r > 0$
1162: \citep{Lebovitz1965,Lebovitz1966}, which is also the condition
1163: in the vertical direction in the disk \citep{Livio1977}.
1164: In the limit of a vanishing gradient of the density,
1165: the Solberg-H\o iland criterion
1166: simplifies to Rayleigh's criterion for rotating fluids.
1167: The axi-symmetric instability in the boundary layer
1168: could possibly lead to a ring-like
1169: structure oscillating and in many ways similar to the ``breathing''
1170: mode of an unstable star. This mode will not lead to turbulence,
1171: but could it explain the short period oscillations observed in CVs?
1172: The mode would depend on the density, temperature and rotation rate
1173: (Solberg-H\o iland criterion), and
1174: its period and coherence would vary with these parameters.
1175: However, dwarf nova oscillations (DNOs) in CVs
1176: exhibit a 180deg phase jump at eclipse \citep{pat79},
1177: and frequency doubling (presence of the first harmonic
1178: \citet{pat81}, and therefore they
1179: cannot be explained by an $m=0$ mode.
1180: On the other hand, Quasi-periodic oscillations (QPOs) in CVs
1181: do not all exhibit the same
1182: characteristics (some are believed to form at larger radii in the
1183: disk), and, therefore,
1184: we cannot completely rule out the $m=0$ mode
1185: in the boundary layer as an additional
1186: mechanism to produce quasi-periodic oscillations (QPOs).
1187:
1188: We now turn our attention to the non-axisymmetric case.
1189: In the case $m\ne 0$, a sufficient condition for stability that does
1190: not involve $k_z$ is \citep{Sung1974}
1191: \begin{equation}
1192: \frac{1}{\rho} g_{eff} S_r -\frac{1}{4} r^2
1193: \left( \frac{d \Omega}{dr} \right)^2 \ge 0.
1194: \end{equation}
1195: This condition is usually written in the form of a modified
1196: Richardson number
1197: \begin{equation}
1198: R_i =
1199: g_{eff}
1200: \left(
1201: \frac{1}{\Gamma_1 p} \frac{dp}{dr} - \frac{1}{\rho}\frac{d \rho}{dr}
1202: \right)
1203: \left( r \frac{d \Omega}{dr} \right)^{-2}
1204: \ge \frac{1}{4}.
1205: \end{equation}
1206: This is a generalization of the Miles-Howard theorem
1207: (\citet{Miles1961,Howard1961}; see also \citet{Chimonas1970}).
1208: The Miles-Howard theorem itself reduces to the original Richardson
1209: criterion \citep{Richardson1920}
1210: when compressibility is omitted and the Schwarzschild
1211: discriminant is replaced by the buoyancy term alone.
1212: The modified Richardson number was considered in a few analytical
1213: studies to analyze the stability of material accreting on the surface
1214: of a white dwarf \citep{Durisen1977,Kippenhahn78,Livio1987}.
1215: Here, we have the opportunity to use results from numerical simulations
1216: to evaluate the modified Richardson number in the boundary layer.
1217: Using the definition of the Ledoux discriminant \citep{Ledoux1958},
1218: $$
1219: A = \frac{1}{\Gamma_1} \frac{d~ln~P}{dr} - \frac{d~ln~\rho}{dr},
1220: $$
1221: the (modified) Richardson condition for stability is then written
1222: \begin{equation}
1223: R_i = N^2_{BV} \left( r \frac{ d\Omega}{dr} \right)^{-2} \ge
1224: \frac{1}{4} ,
1225: \end{equation}
1226: where $A$ is related to the buoyancy (or Brunt-V\"ais\"al\"a)
1227: frequency $N_{BV}$
1228: \citep[e.g.]{Pesnell1986,Livio1987}
1229: $$
1230: N_{BV}^2 = g_{eff}A.
1231: $$
1232: The the flow is unstable for $R_i < 1/4$.
1233: In Fig.8 we show the Richardson number,
1234: $R_i$ in the $\phi$-direction (eqs. 10, 11), for run 91 and 95.
1235: The Richardson number is smaller than 1/4 in most of the domain
1236: and therefore the flow is unstable. It is important to remark
1237: that while the shear could stabilize the flow \citep{Johnson2006},
1238: internal gravity waves can be reflected from a shear layer \citep{Van1982},
1239: and can be over-reflected from a rigid boundary \citep{Sachdev1982}.
1240: It has also been shown that modes can be unstable at the star-disk
1241: interface due to the propagation through the corotation \citep{tsa09}.
1242: Unstable modes could be trapped and grow between (i) the
1243: surface of the WD and the strong shear region; and/or (ii) between the
1244: strong shear region and the corotation radius at larger radii in the
1245: inner disk, in a manner similar to the Papaloizou-Pringle instability
1246: \citep{papa84,papa85,papa87} observed in simulations of disks with a
1247: rigid inner boundary \citep{Godon1997}.
1248: The question of how the instability
1249: will develop cannot be addressed without full 3D simulations, which
1250: are beyond the scope of this work.
1251: However, a Richardson number $<1/4$
1252: in the boundary layer region and {\it at} the stellar surface raises
1253: the possibility of an instability. This instability
1254: could develop in the form of waves in the ``spread layer'' as studied by
1255: \citet{pir04b} to explain DNOs in CVs, and raises the possibility of
1256: strong mixing between the hydrogen rich freshly
1257: accreted material and the WD material.
1258:
1259:
1260:
1261: %\clearpage
1262:
1263: \begin{figure}
1264: %\plotone{f8.ps}
1265: \caption{
1266: The toroidal Richardson Number $R_{\phi}$ as a function of the
1267: radius $r$ in the equatorial plane ($\theta = 90^{\circ}$), for
1268: run 91 and 95. The stellar radius is located at $r=9 \times 10^8$cm.
1269: In most of the domain the Richardson number is smaller than 1/4.
1270: }
1271: \end{figure}
1272:
1273: %\clearpage
1274:
1275:
1276: \subsection{Formation of surface gravity waves}\label{subsec:grav}
1277:
1278: For the $\alpha=0.1$ case, the outer edge of the boundary layer
1279: (where $d\Omega/dr\sim 0$) is located at $1.06r_*$.
1280: At $r=1.03r_*$, $v_\phi$ is about 92\% of $v_K$. This means that
1281: only 84\% of the gravitational pull is supported by the centrifugal
1282: force while the rest is supported by pressure.
1283: The pressure for $\alpha=0.1$ is illustrated in fig.4a and comes from the
1284: build up of density due to the high accretion rate facilitated by the high
1285: viscosity which drags down material from the disk and also keeps it
1286: from moving rapidly towards the poles once it makes contact with the WD
1287: surface.
1288: This results in a dense band at the foot point of the disk which
1289: causes a surface gravity wave of surface matter to spread towards the poles.
1290: Such surface gravity waves have been studied in the context of terrestrial
1291: physics by \cite{Miles57} and the role of turbulence in exciting these
1292: waves has also been studied \cite{Phillips57} and the simulations
1293: show a similar phenomenon on the surface of strongly accreting
1294: white dwarfs.
1295: The matter inflow is, however, sufficiently large to cause the disk
1296: material to overflow the gravity wave supersonically as described
1297: above and shown in fig.5a. Our present simulations do not show any
1298: evidence of wave breaking. However, should wave breaking reveal itself
1299: in simulations where the accretion is more vigorous, it would provide
1300: a further mechanism for directly mixing accreted material with the
1301: CNO-rich material that lies beneath the surface of the white dwarf.
1302:
1303: The propensity for the creation of gravity waves and their magnitude
1304: decreases rapidly with lower values of $\alpha$. Even for $\alpha=0.03$,
1305: the effect is only marginal and for lower values it is no longer present.
1306: The effect is thus only present for large values of $\alpha$, and
1307: possibly during dwarf novae in the outburst stage.
1308: Furthermore the gravity wave might be transient as the surface adjusts to
1309: a fluctuations in the accretion rate. For lower values of $\alpha$, the
1310: matter accretes slowly inwards and spreads towards the poles in a uniformly
1311: thick layer.
1312:
1313:
1314: \subsection{Dissipative heating of the BL}\label{subsec:optical}
1315:
1316: The local rate of energy dissipation depends on the divergence of the angular velocity \citep{Mihalas84}, so fig.6 also indirectly shows where heat
1317: is released.
1318: Fig.9a shows that the supersonic impact of the accretion flow dissipates
1319: the most energy. However, energy is also dissipated at the interface
1320: between the BL and the surface. This is better seen in
1321: Fig.9b which shows that the toroidally rotating
1322: matter dissipates rotational kinetic energy along its entire
1323: interface with the surface as it slides against it.
1324:
1325: In situations where the BL is optically thin, the radiation
1326: will be emitted from the very same regions where the heat is generated
1327: in fig. 6b. This may explain
1328: the optically thin BL and line emission seen in observations of
1329: U Gem during the quiescent phase (see also: \citet{Fisker05c}).
1330: Even in outburst, Fig. 3 shows that the optically thick boundary layer
1331: has an optical depth of ~ 100. We therefore expect that the radiative
1332: diffusion of the heat in the radial direction will be faster than its
1333: advection in the poloidal direction. Consequently, our simulations
1334: provide a dynamically consistent reason for the formation of a warm
1335: belt of accreted matter, as was suggested in the observations of
1336: \citet{Sion04c} and the modeling of \citet{Godon05}.
1337:
1338: %\clearpage
1339:
1340: \begin{figure}
1341: %\plottwo{f9a.eps}{f9b.eps}
1342: \caption{
1343: (a) Left panel.
1344: A logarithmic color plot of the dissipation rate for
1345: the $\alpha=0.1$ case after 3/4 of a Keplerian rotation period.
1346: (b) Right panel.
1347: A logarithmic color plot of the dissipation rate for the
1348: $\alpha=0.001$ case after 3/4 of a Keplerian rotation period.
1349: Notice that the color scale is different for fig. 8a.
1350: }
1351: \end{figure}
1352:
1353: %\clearpage
1354:
1355:
1356: \section{Summary and Conclusion}\label{sec:conclusions}
1357:
1358: We have numerically simulated the structural dynamics of the BL for an
1359: accreting white dwarf surrounded by an $\alpha$-disk for different
1360: values of $\alpha$. The structure and dynamics of the BL is important,
1361: because it determines the specifics of the radiated spectrum that is
1362: emitted from this region. The energy radiated from the
1363: BL could account for up to half of the total energy released in the
1364: accretion process.
1365: For high values of $\alpha$, the BL is optically thick and extends more
1366: than 30 degrees to either side of the disk plane after 3/4 of a
1367: Keplerian rotation period ($t_K$=19s).
1368:
1369: The simulations also show that high values of $\alpha$ result in a
1370: spreading BL which sets off gravity waves in the surface matter.
1371: The accretion rate can be high enough
1372: in the high $\alpha$ cases to cause a depression to form on the surface
1373: of the star. The accretion flow moves supersonically
1374: over the depression that is formed, making it susceptible to the rapid
1375: development of gravity wave and/or Kelvin-Helmholtz instabilities.
1376: The low viscosity cases also show a spreading BL, but here the accretion
1377: flow does not set off gravity waves and it is optically thin.
1378:
1379: We argued in the paper by \citet{Sion04c}
1380: that an accretion belt might be sustained after a
1381: long period of perhaps thousands of dwarf nova events such that
1382: an equilibrium or steady state is established between the Richardson
1383: number and the average rate of accretion. Our 2D simulations in
1384: this paper has not been followed long enough for any steady state
1385: behavior to be observed. Therefore, it may be premature for us
1386: on the basis of the limited extent of these first simulations to
1387: argue in favor or against the formation of an accretion belt.
1388:
1389: If the boundary
1390: layer remains optically thick following an episode of high accretion,
1391: this could explain the "second components" of FUV flux seen in several
1392: dwarf novae during quiescence. It should also be pointed out that
1393: a hot equatorial accretion region could persist even after the
1394: material has reached co-rotation with the white dwarf. For example,
1395: a hot (50,000K) region of low rotation is seen in successive HUT
1396: spectra of U Gem taken early after an outburst and very late after
1397: the same outburst.
1398:
1399:
1400: The susceptibility of the flow in the boundary layer to undergo a
1401: purely hydrodynamic instability (modified Richardson number $<1/4$)
1402: was found assuming an {\it a priori}
1403: alpha viscosity parametrization consistent with that of the disk.
1404: The unstable flow would
1405: be very effective in mixing of the accreted material with the outer
1406: layer (few scale heights) of the WD envelope.
1407: However, it has been shown \citep{papa94,nara94,kato94,god95}
1408: that the turbulent viscosity law for
1409: a non-Keplerian disk cannot be represented by the standard alpha
1410: viscosity prescription ($\alpha=$constant), as alpha is proportional
1411: to the shear \citep{god95,abra96}
1412: $$
1413: \alpha \propto \partial \Omega / \partial r.
1414: $$
1415: Even more so, in the boundary layer,
1416: the turbulent viscosity is proportional to the
1417: square of the turbulent Mach number ${\cal{M}}_t^2$
1418: \citep{shak88,god95}, which decreases quickly for for subsonic turbulence
1419: (supersonic turbulence quickly dissipated due to shocks)
1420: $$
1421: \nu_{BL} = \alpha_{BL} c_s H, ~~~~~~~~
1422: \alpha_{BL}= \alpha {\cal{M}}_t^2 \frac{\delta_{BL}}{H}
1423: \frac{c_s}{v_K-v_*} .
1424: $$
1425: Supersonic radial infall in the boundary layer also leads to a smaller
1426: turbulent viscosity as shown by \citep{papa94,nara94,kato94,god95},
1427: and the formulation of such a viscosity can also be obtained
1428: considering causality in one-dimensional steady-state models
1429: of accretion disk boundary layers \citep{poph92}, namely the
1430: radial infall viscosity cannot be larger than the maximum speed
1431: of the turbulence.
1432: In that case one has \citep{god95}
1433: $$
1434: \nu_{BL} = \alpha_{BL} c_s H, ~~~~~~~~~~~
1435: \alpha_{BL} = \alpha \frac{\delta_{BL}}{H}
1436: \frac{1}{{\cal{M}}^3_r},
1437: $$
1438: where ${\cal{M}}_r > 1$ is the radial infall Mach number, and
1439: it is smaller than the turbulent Mach number ${\cal{M}}_t$
1440: (see also \citet{poph92}).
1441:
1442: In these regions close to the
1443: stellar surface (and even more so away from the equatorial region),
1444: one can use a viscosity of the form
1445: \begin{equation}
1446: \nu \propto (v-v_{*}) z ,
1447: \end{equation}
1448: where $v_*$ is the stellar rotational velocity, $v$ is the velocity
1449: of the flow,
1450: and $z$ is the distance from the surface of the WD. It is clear
1451: that such a viscosity would decrease rapidly
1452: as $z \rightarrow 0$, and would agree with the fact that the size
1453: of the turbulent Eddies be limited by the presence of the stellar
1454: surface and their velocity proportional to the change of $v$ over
1455: a distance $z$ (see also \citet{Landau87}).
1456:
1457:
1458: Piro and Bildsten have shown that the properties of the boundary
1459: (``spread'') layer vary depending on the value of the viscosity,
1460: however, like us, they assume an alpha viscosity in which alpha is not a
1461: function of the coordinates, the radial infall velocity or
1462: or the shear.
1463: It is our aim, in future simulations, to carry out simulation of
1464: the boundary layer, by assuming
1465: a viscosity law that is greatly reduced in the boundary layer.
1466: For this purpose,
1467: one could use a combination of viscosity prescriptions suggested
1468: in the above mentioned works. Since the results presented here
1469: as well as Piro \& Bildsten's results depend strongly on the value
1470: of $\alpha$ (which was assumed to be constant), it is clear that
1471: a modified alpha viscosity prescription can lead to some new
1472: and unexpected results.
1473:
1474:
1475: The evolutionary simulations presented here, while followed for only a
1476: brief interval of time, represent the first successful, fully hydrodynamic
1477: treatment of the the flow of accreted matter {\it into}
1478: the white dwarf surface
1479: layers. All previous attempts to follow accretion hydrodynamically from
1480: the boundary layer into the white dwarf surface layers were either not
1481: computationally viable or the underlying white dwarf was treated as a
1482: solid boundary.
1483:
1484: The successful convergence of our dynamical model simulations opens the
1485: door to going well beyond this first stage and ultimately follow the
1486: dynamical evolution over much longer time intervals for both the high
1487: viscosity and low viscosity cases with the inclusion of radiative processes.
1488:
1489: \acknowledgements
1490: PG wishes to thank Steve Lubow for a discussion on the importance of the
1491: adiabatic term in the modified Richardson number, and
1492: Mario Livio for his kind hospitality at
1493: the Space Telescope Science Institute.
1494: This work is supported by NASA ATFP grant NNX08AG69G to Villanova University
1495: and the University of Notre Dame.
1496: Participation by EMS and PG was also supported in part by NSF grant
1497: AST08-07892 and NASA ADP grant NNX04GE78G to villanova University.
1498:
1499:
1500: \appendix
1501: \section{Numerical model}\label{appendix:model}
1502: The source of the angular momentum transport in the BL is a combination of magnetic fields and turbulence. However, an a priori prescription, in particular in the BL, is a source of disagreement (see \cite{Popham95} and references therein). Instead the efficiency of the angular momentum transport can be parametrized with a coefficient, $\alpha$ \citep{Shakura73}.
1503: Describing the angular momentum transport with a simple shear coefficient
1504: means that the dynamics follows the Navier-Stokes equations.
1505: Here the Navier-Stokes equations are solved in spherical coordinates $i=(r,\theta,\phi)$ as given by \cite{Mihalas84}. The accelerations are defined by the time-derivative of the velocities which are given in spherical coordinates.
1506: \begin{eqnarray}
1507: \rho a_r & = & -\frac{GM\rho}{r}-\frac{\partial p}{\partial r}+\frac{\partial }{\partial r}\left[2\mu \frac{\partial v_r}{\partial r}+\left(\zeta-{2 \over 3}\mu\right)(\nabla\cdot v)\right] + {1\over r}\frac{\partial }{\partial \theta}\left\{\mu\left[r\frac{\partial}{\partial r}\left({v_\theta \over r}\right)+{1 \over r}\frac{\partial v_r}{\partial \theta}\right]\right\} \nonumber \\
1508: &+&{1 \over r \sin \theta} \frac{\partial }{\partial \phi} \left\{\mu\left[{1 \over r \sin \theta}\frac{\partial v_r}{\partial \phi} + r \frac{\partial }{\partial r}\left({v \over v_\phi}\right)\right]\right\}\nonumber \\
1509: &+&{\mu \over r}\left[4r\frac{\partial }{\partial r}\left({v_r \over r}\right)-{2\over r \sin \theta}\frac{\partial }{\partial \theta} (v_\theta \sin \theta ) {2 \over r \sin \theta} \frac{\partial v_\phi}{\partial \phi}+r\cot\theta\frac{\partial }{\partial r}\left({v_\theta \over r}\right)+{\cot \theta \over r}\frac{\partial v_r}{\partial \theta}\right]
1510: \end{eqnarray}
1511: \begin{eqnarray}
1512: \rho a_\theta &=& -{1 \over r}\frac{\partial p}{\partial \theta}+\frac{\partial }{\partial r}\left\{\mu\left[r\frac{\partial }{\partial r}\left({v_\theta \over r}\right)+{1 \over r}\frac{\partial v_r}{\partial \theta}\right]\right\}+{1 \over r}\frac{\partial }{\partial \theta}\left[{2\mu \over r}\left(\frac{\partial v_\theta}{\partial \theta}+v_r\right)+\left(\zeta-{2 \over 3}\mu\right)(\nabla\cdot v)\right]\nonumber \\
1513: &+&{1 \over r \sin\theta}\frac{\partial }{\partial \phi}\left\{\mu\left[{\sin \theta \over r}\frac{\partial }{\partial \theta} \left({v_\theta \over \sin \theta}\right)+{1 \over r \sin \theta}\frac{\partial v_\phi}{\partial \phi}\right]\right\}+{\mu \over r}\left\{{2 \cot \theta \over r}\left[\sin \theta \frac{\partial }{\partial \theta} \left({v_\theta \over \sin \theta}\right)-{1 \over \sin\theta}\frac{\partial v_\phi}{\partial \phi}\right]\right.\nonumber \\
1514: &+&\left.3r\frac{\partial }{\partial r}\left({v_\theta\over r}\right)+{3 \over r}\frac{\partial v_r}{\partial \theta}\right\}
1515: \end{eqnarray}
1516: \begin{eqnarray}
1517: \rho a_\phi &=& - {1 \over r \sin \theta} \frac{\partial p}{\partial \phi}+\frac{\partial }{\partial r}\left\{\mu\left[{1 \over r \sin \theta}\frac{\partial v_r}{\partial \phi}+r\frac{\partial }{\partial r}\left({v_\phi \over r}\right)\right]\right\}+{1 \over r}\frac{\partial }{\partial \theta}\left\{\mu\left[{\sin \theta \over r}\frac{\partial }{\partial \theta}\left({v_\phi \over \sin \theta}\right)+{1 \over r\sin\theta}\frac{\partial v_\theta}{\partial \phi}\right]\right\}\nonumber \\
1518: &+&{1 \over r\sin\theta}\frac{\partial }{\partial \phi}\left[{2\mu\over r}\left({1 \over \sin\theta}\frac{\partial v_\phi}{\partial \phi}+v_R+V_\theta\cot\theta\right)+\left(\zeta-{2 \over 3}\mu\right)(\nabla\cdot v)\right]\nonumber \\
1519: &+&{\mu \over r}\left\{{3 \over r \sin \theta}\frac{\partial v_r}{\partial \phi}+3r\frac{\partial }{\partial r}\left({v_\phi \over r}\right)+2\cot\theta\left[{\sin\theta\over r}\frac{\partial }{\partial \theta}\left({v_\theta\over\sin\theta}\right)+{1 \over r \sin \theta}\frac{\partial v_\theta}{\partial \phi}\right]\right\}
1520: \end{eqnarray}
1521: Here $a_i$ are the accelations, $p$ is the pressure, and $\zeta$ is the coefficient of bulk viscosity, which we set to zero to treat the fluid as a Maxwellian fluid. The kinematic viscosity $\nu=\mu/\rho$ is set by the alpha-disk prescription of \cite{Shakura73}, so $\nu=\alpha c_sH$, where $H$ is the disk scale height which should properly correspond to the turbulent turnover scale with convective bubbles moving at sound speed $c_s$.
1522:
1523:
1524: The viscous dissipation function is given by
1525: \begin{eqnarray}
1526: \Phi & = & 2\mu\left\{\left(\frac{\partial v_r}{\partial r}\right)^2+\left({1 \over r}\frac{\partial v_\theta}{\partial \theta}+{v_r \over r}\right)^2\right.+\left({1 \over r\sin\theta}\frac{\partial v_\phi}{\partial \phi} +{v_r\over r}+{v_\theta\cot\theta\over r}\right)^2\nonumber\\
1527: &+&\left.{1\over 2}\left[r\frac{\partial }{\partial r}\left({v_\theta \over r}\right)+{1\over r}\frac{\partial v_r}{\partial \theta}\right]^2+{1\over 2}\left[r\frac{\partial }{\partial \phi}\left({v_\phi \over r}\right)+{1\over r\sin\theta}\frac{\partial v_r}{\partial \phi}\right]^2+{1\over 2}\left[{\sin\theta\over r}\frac{\partial }{\partial \theta}\left({v_\phi \over \sin\theta}\right)+{1\over r\sin\theta}\frac{\partial v_\theta}{\partial \phi}\right]^2\right\}\nonumber\\
1528: &+&\left(\zeta-{2 \over 3}\mu\right)\left[{1\over r^2}\frac{\partial }{\partial r}(r^2v_r)+{1\over r\sin\theta}\frac{\partial }{\partial \theta}(v_\theta\sin\theta)+{1\over r\sin\theta}\frac{\partial v_\phi}{\partial \phi}\right]^2\,.
1529: \end{eqnarray}
1530: The dissipation function enters the heat equation as
1531: \begin{equation}\label{eq:heatequation}
1532: \rho\left[c_v\frac{dT}{dt}+p{d\over dt}\left({1 \over \rho}\right) - {GM \over r} \right] = \Phi -\nabla\cdot \vec{q}
1533: \end{equation}
1534: where $T$ is the temperature, $c_v=R/(\gamma-1)$ is the specific heat capacity. As mentioned in section 2, the heat equations does not contain transport terms, so $\vec{q}\equiv 0$. This means that dissipated heat can only be lost be pressure work. Conversely, if the dissipation term is neglected as is the case in the optically thin limit, the only way to change the temperature at any given point is through pressure work.
1535: \section{Computational domain setup}\label{appendix:domainsetup}
1536: We set up an accretion disk around the WD. This setup comprises three parts: the disk, the stellar atmosphere and the halo. This improves on the previous simulations of \cite{Kley87a} which did not include a stellar atmosphere. For stability reasons it is important that the pressure of the three components match at their respective interfaces. This is to prevent shocks from dominating the solution. We have therefore carefully designed the domain to ensure the partial pressures are identical at the common interface.
1537:
1538: The radius of the star is $r=9000\textrm{km}$. We assume that the disk is symmetric in the plane and that is it axisymmetric. We use the model of \cite{Balsara04}. In their notation, we have $a=1/2$ and $\gamma=5/3$, so their model effectively reduces to a constant density in the disk
1539: plane for our case. We assume the the initial halo, atmosphere and disk are isothermal and obey the polytropic equation of state
1540: \begin{equation}\label{eq:polytrope}
1541: P=K\rho^\gamma
1542: \end{equation}
1543: We set $\rho_0=1.2\times10^{-5} \textrm{g}\textrm{cm}^{-3}$. Setting a temperature for the disk then fully determines $K$. Above the disk
1544: plane the partial pressure of the disk matter follows the prescription of a plane parallel atmosphere where the pressure falls off with the e-folding distance given by the scale height. The density is calculated according to eq.~[\ref{eq:polytrope}]
1545: (see \cite{Landau87}.
1546:
1547: The temperature of the WD is set to $T_*=300000\textrm{K}$. The atmospheric scale height is given by $H=RT_*/g\mu$, where $g=GM/r^2$, $R$ is the gas
1548: constant and $\mu$ is the mean molecular weight. The inner boundary of the computational domain is set to $r_*-5H_*$. The disk scale height is given by $H_d=c_s\sqrt{r/g}$, where $c_s=\sqrt{\gamma P/\rho}$ is the local sound speed. The outer boundary of the computational domain is set to $r_*+3H_d$. The disk is set up with a Keplerian velocity profile.
1549: The disk temperature is set to $T_d=1000000\textrm{K}$. Pressure matching between the disk and the stellar atmosphere is achieved by selecting $\rho_*$ so that $P_*\equiv P_d$ at the base of the disk at $r=r_*$. The pressure profile of the WD atmosphere follows the standard solution of a plane parallel atmosphere (see \cite{Landau87})
1550:
1551: In these models we do not have a self-consistent disk corona,
1552: which would provide pressure balance at the upper boundary of the accretion
1553: disk. For that reason, we use a very hot, initially static halo that provides
1554: pressure balance both to the upper boundary of the disk as well as to the
1555: WD's atmosphere. The halo's temperature is chosen to be high enough that the
1556: mass in the halo is less than the mass in the disk or the WD atmosphere
1557: by a factor of 35, making it dynamically unimportant.
1558: The temperature of the halo is set so $T_h=50T_d$. The scale height of the halo is determined by the gravitational component in the z-direction. The density of the halo is set to match the pressure of the disk one disk scale height above the disk base at $r_*$.
1559:
1560: The careful pressure matching described above ensures that the dynamical instabilities at $t=0$ are negligible. Therefore, the computational model will quickly stabilize to a dynamical equilibrium.
1561:
1562:
1563:
1564:
1565:
1566: \begin{thebibliography}{}
1567: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
1568:
1569: \bibitem[Abramowicz et al. (1996)]{abra96}
1570: Abramowicz, M., Brandenburg, A., \& Lasota, J.-P. 1996, \mnras, 281, L21
1571:
1572: \bibitem[{Alexakis {et~al.}(2004)Alexakis, Calder, Heger, Brown, Dursl, Truran,
1573: Rosner, Lamb, Timmes, Fryxell, Zingale, Ricker, \& Olson}]{Alexakis04}
1574: Alexakis, A., et al. 2004, \apj, 602, 931
1575:
1576: \bibitem[{Balbus {et~al.}(1994)Balbus, Gammie, \& Hawley}]{Balbus94a}
1577: Balbus, S.~A., Gammie, C.~F., \& Hawley, J.~F. 1994, \mnras, 271, 197
1578:
1579: \bibitem[{Balsara(1998{\natexlab{a}})}]{Balsara98a}
1580: Balsara, D.~S. 1998{\natexlab{a}}, \apjs, 116, 119
1581:
1582: \bibitem[{Balsara(1998{\natexlab{b}})}]{Balsara98b}
1583: ---. 1998{\natexlab{b}}, \apjs, 116, 133
1584:
1585: \bibitem[{Balsara(2004)}]{Balsara04}
1586: ---. 2004, \apjs, 151, 149
1587:
1588: \bibitem[Balsara \& Fisker (2005)]{Balsara05}
1589: Balsara, D.S., \& Fisker, J.L., 2005, in Protostars and Planets V,
1590: p. 8631
1591:
1592: \bibitem[{Balsara \& Shu(2000)}]{Balsara00}
1593: Balsara, D.~S., \& Shu, C.-W. 2000, J. Comput. Phys., 160, 405
1594:
1595: \bibitem[{Balsara \& Spicer(1999{\natexlab{a}})}]{Balsara99a}
1596: Balsara, D.~S., \& Spicer, D.~S. 1999{\natexlab{a}}, J. Comput. Phys., 148, 133
1597:
1598: \bibitem[{Balsara \& Spicer(1999{\natexlab{b}})}]{Balsara99b}
1599: ---. 1999{\natexlab{b}}, J. Comput. Phys., 148, 133
1600:
1601: \bibitem[Bath (1972)]{bat72}
1602: Bath, G.T. 1972, \apj, 173, 121
1603:
1604: \bibitem[Becker \& Marshall (1981)]{Becker81}
1605: Becker, R.H., \& Marshal, F.E. 1981, \apj, 244, L93
1606:
1607: \bibitem[Belloni et al. (1991)]{Belloni91}
1608: Belloni, T. et al. 1991, A\&A, 246, L44
1609:
1610: \bibitem[Beuermann \& Thomas (1993)]{Beuermann93}
1611: Beuermann, K., \& Thomas, H.-C. 1993, Adv.Space Res., 13, 115
1612:
1613: \bibitem[Bhatia (1974)]{Bhatia}
1614: Bhatia, P.K. 1974, \apss, 26, 319
1615:
1616: \bibitem[Brandenburg et al. (1996)]{Brandenburg96}
1617: Brandenburg, A., Lasota, J.-P., \& Abramowicz, M. 1996, \mnras, 281, L21
1618:
1619: \bibitem[{Cannizzo et al.(1988)}]{Cannizzo88}
1620: Cannizzo, J.~K., Shafter, A.W., \& Wheeler, J.C. 1988, \apj, 333, 227
1621:
1622: \bibitem[{Cannizzo(1998)}]{Cannizzo98}
1623: ---. 1998, \apj, 493, 426
1624:
1625: \bibitem[Chandrasekhar (1961)]{Chandra}
1626: Chandrasekhar, S. 1961, Hydrodynamic and Hydromagnetic Stability,
1627: Oxford, Clarendon Press
1628:
1629: \bibitem[Cheng et al. (1997)]{che97}
1630: Cheng, F.H., Sion, E.M., Szkody, P., Huang, M. 1997, \apj, 484, L149
1631:
1632: \bibitem[Chimonas (1970)]{Chimonas1970}
1633: Chimonas, G. 1970, J.Fluid Mech., 43, 833
1634:
1635: \bibitem[{Collins {et~al.}(1998)Collins, Helfer, \& Horn}]{Collins98b}
1636: Collins, T. J.~B., Helfer, H.~L., \& Horn, H. M.~V. 1998, \apj,
1637: 508, L159
1638:
1639: \bibitem[C\'ordova et al. (1980)]{cor80}
1640: C\'ordova, F.A., Chester, T.J., Tuohy, I.R., \& Garmire, G.P. 1980,
1641: \apj, 235, 163
1642:
1643: \bibitem[C\'ordova et al. (1981)]{Cordova81}
1644: C\'ordova, F.A., Mason, K.O., \& Nelson, J.E. 1981, \apj, 245, 609
1645:
1646: \bibitem[C\'ordova \& Mason (1983)]{Cordova83}
1647: C\'ordova, F.A., \& Mason, K.O. 1983, in: Accretion-Driven Stellar X-ray
1648: sources, eds. W.Lewin, E.van den Heuvel, CUP 1983, p.147
1649:
1650: \bibitem[Crawford \& Kraft (1956)]{cra56}
1651: Crawford, J.A., \& Kraft, R.P. 1956, \apj, 123, 44
1652:
1653: \bibitem[Drazin \& Reid (1981)]{Drazin81}
1654: Drazin, P.G., \& Reid, W.H., 1981, {\it{Hydrodynamic Stability}},
1655: Cambridge University Press, Cambridge
1656:
1657: \bibitem[Dubrulle (1993)]{Dubrulle93}
1658: Dubrulle, B. 1993, Icarus, 106, 59
1659:
1660: \bibitem[{Durisen(1973)}]{Durisen73a}
1661: Durisen, R.~H. 1973, \apj, 183, 205
1662:
1663: \bibitem[{Durisen(1977)}]{Durisen1977}
1664: ---. 1977, \apj, 213, 145
1665:
1666: \bibitem[{Ferland {et~al.}(1982)Ferland, Langer, MacDonald, Pepper, Shaviv, \&
1667: Truran}]{Ferland82}
1668: Ferland, G.~J., Langer, S.~H., MacDonald, J., Pepper, G.~H., Shaviv, G., \&
1669: Truran, J. 1982, \apj, 262, L53
1670:
1671: \bibitem[{Fisker \& Balsara(2005)}]{Fisker05c}
1672: Fisker, J.~L., \& Balsara, D.~S. 2005, \apj, 635, L69
1673:
1674: \bibitem[Flannery (1974)]{fla74}
1675: Flannery, B.P. 1974, \mnras, 170, 325
1676:
1677: \bibitem[{Frank {et~al.}(2002)Frank, King, \& Raine}]{Frank02}
1678: Frank, J., King, A., \& Raine, D. 2002, Accretion Power in Astrophysics, 3rd
1679: edn. (Cambridge: Cambridge University Press)
1680:
1681: \bibitem[{Froning {et~al.}(2001)Froning, Long, Drew, Knigge, \&
1682: Proga}]{Froning01}
1683: Froning, C.~S., Long, K.~S., Drew, J.~E., Knigge, C., \& Proga, D. 2001,
1684: \apj, 562, 963
1685:
1686: \bibitem[{Glasner {et~al.}(1997)Glasner, Livne, \& Truran}]{Glasner97}
1687: Glasner, S.~A., Livne, E., \& Truran, J.~W. 1997, \apj, 475, 754
1688:
1689: \bibitem[Godon (1995)]{god95}
1690: Godon, P. 1995, \mnras, 277, 157
1691:
1692: \bibitem[Godon (1996a)]{Godon96}
1693: ---. 1996a, \apj, 462, 456
1694:
1695: \bibitem[Godon (1996b)]{Godon96b}
1696: ---. 1996b, \mnras, 282, 1107
1697:
1698: \bibitem[Godon (1997a)]{Godon1997}
1699: ---. 1997a, \apj, 480, 329
1700:
1701: \bibitem[Godon (1997b)]{Godon97}
1702: ---. 1997b, \apj, 483, 882
1703:
1704: \bibitem[Godon et al. (1995)]{Godon95}
1705: Godon, P., Regev, O., and Shaviv, G. 1995, \mnras, 275, 1093
1706:
1707: \bibitem[{Godon \& Sion(2002)}]{Godon02}
1708: Godon, P., \& Sion, E.~M. 2002, \apj, 566, 1084
1709:
1710: \bibitem[{Godon \& Sion(2003)}]{Godon03}
1711: ---. 2003, \apj, 586, 427
1712:
1713: \bibitem[{Godon \& Sion(2005)}]{Godon05}
1714: ---. 2005, \mnras, 361, 809
1715:
1716: \bibitem[{Godon {et~al.}(2004a)Godon, Sion, Cheng, G{\"a}nsicke, Howell, Knigge,
1717: Sparks, \& Starrfield}]{Godon04}
1718: Godon, P., Sion, E.~M., Cheng, F., G{\"a}nsicke, B.~T., Howell, S., Knigge, C.,
1719: Sparks, W.~M., \& Starrfield, S. 2004a, \apj, 602, 336
1720:
1721: \bibitem[Godon et al. (2004b)]{god04}
1722: Godon, P., Sion, E.M., Cheng, F.H., Szkody, P., Long, K.S., Froning, C.S.
1723: 2004b, \apj, 612, 429
1724:
1725: \bibitem[Hack \& la Dous (1993)]{hac93}
1726: Hack, M., \& la Dous, C. 1993, Cataclysmic Variables and Related
1727: Objects, Monograph series on nonthermal phenomenon in stellar
1728: atmospheres, NASA SP-507, Washington
1729:
1730: \bibitem[Hamilton \& Habernathy (1994)]{Hamilton94}
1731: Hamilton, J.M., \& Habernathy, F.H. 1994, J.Fluid Mech., 264, 185
1732:
1733: \bibitem[Hoare \& Drew (1991)]{Hoare91}
1734: Hoare, M.G., \& Drew, J.E. 1991, \mnras, 249, 452
1735:
1736: \bibitem[Howard (1961)]{Howard1961}
1737: Howard, L.N. 1961, J.Fluid Mech., 10, 509
1738:
1739: \bibitem[{Inogamov \& Sunyeav(1999)}]{Inogamov99}
1740: Inogamov, N.~A., \& Sunyeav, R. 1999, Astron. Lett., 25, 269
1741:
1742: \bibitem[{Jiang \& Shu(1996)}]{Jiang96}
1743: Jiang, G.-S., \& Shu, C.-W. 1996, J. Comput. Phys., 126, 202
1744:
1745: \bibitem[Johnson \& Gammie (2006)]{Johnson2006}
1746: Johnson, B.M., \& Gammie, C.F. 2006, \apj, 636, 63
1747:
1748: \bibitem[{Jos\'e(2005)}]{Jose05}
1749: Jos\'e, J. 2005, Nucl. Phys., A758, 713
1750:
1751: \bibitem[Kato \& Inagaki (1994)]{kato94}
1752: Kato, S., \& Inagaki, S. 1994, \pasj, 46, 289
1753:
1754: \bibitem[King (1997)]{King97}
1755: King, A. 1997, \mnras, 288, 16
1756:
1757: \bibitem[King \& Shaviv (1984)]{King84}
1758: King, A.R., \& Shaviv, G. 1984, Nature, 308, 519
1759:
1760: \bibitem[{Kippenhahn \& Thomas(1978)}]{Kippenhahn78}
1761: Kippenhahn, R., \& Thomas, H.-C. 1978, A\&A, 63, 265
1762:
1763: \bibitem[{Kley(1989)}]{Kley89a}
1764: Kley, W. 1989, A\&A, 208, 98
1765:
1766: \bibitem[{Kley(1991)}]{Kley91}
1767: ---. 1991, A\&A, 247, 95
1768:
1769: \bibitem[{Kley \& Hensler(1987)}]{Kley87a}
1770: Kley, W., \& Hensler, G. 1987, A\&A, 172, 124
1771:
1772: \bibitem[Kley \& Lin (1996)]{Kley96}
1773: Kley, W., \& Lin, D.N.C. 1996, \apj, 461, 933
1774:
1775: \bibitem[Kley \& Lin (1999)]{Kley99}
1776: ---. 1999, \apj, 518, 833
1777:
1778: \bibitem[Klu\'zniak (1987)]{Kluzniak87}
1779: Klu\'zniak, W. 1987, Ph.D. Thesis, Stanford University
1780:
1781: \bibitem[Kraft (1962)]{kra62}
1782: Kraft, R.P. 1962, \apj, 135, 408
1783:
1784: \bibitem[{Landau \& Lifshitz(1987)}]{Landau87}
1785: Landau, L.~D., \& Lifshitz, E.~M. 1987, Fluid mechanics (New York: Pergamon
1786: Press)
1787:
1788: \bibitem[Lebovitz (1965)]{Lebovitz1965}
1789: Lebovitz, N.R. 1965, \apj, 142, 229
1790:
1791: \bibitem[Lebovitz (1966)]{Lebovitz1966}
1792: ---. 1966, \apj, 146, Notes 946
1793:
1794: \bibitem[Ledoux \& Walraven (1958)]{Ledoux1958}
1795: Ledoux, P., \& Walraven, Th. 1958, {\it Handbuch der Physik},
1796: 41, ed. S. Fl\"ugge (Berlin: Springer-Verlag), 353
1797:
1798: \bibitem[{Livio \& Pringle(1992)}]{Livio92}
1799: Livio, M., \& Pringle, J.E. 1992, \mnras, 259, P23
1800:
1801: \bibitem[{Livio \& Pringle(1998)}]{Livio98}
1802: ---. 1998, \apj, 505, 339
1803:
1804: \bibitem[Livio \& Shaviv (1977)]{Livio1977}
1805: Livio, M., \& Shaviv, G. 1977, \aap, 55, 95
1806:
1807: \bibitem[Livio \& Truran (1987)]{Livio1987}
1808: Livio, M., \& Truran, J.W. 1987, \apj, 318, 316
1809:
1810: \bibitem[{Long {et~al.}(1993)Long, Blaier, Bowyers, Sion, \& Hubeny}]{Long93}
1811: Long, K.~S., Blaier, W.~P., Bowyers, C.~W., Sion, E.~M., \& Hubeny, I. 1993,
1812: \apj, 405, 327
1813:
1814: \bibitem[Lubow \& Shu (1975)]{lub75}
1815: Lubow, S.H., \& Shu, F.H. 1975, \apj, 198, 383
1816:
1817: \bibitem[{Lynden-Bell \& Pringle(1974)}]{Lynden-Bell74}
1818: Lynden-Bell, D., \& Pringle, J.~E. 1974, \mnras, 168, 303
1819:
1820: \bibitem[{MacDonald(1983)}]{MacDonald83}
1821: MacDonald, J. 1983, \apj, 273, 289
1822:
1823: \bibitem[Mauche (1996)]{Mauche96}
1824: Mauche, C.W. 1996, in: Proceedings of X-Ray Imaging and Spectroscopy of Comic
1825: Hot Plasmas, Ed.F.Makino
1826:
1827: \bibitem[Mauche (2004)]{mau04}
1828: ---. 2004, \apj, 610, 422
1829:
1830: \bibitem[Mauch et al. (1995)]{mau95}
1831: Mauche, C.W., Raymond, J.C., \& Mattei, J.A. 1995, \apj, 446, 842
1832:
1833: \bibitem[Mauche et al. (1991)]{Mauche91}
1834: Mauche, C.,W., Wade, R.A., Polidan, R.S., van der Woerd, H.,
1835: \& Paerels, F.B.S. 1991, \apj, 372, 659
1836:
1837: \bibitem[{Meyer \& Meyer-Hofmeister(1982)}]{Meyer82}
1838: Meyer, F., \& Meyer-Hofmeister, E. 1982, A\&A, 106, 34
1839:
1840: \bibitem[Meyer \& Meyer-Hofmeister (1994)]{Meyer94}
1841: ---. 1994, A\&A, 288, 175
1842:
1843: \bibitem[{Mihalas \& Mihalas(1984)}]{Mihalas84}
1844: Mihalas, D., \& Mihalas, B.~W. 1984, Foundations of radiation hydrodynamics
1845: (Cambridge: Cambridge Univ. Press)
1846:
1847: \bibitem[{Miles (1957)}]{Miles57}
1848: Miles, J.W. 1957, J. Fluid Mech., 3, 185
1849:
1850: \bibitem[Miles (1961)]{Miles1961}
1851: ---. 1961, J. Fluid Mech., 10, 496
1852:
1853: \bibitem[Mukai \& Patterson (2004)]{muk04}
1854: Mukai, K., \& Patterson, J. 2004, RM AC 20, 244
1855:
1856: \bibitem[Mukai et al. (1997)]{Mukai97}
1857: Mukai, K., Wood, J.H., Naylor, T., Schlegel, E.M., \& Swank, J.H.
1858: 1997, \apj, 475, 812
1859:
1860: \bibitem[Narayan et al. (1994)]{nara94}
1861: Narayan, R., Loeb, A., \& Kumar, P. 1994, \apj, 431, 359
1862:
1863: \bibitem[Narayan \& Popham (1993)]{Narayan93}
1864: Narayan, R. \& Popham, R. 1993, Nature, 362, 820
1865:
1866: \bibitem[Obach \& Glatzel (1999)]{Obach99}
1867: Obach, C. \& Glatzel, W. 1999, \mnras, 303, 603
1868:
1869: \bibitem[Orszag \& Kell (1980)]{Orszag80}
1870: Orszag, S.A., \& Kell, L.C. 1980, J.Fluid Mech., 96, 159
1871:
1872: \bibitem[Pandel et al. (2003)]{Pandel03}
1873: Pandel, D., C\'ordova, F.A., \& Howell, S.B. 2003, \mnras, 346, 1231
1874:
1875: \bibitem[Pandel et al. (2005)]{Pandel05}
1876: Pandel, D., C\'ordova, F.A., Mason, K.O., \& Priedhorsky, W.C. 2005,
1877: \apj, 626, 396
1878:
1879: \bibitem[Papaloizou \& Pringle (1984)]{papa84}
1880: Papaloizou, J.C.B., \& Pringle, J.E. 1984, \mnras, 208, 721
1881:
1882: \bibitem[Papaloizou \& Pringle (1985)]{papa85}
1883: ---. 1985, \mnras, 213, 799
1884:
1885: \bibitem[Papaloizou \& Pringle (1987)]{papa87}
1886: ---. 1987, \mnras, 225, 267
1887:
1888: \bibitem[Papaloizou \& Stanley (1986)]{Papaloizou86}
1889: Papaloizou, J.C.B., \& Stanley, G.Q.G. 1986, \mnras, 220, 593
1890:
1891: \bibitem[Papaloizou \& Szuszkiewicz (1994)]{papa94}
1892: Papaloizou, J.C.B., \& Szuszkiewicz, E. 1994, \mnras, 268, 29
1893:
1894: \bibitem[Patterson (1979)]{pat79}
1895: Patterson, J. 1979, \apj, 234, 978
1896:
1897: \bibitem[Patterson (1981)]{pat81}
1898: Patterson, J. 1981, \apjs, 45, 517
1899:
1900: \bibitem[Patterson \& Raymond (1985)]{Patterson85}
1901: Patterson, J., \& Raymond, J.C. 1985, \apj, 295, 550
1902:
1903: \bibitem[{Phillips (1957)}]{Phillips57}
1904: Phillips, O.M. 1957, J. Fluid Mech., 2, 417
1905:
1906: \bibitem[Pesnell (1986)]{Pesnell1986}
1907: Pesnell, W.D. 1986, \apj, 301, 204
1908:
1909: \bibitem[{Piro \& Bildsten(2004a)}]{Piro04}
1910: Piro, A.~L., \& Bildsten, L. 2004a, \apj, 610, 977
1911:
1912: \bibitem[{Piro \& Bildsten(2004b)}]{pir04b}
1913: Piro, A.~L., \& Bildsten, L. 2004a, \apj, 616, L155
1914:
1915: \bibitem[Piro et al. (2005)]{Piro05}
1916: Piro, A.L., Arras, P., \& Bildsten, L. 2005, \apj, 628, 401
1917:
1918: \bibitem[Polidan et al. (1990)]{Polidan90}
1919: Polidan, R.S., Mauche, C.S., \& Wade, R.A. 1990, \apj, 356, 211
1920:
1921: \bibitem[Ponman et al. (1995)]{Ponman95}
1922: Ponman, T.J., Belloni, T., Duck, S.R., Verbunt, F., Watson, M.G.,
1923: Wheatley, P.J., \& Pfeffermann, E. 1995, \mnras, 276, 495
1924:
1925: \bibitem[Popham (1997)]{Popham97}
1926: Popham, R. 1997, \apj, 478, 734
1927:
1928: \bibitem[Popham (1999)]{Popham99}
1929: ---. 1999, \mnras, 308, 979
1930:
1931: \bibitem[{Popham \& Narayan(1991)}]{pop91}
1932: Popham, R., \& Narayan, R. 1991, \apj, 370, 614
1933:
1934: \bibitem[{Popham \& Narayan(1992)}]{poph92}
1935: ---. 1992, \apj, 394, 255
1936:
1937: \bibitem[{Popham \& Narayan(1995)}]{Popham95}
1938: ---. 1995, \apj, 442, 337
1939:
1940: \bibitem[{Prendergast \& Burbidge(1968)}]{Prendergast68}
1941: Prendergast, K.~H., \& Burbidge, G.~R. 1968, \apj, 151, L83
1942:
1943: \bibitem[{Pringle(1981)}]{Pringle81}
1944: Pringle, J.~E. 1981, ARA\&A, 19, 137
1945:
1946: \bibitem[Pringle \& Savonije (1979)]{Pringle79}
1947: Pringle, J.E., \& Savonije, G.J. 1979, \mnras, 187, 777
1948:
1949: \bibitem[Rayleigh (1880)]{Rayleigh1880}
1950: Rayleigh, Lord, 1880, Proc.London Math.Soc., 11, 57
1951:
1952: \bibitem[Rayleigh (1916)]{Rayleigh16}
1953: ---. 1916, Proc.R.Soc.London, A., 93, 148
1954:
1955: \bibitem[Regev (1983)]{Regev83}
1956: Regev, O. 1983, A\&A, 126, 146
1957:
1958: \bibitem[Regev \& Shara (1989)]{Regev89}
1959: Regev, O., \& Shara, M.M. 1989, \apj, 340, 1006
1960:
1961: \bibitem[Richardson (1920)]{Richardson1920}
1962: Richardson, L.F. 1920, Proc.Roy.Soc.London A, 9F, 354
1963:
1964: \bibitem[Ritter \& Kolb (1998)]{rit98}
1965: Ritter, H., \& Kolb, U. 1998, \aaps, 129, 83
1966:
1967: \bibitem[{Robertson \& Frank(1986)}]{Robertson86}
1968: Robertson, J.~A., \& Frank, J. 1986, \mnras, 221, 279
1969:
1970: \bibitem[{Roe \& Balsara(1996)}]{Roe96}
1971: Roe, P.~L., \& Balsara, D.~S. 1996, SIAM J. Appl. Math., 56, 57
1972:
1973: \bibitem[Rose (1968)]{ros68}
1974: Rose, W.K. 1968, \apj, 152, 245
1975:
1976: \bibitem[{Rosner {et~al.}(2001)Rosner, Alexakis, Young, Truran, \&
1977: Hillebrandt}]{Rosner01}
1978: Rosner, R., Alexakis, A., Young, Y.-N., Truran, J.~W.,
1979: \& Hillebrandt, W. 2001, \apj, 562, L177
1980:
1981: \bibitem[R\"udiger et al. (2002)]{Rudiger2002}
1982: R\"udiger, G., Arlt, R., Shalybkov, D. 2002, \aap, 391, 781
1983:
1984: \bibitem[Saric (1994)]{Saric94}
1985: Saric, W.S. 1994, Annu.Rev.Fluid Mech., 26, 379
1986:
1987: \bibitem[Schwarzschild (1906)]{Schwarzschild1906}
1988: Schwarzschild, K. 1906, G\"ottingen Nachr., 41
1989:
1990: \bibitem[Shachdev \& Satya Narayanan (1982)]{Sachdev1982}
1991: Shachdev, P.L., \& Satya Narayanan, A. 1982, Indian J. Pure Appl.Math.,
1992: 13, 989
1993:
1994: \bibitem[{Shakura \& Sunyaev(1973)}]{Shakura73}
1995: Shakura, N.~I., \& Sunyaev, R.~A. 1973, A\&A, 24, 337
1996:
1997: \bibitem[{Shakura \& Sunyaev(1988)}]{shak88}
1998: Shakura, N.~I., \& Sunyaev, R.~A. 1988, Adv.Space Res., 8, 135
1999:
2000: \bibitem[Shaviv (1987)]{Shaviv87}
2001: Shaviv, G. 1987, ApSS, 130, 303
2002:
2003: \bibitem[Shaviv \& Starrfield (1987)]{Shavivand87}
2004: Shaviv, G., \& Starrfield, S. 1987, \apj, 321, L51
2005:
2006: \bibitem[Sion (1995)]{Sion95}
2007: Sion, E.M. 1995, \apj, 438, 876
2008:
2009: \bibitem[{Sion {et~al.}(2005)Sion, Cheng, G{\"a}nsicke, \& Szkody}]{Sion04c}
2010: Sion, E.~M., Cheng, F., G{\"a}nsicke, B., \& Szkody, P. 2005, \apj,
2011: 614, L61
2012:
2013: %\bibitem[Smak (1971)]{sma71}
2014: %Smak, J. 1971, Acta Astronomica, 21, 15
2015:
2016: \bibitem[Starrfield (1971a)]{sta71a}
2017: Starrfield, S. 1971a, \mnras, 152, 307
2018:
2019: \bibitem[Starrfield (1971b)]{sta71b}
2020: Starrfield, S. 1971b, \mnras, 155, 129
2021:
2022: \bibitem[{Starrfield {et~al.}(1972)Starrfield, Truran, Sparks, \&
2023: Kutter}]{Starrfield72}
2024: Starrfield, S., Truran, J.~W., Sparks, W.~M., \& Kutter, G.~S. 1972,
2025: \apj, 176, 169
2026:
2027: \bibitem[Sung (1974)]{Sung1974}
2028: Sung, C.H. 1974, \aap, 33, 99
2029:
2030: \bibitem[Sung (1975)]{Sung1975}
2031: ---. 1975, \apss, 26, 305
2032:
2033: \bibitem[{Szkody {et~al.}(2002)Szkody, Nishikida, Raymond, Seth, Hoard, Long,
2034: \& Sion}]{Szkody02}
2035: Szkody, P., Nishikida, K., Raymond, J.~C., Seth, A.,
2036: Hoard, D.~W., Long, K.~S.,
2037: \& Sion, E.~M. 2002, \apj, 574, 942
2038:
2039: \bibitem[Tsang \& Lai (2009)]{tsa09}
2040: Tsang, D., \& Lai, D. 2009, \mnras, 396, 589
2041:
2042: \bibitem[Tylenda (1981)]{Tylenda81}
2043: Tylenda, R. 1981, Acta Astronomica, 31, 267
2044:
2045: \bibitem[Van Duin \& Kalder (1982)]{Van1982}
2046: Van Duin, C.A., \& Kalder, H. 1982, J. Fluid Mech., 120, 505
2047:
2048: \bibitem[van Teeseling \& Verbunt (1994)]{vanTeeseling94}
2049: van Teeseling, A. \& Verbunt, F. 1994, A\&A, 292, 519
2050:
2051: \bibitem[van Teeseling et al. (1996)]{vanTeeseling96}
2052: van Teeseling, A. Beuermann, K., \& Verbunt, F. 1996, A\&A, 315, 467
2053:
2054: \bibitem[Warner (1987)]{war87}
2055: Warner, B. 1987, \mnras, 227, 23
2056:
2057: \bibitem[{Warner(1995)}]{Warner95}
2058: ---. 1995, Cataclysmic Variable Stars (Cambridge: Cambridge University
2059: Press)
2060:
2061: \bibitem[{Welsh {et~al.}(2003)Welsh, Sion, Godon, G{\"a}nsicke, Knigge, Long,
2062: \& Szkody}]{Welsh03}
2063: Welsh, W.~F., Sion, E.~M., Godon, P., G{\"a}nsicke, B.~T., Knigge, C., Long,
2064: K.~S., \& Szkody, P. 2003, \apj, 599, 509
2065:
2066: \bibitem[Zahn (1990)]{Zahn90}
2067: Zahn, J.-P. 1990, in Structure and Emission Properties of Accretion
2068: Disks, ed.C.Bertout, S.Collin-Souffrin, J.-P.Lasota, \& V.J.Tran Than
2069: (Paris: Editions Fronti\`eres), 87
2070:
2071: \end{thebibliography}
2072:
2073: \end{document}
2074:
2075: