A single-theory approach to the prediction of solid–liquid and liquid–vapor phase transitions Payman Pourgheysar Department of Chemical Engineering, Amirkabir University of Technology, Tehran Polytechnique, Tehran, Iran G. Ali Mansooria) Department of Chemical Engineering, University of Illinois at Chicago, Chicago, Illinois 60607-7000 Hamid Modarress Department of Chemical Engineering, Amirkabir University of Technology, Tehran Polytechnique, Tehran, Iran ~Received 20 February 1996; accepted 26 August 1996! For simultaneous prediction of solid–liquid and liquid–vapor phase transitions it has been customary to apply two different theories for solid and fluid phases. A single-theory approach will be desirable to answer many of the fundamental problems of molecular theory and their relationship with macroscopic behavior of the matter. Based on a modified version of the cell model of statistical mechanics, a single-theory approach for simultaneous prediction of solid–liquid and liquid–vapor phase transitions is presented here. In developing this theory the order–disorder transition is considered as the essential feature of the fusion and a new function for the potential energy field inside a single-occupancy cell is derived. By reporting the variations of total pressure of the macroscopic system with respect to temperature and volume the nature of the various phase transitions in the system are evaluated and discussed. Variations of the radial distribution function of the molecules in the system with intermolecular distance, temperature, and volume are reported for various phases of matter. © 1996 American Institute of Physics. @S0021-9606~96!50245-X# I. INTRODUCTION Phase transitions and their predictions are of prime importance in the study of matter and they constitute the cornerstone of many processes encountered in the science and technology.1–11 Of the three phase transitions in macroscopic systems consisting of simple molecules, solid–liquid, liquid–vapor, and vapor–solid, the latter two are well understood and the related predictive statistical mechanical models have found numerous practical applications. There has been a wealth of research activity in the development of predictive models for solid–liquid transitions ~both melting and freezing! since the mid-sixties and it has resulted in advancement of a number of quantitatively accurate predictive models.1–5,7–10 Among these the density functional theory has attracted a great deal of attention in recent years.2,3 In predicting simultaneously solid–liquid and liquid– vapor phase transitions it has been necessary to use two different theories to describe the solid and fluid phases, separately. What has been missing from the scene is a singletheory approach through which one can predict all the three phases of matter and the related phase transitions. The single-theory approach presented here is for the purpose of achieving this goal. The cell theory is shown to be able to establish, in a typical way, a connection between liquid and dense gas states.1,14–16 Also, some success has been achieved to do the same in the case of solid and liquid phases using the cell a! Author to whom correspondence should be addressed. 9580 J. Chem. Phys. 105 (21), 1 December 1996 theory.12,14 Considering the above facts we recognize the cell theory to be appropriate for predicting solid–liquid and liquid–vapor phase transitions, simultaneously. The simplicity of the mathematical formulations involved in the singleoccupancy cell theory allows us to pay more attention to the various features of the model and their relationships to the physical phenomenon of phase transitions. This purpose could not be generally achieved with the use of the other more sophisticated theories of statistical mechanics. II. DEVELOPMENT OF THE THEORY In the development of a cell model capable of predicting all the three phase transitions ~solid–liquid, liquid–vapor, solid–vapor!, for simplicity the attention will be focused on the basic single-occupancy version of the cell model. Thus, let us imagine the volume space occupied by particles partitioned into a set of Wigner–Seitz cells corresponding to a face-centered cubic lattice ~fcc! and each cell is supposed to contain exactly one particle. The total intermolecular potential energy of the system, accepting the pairwise additivity assumption, is then given by U5 ( ( 1<i, j<N uij , ~1! where N is the total number of molecules in the system and u i j is the pair intermolecular potential energy function between molecules (I) and ( j) of the system. The classical partition function of this system according to the cell theory is given by 0021-9606/96/105(21)/9580/8/$10.00 © 1996 American Institute of Physics Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions Z5l 23N E E DN ••• D1 e 2 b U dr1 •••drN , ~2! in which: l5[h /(2 p mkT)] 1/2, b51/(kT), and the integration over the position vector ~rI ! is taken over cell (I) ~i.e., Di and not over the whole volume!. Position vector ~rI ! denotes the position of the molecule in cell (I) relative to the center of the cell. To make possible the evaluation of the many-body integral in Eq. ~2!, we seek to approximate (u i j ) by the form 2 u i j ~ r i ,r j ! .w ji ~ r i ! 1w i j ~ r j ! , w ji ~ r i ! 5 21 u i j ~ 0,0! 1 @ u i j ~ r i ,0! 2u i j ~ 0,0!# . ~4! The approach based on this choice is designated as the usual ‘‘free-volume’’ version of the cell theory. Let us now assume that the only important interaction is the one between the first-nearest-neighbor molecules. Then, let us define u i~ r i ! [ (j w ji ~ r i ! 5 (j 1 (j @ u i j ~ r i ,0! 2u i j ~ 0,0!# . ~5! Where the summation is taken over the first-nearestneighbors of molecule (I), and N U[ ( u i~ r i ! . i51 ~6! By substitution of Eq. ~6! in Eq. ~2! we obtain ) E SE e 2 b u i dri N Z5l 23N i51 5l 23N Di Di e 2 b u i dri D ~7! N ~8! , where we have assumed that the function u i (r i ) for different cells is the same. Let us now rewrite Eq. ~8! as the following: FE F ( H ( F ( SE H ( Z5l 23N Di exp 2 b 3exp 2 b j Di 1 2 u i j ~ 0,0! G J G N @ u i j ~ r i ,0! 2u i j ~ 0,0!# dri 5l 23N exp 2 b N 3 j exp 2 b j j 1 2 F Z5l 23N exp 2 b N u i j ~ 0,0! G 3 SE H Di ~9! J D N . ~10! exp 2 b u ik ~ 0,0! G (j @ u i j ~ r i ,0! 2u i j ~ 0,0!# J D N dri ~11! ~12! in which U s5 and ( ( l<i, k<N u ik ~ 0,0! , E H Di exp 2 b (j @ u i j ~ r i ,0! 2u i j ~ 0,0!# J dri , ~13! where ~n f ! is called the free-volume integral.1,15 Evaluation of the individual terms in Eq. ~12!, the freevolume integral ~n f ! and the static potential energy (U s ), are the essential problem in construction of the model which will be addressed next in this report. A. Evaluation of the free-volume integral (nf ) The integration in ~n f ! is to be performed over the volume of the cell which its shape, according to the Wigner– Seitz lattice model for fcc lattice, is a dodecahedron. By using the smearing approximation1,16 we can, for convenience, approximate the potential field inside the cell with a spherically symmetric function depending only on the distance of the molecule from the center of its cell (R i ). Then ~n f ! can be calculated by averaging the potential energy over a sphere of radius (R i ) centered at the cell center. The cell field derived in this way is identical with that derived by regarding the neighboring molecules as ‘‘smeared’’ with uniform probability distribution over the surfaces of concentric spheres of appropriate radius. Also, by regarding this radial symmetry of the potential function, we can, for simplicity, approximate the actual cell with a sphere which its radius (R c ) is obtained from setting the volume per molecule ~n 5V/N! equal to the volume of this sphere @(V) is the total volume and (N) is the total number of molecules in the system#. Therefore we will have nf5 @ u i j ~ r i ,0! 2u i j ~ 0,0!# dri ( ( 1<i, k<N 5l 23N e 2 b U s ~ n f ! N , nf5 u i j ~ 0,0! 1 2 The first exponential term in Eq. ~10! refers to the static potential energy (U s ) of the system, and it can be modified to take into account the interactions of molecule (I) with molecules other than the first-nearest-neighbors of (I) by extending the range of summation to include all the molecules (k) other than (I). Therefore, Eq. ~10! can be written as ~3! which would reduce the multiple integral in Eq. ~2! to a product of a single integrals. One possible choice of the functions (w ji ) is 9581 5 E E Rc 0 Rc 0 H 4 p R 2i exp 2 b (j @ u i j ~ r i ,0! 2u i j ~ 0,0!# 4 p R 2i e 2 bc i ~ r i ! dR i , J dR i ~14! ~15! in which (R i ) is the absolute value of (r i ), 34 p R 3c 5n, and J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp 9582 Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions c i ~ r i ! 5 ( @ u i j ~ r i ,0! 2u i j ~ 0,0!# . ~16! j For fcc lattice, the distance between the first nearest neighbor lattice points ~a 1! is related to ~n! by the relation (a 31 5& v ). Thus, a 31 4 p R 3c 5 3 & or S D Rc .0.55. a1 c ha i ~ ri!5 It should be pointed out that the smearing approximation is actually valid at the high density region. At lower densities the increase of molecular interspace makes its accuracy a bit doubtful. Another noticeable concept, in relation to the assumptions made here, is the ‘‘communal entropy.’’1,15,16 As the temperature approaches infinity, Eq. ~12! takes the form Z T→` 5l 23N SD V N c1 2 0 @ R 2i 1a 21 22a 1 R i cos~ u !# 2 w ~ a 1 ! % sin~ u ! d u , G S D 2 Ri 1 2 a w 9 ~ a 1 ! 1O 2 1 a1 a 1w 8~ a 1 ! 1 4 . ~18! This expansion converges quite rapidly at high densities, where the potential energy well can be represented as a harmonic function. At this stage we choose the Lennard-Jones intermolecular potential energy function to represent w(x): w LJ~ x ! 54 e FS D S D G s x s x 12 2 6 , where s and e are the Lennard-Jones potential parameters. Therefore, c ha i (r i ) given by Eq. ~18! becomes . E $w A p S DF c1 Ri 3 a1 N At this limit, the kinetic energy of each molecule is sufficient to make the effect of intermolecular forces unimportant as it is also the case in the ideal gas state. On the other hand, by comparing the above equation with the partition function of an ideal gas @i.e., Z ideal5l 23N (V N /N!)#, it reveals that the only difference between the two expressions is in the appearance of the factor ~1/N!! in ~Z ideal! instead of ~1/N! in ~Z T→` ! which leads to the appearance of an additional term in entropy formula called the ‘‘communal entropy’’.16 Communal entropy is due to the unlocalization of ideal gas molecules, so that for a molecule in gaseous state, the whole volume is available for its movement. The simplest way to make these two partition functions identical, is entering the factor (e N ) in Eq. ~12! @notice that N!.(N/e) N when N→`, see Ref. 14#. It can be shown that communal entropy has no effect on evaluating the pressure which is the only macroscopic property of the system that will be evaluated, here. In the liquid state, this feature is arisen as the possibility of multiple occupancy of the cells. However, the modifications made to take into account this effect do not produce any noticeable improvement over the single-occupancy version of the cell model considered here. In the solid state, because of the closed-packing of the molecules, this problem is not arisen at all. Therefore, we can ignore the existence of ‘‘communal entropy’’ and multiple-occupancy effects in our calculations. Then with the assumptions of single-occupancy cells, smearing approximation and consideration of only the firstnearest-neighbors interactions in calculation of the freevolume, the following expression for c i (r i ) will be derived:1,14–16 c i~ r i ! 5 (x), and (c I ) is the number of first-nearest-neighbors of a molecule ~for example for fcc lattice c I 512!. Equation ~17! can be solved by a Taylor series expansion in terms of (R i ,). For small values of (R i ,) it can be approximated by an equation which we designate as its ‘‘harmonic approximation’’ 15 @ c ha i (r i ) # 0,R i ,0.55, ~17! where w(x) is the function representing the pair interaction potential energy between every two molecules at the distance c ha i ~ ri!5 S DF F 2 c1 Ri 3 a1 516e y 2i 8 ~ a1!1 a 1 w LJ G 1 2 a w9 ~a ! 2 1 LJ 1 G 16.5 7.5 2 1O ~ y i ! 4 , n *4 n *2 ~19! where ~n*5n/s3! is the dimensionless volume per cell, (y i 5R i /a 1 ) is the dimensionless distance of a molecule from its cell center and (a 31 5& v ). Let us now consider the physical limitation imposed on every molecule in its cell when it is in the solid and liquid states. In such states of matter, every molecule is confined in its cell and, effectively, the expression for c ha i (r i ) will assume the following modified form, c mha (r ), by taking into i i account the repulsive forces at the cell wall, c mha i ~ ri!5 H F G 16.5 7.5 2 , n *4 n *2 0.5,y i . 16y 2i e `, 0<y i <0.5 . ~20! Using the above effective modified form of c mha i (r i ) the expression for the free-volume will assume the following form: nf5 E 0.5 0 mha 4 p y 2i e 2 bc i or, in dimensionless form n *f 5 nf 54 p & n * s3 E 0.5 0 ~ri! ~21! dy i , F y 2i exp 2 S 16y 2i 16.5 7.5 2 T * n *4 n *2 DG dy i . ~22! Equation ~22! can be used to calculate the free-volume integral at various dimensionless temperatures and volumes for the fcc lattice. Similar expressions can be derived for other lattice structures and/or for multi-occupancy cell models. B. Evaluation of the static potential energy ( U s ) Suppose the number of molecules (c i ) and distances (a i ) of the successive neighboring shells with respect to a J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions S D ]n *f ]n * central molecule in terms of the corresponding parameters for the first-nearest-neighbor shell (c 1 ,a 1 ) is given by c i 5n i •c 1 a i 5d i •a 1 , and i51,2,3,... . 54 p & N,T Thus, (U s ), after replacing (u ik ) by wLJ , becomes U s5 (( 1<i,k<N 5 ( ( i51 k51 ~ kÞi ! N 5 ( i51 4e c 2 1 3exp 1 u ~ 0,0! 2 ik FS D ( S D ( G s a1 12 ` s 122 a di 1 ni i51 6 ` ni d 6i i51 ~23! For the fcc lattice we can show that 3exp and S D n 1 51, n 2 51/2, n 3 52, n 4 51, n 5 52, n 6 52/3,... . ]n *f ]n * As a result, it can be shown that ( i51 ` ni .1.01 d 12 i Therefore, N U s5 ( i51 and F S D s 24e 1.01 a1 524N e F ( i51 ni d 6i G S DG ~24! ~25! C. Evaluation of the total pressure By using Eqs. ~21! and ~25! in the partition function, Eq. ~12!, we can evaluate the macroscopic properties of the system such as the total pressure ] ln Z ]V 5kT N,T F H F S 5 2 P *5 ]~ 2bUs! ] ln n f 1N ]V ]V DG ] 1.01 1.205 kT ]n f 24e 2 1 ]n 4 n *4 2 n *2 n f ]n Thus, in dimensionless form F 0 S E 0.5 0 G S D 1.01 1.205 Ps3 T * ]n *f 524 52 3 1 e n* n* n *f ]n * S D D GJ S D D GJ y 2i ~ 16y 2i ! 16y 2i 16.5 7.5 2 2 T n4 n 4p& T* 2 0.5 y 2i ~ 16y 2i ! S 16y 2i 16.5 7.5 2 2 T n4 n D GJ 66 n *4 15 n *2 dy i . ~29! S 54 p & G1 N,T G REP2G ATT T* D ~30! ~31! Finally, substitution of Eqs. ~30! and ~31! in Eq. ~28! gives 6 1.01 1.205 2 . 4 n *4 2 n *2 S D 2 E 16y 2i 16.5 7.5 T n4 n2 n *f 54 p & n * G. By replacing Eqs. ~22! and ~25! into Eq. ~12! one can calculate the partition function, and from that, all the macroscopic properties of the system. In what follows evaluation of the total pressure of the system is reported through which one can demonstrate the prediction of phase transitions. P5kT 4p& T* HF S and .1.205. s 21.205 a1 12 HF y 2i exp We designate the three integrals appearing in the above equation as (G), ~G REP!, and ~C ATT!, respectively. Thus, ( ]n *f / ]n * ) N,T and ( n *f ) can be written in terms of these three integrals as d i 5 Ai ` HF 3dy i 2 . 0.5 0 3dy i 1 u ik ~ 0,0! N N E 9583 J G N,T . ~26! ~27! N,T , S D F S DG 1.01 1.205 T* 1 G REP2G ATT 11 . 52 3 1 n* n* n* T* G ~32! Equation ~32! is plotted for various dimensionless temperature ~T *! values as isothermal ~P *2n*! curves in Figs. 1~a!, 1~b!, and 1~c!. According to Fig. 1~a! for the isotherms T *50.9, 1.5, 2, 3, we observe two phase transitions ~solid– liquid and liquid–vapor!. As we increase the temperatures ~T *57 and 10!, Fig. 1~b!, we observe only the liquid–vapor phase transition and at higher temperatures ~T *.12! no phase transitions are observed. In the other hand, at much lower temperatures ~T *50.1 and 0.2! the resulting isotherms in Fig. 1~c! could indicate the existence of only one phase transition which can be interpreted as the solid–vapor phase transition. According to Figs. 1~a!, 1~b!, and 1~c! the present model is capable of predicting both solid–liquid and liquid–vapor phase transitions, simultaneously at intermediate temperatures. This kind of capability is unique to the present theory, and to our knowledge, no other single theory has been able to predict these transitions together. In order to elucidate further on the capability of this model we also report the radial distribution functions of various phases of matter ~solid, liquid, and vapor! which can be predicted by this single theory approach. P * 524 ~28! N,T where (T * 5kT/ e ) is the dimensionless temperature and ( P * 5 P s 3 / e ) is the dimensionless pressure. But, according to Eq. ~22! we have D. Evaluation of the radial distribution function (RDF) For the modified harmonic approximation model proposed here, Eq. ~20!, we can calculate the RDF using the J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp 9584 Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions FIG. 1. ~a!–~c! Variation of the dimensionless pressure P * vs n* at different values of dimensionless temperatures as calculated by the present model. following expression originally proposed by Corner and Lennard-Jones based on the single-occupancy cell model13 g ~ Y , n * ,T * ! 5 ` 1 ( 16p &G 2 Y k51 mha 3e 2 bc i ~yi! dy i ck dk E E E ` dV 0 1/2 u V2Y u 1/2 u V2d k u mha y j e 2 bc j yi ~y j! dy j , ~33! where, according to Fig. 2, parameter (Y 5D/a 1 ) is the dimensionless distance between any two molecules ~i and j! belonging to two different single-occupancy cells and parameter ~V5S/a 1! is the dimensionless distance of molecule (i) from the center of the cell ( j). Parameter (k) refers to the kth shell and uV2d k u and uV2Y u must be less than 1/2, otherwise the two last corresponding integrals in Eq. ~33! become zero. This guarantees the existence of a molecule in the neighborhood. J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions 9585 By replacing Eq. ~20! in Eq. ~33! we can calculate the RDF of molecules in a substance which obeys the singleoccupancy cell model. In Figs. 3~a!, 3~b!, and 3~c!, the RDFs of solid, liquid, and vapor phases for various dimensionless temperatures and volumes are reported. In calculating the RDF, we have considered up to the 6th nearest-neighbor shells which leads to the conclusion that the ~V! values greater than ~A61 21.2.95! are not effective on the RDF. Therefore, we have substituted the required quantities, mha such as c mha i (r i ) or c j (r j ) by Eq. ~20!, and we have replaced ~`! in the limit of the first integral of Eq. ~33! by ~3.00! ~which is the rounded value of 2.95! and then, we have made numerical calculations. Details about the RDF calculations and other related matters are discussed below. III. DISCUSSION FIG. 2. The intermolecular distance notations used to define the variables in the radial distribution expression ~RDF!, Eq. ~33!. As it is observed in Fig. 1~a!, at relatively moderate ~T *! ranges, there are two S-shaped sections in each isotherm corresponding to a low and a high ~n*!, which can be attributed FIG. 3. ~a!–~c! Variations of the RDF in different phases ~solid, liquid, and vapor! with the dimensionless distance between any two molecules (Y ) at different values of T * and n*. J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp 9586 Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions FIG. 4. ~a! and ~b! Effect of the repulsive ~REP! and attractive ~ATT! parameters on the shape of the P * – n* isotherms and the occurrence of phase transitions. to solid–liquid and liquid–vapor phase transitions, respectively. If we apply the Maxwell’s criterion of phase equilibrium to these two S-shaped sections, we will find that the ratio of equilibrium volume of the low density state to the dense state for liquid–vapor transition is much greater than the corresponding value for the solid–liquid case, which is qualitatively in agreement with the experimental evidence. Also, the increase in equilibrium pressure with temperature in the solid–liquid case is much greater than the corresponding value for liquid–vapor transition, in agreement with the experimental observation. On Fig. 1~a!, it is observed that the equilibrium pressure of liquid–vapor transition predicted by the present cell model is negative. This negative-pressure liquid–vapor phase transition is maintained even at high temperatures, where the real matter according to the experimental observation, must be in its supercritical state. It should be pointed out that at relatively high temperatures, Fig. 1~b!, the pressure can become positive. This deficiency of the present model is an immediate result of the amplified attractive interaction effect of c mha i (y i ) in the neighborhood of the cell wall ~i.e., y i .0.5!. This increases the probability of molecules remaining in the condensed state, and that breaking down this aggregated structure will need sufficient thermal energy. We also notice the negative triple point pressure, Fig. 1~c!, predicted by this model. According to Fig. 1~b!, we observe a liquid–vapor critical point at rather high temperatures ~T *;10!. By investigating the isotherms in Fig. 1~a!, it seems that there is also a critical point corresponding to solid–liquid phase transition. As ~T *! increases, equilibrium dimensionless volume of saturated liquid ~obtained from Maxwell’s criterion! decreases while for saturated solid it increases until they reach at their critical point where they become identical. So far, up to the pressures for which experiments can be performed, it is shown that the decrease of saturated-liquid-volume with ~T *! is correct but saturated-vapor-volume must decrease, also. It should be mentioned that while liquid–vapor critical point can exist, there is no experimental evidence in support of the existence of the solid–liquid critical point. As it can be seen from Figs. 1~a!, 1~b!, and 1~c!, for small values of ~n*!, which correspond to the solid state, the RDF is noticeable only at the shell positions, otherwise it becomes negligible. For large distances between the neighboring shells and the central molecule ~large y values!, the RDF becomes smooth and approaches unity ~not shown here! as we reach at the bulk or macroscopic density domain. By increasing ~T *!, variations in the RDF will be further smoothed out which is another way of saying the solid crystal becomes imperfect ~i.e., the disorder increases!. When ~n*! takes still larger values ~i.e., moving toward the liquid domain!, the RDF takes an oscillatory form with damping amplitude, typical of the liquid state representing short-range order and long-range disorder, so that some peaks of the RDF at the position of the neighboring shells of lattice points in the previous case ~solid! disappear. This is a typical behavior observed during melting that can be attributed to order–disorder phenomena. In the gaseous region ~i.e., at relatively high n*!, the RDF, after the first-nearest-neighbor position, becomes nearly unity in agreement with the theory of RDF. Therefore, as a final result, variations of the RDF, obtained on the basis of this new model can represent solid– liquid and liquid–vapor transitions. There is no special physical significance to coefficients ~16.5 and 7.5! appearing in Eq. ~20! other than the fact that J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp Pourgheysar, Mansoori, and Modarress: Prediction of phase transitions they result from the Lennard-Jones model potential. In order to study the effect of variations in these coefficients on the phase transitions we replace them by ~REP! and ~ATT! notations representing repulsive and attractive energy parameters, respectively. Therefore, we rewrite Eq. ~20! in a more general form c mha i ~ y i!5 H 16y 2i e `, F G REP ATT 2 , n *4 n *2 0<y i <0.5, 0.5,y i . ~34! As it can be seen in Fig. 4~a!, increasing the value of ~ATT! ~i.e., attractive contribution of the pair interactions! compared with ~REP! ~repulsive ones!, promotes the occurance of solid–liquid transition at higher pressures. It also causes the equilibrium pressure of vaporization to increase as expected. It can be also observed that increasing ~REP!, compared with ~ATT!, has the reverse effect as shown in Fig. 4~b!. IV. CONCLUSION In predicting solid–liquid and liquid–vapor phase transitions it has been customary to use two different statistical mechanical models, one for the solid state and another for the fluid state. The modified cell model introduced in this report has the capability of simultaneously predicting, even though qualitively, solid–liquid and liquid–vapor phase transitions of molecular fluids using a single statistical mechanical model. It is based on a modified harmonic cell potential function which is an approximation to the single-occupancy cell theory originally proposed by Lennard-Jones and 9587 Devonshire.1,14–16 The proposed statistical mechanical approach has the capability of being improved to predict all the phase transitions ~solid–liquid, liquid–vapor, and vapor– solid! more quantitatively. Such improvements can be accomplished, possibly, by application of the present approach in reformulation of the more advanced cell models in which there are the possibilities of multi-occupancy and/or vacancy of cells, as well as the possibility of variations of the number of nearest neighbors as a result of changes of state or phase transitions. J. A. Barker, J. Chem. Phys. 63, 632 ~1975!. C. Rascon, G. Navascues, and L. Mederos, Phys. Rev. B 51, 14899 ~1995!. 3 S. M. Osman, J. Phys. Cond. Matter 6, 6965 ~1994!. 4 D. Kuhlmann-Wilsdorf, Phys. Rev. 140, 1599 ~1965!. 5 G. A. Mansoori and F. B. Canfield, J. Chem. Phys. 51, 4967 ~1969!. 6 Molecular Based Study of Fluids, edited by J. M. Haile and G. A. Mansoori Adv. Chem. Series 204 ~1983!. 7 Y. Kato and N. Nagaosa, Phys. Rev. B: Condensed Matter. 47, 2932 ~1993!. 8 M. Hirami, J. Chem. Phys. 99, 8290 ~1993!. 9 D. A. Young, J. Chem. Phys. 98, 9819 ~1993!. 10 S. Doniach, T. Garel, and H. Orland, J. Chem. Phys. 105, 1601 ~1996!. 11 M. R. Ekhtera and G. A. Mansoori ~unpublished!. 12 J. A. Barker, Proc. R. Soc. London Ser. A 240, 265 ~1957!. 13 J. Corner and J. E. Lennard-Jones, Proc. R. Soc. London Ser. A 178, 401 ~1941!. 14 J. E. Lennard-Jones and A. F. Devonshire, Proc. R. Soc. London Ser. A 163, 53 ~1937!; 169, 317 ~1939!; 170, 464 ~1939!. 15 J. O. Hirschfelder, C. F. Curtiss, and R. B. Bird, Molecular Theory of Gases and Liquids ~Wiley, New York, 1964!, Chap. 4, Secs. 5,6,7. 16 T. L. Hill, An Introduction to Statistical Thermodynamics ~Dover, New York, 1986!, Chap. 16. 1 2 J. Chem. Phys., Vol. 105, No. 21, 1 December 1996 Downloaded¬25¬Oct¬2001¬to¬131.193.142.216.¬Redistribution¬subject¬to¬AIP¬license¬or¬copyright,¬see¬http://ojps.aip.org/jcpo/jcpcr.jsp
© Copyright 2026 Paperzz