1: % The following is for the standard submission
2: % \documentclass[12pt,preprint]{aastex}
3: % The following is for the OLD emulation
4: % \documentstyle[emulateapj,apjfonts,epsfig]{article}
5: % This is the NEW emulation
6: \documentclass[apj,numberedappendix]{emulateapj}
7: \usepackage{natbib,apjfonts}
8:
9: \newcommand{\ergsec}{erg sec$^{-1}$}
10: \newcommand{\ta}{triple-$\alpha$ }
11: \newcommand{\ts}{$\theta_{\rm spread}$}
12:
13: \begin{document}
14:
15: \title{A Spreading Layer Origin for Dwarf Nova Oscillations}
16:
17: \author{Anthony L. Piro\altaffilmark{1}
18: and Lars Bildsten\altaffilmark{1,2}}
19: \altaffiltext{1}{Department of Physics, Broida Hall, University of California,
20: Santa Barbara, CA 93106; piro@physics.ucsb.edu}
21:
22: \altaffiltext{2}{Kavli Institute for Theoretical Physics,
23: Kohn Hall, University of California,
24: Santa Barbara, CA 93106; bildsten@kitp.ucsb.edu}
25:
26: \begin{abstract}
27:
28: Dwarf nova outbursts often show coherent ($Q\sim10^4-10^6$)
29: sinusoidal oscillations with the largest pulsed fraction
30: in the extreme
31: ultraviolet. Called dwarf nova oscillations (DNOs),
32: they have periods of $P\approx3-40\textrm{ s}$ and scale with
33: luminosity as $P\propto L^{-\beta}$ with $\beta\approx0.1-0.2$.
34: We propose that DNOs may be produced by nonradial oscillations
35: in a thin hydrostatic layer of freshly accreted material, the
36: ``spreading layer'' (SL), at the white dwarf (WD) equator. This
37: would naturally explain a number of key properties of DNOs,
38: including their frequency range, sinusoidal
39: nature, sensitivity to accretion rate, and why they
40: are only seen during outburst. In support of this hypothesis
41: we construct a simple model that treats the SL as a cavity containing
42: shallow surface waves, each with the same radial structure, but split
43: into three different modes denoted by their azimuthal wavenumber, $m$.
44: The $m=0$ latitudinally propagating
45: mode best matches the periods and scalings associated with
46: most DNOs, and DNOs with periods shorter than the WD Keplerian
47: period are explained by the $m=-1$ prograde mode. We
48: also predict a third set of oscillations, produced by the $m=1$
49: retrograde mode, and show its expected dependence on
50: accretion rate.
51: % Our new idea for explaining DNOs will hopefully
52: % motivate more sophisticated models of nonradial oscillations in the SL
53: % so as to make a detailed test of our hypothesis.
54:
55: \end{abstract}
56:
57: \keywords{accretion, accretion disks ---
58: novae, cataclysmic variables ---
59: stars: oscillations ---
60: white dwarfs}
61:
62: \section{Introduction}
63:
64: Rapid oscillations are often seen in the optical, soft X-ray,
65: and extreme ultraviolet (EUV) flux of cataclysmic variables (CVs)
66: during dwarf nova (DN) outbursts. These oscillations have periods
67: in the range of
68: $P\approx3-40\textrm{ s}$, show a monotonic relationship with
69: EUV luminosity (Mauche 1996) and thus accretion rate, $\dot{M}$,
70: and are fairly coherent ($Q=|dP/dt|^{-1}\sim10^4-10^6$).
71: They are called dwarf
72: nova oscillations (DNOs; Patterson 1981) in contrast to the much
73: less coherent ($Q\sim5-20$) quasi-periodic oscillations (QPOs)
74: also seen from
75: CVs and the highly coherent oscillations ($Q\sim10^{12}$) that
76: reflect the spin of a magnetic primary (e.g., DQ Her). Since their
77: discovery (Warner \& Robinson 1972), DNOs have been seen in
78: $\sim50$ systems (Warner 2004). The DNO periods roughly track
79: the surface Keplerian period of the accreting white dwarf (WD)
80: [$P_{\rm K}= 2\pi/(GM/R^3)^{1/2}$; see Fig. 19 of Patterson 1981,
81: or Knigge et al. 1998],
82: so that it is generally believed that the oscillations are created near
83: or on the WD surface.
84:
85: A number of explanations have been proposed for DNOs, but none
86: have quantitatively explained the majority of their key
87: features. The characteristic period of the DNOs, $P\sim P_{\rm K}$,
88: implies a pulsational, rotational, or even perhaps inner accretion
89: disk origin. Since the period changes on the same
90: timescale as accretion, it is improbable that the entire WD takes
91: part in creating DNOs, ruling out global {\it g}-modes or the WD spin
92: modulated by hot spots. This led Paczy\'{n}ski (1978) to suggest
93: that if accretion torques were causing the period drifts then only
94: a small amount of material is involved in making DNOs
95: ($\lesssim10^{24}\textrm{ g}$). Papaloizou \& Pringle (1978) studied
96: rotationally modified global {\it g}-modes and {\it r}-modes on WDs.
97: Since truly global modes are not sensitive to the instantaneous
98: accretion rate, they proposed that DNOs
99: are high radial-order modes that have large amplitudes near the WD
100: surface. Their work explains many of the properties of DNOs, but
101: some questions still remain, including: (1) what physical mechanism
102: favors modes that are strongly concentrated in the outer layers of
103: the star, (2) why the radiating area associated with DNOs is such
104: a small fraction of the total WD surface area ($\sim10^{-4}$ to $10^{-2}$ as
105: inferred from the EUV emission of SS Cyg during DN outbursts;
106: C\'{o}rdova et al. 1980; Mauche \& Robinson 2001, hereafter MR01;
107: Mauche 2004),
108: and (3) why the modes are only excited during outbursts. These
109: failings led others to suggest alternatives, including
110: magnetic accretion onto a slipping belt (Warner \& Woudt 2002) and
111: non-axisymmetric bulges on the inner part of accretion disks
112: (Popham 1999).
113:
114: Inogamov \& Sunyaev (1999) presented a new way of studying disk
115: accretion close to an accreting neutron star (NS), which Piro \& Bildsten
116: (2004; hereafter PB04) extended to the case of WDs. Instead of using
117: a boundary layer model, they follow the latitudinal flow of accreted
118: material over the stellar surface, starting at the equator and streaming
119: up toward the pole, using a spreading layer (SL) model. In the case of
120: WDs the covering fraction (which is equivalent to the spreading angle)
121: is found to be small, $f\sim10^{-3}$ to $10^{-1}$
122: (for $\dot{M}\approx10^{16}-10^{18}\textrm{ g s}^{-1}$),
123: so that the size of the freshly accreted layer
124: is set by the thickness of the accretion disk.
125: %This model provides the
126: %first calculation of the properties of recently accreted material in
127: %hydrostatic balance on the WD surface.
128:
129: The material in the SL is much hotter in temperature and lower in
130: density than the underlying WD. This contrast allows waves
131: in the SL to travel freely, unencumbered by the material below. We
132: propose that DNOs are shallow surface waves in the layer of recently
133: accreted material confined to the WD equator. We argue that the $m=0$
134: mode provides the period that is most often identified as a DNO, and
135: we show that the $m=-1$ prograde mode explains the occurrence
136: of some of the higher frequency DNOs, including the
137: extra periods seen from SS Cyg and VW Hyi (MR01; Woudt \& Warner 2002).
138: We also discuss the $m=1$ retrograde mode and whether
139: it also corresponds to an observed frequency.
140: % Qualitatively, this
141: % model explains the basic characteristics of DNOs,
142: % including their sinusoidal nature, why they only involve a small
143: % amount of mass, their small radiating area, and why they are only
144: % seen during outburst.
145:
146: In \S 2 we describe a simple model to estimate the shallow surface wave
147: periods and show how they scale with $\dot{M}$ and the WD mass and
148: radius. We then compare these periods to observed DNOs from CVs in
149: \S 3. We conclude in \S 4 with a summary of the DNO properties
150: successfully understood using our model
151: along with a discussion of problems that are still left unresolved.
152:
153: \section{Nonradial Oscillation Periods in the Spreading Layer}
154:
155: Patterson (1981) showed that (at the time) all known DNOs on WDs with
156: measured masses have $P\gtrsim P_{\rm K}$. This relation is an important
157: constraint for any explanation of DNOs, so we begin by showing why a
158: SL mode should mimic such a period.
159: The geometry of the SL is a thin, quickly spinning, layer in hydrostatic
160: balance, which is predicted both observationally (Mauche 2004) and theoretically
161: (PB04) to cover a fraction of the surface area $f\sim10^{-3}-10^{-1}$.
162: In situations where there is a large entropy contrast between a surface
163: layer and the underlying material, nonradial oscillations can be confined to
164: high altitude regions with little or no pulsational energy extending
165: deeper into the WD. When the horizontal wavelength is much greater than
166: the layer depth, these shallow surface waves have a frequency
167: \begin{eqnarray}
168: \omega^2 = g_{\rm eff}hk^2,
169: \end{eqnarray}
170: where
171: $g_{\rm eff}=GM/R^2-v_\phi^2/R-v_\theta^2/R\approx GM/R^2-v_\phi^2/R$
172: is the surface gravitational acceleration decreased by centrifugal effects,
173: $h=P/(\rho g_{\rm eff})$ is the pressure scale height of the layer,
174: and $k$ is the transverse wavenumber. The SL can be thought of as a waveguide
175: with latitudinal width $2fR$ and azimuthal length $2\pi R$, but since
176: $f\ll1$ the latitudinal contribution dominates so that
177: $k\sim1/(fR)$. Setting $g_{\rm eff}=\lambda GM/R^2$, where
178: $\lambda\lesssim1$ is a dimensionless parameter that depends
179: on the spin of the layer, we rearrange the terms in equation (1) to find
180: \begin{eqnarray}
181: \omega =
182: \left( \frac{GM}{R^3} \right)^{1/2}
183: \left( \frac{\lambda h}{f^2R} \right)^{1/2},
184: \end{eqnarray}
185: which shows that the mode's frequency is the Keplerian frequency times
186: a factor less than unity (as long as $\lambda h\lesssim f^2R$), and
187: therefore explains why these shallow waves are consistent with the
188: findings of Patterson (1981).
189:
190: To find how these modes scale with $\dot{M}$ and the WD mass we
191: consider a simple model to estimate $f$ and $h$.
192: The theoretical studies of PB04
193: suggest that in the $\dot{M}$ range of DN
194: the covering fraction, $f$, may be set by the thickness of
195: the accretion disk resulting in (Shakura \& Sunyaev 1973;
196: in the limit of gas pressure much greater than
197: radiation pressure and using Kramer's opacity)
198: \begin{eqnarray}
199: f = 1.8\times10^{-2}
200: \alpha_{\rm disk,2}^{-1/10}
201: \dot{M}_{17}^{3/20}
202: M_1^{-3/8}
203: R_9^{1/8},
204: \end{eqnarray}
205: where $\alpha_{\rm disk}$ is the viscosity parameter for the accretion disk
206: and $\alpha_{\rm disk,2}\equiv\alpha_{\rm disk}/10^{-2}$,
207: $\dot{M}_{17}\equiv\dot{M}/(10^{17} \textrm{ g s}^{-1})$,
208: $M_1\equiv M/M_\odot$, $R_9\equiv R/(10^9 \textrm{ cm})$,
209: and we set the factor $[1-(r/R)^{-1/2}]\approx1$.
210:
211: As long as the WD
212: is rotating at much less than breakup (G\"{a}nsicke et al. 2001), half of
213: the accretion luminosity is released in a layer at the WD surface, so that
214: the flux of this layer is given by $4\pi fR^2 F = GM\dot{M}/(2R)$. In
215: a one-zone layer the radiative flux equation can be integrated
216: to give $F=acT^4/(3\kappa y)$, where $a$ is the radiation constant,
217: $\kappa$ is the opacity (set to $0.34\textrm{ cm}^2\textrm{ g}^{-1}$
218: for Thomson scattering with a solar composition), and $y$
219: is the column depth of the layer
220: (measured in units of $\textrm{g cm}^{-2}$). The column depth is set by
221: continuity to be $y=\dot{M}t_{\rm visc}/(4\pi fR^2)$,
222: where $t_{\rm visc}=h^2/\nu$ is the timescale for viscous dissipation
223: in the SL. The viscosity between the fresh, quickly spinning material and the
224: underlying WD is given by $\nu=\alpha_{\rm SL}v_\phi h$ (PB04),
225: where $\alpha_{\rm SL}$ is the SL's viscosity parameter, and
226: $v_\phi\approx(GM/R)^{1/2}$ is the initial azimuthal velocity of the
227: material in the SL.
228: Using an ideal gas equation of state,
229: we find the lowest order mode, which does not
230: propagate in the azimuthal direction so we denote it with an azimuthal
231: wavenumber $m=0$, has a period
232: \begin{eqnarray}
233: P_{m=0} = 30\textrm{ s }\alpha_{\rm disk,2}^{-2/15}
234: \alpha_{\rm SL,3}^{1/6}\lambda_1^{1/6}\dot{M}_{17}^{-2/15}
235: M_1^{-1/3}R_9^{19/12},
236: \end{eqnarray}
237: where $\alpha_{\rm SL,3}\equiv\alpha_{\rm SL}/10^{-3}$ and
238: $\lambda_1\equiv\lambda/10^{-1}$.
239: Equation (4) provides both a scaling and period suggestive of DNOs.
240: We discuss these similarities in more detail when we compare to observations
241: in \S 3. If we instead take $\nu=\alpha_{\rm SL}c_sh$
242: (in analogy to a Shakura \& Sunyaev accretion disk), where $c_s$ is
243: the speed of sound, we find
244: $P\propto\dot{M}^{-13/140}M^{-27/56}R^{89/56}$, also similar to observations.
245: This shows the robustness of this idea in replicating the general scalings of DNOs.
246:
247: The arguments from above suggest that the SL can contain additional modes,
248: most notably those that propagate in
249: the azimuthal direction. These modes have an
250: observed frequency of $\omega_{\rm obs}=|\omega-m\omega_{\rm SL}|$, where
251: $\omega_{\rm SL}$ is the spin of the SL and $m$ is the azimuthal
252: wavenumber. Since the layer is spinning quickly, it is possible that
253: Coriolis effects will modify $k$, which has been studied in the limit
254: of a thin layer using the ``traditional approximation''
255: (see Bildsten, Ushomirsky \& Cutler 1996 and references therein).
256: Using this analysis, Coriolis effects are negligible since
257: $f\lesssim\omega/(2\omega_{\rm SL})$. If we set
258: $P_{\rm SL}=2\pi/\omega_{\rm SL}$, where $P_{\rm SL}\gtrsim P_{\rm K}$,
259: the next lowest order modes (still $n=1$) have periods
260: \begin{eqnarray}
261: \frac{1}{P_{m=\pm1}}
262: = \left|\frac{1}{P_{m=0}}\mp\frac{1}{P_{\rm SL}}\right|.
263: \end{eqnarray}
264: Since there are a few DNOs with $P\lesssim P_{\rm K}$
265: (as we discuss in \S 3), we propose that the prograde $m=-1$ mode
266: is also necessary for explaining some short period DNOs. It is less clear whether the
267: $m=1$ retrograde mode is consistent with an oscillation observed
268: during a DN outburst, and in \S 3 we speculate whether this mode may be related to
269: the long-period DNOs (lpDNOs; Warner, Woudt \& Pretorius 2003).
270:
271: \section{DNOs in the CV Population}
272:
273: We compare the most recent compilations of DNO periods (Table 1 of
274: Warner 2004) and WD masses (Ritter \& Kolb 2003) of CVs in Figure 1.
275: The period error bars
276: indicate the range of periods that have been seen from each object,
277: while the mass error bars indicate the errors. We plot
278: systems that have shown both high and low DNO periods as
279: two separate points (these CVs are SS Cyg, CN Ori, and VW Hyi).
280: The thick, dashed line denotes the surface Keplerian period, $P_{\rm K}$,
281: for a given WD mass (using the mass-radius relation of Truran
282: \& Livio 1986). This demonstrates that there are DNOs both above
283: and below $P_{\rm K}$, which is why we consider SL nonradial
284: oscillations with $m=0$ and $m=-1$.
285: For each mode we present periods for accretion rates
286: $\dot{M}=10^{16}-10^{18}\textrm{ g s}^{-1}$, the range
287: expected during a DN outburst. To calculate the
288: $m=-1$ mode we must assume a period for the SL's spin,
289: $P_{\rm SL}$, so we plot mode periods for both $P_{\rm SL}=P_{\rm K}$
290: and $P_{\rm SL} = 2P_{\rm K}$. This shows that our model is
291: insensitive to the choice of the SL spin rate.
292: \begin{figure}
293: \epsscale{1.2}
294: \plotone{f1.eps}
295: \figcaption{DNO periods vs. measured WD masses. The vertical bars show
296: the range of periods observed from each CV, while the horizontal bars correspond to the
297: mass measurement errors. CVs that show two distinct DNO periods are
298: plotted as two separate points.
299: The thick dashed line is the Keplerian period, $P_{\rm K}$.
300: The $m=0$ mode ({\it wide, dark-shaded region}) is plotted for
301: $\dot{M}=10^{16}-10^{18}\textrm{ g s}^{-1}$
302: ({\it from top to bottom}), using eq. (4). The $m=-1$ mode
303: is also plotted for $\dot{M}=10^{16}-10^{18}\textrm{ g s}^{-1}$
304: ({\it from top to bottom}),
305: for both $P_{\rm SL}=P_{\rm K}$ ({\it light-shaded region})
306: and $P_{\rm SL}=2P_{\rm K}$
307: ({\it medium-shaded region}), using eq. (5).
308: The $m=-1$ modes show less variation with $\dot{M}$ in
309: comparison to the $m=0$ mode.}
310: \end{figure}
311:
312: Besides predicting multiple classes of DNOs with different period
313: ranges, we also predict that each of these groups will have
314: a different dependence on $\dot{M}$. This can be seen from the size
315: of each shaded band, which is much wider for the $m=0$ mode than the
316: $m=-1$ mode. On average the
317: DNOs with $P\lesssim P_{\rm K}$ show less variation with $\dot{M}$,
318: qualitatively consistent with the $m=-1$ mode of our model.
319: % Our model implies
320: % that the reason many DNO periods lie in the range of
321: % $20-30\textrm{ s}$ (see Warner 2004 since Table 1 is incomplete) is
322: % due to the WD masses being grouped around $0.6-1.0M_\odot$.
323: % and not necessarily indicative of other properties, such as the WD spin.
324:
325: To investigate the $\dot{M}$ dependence in more detail we
326: plot our predicted DNO periods versus $\dot{M}$ for an $M=1.0M_\odot$
327: WD in Figure 2. This illustrates the shallower dependence for the
328: $m=-1$ mode. The separation and relative slopes of the $m=-1$ and
329: $m=0$ modes are suggestive of the frequency doubling
330: that was seen by MR01 during the 1996 October outburst
331: of SS Cyg. In both the beginning and tail of the burst, they saw a DNO
332: with $P=6.59-8.23\textrm{ s}$ and following a
333: $P\propto L^{-\beta}$ power law with $\beta=0.097$.
334: Near the burst peak, $P$ suddenly shifted to $2.91\textrm{ s}$
335: and then decreased with a shallower power law with $\beta=0.021$.
336: Indeed, we find similar periods and power laws when we use a
337: larger WD mass.
338: \begin{figure}
339: \epsscale{1.2}
340: \plotone{f2.eps}
341: \figcaption{SL nonradial oscillation period compared to
342: $\dot{M}$ for an $M=1.0M_\odot$ WD.
343: The thick solid line is the $m=0$ mode, while the
344: shaded region shows the $m=-1$ mode for a range of
345: $P_{\rm SL}=(1.0-1.5)\times P_{\rm K}$ ({\it bottom to top}). The dashed lines
346: show the $m=1$ mode for the same range of $P_{\rm SL}$.}
347: \end{figure}
348:
349: In Figure 2, we also plot the $m=1$ retrograde mode for
350: $P_{\rm SL}=(1-1.5)\times P_{\rm K}$.
351: Owing to the difference taken in equation (5), this mode has a much
352: more complicated dependence on $\dot{M}$, no longer being
353: a power law nor monotonic.
354: This mode has a higher period than typical DNOs,
355: and may be relevant for the lpDNOs.
356: Warner (2004) identified 17 CVs as showing such oscillations.
357: The lpDNOs typically have larger amplitudes than DNOs and,
358: like DNOs, are not seen in every DN outburst.
359: Often, both of these oscillations
360: are seen simultaneously. To positively identify these as modes requires
361: checking for the predicted scalings between $P$ and $\dot{M}$.
362: This test may negate a mode explanation for lpDNOs since
363: Warner, Woudt \& Pretorius (2003) claim to find no such
364: correlation for these oscillations and thus favor a spin-related
365: mechanism. On the other hand, on separate occasions there
366: have been $32-36\textrm{ s}$
367: (Robinson \& Nather 1979) and $83-110\textrm{ s}$ (Mauche 2002b)
368: oscillations seen from SS Cyg, in the domain expected for these modes.
369: This wide spread of periods may be explained by
370: the $m=1$ mode's steeper dependence on $\dot{M}$.
371: The $32-36\textrm{ s}$ oscillations showed much less coherence than
372: typical DNOs, which could be a result of the retrograde oscillations
373: beating against the accretion disk.
374: Other CVs such as VW Hyi show similar modes in the range of $\sim90\textrm{ s}$
375: (Haefner, Schoembs \& Vogt 1977, 1979), which may be of similar origin.
376: % Further time dependent studies are required to understand
377: %whether a $m=1$ SL oscillation is observed.
378:
379: \section{Discussion and Conclusions}
380:
381: We propose that DNOs in outbursting CVs are nonradial oscillations
382: in a hot layer of freshly accreted material near the WD equator.
383: A large number of DNO properties are then simply understood:
384: (1) the highly sinusoidal nature of the oscillations is consistent with
385: nonradial oscillations, (2) the periods can change on the timescale of
386: accretion because
387: there is little mass in the layer ($\lesssim10^{21}\textrm{ g}$; PB04),
388: (3) the periods vary inversely with $\dot{M}$ because they have the
389: temperature scaling of shallow surface waves, (4) the covering fraction is
390: naturally small for the SL, (5) the DNOs are only seen during DN outbursts
391: because this is the only time when an optically thick layer of material
392: can build up at the equator, and (6) the largest pulsed amplitude is in the
393: EUV, consistent with the SL temperature.
394:
395: In support of our hypothesis, we presented a simple, phenomenologically
396: motivated model that quantitatively explains many of the
397: DNO's features. We compared our model with the overall population of
398: DNOs, both as a function of WD mass and accretion rate. In each case, it
399: is necessary to consider both the $m=0$ mode
400: and the $m=-1$ (prograde) mode to explain the range of periods
401: observed and the different $P$-$\dot{M}$
402: scalings.
403: The majority of DNOs are consistent with the period of the $m=0$ mode,
404: and this may be related to its latitudinal propagation. The speed
405: of shallow surface waves is $\approx gh$ so that they slow down as
406: the scale height decreases. This steers an initially
407: azimuthally propagating mode to instead travel perpendicular
408: to the SL edge, similar to ocean waves at the beach, and
409: may explain why $m=0$ modes are favored over $m\neq0$
410: modes. In the 1996 DN outburst of
411: SS Cyg (MR01), the $m=-1$ mode is only seen
412: near the outburst peak. This may indicate that a wide enough region
413: without differential rotation is only present when there
414: is sufficient spreading at high $\dot{M}$.
415:
416: There are still many difficulties that must be addressed about this
417: idea for explaining DNOs. From the SL model (PB04) we borrowed the
418: concept of hot material in hydrostatic balance covering a small fractional
419: area of the WD, but we ignored important details of these calculations,
420: such as differential rotation and the change of scale height with
421: latitude. The data do not require such additional complications,
422: so we refrain from including them for now. In a more
423: sophisticated model, it would probably still be true that
424: $k\propto1/(fR)$,
425: but an eigenvalue calculation would determine the
426: constant of proportionality. A full calculation may
427: also help explain the variety of power laws,
428: $\beta\equiv-d\log P/d\log L$,
429: for various CVs, since our model only predicts one unique
430: power law index for the $m=0$ mode. The WD spin
431: (which has so far been neglected) may be an
432: important differentiating variable as well.
433:
434: Another important property of WDs that would affect the modes is
435: a magnetic field. A strong field inhibits shearing between the
436: SL and WD and modifies the frequency of shallow surface
437: waves. It is therefore interesting that no intermediate polars
438: (IPs) have shown DNOs or lpDNOs. Even LS Peg and
439: V795 Her, neither of which are IPs, but both showing polarization modulations
440: (Rodr\'{i}quez-Gil et al. 2001, 2002) indicative of a reasonably strong
441: magnetic field, are without DNOs or lpDNOs.
442: Further studies should also work toward an understanding of
443: the excitation mechanism for the modes, which would help explain
444: the high coherence typical of DNOs. Material deposited at the WD equator
445: spreads fairly quickly, $10-100\textrm{ s}$ (PB04), so that this material must
446: be moving through the oscillating region on timescales
447: of order the mode period. In light of the many strengths of the SL mode
448: explanation of DNOs we do not abandon it because of this problem,
449: but this is definitely a weakness that puts limits
450: any proposed excitation mechanism.
451:
452: Our explanation of DNOs raises interesting questions about the
453: relationship between oscillations originating from accreting compact
454: objects. Mauche (2002b) showed that there is a strong correlation in
455: the high to low oscillation frequency ratio of WDs, NSs,
456: and black holes (BHs). Using this picture, DNOs are associated with
457: the kilohertz QPOs of low mass X-ray binaries. Suggestively, the
458: Fourier frequency resolved spectroscopy of NSs
459: (Gilfanov, Revnivtsev \& Molkov 2003)
460: imply that both the normal branch oscillations and the kilohertz
461: QPOs are created in the NS boundary layer, strengthening the
462: possibility that our SL mode model may also apply in this case.
463: For a typical NS mass and
464: radius our model results in frequencies in the range
465: expected for kHz QPOs, but it does not explain their complicated
466: $P$-$L$ relation (the ``parallel
467: tracks;'' van der Klis 2000). Interestingly, DNOs may
468: also show the parallel tracks phenomenon, as seen in three observations
469: of SS Cyg (Mauche 2002a), once again supporting the correlation.
470: On the other hand, continuing the analogy to
471: WD and BH oscillations is problematic because in the case of
472: BHs there is no surface for nonradial oscillations.
473:
474: We thank Phil Arras and Christopher Mauche for many helpful
475: discussions. We also thank the referee for asking the right questions,
476: which inspired a significantly revised manuscript.
477: This work was supported by the National Science Foundation under
478: grants PHY99-07949 and AST02-05956, and by the Joint Institute
479: for Nuclear Astrophysics through NSF grant PHY02-16783.
480:
481: \begin{thebibliography}{}
482:
483: \bibitem[]{}
484: Bildsten, L. Ushomirsky, G. \& Cutler, C. 1996, \apj, 460, 827
485:
486: \bibitem[]{}
487: C\'{o}rdova, F. A., Chester, T. J., Tuohy, I. R. \& Garmire, G. P. 1980,
488: \apj, 235, 163
489:
490: \bibitem[]{}
491: G\"{a}nsicke, B. T. et al. 2001, \aap, 374, 656
492:
493: \bibitem[]{}
494: Gilfanov, M., Revnivtsev, M. \& Molkov, S. 2003, \aap, 410, 217
495:
496: \bibitem[]{}
497: Haefner, R., Schoembs, R. \& Vogt, N. 1977, \aap, 61, 37
498:
499: \bibitem[]{}
500: Haefner, R., Schoembs, R. \& Vogt, N. 1979, \aap, 77, 7
501:
502: \bibitem[]{}
503: Inogamov, N. A. \& Sunyaev, R. A. 1999, AstL, 25, 269
504:
505: \bibitem[]{}
506: Knigge, C., Drake, N., Long, K. S., Wade, R. A., Horne, K. \&
507: Baptista, R. 1998, \apj, 499, 429
508:
509: \bibitem[]{}
510: Mauche, C. W. 1996, in Cataclysmic Variables and Related Objects, ed.
511: A. Evans \& J. H. Wood (Dordrecht: Kluwer), 243
512:
513: \bibitem[]{}
514: Mauche, C. W. 2002a, in ASP Conf. Ser. 264, Continuing the Challenge
515: of EUV Astronomy: Current Analysis and Prospects for the Future, ed.
516: S. Howell, J. Dupuis, D. Golombek \& J. Cullison (San Francisco: ASP),
517: 75
518:
519: \bibitem[]{}
520: Mauche, C. W. 2002b, \apj, 580, 423
521:
522: \bibitem[]{}
523: Mauche, C. W. 2004, \apj, 610, 422
524:
525: \bibitem[]{}
526: Mauche, C. W. \& Robinson, E. L. 2001, \apj, 562, 508 (MR01)
527:
528: \bibitem[]{}
529: Paczy\'{n}ski, B. 1978, in Nonstationary Evolution of Close Binaries,
530: ed. A. Zytkow (Warsaw: Polish Scientific Publishers), p. 89
531:
532: \bibitem[]{}
533: Papaloizou, J. \& Pringle, J. E. 1978, \mnras, 182, 423
534:
535: \bibitem[]{}
536: Patterson, J. 1981, \apj, 45, 517
537:
538: \bibitem[]{}
539: Piro, A. L. \& Bildsten L. 2004, \apj, 610, 977
540:
541: \bibitem[]{}
542: Popham, R. 1999, \mnras, 308, 979
543:
544: \bibitem[]{}
545: Ritter, H. \& Kolb, U. 2003, \aap, 404, 301
546:
547: \bibitem[]{}
548: Robinson, E. L. \& Nather, R. E. 1979, \apjs, 39, 461
549:
550: \bibitem[]{}
551: Rodr\'{i}quez-Gil, P., Casares, J., Mart\'{i}nez-Pais, I. G. \& Hakala, P.
552: 2002, in ASP Conf. Ser. 261, The Physics of Cataclysmic Variables
553: and Related Objects, ed. B. T. G\"{a}nsicke, K. Beuermann \&
554: K. Reinsch (San Francisco: ASP), 533
555:
556: \bibitem[]{}
557: Rodr\'{i}quez-Gil, P., Casares, J., Mart\'{i}nez-Pais, I. G., Hakala, P.
558: \& Steeghs, D. 2001, \apj, 548, L49
559:
560: \bibitem[]{}
561: Shakura, N. I. \& Sunyaev, R. A. 1973, \aap, 24, 337
562:
563: \bibitem[]{}
564: Truran, J. W. \& Livio, M. 1986, \apj, 308, 721
565:
566: \bibitem[]{}
567: van der Klis, M. 2000, ARA\&A, 38, 717
568:
569: \bibitem[]{}
570: Warner, B. 2004, \pasp, 116, 115
571:
572: \bibitem[]{}
573: Warner, B. \& Robinson, E. L. 1972, Nature Phys. Sci., 239, 2
574:
575: \bibitem[]{}
576: Warner, B. \& Woudt, P. A. 2002, \mnras, 335, 84
577:
578: \bibitem[]{}
579: Warner, B., Woudt, P. A. \& Pretorius, M. L. 2003, 344, 1193
580:
581: \bibitem[]{}
582: Woudt, P. A. \& Warner, B. 2002, \mnras, 333, 411
583:
584: \end{thebibliography}
585:
586: \end{document}
587:
588:
589:
590: