1: \documentclass{emulateapj}

2: \usepackage{natbib}

3: 

4: \def\gsim{\;\rlap{\lower 2.5pt

5:  \hbox{$\sim$}}\raise 1.5pt\hbox{$>$}\;}

6: \def\lsim{\;\rlap{\lower 2.5pt

7:    \hbox{$\sim$}}\raise 1.5pt\hbox{$<$}\;}

8: \def\msolarunits{$h^{-1}$ $M_{\odot}$ }

9: \def\lengthunits{$h^{-1}$ kpc }

10: \def\velocityunits{km s$^{-1}$ }

11: \def\timeunits{$h^{-1}$ Gyr }

12: 

13: \begin{document}

14: 

15: \title{Antitruncated Stellar Disks via Minor Mergers}

16: \author{Joshua D. Younger, T. J. Cox, Anil C. Seth, \& Lars Hernquist}

17: 

18: \affil{Harvard--Smithsonian Center for Astrophysics,

19: 	 60 Garden Street, Cambridge, MA 02138}

20: 

21: \email{jyounger@cfa.harvard.edu}

22: 

23: \begin{abstract}

24: 

25: We use hydrodynamic simulations of minor mergers of galaxies to

26: investigate the nature of surface brightness excesses at large radii

27: observed in some spiral galaxies: antitruncated stellar disks.  We find that

28: this process can produce the antitruncation via two competing

29: effects: (1) merger-driven gas inflows that concentrate mass in the

30: center of the primary galaxy and contract its inner density profile;

31: and (2) angular momentum transferred outwards by the interaction,

32: causing the outer disk to expand.  In our experiments, this requires

33: both a significant supply of gas in the primary disk, and that the

34: encounter be prograde with moderate orbital angular momentum.  The

35: stellar surface mass density profiles of our remnants both

36: qualitatively and quantitatively resemble the broken exponentials

37: observed in local face--on spirals that display antitruncations.

38: Moreover, the observed trend towards more frequent antitruncation

39: relative to classical truncation in earlier Hubble types is consistent

40: with a merger-driven scenario.

41: 

42: \end{abstract}

43: 

44: \keywords{galaxies: kinematics and dynamics, galaxies: interactions,

45: galaxies: formation, galaxies: evolution, galaxies: spiral,

46: galaxies: structure, methods: $n$--body simulations}

47: 

48: \section{Introduction}

49: 

50: Through the pioneering studies of \citet{patterson1940} and

51: \citet{deVaucouleurs1959}, it was first recognized that the stellar

52: disks of most spiral galaxies are well-approximated by an exponential

53: surface brightness profile.  However, \citet{vanderKruit1979} later

54: found that in many cases this description breaks down at large radius,

55: where the disk surface density appears to be truncated

56: \citep[see][and references therein]{pohlen2006}.

57: 

58: Many studies of truncated disks have examined the surface brightness

59: profiles of local edge--on systems \citep{vanderKruit1979,barnaby1992,

60: barteldrees1994, pohlen2000,deGrijs2001,vanderKruit2001,kregel2002,Florido2006b,Florido2006a} and at higher redshifts $z\sim 1$ \citep{perez2004,trujillo2005}.  This

61: choice of inclination facilitates detection of truncations, but is

62: subject to potential biases owing to the effects of dust extinction

63: and line--of--sight integration \citep[see, e.g.,][]{pohlen2006}.

64: Observations of face--on systems, which mitigate these complications,

65: were also successful at detecting truncation, but found that the

66: surface brightness profiles are better represented by a broken

67: exponential than a hard break \citep{pohlen2002,erwin2005,pohlen2006}.  \citet{hunter2006}

68: also note the existence of double--exponentials in their sample of 

69: very late--type spirals and dwarfs.

70: 

71: These studies also uncovered a broad range of behaviors, including

72: some disks that follow a pure exponential profile out to very large

73: radius \citep{BlandHawthorn2005,pohlen2006}, and -- as \citet{erwin2005} first

74: observed -- some that are {\it antitruncated}, with an excess surface brightness relative to an

75: exponential profile fitted to the inner disk

76: \citep{erwin2005,pohlen2006,pohlen2007,erwin2007}.  These extended

77: stellar disks dominate the light past 4--6 scale lengths, and have

78: flatter profiles with scale lengths that are $\sim50\%$ larger than

79: the inner disk.

80: 

81: A number of authors have proposed theoretical explanations for

82: truncated disks.  Dynamical arguments for star--formation thresholds

83: have been successful in motivating the locations of truncations

84: \citep{kennicutt1989,schaye2004,naab2006}.  However, observations of

85: ultraviolet emission

86: \citep{thilker2005,GildePaz2005,GildePaz2006,boissier2006}, young

87: stellar populations \citep{Cuillandre2001}, and HII regions

88: \citep{Ferguson1998} in extended disks suggest that there is indeed

89: star formation occurring beyond implied thresholds.

90: More recently, star formation models including a

91: variety of triggering mechanisms \citep[gravitational instabilities, spiral wave shocks, and stellar and turbulent compression;][]{Elmegreen2006} and N--body

92: simulations of angular momentum redistribution via bar instabilities 

93: \citep{Debattista2006} have been more successful at explaining

94: classically truncated disks. Despite these successes, theoretical mechanisms

95:  for producing antitruncated disks have received comparatively little attention.

96: 

97: While it is clear that secular processes can influence disk structure

98: and may produce truncated stellar disks, it has also become apparent

99: that disk galaxies exist within a hierarchical universe, in

100: which mergers are a frequent occurrence \citep{laceycole1993,somerville1999,

101: somerville2000}.  Furthermore, these mergers are likely to play an important

102: role in shaping the appearance of galaxies.

103: This is certainly true of collisions

104: between spiral galaxies of equal mass, so called major mergers, which have

105: been suggested as the dominant formation mechanism for present-day elliptical

106: galaxies \citep{toomre1972,toomre1977,negroponte1983,barnes1992,hernquist1992,

107: hernquist1993a,silk1993,naab2003,robertson2006b,robertson2006a,coxdutta2006}.

108: 

109: There has also been considerable study of the effects of minor mergers

110: ($M_{prim}/M_{sec} \gsim 3$) on the vertical structure and dynamics of

111: stellar disks \citep{quinn1986,quinn1993,walker1996,huang1997,

112: sellwood1998,velazquez1999,font2001,ardi2003,brook2004,brook2005,brook2006,

113: gauthier2006,hayashi2006,kazantzidis2007},

114: in addition to observational evidence for past interactions with

115: satellites as the origin of the Milky Way's thick disk

116: \citep{freeman2002,gilmore2002,wyse2006}.  Moreover, tidal

117: structures indicative of recent minor mergers have been observed in

118: both the Milky Way \citep{newberg2002,ibata2003} and M31

119: \citep{ibata2001,mcconnachie2003}.

120: 

121: In this work, we explore the effects of minor mergers on the structure

122: of stellar disks at large radius and find that, under certain conditions,

123: this provides a viable physical mechanism

124: for producing antitruncated disks.  We demonstrate this process using a

125: set of hydrodynamical simulations, which are described in \S~\ref{sec:sims}.

126: An overview of the merger process and specifically the dynamical response of the

127: stellar disk during a minor merger is provided in \S~\ref{sec:resp}.

128: \S~\ref{sec:fits} summarizes the surface density profile fitting

129: procedure, and \S~\ref{sec:gf} and \ref{sec:param} describe the dependence on

130: parameters of the interaction.  In \S~\ref{sec:discuss} we discuss our results

131: in comparison to observations of antitruncated disks and within the context of

132: hierarchical galaxy formation.  Finally, we conclude in \S~\ref{sec:conclude}.

133: 

134: 

135: 

136: \section{The Simulations}

137: \label{sec:sims}

138: 

139: For this study, we consider the effects of a 1:8 merger on the

140: stellar surface density of the primary component's

141: disk.  These interactions are both cosmologically common

142: \citep[see, e.g.,][]{laceycole1993,somerville1999,somerville2000} and

143: kinematically important enough to play a significant role in

144: determining the appearance of most present--day stellar disks, while at

145: the same time largely preserving the overall disk structure \citep[see,

146: e.g.,][]{quinn1993,walker1996,velazquez1999,font2001,kazantzidis2007}.

147: 

148: We consider the idealized case of an isolated interaction, in contrast

149: to much work on disk galaxy formation done in a full cosmological

150: context

151: \citep{font2001,ardi2003,brook2004,brook2005,gauthier2006,kazantzidis2007}.

152: The cosmological approach has the relative advantage of a more realistic accretion

153: history.  However, the isolated interactions analyzed here offer the

154: alternative benefit of examining the individual effects of a single

155: encounter, and allow us to efficiently sample the parameter space of interactions.  Furthermore, our approach allows the simulation to be performed at much higher resolution.  This helps capture the dynamical effects of minor mergers on the surface density profile at large radius, where resolution is critical.

156: 

157: \begin{table}

158: \begin{center}

159: \caption{Model Galaxy Parameters}

160: \begin{tabular}{ccc}

161: \hline

162: \hline

163: & {\sc Sb} & {\sc Im} \\

164: \hline

165: $M$ ($10^{10}$ \msolarunits) & 95 & 12 \\

166: $V_c$ (\velocityunits) & 160 & 80 \\

167: $c$ & 9 & 12 \\

168: $h_{D}$ (\lengthunits) & 4.1 & 1.6 \\

169: $M_D/M$ & 0.05 & 0.05 \\

170: $N_H$ & $1.0\times 10^6$ & $1.3\times 10^5$ \\

171: $N_B$ & $8\times 10^5$  & $4.3\times 10^4$ \\

172: \hline

173: \hline

174: \end{tabular}

175: \label{tab:models}

176: \end{center}

177: \end{table}

178: 

179: The simulations presented in this study were performed with {\sc

180: Gadget2} \citep{springel2005}, an N--Body/SPH (Smooth Particle

181: Hydrodynamics) code using the entropy conserving formalism of

182: \citet{springelhernquist2002}.  We include the effects of radiative

183: cooling and star formation, tuned to fit the observed Schmidt Law

184: \citep{schmidt1959,kennicutt1998}.  We also incorporate a

185: sub--resolution multi--phase feedback model of the interstellar medium

186: (ISM)

187: \citep{springelhernquist2003} -- softened ($q_{EOS} = 0.25$)

188: such that the mass--weighted ISM temperature is $\sim 10^{4.5}$ -- and

189: sink particles representing supermassive black holes that can accrete

190: gas and release isotropic thermal energy to the surrounding medium

191: \citep{springeldimatteo2005a}.  For further details on the progenitor

192: galaxy models, we refer to \citet{springeldimatteo2005b}, and to other

193: work done as part of a larger study of the effects of galaxy

194: interactions on the formation and evolution of galaxies

195: \citep{dimatteo2005, hopkinshernquist2005c,

196: hopkinshernquist2005b,hopkinshernquist2005a,

197: hopkinshernquist2005d,hopkinsall2006,hopkinsred2006,HopkinsSommerville2006,

198: Hopkins2007a,

199: robertson2006b,robertson2006a,CoxDimatteo2006,coxdutta2006,CoxJonsson2006}.

200: 

201: \begin{table}

202: \begin{center}

203: \caption{Simulation Orbital Parameters}

204: \begin{tabular}{ccccc}

205: \hline

206: \hline

207:  Name & $i$ & $f_g$ & $R_p$  \\

208:              &        &            & (\lengthunits) \\ 

209: \hline

210: \hline 

211: {\sc Sb0Im30Rp1} & $30^\circ$ & 0.0 & 5.0 \\

212: {\sc Sb2Im30Rp1} & $30^\circ$ &  0.2 & 5.0 \\

213: {\sc Sb4Im30Rp1} & $30^\circ$ & 0.4 & 5.0 \\

214: {\sc Sb8Im30Rp1} & $30^\circ$ & 0.8 & 5.0 \\

215: \hline

216:  {\sc Sb2Im0Rp1} & $0^\circ$  & 0.2 & 5.0 \\

217:  {\sc Sb2Im90Rp1} & $90^\circ$ &  0.2 & 5.0 \\

218:  {\sc Sb2Im150Rp1} & $150^\circ$ & 0.2 & 5.0 \\

219:  {\sc Sb2Im180Rp1} & $180^\circ$ & 0.2 & 5.0 \\

220: \hline

221: {\sc Sb2Im30Rp2} & $30^\circ$ & 0.2 & 2.5 \\

222: {\sc Sb2Im30Rp3} & $30^\circ$ & 0.2 & 10.0 \\

223: \hline

224: \hline

225: \end{tabular}

226: \label{tab:sims}

227: \end{center}

228: \end{table}

229: 

230: A summary of the galaxy models used in the simulations is provided in

231: Table~\ref{tab:models}, including the total (baryons and dark matter)

232: mass $M$, circular rotation velocity $V_c$, concentration parameter

233: $c$, initial disk scale length $h_D$, disk (stars and gas) mass

234: fraction $M_D/M$, number of dark matter particles in the halo $N_H$,

235: and baryonic (stars and gas) particles $N_B$.  Both are designed to be

236: representative of their eponymous local hubble types \citep[see

237: e.g.,][]{roberts1994}.

238: 

239: The different encounter configurations

240: considered are summarized in Table~\ref{tab:sims}. 

241: We assume zero--energy parabolic orbits

242: ($e=1$), as motivated by cosmological simulations

243: \citep{benson2005,khochfar2006}, with radius of pericenter $R_p$,

244: orbital inclination ($i$), and primary disk gas fraction $f_g$.  For

245: the primary disk ({\sc Sb}), we consider 20\% gas disks ($f_g=0.2$),

246: which are intended to be representative of disks in the local

247: universe \citep{mcgaugh1997,bell2000}, and higher gas fractions

248: ($f_g=0.4,0.8$) which are consistent with both the more gas--rich local

249: systems \citep{mcgaugh1997} and high redshift ($z\sim 2$) spirals

250: \citep{erb2006a}.  The secondary disk has a fixed gas fraction of 40\%

251: ($f_g=0.4$) -- with the exception of the purely collisionless

252: interaction {\sc Sb0Im30Rp1} which has $f_g=0$ -- and is consistent

253: with observations of dwarf galaxies and low--mass disks in the local

254: universe \citep{schombert2001,geha2006}.  The initial spins of the two disks

255: are not aligned, and are the same in all our simulations.

256: 

257: \section{Dynamical Response of the Stellar Disk to a Minor Merger}

258: \label{sec:resp}

259: 

260: We find that minor mergers can create antitruncated stellar disks in

261: face--on spirals.  To examine this effect in detail and

262: illustrate some of the generic features of a minor merger, we

263: concentrate on {\sc Sb2Im30Rp1} and its collisionless counterpart {\sc

264: Sb0Im30Rp1}.  In all cases, the stellar mass surface density profiles 

265: of the remnant are measured 1 \timeunits after the final coalescence -- or several orbital periods at the half mass radius -- to allow the remnant disk to reach a state of approximate dynamical equilibrium.

266: 

267: In our experiments, antitruncations are produced only when there is a

268: significant supply of gas in the primary disk.  The driving physical mechanism for

269: producing this outcome represents a competition between merger

270: driven inflows and transfer of angular momentum to large radius in the

271: remnant stellar disk; i.e., gas moves inwards while stars move outwards.  This

272: effect is manifest in the changes induced in both the gravitational

273: potential and angular momentum profile during the encounter.

274: 

275: \begin{figure}

276: \plotone{f1.eps}

277: \caption{Stellar surface density profiles ($\Sigma$) as a function of radial distance from the stellar center of mass ($R$) for an isolated 20\% gas disk {\sc Sb2}, and the remnant {\sc Sb2Im30Rp1}.  Included are the initial {\sc Sb} disk (dotted), an {\sc Sb2} disk evolved in isolation for 2 \timeunits (dashed), and the {\sc Sb2Im30Rp1} remnant (solid).  Also shown is contribution of stars from the secondary disk to the total stellar surface density in the remnant (dash--dot).}

278: \label{fig:min30_prof}

279: \end{figure}

280: 

281: In Figure~\ref{fig:min30_prof}, we present the stellar mass surface

282: density profiles for {\sc Sb2Im30Rp1}, compared to both the initial

283: primary disk {\sc Sb2} and the same initial disk evolved in isolation

284: for 2 \timeunits, as a function of radial distance from the stellar

285: center of mass $R$.  The primary disk is stable; when it is evolved in

286: isolation over several orbital periods, the inner scale length is only

287: marginally shorter owing to preferential star formation occurring near

288: the center \citep[SFR $\sim \rho_g^{1.4}$:][]{kennicutt1998}.  At

289: large radius, there is some fluctuation of the evolved, isolated disk

290: about the initial stellar mass surface density.  This owes to Poisson

291: noise arising from low particle counts at large $R$ and the

292: development of spiral structure from numerical noise associated with

293: the discretized dark matter halo \citep{hernquist1993}.

294: 

295: The surface density profile of the {\sc Sb2Im30Rp1} merger remnant shown

296: in Figure~\ref{fig:min30_prof} displays three key features.  First, within

297: $4$ \lengthunits, the surface density profile is steep, indicative of

298: a bulge component produced by the merger.  Second, the surface density

299: profile from $4 - 20$ \lengthunits is nearly identical to the primary disk.

300: Third, beyond $20$ \lengthunits, there is a clear excess of surface density

301: relative to the initial disk of the primary.

302: 

303: \begin{figure}

304: \plotone{f2.eps}

305: \caption{Same as Figure~\ref{fig:min30_prof}, but split into the two component stellar populations: ``progenitor'' stars which are present at the start of the simulation, and ``new'' stars which are formed during the interaction.  The dotted line in the right panel shows the initial gas distribution.}

306: \label{fig:min30_prof_pops}

307: \end{figure}

308: 

309: The excess surface density in the outer profile of the remnant {\sc Sb2Im30Rp1},

310: i.e., the antitruncation of its disk, qualitatively -- and, as we will see in

311: \S~\ref{sec:gf} and \S~\ref{sec:param}, quantitatively -- resembles

312: the broken exponentials of \citet{pohlen2006}, with an inner scale

313: length close to that of the initial disk.  In Figure~\ref{fig:min30_prof_pops}, 

314: we separate out the ``progenitor''

315: stellar particles that are initialized with the disks and ``new''

316: stellar particles formed from the gas during the interaction.  We

317: find that the antitruncation is dominated by progenitor stars.  An

318: antitruncated disk in {\sc Sb2Im30Rp1} is produced independent of a

319: fitted profile; a robust result with respect to any fitting procedure.

320: Separating out the stellar particles that originate in the secondary

321: stellar disk (see Figure~\ref{fig:min30_prof}), we find that the

322: increased outer surface density ($R \gsim 20$ \lengthunits)

323: is dominated by progenitor stars from the primary disk that have been

324: transferred to larger radius by the interaction.  Furthermore, we find

325: that these large $R$ features in the profile are rotationally

326: supported -- their median circular velocity in circular annuli is

327: $\sim 0.8-0.9$ times the Keplerian orbital velocity at that radius --

328: and thus long--lived.  We confirm this by evolving our remnant in

329: isolation for $\sim 10^{10}$ years, and find that the antitruncation

330: is not a transient feature.

331: 

332: \begin{figure} 

333: \plotone{f3.eps}

334: \caption{Stellar surface density profiles ($\Sigma$) as a function of radial distance from the stellar center of mass ($R$) for an isolated collisionless disk {\sc Sb0}, and the collisionless remnant {\sc Sb0Im30Rp1}.  Included are the initial {\sc Sb} disk (dotted), an {\sc Sb0} disk evolved in isolation for 2 \timeunits (dashed), and the {\sc Sb0Im30Rp1} remnant (solid).  The grey dashed line is an exponential fit to the collisionless remnant {\sc Sb0Im30Rp1} for $R > 10$ \lengthunits, meant to highlight that this profile, in contrast to the 20\% gas remnant  {\sc Sb2Im30Rp1} (see Figure~\ref{fig:min30_prof}), would 

335: not likely be observed as significantly antitruncated.} 

336: \label{fig:min30_prof_nocoll}

337: \end{figure}

338: 

339: The surface density profiles shown in Figure~\ref{fig:min30_prof_nocoll}

340: demonstrate that the collisionless interaction {\sc Sb0Im30Rp1} displays no

341: antitruncation in the stellar mass surface density profile of its remnant.

342: Rather, its surface density profile,

343: which has been tilted, increasing the scale length at all

344: radii, does not have a well--defined break. Features at large $R$ are, as in the previous case, also rotationally supported.

345: 

346: Although there is some evidence for bulge formation in {\sc

347: Sb2Im30Rp1} at $R\lsim 2$ \lengthunits, {\sc Sb0Im30Rp1} shows a much

348: more pronounced bulge which remains prominent out to larger radii.

349: This is expected; phase space conservation in a collisionless

350: interaction leads to lower phase space densities in the core of the

351: collisionless remnant, and accordingly a more diffuse spheroid

352: component \citep{hsh1993}.  While bulge growth via minor mergers is a

353: topic worthy of further study, we postpone a detailed analysis to

354: future work and point out the qualitative difference between the

355: surface density profiles of {\sc Sb2Im30Rp1} and {\sc Sb0Im30Rp1}.

356: 

357: \begin{figure}

358: \plotone{f4.eps}

359: \caption{Gravitational potential relative to the initial disk, as a function of radial distance from the stellar center of mass ($R$), for the {\sc Sb2Im30Rp1} remnant (solid) and the collisionless {\sc Sb0Im30Rp1} remnant (dashed).  We find that dissipation efficiently deepens the central potential of the remnant relative to the initial disk.}

360: \label{fig:min30_gravpot}

361: \end{figure}

362: 

363: To investigate the physical processes driving the antitruncation in

364: our simulations, we first consider the gravitational potential as a

365: function of radial distance from the stellar center of mass of

366:  {\sc Sb0Im30Rp1}, and {\sc Sb2Im30Rp1} as compared to the initial disk in

367: Figure~\ref{fig:min30_gravpot}.  For certain orbits, a minor merger

368: can drive nuclear inflows of gas

369: \citep{hernquist1989,mihos1994,mihos1995}, fueling a centrally

370: concentrated starburst and creating a deeper potential well there

371: owing to the effects of gas disspation (see

372: Figure~\ref{fig:min30_prof_pops}).  This deeper potential will

373: contract the remnant profile, counteracting the broadening of the profile 

374: owing to 

375: angular momentum transfer and maintaining an inner scale length

376: similar to the initial primary disk.  Furthermore, the newly formed stars will be more

377: concentrated than the progenitor stars (see

378: Figure~\ref{fig:min30_prof_pops}), which will also tend to contract

379: the inner scale length and populate the stellar mass surface density

380: at small $R$.

381: 

382: At the same time, the interaction transfers angular momentum and

383: stellar mass to the outer disk.  In Figure~\ref{fig:min30_ltot}, we

384: show the total angular momentum in circular annuli as a function of

385: $R$.  In both {\sc Sb2Im30Rp1} and {\sc Sb0Im30Rp1}, the angular

386: momentum at large $R$ is nearly double that of the initial and evolved

387: disks; this transfer occurs whether or not gas is included.  In {\sc

388: Sb2Im30Rp1}, the magnitude of the angular momentum 

389: in shells at $R < 10$ \lengthunits is also higher, owing to the

390: more efficient inflows generated by dissipation during the

391: interaction.  In the collisionless case, because the inner potential

392: is not as deep, the disk expands more uniformly in response to this

393: transfer, and therefore does not show an antitruncation.  Thus,

394: the antitruncation at large $R$ results from expansion of the outer

395: disk -- similar to that first noted by \citet{quinn1993} -- in response to a net transfer of angular momentum.  When gas is present, the inner potential is deep enough to contract this inner

396: profile and maintain an inner scale length similar to that of the initial primary disk.

397: 

398: \section{Fitting the Surface Density Profile}

399: \label{sec:fits}

400: 

401: \begin{figure}

402: \plotone{f5.eps}

403: \caption{Total angular momentum in the stellar disk, as a function of radial distance from the stellar center of mass ($R$).  Shown are: the initial disk (dotted), an isolated disk evolved for 2 \timeunits (dashed), and (solid) the remnant for both a gas fraction of $f_g=0.2$ (left: {\sc Sb2} disk and {\sc Sb2Im30Rp1} remnant) and a collisionless interaction (right: {\sc Sb0} disk and {\sc Sb0Im30Rp1} remnant).  We find that angular momentum is transferred to the outer disk, which increases its scale length relative to the inner disk, and creates the antitruncation.}

404: \label{fig:min30_ltot}

405: \end{figure}

406: 

407: In \S~\ref{sec:resp}, we find antitruncated stellar disks independent

408: of the fitted surface density profile.  However, to facilitate

409: comparison to the observational constraints on antitruncated disks, we

410: fit profiles to the stellar mass density profiles of our remnants,

411: projected face--on, and include both progenitor and new stars.  The

412: fit is performed, as in \S~\ref{sec:resp}, 1 \timeunits after the

413: final coalescence, so the remnant reaches a state of approximate

414: dynamical equilibrium.  

415: 

416: Following \citet{pohlen2006}, we mask out the

417: inner 5 \lengthunits of the disk and first fit an exponential profile

418: \citep{patterson1940,freeman1970} with scale length $h_{D1}$ to the

419: inner disk, then fit a second exponential profile with scale length

420: $h_{D2}$ to the outer disk\footnote{The fitting range for the outer disk was

421: set by an initial guess for $R_{br}$.  We found that the fitted value for

422: $R_{br}$ and the outer disk parameters were largely insensitive to this choice.}.  

423: The break radius $R_{br}$ is defined at

424: the intersection of the inner and outer disk profiles.  The fits are

425: performed using a Levenberg-Marquardt least-squares minimization

426: routine, with bins weighted by the Poisson error ($\sigma_{i} \sim

427: \Sigma_{i}$).  The results are tabulated in Table~\ref{tab:fits}.  We

428: note as a caveat that the values of the parameters in our fits are

429: somewhat sensitive to the manner in which the bins are weighted.

430: However, using slightly different weights, such as ``flux'' ($\Sigma

431: r^2$), does not qualitatively affect our results.

432: 

433: Since we are considering the stellar mass distribution out to large

434: radius, resolution effects are particularly important.  We

435: note that all of our simulations have more than $1.5\times 10^{4}$

436: stellar particles in the ``outer'' disk ($4h_{D1} < R < 10h_{D1}$),

437: and a majority have $\gsim 3.0\times 10^{4}$ over the same

438: range. Furthermore, tripling the number of particles did not change

439: the surface density profile at large radius ($R \gsim 20$

440: \lengthunits) by more than 15\%.  Therefore, we find that our

441: resolution is sufficient to make robust claims about the the stellar

442: surface mass density profiles at large radius.

443: 

444: 

445: \begin{table}

446: \begin{center}

447: \caption{Fitted Surface Density Profiles}

448: \begin{tabular}{ccccc}

449: \hline

450: \hline

451:  Name & $h_{D1}$ &  $h_{D2}/h_{D1}$ & $R_{br}/h_{D1}$  \\

452:              &  (\lengthunits) & (\lengthunits) &  &   \\ 

453: \hline

454: \hline 

455: {\sc Sb0Im30Rp1} & 5.99 & \nodata & \nodata \\

456: {\sc Sb2Im30Rp1} & 4.00 & 1.69 & 3.27 \\

457: {\sc Sb4Im30Rp1} & 3.54 & 1.85 & 3.33 \\

458: {\sc Sb8Im30Rp1} & 3.95 & 1.26 & 3.85 \\

459: \hline

460:  {\sc Sb2Im0Rp1} & 3.38 & 1.72 & 2.47 \\

461:  {\sc Sb2Im90Rp1} & 4.10 & 1.24 & 2.64 \\

462:  {\sc Sb2Im150Rp1} & 4.64 & 1.61 & 5.45 \\

463:  {\sc Sb2Im180Rp1} & 4.22 & 1.27 & 3.49 \\

464: \hline

465: {\sc Sb2Im30Rp2} & 3.70 & 1.80 & 3.10 \\

466: {\sc Sb2Im30Rp3} & 4.28 & 1.61 & 2.87 \\ 

467: \hline

468: \hline

469: \end{tabular}

470: \label{tab:fits}

471: \end{center}

472: \end{table}

473: 

474: \section{Dependence on the Gas Content}

475: \label{sec:gf}

476: 

477: Because antitruncation appears to be a dissipational effect, we expect

478: the degree and location of the break to depend on the gas content of

479: the primary disk.  Therefore, we perform a set of experiments varying

480: the gas fraction of the primary disk, while holding the orbital

481: parameters fixed.  In Figure~\ref{fig:gf_scale}, we present the

482: stellar surface mass density profiles of the remnant, including both

483: the progenitor and new stellar particles, for the four different gas

484: fractions listed in \S~\ref{sec:sims}.  In addition, we show the

485: broken exponential disk profiles listed in Table~\ref{tab:fits}.

486: 

487: We find that {\sc Sb0Im30Rp1} -- the collisionless interaction, see

488: also Figure~\ref{fig:min30_prof_nocoll} -- is not well--fitted by a

489: broken exponential as in \citet{erwin2005} and \citet{pohlen2006}, and therefore would not

490: be observed to be antitruncated.  Rather, the scale length increases

491:  to $h_{D1} \approx 6$\lengthunits, relative to $h_{D} = 4.14$

492: \lengthunits initially, with a substantial bulge--like component following

493: an $R^{1/4}$ \citep{deVaucouleurs1959} profile, which is shown in the

494: fit presented in Figure~\ref{fig:gf_scale}.

495: 

496: Figure~\ref{fig:gf_pot} shows 

497: that the inner potential is deeper -- and therefore

498: the inner scale length shorter -- for increased gas fractions (see

499: Table~\ref{tab:fits}).  The antitruncation is strongest at $f_g=0.2$

500: and 0.4, while for the highest gas fraction $f_g=0.8$, the relative scale

501: length $h_{D1}/h_{D2}$ is significantly flatter, resulting in a less

502: pronounced break.  This is related to the composition of the initial

503: disk; very gas rich disks have fewer stars remaining in the outer disk

504: after the initial inflow of gas.  Therefore, as the outer disk

505: expands, less stellar mass resides at large radii.

506: 

507: \begin{figure*}

508: \plotone{f6.eps}

509: \caption{Stellar mass surface density of the remnant as a function of radial distance from the center of mass $R$ in units of the inner scale length $h_{D1}$ (see Table~\ref{tab:fits} for fitted values), and its dependence on the initial gas fraction of the primary disk $f_g$: from left to right $f_g = $ 0.0 ({\sc Sb0Im30Rp1}), 0.2 ({\sc Sb2Im30Rp1}), 0.4 ({\sc Sb4Im30Rp1}), and 0.8 ({\sc Sb8Im30Rp1}).  We include both the binned simulation data (crosses), and fitted disk profiles (solid line).  The collisionless ($f_g = 0.0$) remnant is not well--described by a double exponential, and instead has been fit with a combination disk and \citet{deVaucouleurs1959} profile.}

510: \label{fig:gf_scale}

511: \end{figure*}

512: 

513: \begin{figure}

514: \plotone{f7.eps}

515: \caption{Response of the gravitational potential to a minor merger as a function of radial distance from the stellar center of mass $R$, and its dependence on the initial gas fraction of the primary disk $f_g$: $f_g = $ 0.0 (dashed, {\sc Sb0Im30Rp1}), 0.2 (solid, {\sc Sb2Im30Rp1}), 0.4 (dash--dot, {\sc Sb4Im30Rp1}), and 0.8 (long dash, {\sc Sb8Im30Rp1}).  The dotted line shows the initial potential of the primary disk.}

516: \label{fig:gf_pot}

517: \end{figure}

518: 

519: \section{Dependence on the Orbital Parameters}

520: \label{sec:param}

521: 

522: Since the processes that create the antitruncation are dynamical, 

523: the degree and location of the break should be sensitive to

524: the orbital parameters of the encounter.  To investigate this,

525: we vary the orbital parameters -- the orbital inclination

526: $i$ and radius of pericenter $R_p$ -- holding the gas content of the

527: primary disk fixed.

528: 

529: \subsection{Orbital Angular Momentum}

530: \label{sec:lorbit}

531: 

532: To test the ability of the deeper potential to mitigate against the

533: expansion of the inner disk owing to angular momentum transfer,

534: we vary the total angular momentum of the secondary's orbit, while

535: fixing the orbital inclination.  This is done by adjusting the location of 

536: the pericenter of the secondary's orbit.

537: 

538: We find that the antitruncation is largely insensitive to increasing the

539: orbital angular momentum.  Figure~\ref{fig:rp_scale} shows the surface

540: density profiles for three different radii of pericenter ($R_p$)

541: spanning a factor of four in orbital angular momentum.  The broken exponential fits

542: are listed in Table~\ref{tab:fits}. Though the inner scale length does increase

543: over this range -- despite a deeper inner potential (see

544: Figure~\ref{fig:rp_pot}) -- it does so by less than 20\%.  Over the same

545: range, the break radius and relative scale lengths decrease by less

546: than 10\%.  Therefore, we expect that at fixed inclination, most

547: orbits would create similar antitruncations.

548: 

549: \begin{figure*}

550: \plotone{f8.eps}

551: \caption{Stellar surface mass density of the remnant as a function of radial distance from the center of mass $R$ in units of the inner scale length $h_{D1}$ (see Table~\ref{tab:fits} for fitted values), and its dependence on the radius at pericenter $R_p$ of the orbit of the secondary: from left to right $R_p = $ 2.5 ({\sc Sb2Im30Rp2}), 5.0 ({\sc Sb2Im30Rp1}), and 10.0 ({\sc Sb2Im30Rp3}) \lengthunits.  We include both the binned simulation data (crosses), and fitted disk profiles (solid line).}

552: \label{fig:rp_scale}

553: \end{figure*}

554: 

555: \begin{figure}

556: \plotone{f9.eps}

557: \caption{Response of the gravitational potential to a minor merger as a function of radial distance from the center of mass $R$, and its dependence on the radius at pericenter $R_p$ of the orbit of the secondary: $R_p = $ 2.5 (dashed, {\sc Sb2Im30Rp2}), 5.0 (solid, {\sc Sb2Im30Rp1}), and 10.0 (dot--dash, {\sc Sb2Im30Rp3}) \lengthunits.}

558: \label{fig:rp_pot}

559: \end{figure}

560: 

561: \subsection{Orbital Inclination}

562: \label{sec:iorbit}

563: 

564: Varying the orbital inclination of the interaction introduces two

565: competing effects.  First, prograde minor mergers are more efficient

566: than retrograde mergers at coupling to the rotation of the primary

567: disk.  At the same time, coplanar minor mergers are more efficient at

568: transferring angular momentum to the stellar orbits, inducing bar

569: formation, and centrally concentrating gas and stars.  As a result,

570: the closer the interaction is to coplanar, the deeper the remnant's

571: inner potential.  We present results for five different inclinations,

572: as outlined in \S~\ref{sec:sims} and Table~\ref{tab:sims}, in

573: Figures~\ref{fig:inc_scale} and \ref{fig:inc_pot}.

574: 

575: We first consider the prograde and polar interactions -- {\sc

576: Sb0Im0Rp1}, {\sc Sb0Im30Rp1}, and {\sc Sb0Im90Rp1} -- to illustrate

577: the combined effects of the merger-driven inflow.  As the inclination

578: increases from $i=0$ (coplanar prograde) to $i=90$ (polar), the

579: potential at small scale radius is shallower (see

580: Figure~\ref{fig:inc_pot}).  Accordingly, the inner scale radii of {\sc

581: Sb0Im0Rp1}, {\sc Sb0Im30Rp1}, and {\sc Sb0Im90Rp1} are successively

582: larger with flatter relative outer to inner scale lengths

583: $h_{D2}/h_{D1}$.  As a result, the antitruncation is less pronounced

584: for increasing $i$.

585: 

586: When the interaction is retrograde, as with {\sc Sb150Im0Rp1} and {\sc

587: Sb180Im0Rp1}, the orbital frequency of the secondary is out of resonance

588: with the orbits of particles in the inner disk.  This makes the encounter

589: much less efficient at transferring angular momentum to the outer disk.  As a 

590: result, though there are antitruncations in our retrograde experiments, 

591: they are far less pronounced than those in prograde encounters.  

592: Therefore, our simulations suggest that prograde minor mergers will be most

593: effective at producing antitruncations.

594: 

595: \section{Discussion}

596: \label{sec:discuss}

597: 

598: We find that minor mergers can create antitruncated stellar disks in

599: face--on spiral galaxies, and that this antitruncation is produced by

600: a competition between merger driven inflows of gas and transfer of

601: angular momentum to large $R$ in the remnant stellar disk that moves

602: stars outwards.  Because this process requires gaseous inflows, 

603: antitruncated stellar disks are produced only when gas dissipation and star

604: formation are included.  Moreover, the magnitude and location of

605: antitruncation is related to both the gas content of the primary disk

606: and the orbital parameters of the interaction.  These features in the

607: surface stellar mass density profile are rotationally supported, and

608: therefore long--lived and likely to be observed in local spirals.

609: 

610: This merger-driven scenario for the production of antitruncated disks

611: is supported by observations of face--on spirals that find antitruncated

612: disks occur more frequently in earlier--type spirals and in higher density

613: environments \citep{pohlen2006}.  The authors note

614: that the frequency of antitruncated disks

615: increases from 20\% in Sd types to 50\% in Sb types, while the

616: fraction of classically truncated disks decreases from 40\% to 10\%

617: over the same range.  This agrees qualitatively with minor mergers as the 

618: physical mechanism driving disk antitruncation: spirals in higher density environments are

619: more likely to have undergone minor mergers which create

620: systematically earlier Hubble types \citep{naab2003}.  At the same time, both 

621: \citet{erwin2005} and \citet{pohlen2006} show observational evidence for asymmetries

622: or recent interactions in antitruncated systems, which further supports a merger--driven scenario.

623: 

624: Using the extended Press-Schechter formalism

625: \citep{jenkins2001} and the method of \citet{laceycole1993} to

626: estimate halo merger histories, and assuming the cosmology of

627: \citet{spergel2003}, we find that {\sc Sb}--type halos ($M_{tot}\sim

628: 10^{12}$ \msolarunits) are likely to experience of order one 1:8 minor

629: merger from $z=1$ to the present--day

630: \citep{hopkins2007c,hopkins2007b}.  In our simulations, strong

631: antitruncations are produced when the orbit of the secondary is

632: inclined ($0^\circ \lsim i \lsim 90^\circ$), prograde, and has

633: moderate angular momentum ($R_p \gsim h_{D}$).  So, if all orbits are

634: distributed isotropically -- i.e., equally likely in bins of

635: $d\cos{i}$ -- and follow the distribution of $R_p$ as inferred from

636: cosmological N--body simulations \citep[e.g.,][]{benson2005}, then we

637: would expect $\sim 40\%-50\%$ of Sb type spirals to have pronounced

638: antitruncations.

639: 

640: At the same time, our fits agree quantitatively with the relative

641: scale lengths $h_{D2}/h_{D1}$ observed by \citet{pohlen2006} and \citet{erwin2005}.  We

642: find, however, that the break radius in our simulations is at the low end of the observed range; 

643: \citet{pohlen2006} and \citet{erwin2005} find $R_{br}/h_{D1} = 3-6$ 

644: while in our simulations $R_{br}/h_{D1} = 3-4$.

645: This may owe either to: (1) the smaller mass of a typical galaxy in the \citet{pohlen2006} sample, or

646: (2) the limited range of parameter space spanned by our simulations.

647: The observations of \citet{pohlen2006} appear to be dominated by

648: somewhat lower--mass spirals than the Milky--Way mass primary disk in the

649: interactions examined here.  This could potentially lead to shorter inner scale

650: lengths \citep[e.g.,][]{courteau1996,dejong1996}, which may tend to

651: increase the average observed $R_{br}/h_{D1}$ ratio.  Also, our

652: simulations sample only a small subset of the parameter space for

653: individual interactions.  The real merging history of galaxies

654: likely involves a variety of mass fractions and multiple mergers

655: that may produce subtly different effects.  However, we find that our

656: simulations of minor mergers generically lead to antitruncated disks

657: for a range of orbital geometries, and therefore represent a viable

658: mechanism for producing these features.

659: 

660: \begin{figure*}

661: \plotone{f10.eps}

662: \caption{Stellar surface mass density of the remnant as a function of radial distance from the center of mass $R$ in units of the inner scale length $h_{D1}$ (see Table~\ref{tab:fits} for fitted values), and its dependence on the orbital inclination $i$ of the interaction: from left to right $i = $ 0 (coplanar prograde, {\sc Sb2Im0Rp1}), 30 (prograde, {\sc Sb2Im30Rp1}), 90 (polar, {\sc Sb2Im90Rp1}), 150 (retrograde, {\sc Sb2Im150Rp1}), and 180 (coplanar retrograde, {\sc Sb2Im180Rp1}).  We include both the binned simulation data (crosses), and fitted disk profiles (solid line).}

663: \label{fig:inc_scale}

664: \end{figure*}

665: 

666: \begin{figure}

667: \plotone{f11.eps}

668: \caption{Response of the gravitational potential to a minor merger as a function of radial distance from the center of mass $R$, and its dependence on the orbital inclination $i$ of the interaction: $i = $ 0 (coplanar prograde, dashed, {\sc Sb2Im0Rp1}), 30 (prograde, solid, {\sc Sb2Im30Rp1}), 90 (polar, dot--dash, {\sc Sb2Im90Rp1}), 150 (retrograde, double dot--dash, {\sc Sb2Im150Rp1}), and 180 (coplanar retrograde, long dash, {\sc Sb2Im180Rp1}).  The dotted line represents the initial potential of the primary disk.}

669: \label{fig:inc_pot}

670: \end{figure}

671: 

672: More locally, \citet{ibata2005} and \citet{ibata2007} recently observed an extended stellar disk in M31.  Though it is not entirely clear whether or not this feature is preceded by a well defined break in the surface brightness profile, it is possible that the extended disk represents an antitruncation of the type observed by \citet{erwin2005} and \citet{pohlen2006}.  \citet{ibata2005} estimate that it contains roughly 10\% of the stellar mass and 30\% of the angular momentum of the total stellar disk, as compared to 5\% of the mass and 45\% of the angular momentum in the antitruncated disk of {\sc Sb2Im30Rp1}.  Furthermore, the kinematics -- specifically, the dispersion of circular velocity lags relative to Keplerian rotation -- of resolved stars in the extended disk show evidence of dynamical heating which could have been caused by a minor merger \citep[e.g.][]{quinn1993,walker1996,velazquez1999}.  Therefore, although we cannot say with certainty that the extended disk of M31 represents a local example of an antitruncation, it is broadly consistent with our modeling.

673: 

674: \section{Conclusion}

675: \label{sec:conclude}

676: 

677: We use hydrodynamic simulations to investigate minor mergers as a

678: physical mechanism for creating antitruncated disks in face--on

679: spirals.  We find that the antitruncation is produced by two competing

680: effects: merger--driven gas inflows deepen the central potential and

681: contract the inner profile, while at the same time angular momentum is

682: transferred to large radius and causes the outer disk to expand.

683: Because the inflows are far more efficient when gas dissipation is

684: included, the antitruncation is produced in our experiments only when

685: a significant gas supply is present in the initial primary disk.  This

686: effect is also only seen when the interaction is prograde, rather

687: than polar or retrograde, with moderate ($R_p \sim h_{D}$) orbital

688: angular momentum.  

689: 

690: Our merger-driven scenario for producing antitruncated disks yields

691: results that agree with observations of local face--on spirals \citep{erwin2005,pohlen2006}, 

692: both in terms of the parameters of the antitruncation and its frequency with Hubble type.  Therefore, we find that minor mergers are a viable physical mechanism for producing antitruncated disks.

693: \newpage

694: \acknowledgements

695: 

696: Particular thanks to the referee, Michael Pohlen, for his comments and suggestions that improved this manuscript.  Also thanks to Peter Erwin, Yuexing Li, Phillip F. Hopkins, Beth Willman, and Du\v{s}an Kere\v{s} for helpful discussions.  These simulations were performed at the Harvard Institute for Theory and Computation at the Smithsonian Center for Astrophysics.

697: 

698: \bibliographystyle{apj}

699: \bibliography{../../minors}

700: 

701: \end{document}

702: