1: \documentstyle[preprint,aps,prl,epsf]{revtex}

2: %\documentstyle[aps,prl,epsf,twocolumn]{revtex}

3: \begin{document}

4: \def\bea{\begin{eqnarray}}

5: \def\eea{\end{eqnarray}}

6: \def\a{\alpha}

7: \def\d{\delta}

8: \def\p{\partial} 

9: \def\nn{\nonumber}

10: \def\r{\rho}

11: \def\rv{\bar{r}}

12: \def\la{\langle}

13: \def\ra{\rangle}

14: \def\e{\epsilon}

15: \def\o{\omega}

16: \def\n{\eta}

17: \def\g{\gamma}

18: \def\break#1{\pagebreak \vspace*{#1}}

19: \def\f{\frac}

20: %\twocolumn[\hsize\textwidth\columnwidth\hsize\csname

21: %@twocolumnfalse\endcsname

22: \draft

23: \title{Inequivalence of Statistical Ensembles \\in\\ 

24: Single Molecule Measurements}

25: \author{Supurna Sinha$^{}$\cite{SUP} and Joseph Samuel$^{}$\cite{SAM}}

26: \address{Harish Chandra Research Institute,\\

27: Chhatnag Road, Jhunsi, Allahabad 211 019, India\\

28: and\\

29: Raman Research Institute, Bangalore 560 080, India\\}

30: %\date{\today}

31: \maketitle

32: \widetext

33: \begin{abstract}

34: We study the role of fluctuations in 

35: single molecule experimental measurements of force-extension ($f-\zeta$)

36: curves. We use the Worm Like Chain (WLC) model to bring out the 

37: connection between the Helmholtz ensemble characterized by the 

38: Free Energy ($F(\zeta)$) and the Gibbs ensemble characterized by 

39: the Free Energy ($G(f)$). 

40: We consider the rigid rod limit of the 

41: WLC model as an instructive special case to bring out the issue of 

42: ensemble inequivalence. 

43: We point out the need for taking into account the free energy of

44: transition when one goes from one ensemble to another. 

45: We also comment on the ``phase transition'' noticed in an 

46: isometric setup for semiflexible 

47: polymers and propose a realization of its thermodynamic limit.  

48: We present general arguments which rule out non-monotonic force-extension

49: curves in some ensembles

50: and note that these do {\it not} apply to the isometric ensemble. 

51: 

52: \end{abstract}

53: \pacs{PACS numbers: 87.15.-v,05.40.-a,36.20.-r}

54: \narrowtext

55: \textwidth=12cm

56: {\bf Introduction:}

57: 

58: In the past, experiments on polymers were confined to studying their

59: bulk properties, which involved probing 

60: large numbers of molecules\cite{bulk}. 

61: The results of these experiments could be analyzed by using 

62: the traditional tools of thermodynamics. 

63: In recent years, however, researchers have been successful in 

64: micromanipulating {\it single}

65: biological molecules such as DNA, proteins and RNA to probe 

66: their elastic

67: properties \cite{bust}. 

68: Such studies serve a twofold role. 

69: On one hand, they shed light on 

70: mechanical properties of semiflexible polymers,

71: which are of clear biological importance in processes such as 

72: gene regulation and transcription \cite{Strick,twist,writhe}.  

73: On the other hand, they provide physicists with a concrete testing

74: ground for understanding some of the fundamental ideas of 

75: statistical mechanics.

76: In statistical mechanics, an isometric setup would be described by  

77: the 

78: Helmholtz free energy, whereas an isotensional setup would be

79: described by the Gibbs free energy\cite{landau}. In the thermodynamic 

80: limit 

81: these 

82: two descriptions agree, but semiflexible polymers (those with contour 

83: lengths comparable to their persistence lengths) are {\it not} at the 

84: thermodynamic limit. Experimentally, both isometric and isotensional 

85: ensembles are realizable. 

86: Typically the polymer molecule is suspended (in a suitable medium) between 

87: a translation stage

88: and a force sensor.

89: The force sensor could be realized by using an Atomic 

90: Force

91: Microscope (AFM) cantilever or by optical or magnetic forces. 

92: As noted by Kreuzer and Payne \cite{kreuz},

93: an isometric setup can be realized using a stiff trap and an isotensional 

94: setup by using a soft trap. 

95: In a more sophisticated

96: version, an electronic feedback circuit is used to control the 

97: force (or the extension) and one measures the fluctuations in the 

98: extension

99: (or the force) \cite{keller}.

100: Here we will focus on the role of 

101: fluctuations in single molecule experiments.

102: In order to correctly interpret such experiments one needs to 

103: understand the effect of fluctuations on the measured quantities. 

104: For instance, it turns out, that an experiment in which the 

105: ends of a polymer molecule are fixed (isometric) and the 

106: force fluctuates yields a different

107: result from one in which the force between the ends is held fixed 

108: (isotensional) and 

109: the end-to-end distance fluctuates\cite{keller,neumann}. 

110: This difference can be traced to 

111: large fluctuations about the mean value of the force or the extension,

112: depending on the experimental setup. These fluctuations

113: vanish only in the thermodynamic limit of very long polymers. 

114: 

115: Here we use the Worm Like Chain (WLC) model \cite{kratky,bend,Kleinert} to 

116: study the inequivalence of ensembles due to 

117: finite size effects.

118: The WLC model 

119: has been very successful in achieving quantitative agreement with  

120: experimentally measured force-extension curves\cite{bust,Marko}. 

121: The paper is organized as follows. In Sec. $II$ we discuss the Helmholtz

122: and Gibbs ensembles. In Sec. $III$ we consider the rigid rod limit which

123: forcefully brings out the main issues dealt with in this paper. 

124: In Sec. $IV$ we draw attention to the importance of taking into 

125: account the free energy of transition in going from one ensemble to 

126: another. In Sec. $V$ we discuss the thermodynamic

127: limit of a ``phase transition'' recently seen in semiflexible polymers. 

128: Finally, we end the paper with a discussion in Sec. $VI$. 

129:   

130: {\bf II Helmholtz and Gibbs Ensembles:}

131: 

132: Consider an idealized experiment in which one end of a molecule is 

133: held fixed at $(x_0,y_0,0)$ and the other end is attached to a dielectric

134: bead confined to a harmonic optical trap described by the potential

135: \begin{equation}

136: V(x,y,z) = A\frac{[(x-x_0)^2+(y-y_0)^2]}{2} + C\frac{(z-z_0)^2}{2}.

137: \label{trap}

138: \end{equation}

139: with $(x_0,y_0,z_0)$ defining the center of the trap. 

140: Consider $A$ to be small 

141: so that the bead is free to move in the plane $z=z_0$.

142: For a polymer of contour length $L$ and persistence length $L_P$

143: it is convenient to introduce the following dimensionless variables:

144: $\zeta =\frac{z}{L}$, $\zeta_0 =\frac{z_0}{L}$, $\beta =\frac{L}{L_P}$ and 

145: $f=\frac{FL_P}{k_BT}$ where $F$ is an applied stretching force and

146: $k_BT$ is the thermal energy at a temperature $T$.

147: Consider $P(\zeta)d{\zeta}$, the number of 

148: configurations (counted with 

149: Boltzmann weight)\cite{bend} for a polymer of length $L$ 

150: starting from the origin and ending anywhere on the $x-y$ plane

151: in an interval $d\zeta$ of $\zeta$. 

152: The free energy defined by ${\cal F}(\zeta)= -\frac{1}{\beta}ln P(\zeta)$ 

153: is the 

154: Helmholtz

155: free energy.  

156: The partition function\cite{partfoot} for the combined system consisting 

157: of the 

158: polymer molecule

159: {\it and} the trap is given by:

160: \begin{equation}

161: Z(\zeta_0,\beta)=\sqrt{\frac{\tilde C}{2 

162: \pi}}\int_{-\infty}^{+\infty}{d\zeta e^{-\beta 

163: {\cal F}(\zeta)}e^{-\tilde{C}\frac{(\zeta 

164: -\zeta_0)^2}{2}}}

165: \label{part}

166: \end{equation}

167: where $\tilde{C} =C L^2$.

168: By tuning the longitudinal stiffness $\tilde{C}$ we can realize the two 

169: limiting

170: cases. 

171: ${\it Helmholtz:}$

172: In the limit of a stiff trap ($\tilde{C} \rightarrow \infty$), the 

173: Gaussian factor 

174: pertaining to the trap approaches a delta function and one gets

175: \begin{equation}

176: Z(\zeta,\beta)= e^{-\beta {\cal F}(\zeta)}

177: \label{stifftrap}

178: \end{equation} 

179: Here we have switched notation to write $\zeta$ in place of $\zeta_0$.

180: Thus a stiff trap realizes the Helmholtz 

181: ensemble by constraining fluctuations in the $\zeta$ coordinate.

182: To extend the molecule from $\zeta$ to $\zeta+d\zeta$ one needs

183: to apply a force $<f>=\frac{\partial F}{\partial \zeta}$

184: in order to compensate for the change of entropy. Plotting

185: $<f>$ versus $\zeta$ we find the $(<f>,\zeta)$ force-extension relation.

