The purpose is an exact study of the unbinding transition from two interacting manifolds (strings or bilayer membranes). These systems have similar scaling behavior, and then it is sufficient to consider only the strings’ problem. We assume that the manifolds interact via a realistic potential of Morse type. To this end, the use is made of the transfer matrix method, based on the resolution of a Schrödinger equation. We first determine the associated bound states and energy spectrum. Second, the exact ground state energy gives the free energy density, from which we extract the expression of the unbinding temperature. Third, we determine the contact probability between manifolds, from which we compute the (diverging) average separation and roughness of the manifolds. It is found that their critical behavior is close to that obtained using Field-Theoretical Renormalization-Group. The conclusion is that these analytical studies reveal that the Morse potential is a good candidate for the study of the unbinding phenomenon within manifolds. Finally, the discussion is extended to generalized Morse potential.
22 April 2013 - 25 April 2013The Cape Verde International Days on Mathematics 2013Praia, Cape Verde1. Introduction
Unbinding is a phenomenon that occurs in soft elastic manifolds, including strings and bilayer membranes [1–3]. The mechanism that rules adhesion and unbinding of membranes is analog to that of governing interfacial wetting [4] and adsorption-desorption transitions of polymers [5]. The common feature of these interfacial phenomena is that they have a behavior similar to that of 1-dimensional lines or strings of finite tension [6], such as ledges on crystal surfaces, stretched (or directed) polymers, and vortex lines in superconductors.
The major result is that renormalization-group (RG) calculations predicted that fluid membranes possess similar scaling properties as strings in the vicinity of the critical potential depth [7]. Such a property will be the key for the investigation of the unbinding transition in bilayer membranes.
The unbinding transition from strings and bilayer membranes is often driven by steric-shape fluctuations [8] whose amplitude increases with temperature. These undulation forces not only act between lipid bilayers in the Lα-phase but also prevent vesicles from coagulation and stabilize emulsions [9, 10]. For a membrane pair, the induced fluctuation interactions are balanced, at short separation, by repulsive hydration [11] and attractive van der Waals forces [12]. The hydration potential (per unit area) acts at small separations of the order of 1 nm. Its amplitude (hydration pressure), Ph, and potential range, λh, are about Ph≃0.2J/m2 and λh≃0.3nm, respectively. While the attractive van der Waals potential originates from polarizabilities of lipid and water molecules. The associated amplitude (Hamaker constant), W, is in the range 10-22–10-21J, and the bilayer thickness is about δ≃4nm. The total interaction potential is then the sum of these three interactions. The bilayer membrane remains stable at the minimum of the potential, provided that the potential depth is comparable to the thermal energy. In particular, this depth depends on the value of amplitude W of the direct van der Waals energy. In fact, the Hamaker constant W may be varied changing the polarizability of the aqueous medium.
In a seminal paper, Lipowsky and Leibler [13] demonstrated that there exists a certain threshold beyond which the van der Waals attractive interactions are sufficient to bind the membranes together, while below this characteristic amplitude, the membrane undulations dominate the attractive forces, and then the membranes separate completely. In particular, it was found [2, 3, 13] that when the unbinding temperature, Tc, is approached from below, the mean-separation between the two membranes, 〈l〉, and their roughness, ξ⊥, diverge according to 〈l〉~ξ⊥~(Tc-T)-ψ, where the roughness exponent ψ is very close to 1. Afterwards, the results were extended to charged membranes [13], where the unbinding is entirely controlled by the ionic concentration of the aqueous medium. Of course, the latter depends on the nature of the lipid system. The theoretical predictions for charged bilayer membranes were tested in some experiment on DPPC in CaCl2 solutions [14]. Finally, in all cases, beside ψ, one defines other exponents [1], such as roughness, roughening, and contact exponents.
An interesting alternative tool is the transfer matrix method (TMM) usually encountered in quantum mechanics [15–17] and critical phenomena [18, 19]. Such an approach was successfully applied to bilayer membranes [20, 21]. Very recently, TMM was applied to the unbinding transition with single and double well potentials [21].
In this paper, the goal is the application of TMM to some popular potential, that is, the Morse potential (MP) [22]. The latter was introduced to describe vibrations of covalent diatomic molecules of very weak van der Waals complexes [23, 24]. The advantage of MP is that it is not singular and simply presents as a difference of two exponential tails (see hereinafter).
We recall that MP was applied recently to unbinding membranes [25] using the traditional variational method. This use can be motivated by the fact that this potential reflects the two essential features described above, namely, a repulsion at short distance and an attraction at intermediate distance. It is characterized by only three parameters, which are its depth (dissociation energy), D, its range α-1, and the position of its minimum point, lm (see hereinafter). The shape of such a potential models well the real interaction potential discussed above. Another virtue of this potential is that its associated energy spectrum is discrete, and then all its eigenfunctions are bound states. The existence of these states makes it a good candidate for the study of the unbinding transition from strings and membranes. The adjustable parameters D,α-1, and lm can be related to those of the real potential. More precisely, we have functional relations of types x=gx(W,δ,κ,Ph,λh), with x=D,α,lm, where κ is the bending modulus. In addition, MP parameters are temperature dependent.
The variational method is a mean-field theory that underestimates strong fluctuations near criticality. In this paper, however, we present an exact study of the unbinding transition with MP using TMM. In particular, we compute exactly the unbinding temperature, contact probability, from which we extract all critical properties, namely, the average separation and roughness. Therefore, the unbinding exponents are exactly known. The idea will consist to determine all critical behaviors for strings with a MP; then, the results will be translated to fluid membranes since the two systems are governed by the same scaling.
This paper is organized as follows. In Section 2, we present the general string model. A rigorous study of the unbinding transition is presented in Section 3. Finally, some concluding remarks are drawn in the last section.
2. String Model
Strings are commonly defined as one-dimensional lines whose shape fluctuations are governed by finite tension [20]. Consider two interacting strings that move on a two-dimensional space. We assume that, on average, they fluctuate around a line reference, say x-axis, and in addition, their associated elongations remain perpendicular to this axis. The conformation of strings can be described by the local field separation, l(x)>0, which is perpendicular to the line reference. The statistical mechanics of strings is based on the Hamiltonian:
(1)H[l]=∫0L[σ2(dldx)2+V(l)]dx,
where L denotes the string length, σ is the effective tension, and V(l) represents the interaction potential, whose form will be specified hereinafter.
In the thermodynamic limit, that is, for L→∞, the statistical properties of model (1) can be studied using TMM that is, based on the resolution of a Schrödinger equation type [15–17]:
(2)-(kBT)22σd2φndl2+V(l)φn=Enφn,
where T is the absolute temperature and kB is the Boltzmann’s constant. In the above differential equation, the parameters En and φn denote the set of eigenvalues and wave functions, respectively. The eigenvalues En’s are ordered in such a way that E0≤E1≤E2≤⋯. The ground state eigenvalue E0 defines the free energy density, f, that is, f=E0, while the corresponding eigenvector, φ0(l), determines the probability distribution, P(l). In fact, P(l)dl represents the probability of finding two strings at a separation between l and l+dl. The probability distribution is then given by
(3)P(l)=|φ0(l)|2∫|φ0(l)|2dl.
With the help of this distribution, one can calculate, for example, its first and second moments, (3a)〈l〉=∫lP(l)dl,〈l2〉=∫l2P(l)dl.
The string roughness is given by
(3b)ξ⊥=[〈l2〉-〈l〉2]1/2.Before studying their unbinding transition, it will be convenient to recall some backgrounds concerning the general statistical properties of the string pair.
First, we recall the definition of the roughness exponent. Consider a fluctuating manifold, and notice that the latter makes large transverse excursions from its average position. The manifold is rough if the typical size, ξ⊥, of its transverse excursions grows with its lateral size, ξ∥. It is admitted that the two lengths obey the following scaling relation [1–3]:
(4)ξ⊥~ξ∥ζ.
This relation then defines the roughness exponent ζ. The latter crucially depends on the nature of the considered manifold, for example, ζ=1/2, for string, and ζ=1, for almost flat fluid membranes.
The second exponent of our interest is the unbinding exponent. To definite it, we consider two interacting manifolds of mean separation 〈l〉. When they exhibit an unbinding transition, this average separation diverges according to
(5)〈l〉~(Tc-TTc)-ψ.
This behavior defines the unbinding exponent ψ, where Tc is the transition temperature. Also, this exponent generally depends on the nature of the interaction potential experienced by the manifolds. When the unbinding transition is driven by the shape fluctuations, one has 〈l〉~ξ⊥. In this case, the unbinding exponent ψ coincides with the roughening exponent ν⊥ (we do not recall here), except for special cases as discussed in [1].
The last and interesting exponent to recall is the contact exponent. This characterizes the singular behavior at roughening and unbinding transitions of the contact probability of two adjacent manifolds. The latter scales as [1]
(6)P~ξ∥-ζ0~ξ⊥-ζ0/ζ.
Close to criticality, this scaling relation defines the contact exponent ζ0. Also, the value of this exponent primarily depends on the nature of the interaction potential between manifolds.
On the other hand, we can put the probability distribution on the following scaling form:
(7)P(l)=ξ⊥-1Ω(lξ⊥).
Here, the explicit factor ξ⊥-1 arises from normalization. To recover relationship (6), the scaling function Ω(s) must behave as
(8)Ω(s)~s-1+ζ0/ζ,forsmalls(withζ0>0).
This scaling behavior remains valid as long as the mean separation is such that 〈l〉~ξ⊥.
In the following section, we shall determine all these critical properties in an exact way for MPs.
3. Exact Study of Unbinding Transition3.1. Morse Potential Form
First, we recall the expression of the used one-dimensional MP [21–23]:
(9)V(l)=D[e-2α(l-lm)-2e-α(l-lm)],
where D, lm, and α are positive adjustable parameters. The Morse potential has a minimum value of -D at l=lm and vanishes at l=∞. The particularity of this potential is that at l=0, V(0) has finite value of D(e2αlm-2eαlm), which is positive if only if αlm>ln2. We recall that the MP energy function is frequently used to describe the vibration motion of diatomic molecules. In this case, D is the dissociation energy (the potential depth is -D), and lm accounts for the equilibrium internuclear distance of a diatomic molecules. The former is of the order of 10-5J/m2 (for fluid membranes), and the second is about ten nanometers. The value V(0) is always positive for covalent diatomic molecules [22].
In Figure 1, we report MP versus distance for various values of parameters (D,lm,α).
Reduced Morse potential,V(l)/D, versus distance (in nanometer unit), at fixed minimum point (lm=10nm), for three values of parameter α: 0.1 (solid line), 0.15 (dashed line), and 0.2 (dotted line).
3.2. Resolution of the Schrödinger Equation
First, we start by recalling the useful Katos’s mathematical theorem [26].
Suppose that the potential V(x) of the Schrödinger operator H=-Δ+V(x) is bounded, and that
(10)limx→∞V(x)=0.
Then H has no positive eigenvalues; that is, Hφ=Eφ, where E>0, φ∈L2(Rn); then, φ=0.
Therefore, all eigenfunctions of the Schrödinger equation are bound states, of negative eigenvalues. The eigenvalues spectrum is then discrete.
As remark [26], the point E=0 can be an eigenvalue of the operator H=-Δ+V(x) even when V(x)∈C0∞(Rn).
The considered MP effectively satisfies the above theorem conditions. First, it is locally bounded, since V(l)<V(0), for l<l0, and |V(l)|<D, for l>l0, where l0=lm-α-1ln2 is the unique positive zero of this potential. Second, the limit (10) is also fulfilled due to the presence of exponential tails being the potential expression (9). In addition, according to the above remark, this potential may have a ground state since it is infinitely differentiable. Thus, the system can undergo an unbinding transition that takes place when a bound state appears. The conclusion is that the Schrödinger equation with a Morse potential has only bound states as solutions and a discrete spectrum.
Now to solve the differential equation (2), we first note that this is very similar to the traditional Schrödinger equation:
(11)-ℏ22μd2φndl2+V(l)φn=Enφn,
with μ is the reduced mass and ℏ is the renormalized Planck’s constant, making the substitution(11a)(kBT)2σ⟶ℏ2μ.
The Schrödinger equation has been exactly solved [27], and it will be convenient to recall briefly the essential steps of its resolution.
The first step consists of rewriting it in terms of a new variable r,
(12)-d2φndr2+V(r)φn=εnφn,
with the dimensionless quantities(12a)r=αl,re=αlm,(12b)λ=2μDαℏ,εn=2μα2ℏ2En,
and the rescaled Morse potential
(12c)V(r)=λ2[e-2(r-re)-2e-(r-re)].The minimal value of V(r) is -λ2, reached for r=re. The reduced energies εn (or En) are negative for bound states.
In the second step, the variable
(13)z=2λe-(r-re)
is introduced. It was found in [27] that the orthonormalized bound state wave functions φn(r) are as follows:
(14)φn(r)=Anzλ-n-1/2e-z/2Ln2λ-2n-1(z),
with the normalization constant(14a)An=[Γ(n+1)Γ(2λ-n)]-1/2
and the orthogonal Laguerre polynomial
(14b)Lnγ(z)=z-γezdndzn(zn+γe-z).The associated eigenenergies are
(15)εn=-(λ-n-12)2.
These were obtained using TMM [27]. The existence of solutions to the Schrödinger equation requires that the quantum number n must be smaller than λ. Therefore, the vibrational quantum n is(15a)n=0,1,2,…,nmax,where nmax is the largest integer less than λ.
In terms of the original variables and using substitution (11a), we find that the energies En (per unit length) write
(16)En=-Dλ2(λ-n-12)2,
with the notation(16a)λ=2DσαkBT.
3.3. Probability Density
Applying the general formula (14) to the particular value n=0 and using the fact that L0γ(z)=1, whatever are the values of indice γ and argument z, we find that the ground state is exactly given by
(17)φ0(l)=A0e-α(λ-1/2)(l-lm)-λe-α(l-lm).
This gives the probability density
(18)P(l)=K0e-α(2λ-1)(l-lm)-2λe-α(l-lm),
with the normalization constant(18a)K0=α(2λ)1-2λγ(2λ-1,2λeαlm),
with the condition λ>1/2. We have used the established formula
(18b)∫0cyν-1e-μy(lny)mdy=dmdνm[μ-νγ(ν,cμ)],m=0,1,…,Reμ>0,Reν>0,where γ(ν,x) is the incomplete gamma function [28].
With the help of the determined probability, we compute quantities of our interest, such as 〈l〉 and ξ⊥=[〈l2〉-〈l〉2]1/2.
3.4. Average Separation
We start from its definition
(19)〈l〉=∫0∞lP(l)dl=lm+〈l-lm〉,
with(19a)〈l-lm〉=K0∫-lm∞xe-α(2λ-1)x-2λe-αxdx.
To calculate this integral, we introduce the new variable y=e-αx. Then, we have
(19b)〈l-lm〉=-K0α2∫0eαlmy2λ-2lnye-2λydy.Using the general formula (18b), we find
(20)〈l〉=lm-1α(2λ)1-2λγ(2λ-1,2λeαlm)〈l〉=×ddν[μ-νγ(ν,μeαlm)]|μ=2λ,ν=2λ-1.
The above exact expression has a mathematical sense only if λ>1/2.
3.5. String Roughness
The starting point is its definition:
(21)ξ⊥2=∫0∞l2P(l)dl-(∫0∞lP(l)dl)2.
Using the same techniques yields
(22)ξ⊥2=1α2(2λ)1-2λγ(2λ-1,2λeαlm)ξ⊥2=×d2dν2[μ-νγ(ν,μeαlm)]|μ=2λ,ν=2λ-1ξ⊥2=-(1α(2λ)1-2λγ(2λ-1,2λeαlm)ξ⊥2===×ddν[μ-νγ(ν,μeαlm)]|μ=2λ,ν=2λ-1)2.
This exact expression makes sense if and only if λ>1/2.
3.6. Discussion and Scaling
Let us first note that the system undergoes an unbinding transition if the ground state energy, E0=-(D/λ2)(λ-1/2)2, vanishes. This produces for the typical value λc=1/2. This value then defines the unbinding temperature
(23)Tc=22DσkBα
that naturally depends on the parameters of the problem, which are σ, D, and α. At this temperature, the system exhibits a transition from the bound state (T<Tc or D>Dc) to the unbound one (T>Tc or D<Dc). The above definition then defines a critical line in the plane of variables T and σ (Figure 2), which separates unbound and bound states domains.
Critical unbinding line in the (σ,T)-plane, where the parameters D and σ are fixed to some values.
The above expression also defines a critical potential amplitude, Dc, at fixed temperature,
(24)Dc=α2(kBT)28σ.
In particular, this expression indicates that the unbinding transition is more accentuated for those high tension strings (and high-bending modulus fluid membranes).
Now, come back to the probability distribution expression (18), and notice that its discussion depends on the values of the parameter λ (at fixed α). First, the condition that the probability must fail at infinity (l→∞) requires the inequality λ>1/2. Second, this probability reaches its maximal value for the particular distance
(25)lmax=lm+1αln(2λ2λ-1).
The fact that lmax>lm expresses the tendency of the system to unbind. But as the temperature transition is approached from above (T→Tc- or λ→λc+), the maximum point is shifted towards its higher values, and the probability distribution then becomes more and more boarder; hence the average or mean string position as well as the string roughness increase (see discussion hereinafter). At the transition temperature, the probability vanishes, in agreement with behavior (6) in the ξ⊥→∞ limit.
In Figure 3, we report the probability distribution upon distance for some values of λ close to the critical value λc=1/2.
Contact probability versus renormalized distance, αl, for some values of λ:0.6 (solid line), 0.7 (dashed line), and 0.8 (dotted line). These curves are drawn choosing α=0.1nm-1 and lm=10nm. Remark that the probability is more and more boarder as λ approaches its critical value λc=1/2.
Consider now the mean separation defined in (20). Using the limit γ(ν,x)~ν-1xν (ν=2λ-1→0) [28], we find that near the transition temperature, this average position scales exactly as
(26)〈l〉=1α12λ-1=1α(Tc-TTc)-1,T⟶Tc-.
Therefore the unbinding exponent ψ has the exact value(26a)ψ=1that is very close to the computed one by RG [13].
On the other hand, from relation (22), we find that the string roughness is given by
(27)ξ⊥=1α12λ-1=1α(Tc-TTc)-1,T⟶Tc-.
Therefore, asymptotically, we have the exact equality ξ⊥=〈l〉.
We show that the probability distribution obeys the exact scaling law
(28)P(l)=1ξ⊥Ω(l-lmξ⊥),
with the scaling function(28a)Ω(s)=e-s-e-αξ⊥s.
Its behavior as s→0 is Ω(s)=e-1s0=e-1. Thus, the associated contact exponent exactly reads
(28b)ζ0=ζ.
Then, ζ0=1/2, for strings, and ζ0=1, for fluid membranes. Incidentally, the above equality is that found for those interfaces characterized by a Gaussian probability distribution ξ⊥-1exp(-l2/2ξ⊥2) for the fluctuating field l.
Near criticality, the maximum point of the probability distribution is
(28c)lmax=lm-1αln(Tc-TTc),T⟶Tc-.Thus, the shift lmax-lm diverges logarithmically as the unbinding temperature is approached. Incidentally, this shift plays the role of the wetting-layer size [4].
Finally, the free energy density scales as
(29)f=-Dα21ξ⊥2=-D(Tc-TTc)2,T⟶Tc-.
The corresponding disjoining pressure, Π, can be obtained from the standard definition Π=∂f/∂ξ⊥. Then, we have Π=2(D/α2)ξ⊥-3. The latter can be interpreted as the pressure required to maintain the two strings at the average distance 〈l〉=ξ⊥. Naturally, this pressure is directly proportional to the potential amplitude D. This result reveals that the unbinding phenomenon is entirely governed by the shape fluctuations of strings and bilayer membranes.
4. Conclusions
The goal of this paper is an analytical study of the unbinding transition undergone by strings or bilayer membranes, with a potential of Morse type. Using the transfer matrix method, based on the resolution of a Schrödinger equation, their solutions were found to be bound states. From the exact ground state expression, we first derived the expression of the unbinding temperature in terms of the parameters of the potential. Second, we computed the contact probability that is defined as the probability to find the two interacting manifolds at a (finite) distance apart each other. This probability allowed us to extract all length scales, which are the average separation and roughness, and the associated exponents were exactly deduced. These were found to be very close to those derived using RG machineries.
The main conclusion is that our analytical studies reveal that the Morse potential is a good candidate for the description of the unbinding phenomenon within manifolds (strings and membranes). This potential choice overcomes the difficulty related to the singularity of the real potential at the origin. As a matter of fact, this singularity cannot be seen since one is usually interested in the high scales critical behavior.
We emphasize that once our work is achieved, we learned that the Morse potential was already used for the study of DNA denaturation problem [29, 30].
Such a study can be extended to a generalized Morse potential (GMP). The latter was introduced by Deng and Fan [31] and related to the Manning-Rosen potential [32, 33] (also called Eckart potential by some authors [34, 35]). It is defined by V(l)=D[(1-b/(eal-1))2-1], with the notation b≡ealm-1, where lm is the maximum of the potential. The specificity of GMP is that it behaves as a hard-sphere potential at the origin l=0. It has been demonstrated [36] that the Schrödinger equation with this potential is exactly solvable. In our context, in particular, we find that the unbinding temperature reads Tc=kB-1b/(b+2)2Dσ. Other developements will be presented elsewhere [37].
Finally, the present analysis may be extended to more than two manifolds.
Acknowledgments
The authors are much indebted to Professor T. Bickel for the fruitful correspondence and to Professors J.-F. Joanny and C. Marques for the helpful discussions during the Second International Workshop On Soft-Condensed Matter Physics and Biological Systems, 28–30 November 2010, Fez, Morocco.
LässigM.LipowskyR.van BeijerenH.ErnstM. H.Universal aspects of interacting lines and surfaces1994Elsevier Science B.V169206SackmannE.LipowskyR.SackmannE.Physical basis of self-organization and function of membranes: physics of vesicles1995Amsterdam, The NetherlandsElsevier213304AlmeidaP. F. F.VazW. L. C.LipowskyR.SackmannE.Lateral diffusion in membranes1995Amsterdam, The NetherlandsElsevier305357DietrichS.DombC.LebowitzJ.Wetting phenomena198812London, UKAcademic Press1218EisenrieglerE.1993SingaporeWorld ScientificFisherM. E.Interface wandering in adsorbed and bulk phases, pure and impure1986821569160310.1039/f29868201569LipowskyR.Lines of renormalization group fixed points for fluid and crystalline membranes198873255262HelfrichW.Elastic properties of lipid bilayers-theory and possible experiments197328693703SafinyaC. R.RouxD.SmithG. S.SinhaS. K.DimonP.ClarkN. A.BellocqA. M.Steric interactions in a model multimembrane system: a synchrotron X-ray study198657212718272110.1103/PhysRevLett.57.2718SafinyaC. R.SirotaE. B.RouxD.SmithG. S.Universality in interacting membranes: the effect of cosurfactants on the interfacial rigidity198962101134113710.1103/PhysRevLett.62.1134RandR. P.ParsegianV. A.Hydration forces between phospholipids bilayers198998833513762-s2.0-002437891810.1016/0304-4157(89)90010-5IsraelachviliJ. N.19912ndLondon, UKAcademic PressLipowskyR.LeiblerS.Unbinding transitions of interacting membranes19865623254125442-s2.0-000142852410.1103/PhysRevLett.56.2541LisL. J.LisW. T.ParsegianV. A.RandR. P.Adsorption of divalent cations to a variety of phosphatidylcholine bilayers1981207177117772-s2.0-0019891896WiegelF.1986Philadelphia, Pa, USAWorld ScientificSchulmanL. S.2005DoverKleinertH.20095thLondon, UKWorld ScientificItzyksonC.DrouffeJ.-M.1989Cambridge, Mass, USACambridge University PressZinn-JustinJ.1989Oxford, UKClarendon PressLipowskyR.Typical and exceptional shape fluctuations of interacting strings19917703708AsfawM.Unbinding transitions of membranes and strings from single and double-well potential2008387143526353610.1016/j.physa.2008.02.048MorseP. M.Diatomic molecules according to the wave mechanics. II. Vibrational levels1929341576410.1103/PhysRev.34.57HerzbergG.1950New York, NY, USAVan Nostrand-ReinholdChoS.SunH.Rotational state distributions of I2(B) from vibrational predissociation of I2(B)-Ne200425913971402ManghiM.DestainvilleN.Statistical mechanics and dynamics of two supported stacked lipid bilayers2010266405740682-s2.0-7795792183310.1021/la903504nBerezinF. A.ShubinM. A.1991Kluwer Academic PublishersSunH.The Morse potential eigenenergy by the analytical transfer matrix method200533830931410.1016/j.physleta.2005.02.054GradshteynI. S.RyzikI. M.1980New York, NY, USAAcademic PressPeyrardM.BishopA. R.Statistical mechanics of a nonlinear model for DNA denaturation19896223275527582-s2.0-424318731910.1103/PhysRevLett.62.2755DauxoisT.PeyrardM.BishopA. R.Dynamics and thermodynamics of a nonlinear model for DNA denaturation19934716846952-s2.0-1214427997010.1103/PhysRevE.47.684DengZ. H.FanY. P.A potential function of diatomic molecules19577162166InfeldL.HullT. E.The factorization method1951231216810.1103/RevModPhys.23.21ManningM. F.RosenN.Minutes of the Middletown meeting, October 14, 193319334411951954DabrowskaJ. W.KhareA.SukhatmeU. P.Explicit wavefunctions for shape-invariant potentials by operator techniques1988214L195L2002-s2.0-000171869610.1088/0305-4470/21/4/002CooperF.KhareA.SukhataneU.Supersymmetry and quantum mechanics19952515-626738510.1016/0370-1573(94)00080-Mdel Sol MesaA.QuesneC.SmirnovY. F.Generalized Morse potential: symmetry and satellite potentials19983113213352-s2.0-003249805010.1088/0305-4470/31/1/028BenhamouM.El KinaniR.KaidiH.Exact study of the unbinding transition of strings and biomembranes from generalized short-ranged attractive potentials: proof of universalityin preparation