186: 

187: ${\it Gibbs:}$ 

188: In the opposite limit of a soft trap ($\tilde{C} \rightarrow 0$ and 

189: $\zeta_0 

190: \rightarrow \infty$ such that $\tilde{C}\zeta_0 =\beta f$ 

191: remains finite),

192: one gets\cite{partfoot}

193: \begin{equation}

194: {\tilde Z}(f,\beta)=\int_{-\infty}^{+\infty}{d\zeta e^{-\beta 

195: F(\zeta)}e^{\beta 

196: f\zeta}}

197: \label{softrap}

198: \end{equation}

199: Thus a soft trap permits fluctuations in the $\zeta$ coordinate but 

200: constrains

201: the force fluctuations\cite{kreuz} and thus realizes the 

202: Gibbs ensemble.

203: ${\tilde Z}(f)$ is the generating function for the $\zeta$ distribution.

204: Defining the Gibbs free energy $G(f)={-\beta}ln{\tilde{Z}}(f)$ we

205: can work out the mean extension $<\zeta>=-\frac{\partial G}{\partial f}$

206: and the $(<\zeta>,f)$ force-extension relation.

207: 

208: Notice that ${\tilde Z}(f)$ is the Laplace transform 

209: of 

210: $Z(\zeta)$.    

211: In the thermodynamic limit of long polymers ($\beta 

212: \rightarrow \infty$) the Laplace transform integral Eq.(\ref{softrap}) 

213: is dominated by 

214: the saddle point value and therefore  

215: ${\cal F}(\zeta)$ and $G(f)$ are related by a Legendre transform:

216: \begin{equation}

217: {\cal F}(\zeta) = G(f) +f\zeta.

218: \label{legendre}

219: \end{equation}

220: For finite $\beta$ i.e. for a polymer of finite extent, 

221: the saddle point approximation 

222: no longer holds true and fluctuations about the saddle point value

223: of the free energy become important.  

224: Thus one finds that ${\cal F}(\zeta)$ and $G(f)$ are {\it not} 

225: Legendre transforms of each other.

226: We notice that this difference between the Legendre transform [Eq. 

227: (\ref{legendre})] and 

228: the Laplace transform [Eq. (\ref{softrap})] is the mathematical origin 

229: of the finite size fluctuation effects described in this paper. 

230: These fluctuations are of thermal origin and 

231: can ultimately be traced to 

232: collisions of the polymer molecule

233: with the molecules of the suspending medium. 

234: In this section we have recovered the results of \cite{kreuz}.

235: We have also gone beyond \cite{kreuz} and 

236: traced the origin of the 

237: difference between the two 

238: ensembles to the difference between the 

239: Legendre and Laplace transforms.

240: 

241: {\bf III Rigid Rod Limit: An Instructive Extreme Case}

242: 

243: We noticed in the last section that because of fluctuation effects

244: the Helmholtz and Gibbs free energies are not related by a 

245: Legendre Transform. 

246: An important consequence of this is that the $(F,<z>)$ relation 

247: is different from the $(<F>, z)$ relation. In other words, the 

248: force-extension curves plotted in the two ensembles are {\it distinct}

249: due to finite size fluctuation effects.  

250: Fluctuations dominate at finite $\beta$ 

251: and disappear in the thermodynamic limit ($\beta \rightarrow \infty$) of 

252: flexible polymers. 

253: We bring out the ensemble dependence 

254: of the force-extension 

255: relations explicitly and most dramatically 

256: in the limiting case of a very stiff polymer ($\beta \rightarrow 0$).

257: 

258: In this extremely rigid limit\cite{frey}, the end of the polymer is uniformly 

259: distributed over the sphere of directions. In the Helmholtz ensemble we  

260: thus have $P(z)d{z}= \frac{d z}{2L}$ for $-L<z<L$ and 

261: $P(z) =0$ otherwise. 

262: Since the free energy is constant in the range $-L<z<L$ and diverges 

263: otherwise, we find that the average force $<F>$ vanishes for 

264: $|z|<L$ and diverges for 

265: $|z| =L$ (See Fig. $1 a$). In the Gibbs ensemble we find by standard 

266: manipulations that 

267: \begin{equation}

268: <z>=[LCoth{FL}-\frac{1}{F}]

269: \label{zfrel}

270: \end{equation}

271: which differs from the $<F>-z$ relation determined above in 

272: the Helmholtz ensemble (See Fig. $1 b$). 

273: The theoretical analysis of the ensemble dependence of the force-extension 

274: relation based on the rigid rod limit is a new result of this paper.

275: 

276: Thus an experimenter making force-extension measurements on, 

277: for instance, Actin filaments\cite{loic}, would find that a measurement

278: in which the force is controlled and the end-to-end distance is measured

279: leads to a different force-extension curve 

280: from a measurement in which the end-to-end separation is controlled and 

281: the force is measured. A theorist interpreting the curves also needs 

282: to keep in mind whether the curves are obtained in a constant-force

283: setup or a constant-extension setup since a proper interpretation of

284: the curves requires a knowledge of the ensemble used in the 

285: measurement.   

286: 

287: \vspace{.6 cm}

288: \vbox{

289: \epsfxsize=13.0cm

290: \epsfysize=6.0cm

291: \epsffile{figure1.eps}

292: \begin{figure}

293: \caption{Force-Extension Curve in the Helmholtz (Fig. 1a) and Gibbs 

294: (Fig. 1b)

295: ensembles for $\beta=0.$ We have set $L=1$. }

296: \label{contrast}

297: \end{figure}} 

298: 

299: 

300: {\bf IV Inequivalence Of Ensembles and the Free Energy Of Transition}

301: 

302: In an isotensional setup one controls the force and one 

303: measures the mean extension and plots

304: it as a function of force.

305: In an isometric setup the roles of extension and force are interchanged.

306: In both setups zero force corresponds to zero extension ($\zeta=0$) 

307: and a large force corresponds to maximal extension (i.e. $\zeta = 1$).

308: Imagine going from zero force to a large force via the 

309: isotensional setup and returning from maximal extension to 

310: zero extension via the isometric setup.

311: Since the ``equation of state'' depends on the

312: chosen ensemble, in general there will be two distinct curves describing

313: the extension in one ensemble followed by contraction in the other 

314: \cite{foot}.

315: In such a situation there could be a net area enclosed in the

316: Force-Extension plane. 

317: This poses a puzzle because it appears that a cyclic process can 

318: extract work from the system. This puzzle is easily resolved. 

319: In completing the cycle and returning to the initial

320: state one is in fact changing ensembles twice at the two end points.

321: These correspond to {\it finite} free energy changes which need to be

322: taken into account. 

323:  

324: Let $(\zeta_1,f_1)$ and $(\zeta_2,f_2)$ be two points which lie on 

325: both isometric and isotensional force-extension curves. In our example

326: $(\zeta_1,f_1)=(0,0)$ and $(\zeta_2,f_2)=(1,\infty)$.

327: Let us suppose that we go from $(\zeta_1,f_1)$ to $(\zeta_2,f_2)$

328: in the isotensional ensemble and return via the isometric ensemble.

329: We find that 

330: \begin{equation}

331: G(f_2)-G(f_1) =\int_{f_1}^{f_2}{\frac{\partial G}{\partial 

332: f}df}

333: =-\int_{f_1}^{f_2}{<\zeta> df}. 

334: \end{equation}

335: Similarly,

336: \begin{equation}

337: F(\zeta_2)-F(\zeta_1) =\int_{\zeta_1}^{\zeta_2}{\frac{\partial 

338: F}{\partial 

339: \zeta}d\zeta}

340: =\int_{\zeta_1}^{\zeta_2}{<f> d\zeta}.

341: \end{equation}

342: The area enclosed between the two curves is given by 

343: $$W=\int_{\zeta_1}^{\zeta_2}{f d<\zeta>} 

344: -\int_{\zeta_1}^{\zeta_2}{<f> d\zeta}$$

345: which can be rewritten as 

346: \begin{eqnarray}

347: W=\int_{\zeta_1}^{\zeta_2}{d(f <\zeta>)}

348: -\int_{f_1}^{f_2}{<\zeta> df} - 

349: \int_{\zeta_1}^{\zeta_2}{<f> d{\zeta}}\\{\nonumber}

350: =f_2\zeta_2 - f_1\zeta_1 +G(f_2) - G(f_1)-F(\zeta_2) + 

351: F(\zeta_1)\\{\nonumber}

352: =[f\zeta+G(f)-F(\zeta)]_1^{2} 

353: =[\tilde{F}(\zeta)-F(\zeta)]_1^{2}\\{\nonumber}

354: =\Delta F^{tr}(\zeta)|_1^{2} 

355: \label{freetr}

356: \end{eqnarray}

357: 

358: where we define $\Delta F^{tr}(\zeta)$ as the free energy of transition. 

359: $\Delta F^{tr}(\zeta)$ is the difference between 

360: $\tilde{F}(\zeta)=f\zeta+G(f)$,

361: the Legendre transform of $G(f)$ and the Helmholtz 

362: free energy $F(\zeta)$. Since these are not equal (except in the limit

363: of long polymers) the free energy of transition between ensembles must

364: be considered in order that the total free energy change in a cyclic

365: process vanishes. 

366:    

367: In order to understand this issue more explicitly we consider 

368: corrections to the saddle point approximation which is valid 

369: in the long polymer limit.

370: Let us expand $\phi({\zeta}) = {\cal F}(\zeta)-f\zeta$, the argument of 

371: the exponential appearing 

372: on the 

373: right hand side of Eq. (\ref{softrap})

374: around the saddle point value $\zeta = \zeta_{*}$ (which dominates the 

375: integral in 

376: the long polymer limit) and retain terms upto second order in 

377: the fluctuations about the saddle point value: 

378: $$\phi({\zeta}) = \phi({\zeta_{*}}) 

379: +\frac{1}{2}{\phi}^{''}(\zeta)|_{\zeta=\zeta_{*}}(\zeta 

380: -\zeta_{*})^2.$$

381: If we plug in this expansion in Eq. (\ref{softrap}) and identify ${\tilde 

382: Z}(f,\beta)$ with 

383: $e^{-\beta 

384: G(f)}$

385: we arrive at the following equation:

386: \begin{equation}

387: G(f)=[{\cal F}(\zeta_{*})-f\zeta_{*}]+\frac{1}{2\beta}

388: ln{{\cal F}^{''}(\zeta_{*})}

389: \label{saddle}

390: \end{equation}

391: The free energy due to fluctuations around the saddle point value is

392: $\frac{1}{2\beta}ln{{\cal F}^{''}(\zeta_{*})}$. 

393: Notice that in the long polymer limit of $\beta \rightarrow \infty$,

394: this term vanishes. For finite $\beta$, this nonzero contribution 

395: to the free energy accounts for the transition between the constant

396: extension ensemble and the constant force ensemble.

397: In going from a soft trap to a stiff trap work is done 

398: on the bead by the trap. Similarly in going from a stiff trap to a soft 

399: trap work is extracted from the bead by the trap. The net work done is the 

400: difference between the work done at the two ends

401: of the force-extension curves in switching ensembles.

402: This net work exactly cancels out the nonzero 

403: area enclosed in the force-extension plane. 

404: 

405: {\bf V ``First Order Phase Transition'' and the Thermodynamic Limit in 

406: Semiflexible Polymers}

407: 

408: In Sec. $II$ we considered the bead to be in a potential which was 

409: soft in the $x$ and $y$ directions. Let us now consider what happens

410: when the trap is stiff in all three directions

411: ($A$ as well as $C$ in Eq. ({\ref{trap}}) are large) and the vector 

412: position 

413: of the bead is constrained to be at $(x_0,y_0,z_0)$.

414: Let $Q({\vec r})$ be the number of polymer configurations 

415: which start at the origin and end in the volume element $d{\vec r}$ 

416: centered at ${\vec r}$ \cite{bend,dabhi}. 

417: $Q({\vec r})$ is related to $P(z)$ via the equation 

418: \begin{equation}

419: P(z)=\int{d\vec{r} Q(\vec{r})\delta(r_3 -z)} 

420: \label{pofz}

421: \end{equation}

422: which in words, means that $P(z)$ is obtained by 

423: integrating $Q({\vec r})$ over a plane of constant $z$\cite{bend}.

424: The distribution $Q({\vec r})$ was studied in \cite{dabhi} 

425: where it was noticed that 

426: in an intermediate range (around $3.8$) of $\beta$

427: the free energy 

428: $A({\vec r})= {\frac{-1}{\beta}}log{Q(\vec r)}$ had multiple minima.

429: For a fixed contour length as one varies $\beta$ by tuning the persistence

430: length $L_P$ one finds a competition between flexible and rigid phases of 

431: the polymer for intermediate values of $\beta$. Thus the polymer 

432: undergoes a flexible to rigid ``first order phase transition''

433: via a two peaked profile of $Q({\vec r})$. 

434: This leads to a curious 

435: force-extension relation. 

436: As one pulls the bead, the restoring force at first increases, 

437: then decreases to zero and then goes 

438: negative and becomes a destabilizing force. 

439: The 

440: molecule is unstable and goes to a new extended state. 

441: 

442: The words ``first order phase transition'' above were in quotes as a

443: finite system does not show phase transitions. If one takes the 

444: thermodynamic limit by taking the length of the polymer to 

445: infinity ($\beta \rightarrow \infty$) one loses the multiple minima

446: structure which is present only in a small range of $\beta$ around

447: $3.8$. Is it possible for this phase transition to survive the 

448: thermodynamic limit? As we will see below, this is indeed possible 

449: provided one takes the replica thermodynamic limit. 

450: We take $N$ replicas of the molecule with fixed $\beta$ and let $N$ 

451: tend to $\infty$. Consider $N$ identical polymers with $\beta= 

452: \frac{L}{L_P}$ fixed, their two ends anchored to flat surfaces 

453: $S_1$ and $S_2$  (See Fig. $2$). One could realize the above

454: arrangement by using 

455: $(i)$ two planar arrays of optical traps or

456: $(ii)$ by introducing suitably synthesized supramolecular lamellar

457: structures. 

458: The anchoring is such that the tangent vectors to the molecule 

459: at the fixed ends are free to swivel. If one applies a 

460: force $F$ to pull $S_1$ and $S_2$ 

461: apart the $N$ molecules are also stretched. We consider the molecules

462: to be well separated so that they can be regarded as independent.

463: One could now look

464: at the mean force $F$ needed to maintain the separation $r$. It is easily

465: seen

466: that the force is proportional to $N$ and also the mean square

467: fluctuation

468: $<(\Delta F)^2>$ in the force is proportional to $N$. 

469: This is because the mean force and its variance are 

470: respectively the first and second derivatives of the free energy,

471: which being an extensive quantity is proportional to $N$. 

472: %Heuristically, this

473: %is because of cancellations of the force fluctuations from different

474: %polymers.

475: It follows that as

476: $N$ goes to infinity, the fluctuations ($\Delta F/F$) in $F$ die out as

477: $1/\sqrt{N}$. We can now regard the mean force $F$ as a control parameter

478: (i.e. consider a constant $F$ ensemble)

479: and observe that if we tune the applied force $F$, there is a

480: discontinuous change in the

481: separation $r$ between the two sets of optical traps 

482: signalling a first order phase transition with the inter-trap separation

483: $r$ as the order parameter. Thus, 

484: in the replica way of taking the thermodynamic limit 

485: the double humped

486: form of the distribution function  $Q({\vec r})$ results in a true first 

487: order phase transition. 

488: 

489: \vbox{

490: %\vspace{-0.5cm}

491: \epsfxsize=6.0cm

492: \epsfysize=6.0cm

493: \epsffile{figure2.eps}

494: \begin{figure}

495: \caption{Schematic Experimental Design for Replica Thermodynamic Limit,

496: shown above for 

497: $N=4.$}

498: \label{replica}

499: \end{figure}}

500: 

501: {\bf VI Discussion:}

502: 

503: In this paper we point out the importance of considering the free energy 

504: of transition in going between the Helmholtz and the Gibbs ensembles

505: in the context of single molecule force-extension measurements. 

506: We also study two distinct ways of taking the thermodynamic limit

507: $(i)$ by letting the length of the polymer tend to infinity

508: (i. e. $\beta \rightarrow \infty$) and 

509: $(ii)$ by considering replicas. In particular we notice that 

510: the flexible to rigid transition mentioned in Sec. $V$ for a 

511: single semiflexible polymer survives the replica thermodynamic limit. 

512: In contrast, this feature disappears in the 

513: usual thermodynamic limit of $\beta \rightarrow \infty$. 

514: 

515: The non-monotonic behavior of $Q(\vec r)$ is an intriguing feature of 

516: molecular elasticity. 

517: It was noticed in recent simulations\cite{dabhi} and subsequently in 

518: a semi-analytical treatment\cite{bend}. It was also commented on in a 

519: recent paper\cite{Kleinert2}.

520: This double humped form of $Q({\vec r})$

521: has gained renewed interest in the context of cyclization 

522: of polymers and its significance to gene regulation \cite{gautam,CELL}.

523:  

524: 

525: We note that this remarkable qualitatively

526: distinctive feature is special to the isometric ensemble (fixed vector

527: end-to-end separation) and is not shared by other ensembles. 

528: In other ensembles one can argue generally that

529: such non-monotonic behavior cannot occur. For example, the conjugate

530: distribution 

531: ${\tilde P}({\vec F})=\int Q({\vec r})e^{{\vec F}.{\vec 

532: r}}d{\vec r}$ 

533: is monotonic. This follows from noticing that the $3\times 3$ matrix  

534: \begin{equation}

535: \frac{\partial{ln {\tilde P}}}{\partial{F_i}\partial{F_j}}

536: =<(r_i -<r_i>)(r_j -<r_j>)> 

537: \label{nonmon}

538: \end{equation}

539: is positive definite. Such 

540: arguments do not apply to $Q({\vec r})$ since there is no analogous

541: formula to Eq. (\ref{softrap}) in the conjugate distribution. Indeed,

542: if there were, a double humped form could not appear in $Q({\vec r})$,

543: for one could express the second derivative of $Q({\vec r})$ as the 

544: variance of the force.

545: Can $P(z)$ 

546: show non-monotonic behavior? The answer is no,

547: for it has been shown in \cite{bend} that  

548: $${\frac{-2}{z}}{\frac{d P}{d z}} = Q(z)$$

549: Since $Q(\vec r)$ is a probability density and therefore non-negative,

550: it follows that $\frac{d P}{d z} \leq 0$ for $z>0$, thus 

551: ruling out multiple peaks in $P(z)$. 

552: This argument which rules out multiple peaks in $P(z)$ is a new 

553: observation. 

554: Note that a $P(z)$ measurement differs from $Q(\vec r)$ only in the 

555: transverse stiffness $A$ of the trap. By tuning $A$ we can permit 

556: fluctuations in the transverse direction and therefore 

557: destroy the 

558: phase transition present in the stiff $A$ limit.

559: One would expect to see a critical stiffness $A=A_c$ below which the 

560: phase transition is destroyed. Alternately, one could tune the mean 

561: force and expect to see the phase transition vanishing below a critical 

562: mean force $F=F_c$ for a fixed value of $\beta$ in the intermediate

563: range of $\beta$.  

564: We emphasize that the non-monotonic 

565: features of $Q({\vec r})$ in the semiflexible range $\beta \approx 3.8$

566: are predictions of the WLC model which can be tested against experiments.

567: A single molecule with its ends confined in optical traps is expected

568: to show this flexible to rigid transition. The effect can be dramatic

569: however, if a large number of molecules co-operatively show such a 

570: transition. One could attach the ends of 

571: a collection of semiflexible

572: polymers to supramolecular 

573: layers\cite{drozdov} and detect the 

574: flexible to rigid

575: transition signalled by a change in the inter-layer 

576: separation via a suitable probe. 

577: It may be possible to exploit this dramatic transition from flexible to 

578: rigid behavior in technological applications.   

579: 

580: If one considers the replica thermodynamic limit of a semiflexible polymer

581: one sees that force-extension curves continue to remain distinct in 

582: the Helmholtz and the Gibbs ensembles. 

583: So while interpreting a force-extension curve obtained for a collection 

584: of semiflexible polymers suspended between two arrays of traps,

585: one needs to know if the curve is obtained using

586: the soft trap setup or a stiff trap setup.

587: However, for $\beta \rightarrow 

588: \infty$, which is the usual thermodynamic limit, the two ensembles 

589: give rise to the {\it same} force-extension curve.

590: This observation is consistent with 

591: the fact that inequivalence of ensembles can 

592: survive 

593: at the thermodynamic limit for systems with long range 

594: interactions\cite{kreuz,keller,Lynd}. 

595: In the context of semiflexible polymers,

596: the persistence length $L_P$ plays the role of 

597: the range of interactions.

598: There has been some work\cite{baro} on the thermodynamics

599: of particle systems in the presence of external macroscopic 

600: fields in classical and quantal contexts. In these papers

601: the authors have dealt with the macroscopic limit of the 

602: definition of pressure which is analogous to the thermodynamic 

603: limit of the definition of force in our work. 

604: In particular the authors of Ref. \cite{baro} discuss the 

605: connection between the values of the

606: pressure defined by two different thermodynamic limit procedures:

607: in the first, the system is confined successively in a

608: sequence of boxes which grows to fill up the whole space.  In the 

609: second,

610: the system is in an external potential 

611: similar to the present context.

612: In the case of a

613: semiflexible polymer the force corresponding to a given extension is 

614: the same in the Gibbs and Helmholtz ensembles only in the thermodynamic

615: limit of $\beta \rightarrow \infty$. This is analogous to the second

616: procedure of taking the thermodynamic limit 

617: given in \cite{baro}, where they recover the thermodynamic notion 

618: of pressure from an underlying microscopic definition when they 

619: let the scale factor go to infinity in the macroscopic limit. 

620: The results of this paper are therefore, consistent with the general 

621: treatment in \cite{baro}. 

622: 

623: The fluctuation effects mentioned here also 

624: have some biological significance.

625: In particular, the process of gene regulation involves interaction

626: between parts of a DNA molecule which are about 

627: less than one persistence length

628: apart ($\approx 34 nm$)\cite{CELL}. Over such short segments of the 

629: DNA fluctuation effects would be significant.

630: The replica thermodynamic

631: limit would also play a role in the concrete biological 

632: context of a network of  

633: actin filaments forming the cytoskeletal structure. 

634: 

635: 

636: {\it Acknowledgements:}

637: We thank Dipanjan Bhattacharya for drawing attention to Ref. 

638: \cite{keller}

639: and Abhishek Dhar, Deepak Dhar, Erwin Frey and Richard Neumann for 

640: discussions.

641: We thank one of the referees for drawing attention to Ref.\cite{baro}.

642: We thank Harish-Chandra Research Institute for their hospitality and 

643: the wonderful working atmosphere in which this paper was written.

644: \begin{references}

645: %\vspace{-1.5cm}

646: \bibitem[\dagger]{SUP} e-mail\,:supurna@mri.ernet.in

647: \bibitem[\star]{SAM} e-mail\,:sam@mri.ernet.in

648: \bibitem{bulk} M. Doi, {\it Introduction to Polymer Physics}, Oxford,

649: Clarendon Press (1996). 

650: \bibitem{bust} C. Bustamante et al,

651: Current Opinion in Structural Biology {\bf 10}, 279 (2000).

652: \bibitem{Strick}T. R. Strick et al, {\it Science} {\bf 271}, 1835 (1996).

653: \bibitem{twist} J. Samuel and S. Sinha, {\it Phys. Rev. Lett.} {\bf 90},

654: 098305 (2003).

655: \bibitem{writhe} Supurna Sinha, {\it Physical Review E} {\bf 70 },

656: 011801 (2004). 

657: %\bibitem{Jarz} C. Jarzynski, {\it Phys. Rev. Lett.} {\bf 78}, 

658: %2690 (1997); G. Gallavotti and E. G. D. Cohen, {\it Phys. Rev. Lett.}

659: %{\bf 74}, 2694 (1995); G. E. Crooks, {\it J. Stat. Phys.} {\bf 90}, 1481

660: %(1998).

661: \bibitem{landau} L. D. Landau and E. M. Lifshitz, {\it Statistical 

662: Physics, Part I }, Pergamon Press, {\it 3rd ed.} (1980).  

663: \bibitem{kreuz}

664: H. J. Kreuzer and S. H. Payne, {\it Phys. Rev.} {\bf E 63}, 021906 (2001).

665: \bibitem{keller} D. Keller, D. Swigon and C. Bustamante, 

666: {\it Biophysical Journal} {\bf 84}, 733 (2003).

667: \bibitem{neumann} Richard M. Neumann, 

668: {\it Biophysical Journal} {\bf 85}, 3418 (2003) and references therein.

669: \bibitem{kratky}

670: O. Kratky and G. Porod, {\it Recl. Tral. Chim,} {\bf 68}, 1106 (1949).

671: \bibitem{bend} J. Samuel and S. Sinha, 

672: {\it Physical Review E}, {\bf 66} 050801(R) (2002). 

673: \bibitem{Kleinert}

674: H. Kleinert, {\it Path Integrals in Quantum Mechanics, Statistics and 

675: Polymer Physics},

676: {\it World Scientific, Singapore}, Ist ed.,1990.

677: \bibitem{Marko}

678: J.Marko and E.D.Siggia, Macromolecules {\bf 28}, 8759 (1995);

679: H.E.Daniels, Proc.R.Soc.Edinb. {\bf 63A}, 29 (1952).

680: \bibitem{partfoot} We drop additive constants in the free energies or

681: equivalently, multiplicative constants in the partition functions.

682: \bibitem{frey}

683: J. Wilhelm and E. Frey, {\it Phys. Rev. Lett.} {\bf 77}, 2581 (1996). 

684: \bibitem{loic}

685: L. Le Goff, O. Hallatschek, E. Frey and F. Amblard,

686: {\it Phys. Rev. Lett.} {\bf 89}, 258101 (2002).    

687: \bibitem{foot}In fact we have illustrated this point explicitly in the 

688: context of the rigid rod limit in Sec. $III$.

689: \bibitem{dabhi} A. Dhar and D. Chaudhuri, {\it Phys. Rev. Lett.} {\bf 89}, 

690: 065502-1 (2002).

691: \bibitem{Kleinert2}

692: B. Hamprecht, W. Janke and H. Kleinert, cond-mat/0307530.

693: \bibitem{gautam} P. Ranjith, P. B. Sunil Kumar and G. I. Menon

694: cond-mat/0411723.

695: \bibitem{CELL} B. Alberts et al, {\it Molecular Biology Of The Cell}

696: (Garland Publishing, New York, 1994), 3rd ed; T. E. 

697: Cloutier and J. Widom, {\it Mol. Cell}, {\bf 14}, 355 (2004).

698: \bibitem{drozdov}

699: A. D. Drozdov, cond-mat/0408474.

700: \bibitem{Lynd}

701: R. M. Lynden-Bell, {\it Mol. Phys.} {\bf 86}, 1353 (1995);

702: J. Barre, D. Mukamel and S. Ruffo, {\it Phys. Rev. Lett} {\bf 87}, 030601

703: (2001);

704: F. Gulminelli and Ph. Chomaz, {\it Physical Review E}, {\bf 66}, 046108 

705: (2002).

706: \bibitem{baro} C. Marchioro and E. Presutti,{\it Communications

707: in Mathematical Physics}, {\bf 27}, 146 (1972);

708: C. Marchioro and E. Presutti,{\it Communications

709: in Mathematical Physics}, {\bf 29}, 265 (1973).

710: \end{references}

711: \end{document}

712: 

713: