cond-mat0005118/part1
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%%         The following is a LaTeX file        %%% 
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \documentstyle[12pt,epsfig]{article}             %%%
5: %%%%%%%%%%%%%%%%  Definitions  %%%%%%%%%%%%%%%%%%%%%
6: \textheight 237mm                               %%
7: \textwidth 160mm                                %%
8: \headheight 0mm                                 %%
9: \setlength{\oddsidemargin}{0mm}                 %% 
10: \topmargin -1cm                                 %%
11: %\hbadness=5000                                  %%
12: %\vbadness=8000                                  %%
13: %\addtolength{\voffset}{-1in}                    %%
14: %\addtolength{\hoffset}{-0.3in}                  %%
15: %\addtolength{\textheight}{1in}                  %%
16: %\addtolength{\textwidth}{0.6in}                 %%
17: %\newfont{\ffont}{msym10}                        %%
18: \newcommand{\beq}{\begin{equation}}             %%
19: \newcommand{\eeq}{\end{equation}}               %%
20: \newcommand{\bqry}{\begin{eqnarray}}            %%
21: \newcommand{\eqry}{\end{eqnarray}}              %%
22: \newcommand{\bqryn}{\begin{eqnarray*}}          %%
23: \newcommand{\eqryn}{\end{eqnarray*}}            %%
24: \newcommand{\NL}{\nonumber \\}                  %%
25: \newcommand{\preprint}[1]{\begin{table}[t]      %%
26:             \begin{flushright}                  %%
27:             \begin{large}{#1}\end{large}        %%
28:             \end{flushright}                    %%
29:             \end{table}}                        %%
30: \newcommand{\eref}[1]{(\ref{#1})}               %%
31: \newcommand{\DD}[2]{\frac{d^{#2}}{d#1^{#2}}}    %%
32: \newcommand{\PD}[2]                             %%
33:     {\frac{\partial^{#2}}{\partial #1^{#2}}}    %%
34: \newcommand{\cc}{\mbox{\ffont C}}               %%
35: \newcommand{\rr}{\mbox{\ffont R}}               %%
36: \newcommand{\kk}{\mbox{\ffont K}}               %%
37: \newcommand{\zz}{\mbox{\ffont Z}}               %%
38: \newcommand{\ii}{\mbox{\ffont I}}               %%
39: \newcommand{\nn}{\mbox{\ffont N}}               %%
40: \newcommand{\qq}{\mbox{\ffont Q}}               %%
41: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
42: %%%%%%%%% numbering equations by sections %%%%%%%%%%
43: %%\catcode`\@=11 \@addtoreset{equation}{section}  %%
44: %%\renewcommand{\theequation}                     %%
45: %%         {\arabic{section}.\arabic{equation}}   %%
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: \begin{document} 
48: \preprint{LA-UR-99-6772} 
49: \title{Analysis of Dislocation Mechanism for Melting \\ of Elements: 
50: Pressure Dependence} 
51: \author{\\ Leonid Burakovsky\thanks{E-mail: BURAKOV@T5.LANL.GOV}, \
52: Dean L. Preston\thanks{E-mail: DEAN@LANL.GOV}, \
53: and Richard R. Silbar\thanks{E-mail: SILBAR@WHISTLESOFT.COM. Also at 
54: WhistleSoft, Inc., %168 Dos Brazos, 
55: Los Alamos, NM 87544, USA
56: %\hspace*{0.5cm} NM 87544, USA
57: }
58:  \\  \\ 
59: %Theoretical Division, MS B283 \\  
60: Los Alamos National Laboratory \\ Los Alamos, NM 87545, USA }
61: \date{ }
62: \maketitle
63: \begin{abstract}
64: In the framework of melting as a dislocation-mediated phase transition we 
65: derive an equation for the pressure dependence of the melting temperatures 
66: of the elements valid up to pressures of order their ambient bulk moduli. 
67: Melting curves are calculated for Al, Mg, Ni, Pb, the iron group (Fe, Ru, Os), 
68: the chromium group (Cr, Mo, W), the copper group (Cu, Ag, Au), noble gases 
69: (Ne, Ar, Kr, Xe, Rn), and six actinides (Am, Cm, Np, Pa, Th, U). These 
70: calculated melting curves are in good agreement with existing data. We also 
71: discuss the apparent equivalence of our melting relation and the Lindemann 
72: criterion, and the lack of the rigorous proof of their equivalence. We show 
73: that the would-be mathematical equivalence of both formulas must manifest 
74: itself in a new relation between the Gr\"{u}neisen constant, bulk and 
75: shear moduli, and the pressure derivative of the shear modulus. 
76: \end{abstract}
77: \bigskip 
78: \centerline{{\it Key words:} melting, string, dislocation, melting curve, 
79: equation of state, high} %pressure, 
80: \hspace*{3cm} pressure, elements, actinides 
81: 
82: \centerline{PACS: 62.50.+p, 64.10.+h, 64.70.Dv, 64.90.+b, 74.62.Fj, 77.84.Bw, 
83: 91.60.Gf} 
84: \bigskip
85: 
86: \section{Introduction} 
87: 
88: The idea that a proliferation of dislocations is associated with melting dates 
89: back to Mott \cite{Mott}. The very first theory of dislocation-mediated 
90: melting \cite{MO} was a success, inasmuch as it predicted a first-order 
91: transition, as a consequence of incorporating the mutual screening of 
92: dislocations, in agreement with observations. Molecular dynamics \cite{MD} 
93: and Monte Carlo \cite{MC} calculations have more recently provided further 
94: evidence for the notion that dislocations drive the melting transition in 
95: three dimensions. There is also some experimental evidence that line defects 
96: are present in solids near melting \cite{Craw}. 
97: 
98: In refs.\ \cite{prev1,prev2} we formulated a dislocation theory of melting in 
99: which dislocations near melt were modeled as non-interacting strings on a 
100: lattice. The possible configurations of a dislocation were taken to be closed 
101: random walks. Screening of long-range strain fields by other dislocations in 
102: a dense ensemble %\cite{HL,Sar}, 
103: results in a $-\rho \ln \rho $ dependence of the free energy on the 
104: dislocation density, $\rho ,$ and thus a first-order transition. We obtained 
105: the following relation between the melting temperature $T_m,$ the shear 
106: modulus, $G,$ the Wigner-Seitz volume, $v_{WS},$ the coordination number, 
107: $z,$ and the critical density of dislocations, $\rho (T_m)$ (in units where 
108: $k_B=1):$ 
109: \beq 
110: T_m=\frac{\kappa \lambda Gv_{WS}}{8\pi \ln (z-1)}\;\!\ln \left( \frac{\alpha 
111: ^2}{4b^2\rho (T_m)}\right) . 
112: \eeq 
113: Here $b$ is the length of the shortest perfect-dislocation Burgers vector, 
114: $\kappa $ is 1 for a screw dislocation and $(1-\nu )^{-1}\approx 3/2$ for an 
115: edge dislocation $(\nu $ is the Poisson ratio), $\lambda \equiv b^3/v_{WS}$ 
116: and $\alpha ,$ which accounts for non-linear effects in the dislocation core, 
117: has a value of 2.9 \cite{prev2}. Experimental data on 51 elements show that 
118: \beq 
119: \frac{Gv_{WS}}{4\pi T_m\ln (z-1)}=1.01\pm 0.17 
120: \eeq 
121: at zero pressure \cite{prev1}. Eqs.\ (1) and (2) imply that the critical 
122: dislocation density at zero pressure is 
123: \beq
124: \rho (T_m)=(0.61\pm 0.20)\;\!b^{-2}. 
125: \eeq
126: This value is in good agreement with the critical density
127: \beq
128: \rho (T_m)=(0.66\pm 0.11)\;\!b^{-2}, 
129: \eeq 
130: obtained by applying our relation for the latent heat of fusion \cite{prev2}, 
131: \beq 
132: L_m=\frac{1}{\lambda}\;\!b^2\rho (T_m)R\;\!T_m\ln (z-1), 
133: \eeq 
134: to data on latent heats for 75 elements. Hence, $b^2\rho (T_m)$ is 
135: approximately constant across the Periodic Table with the numerical value 
136: \beq
137: b^2\rho (T_m)=0.64\pm 0.14, 
138: \eeq
139: which is the uncertainty-weighted average of Eqs.\ (3) and (4). 
140: 
141: In this paper we investigate the validity of our melting relation, Eq.\ (1), 
142: up to pressures of order 100 GPa, by comparing to experimental melting curves, 
143: i.e., melting temperatures versus pressure, $p.$ This comparison requires 
144: $v_{WS}(p),$ or its equivalent, the pressure dependence of the compression, 
145: $\eta \equiv V_0/V.$ We obtain $\eta (p)$ from the bulk modulus, $B(p),$ which 
146: is extrapolated to high pressure using only its value and first pressure 
147: derivative at ambient conditions, viz., room temperature and zero pressure. 
148: The shear modulus $G(p)$ is similarly extrapolated to high pressure. Pressure 
149: derivatives of $G$ and $B$ are typically $O(1),$ so the extrapolation of the 
150: bulk modulus is expected to break down at pressures of order the ambient bulk 
151: modulus. The parameter $\kappa $ in Eq.\ (1), which depends on the Poisson 
152: ratio, varies by only a few percent between $p=0$ and 100 GPa (we discuss this 
153: in more detail in Section 2). Since the accuracy of our melting relation at 
154: zero pressure is 17\%, we take $\kappa $ to be a constant. We also make the 
155: necessary but reasonable assumption that $b^2\rho (T_m)$ is also a 
156: pressure-independent constant. With this assumption, we find that our melting 
157: relation agrees well with experimental melting curves up to pressures $\approx 
158: B,$ and, in fact, our extrapolation of $T_m$ is often in good agreement with 
159: data to pressures $\approx 2B.$ In addition to the good agreement with the 
160: existing melting curve data, we also predict the high-pressure melting 
161: curves of Ag, Au, Cr, Cu, Mo, Os, Ru, W, and several actinides. 
162:  
163: \section{Melting curve equation}
164: 
165: We now consider the pressure dependences of the factors appearing in our 
166: melting relation, Eq.\ (1). The parameter $\lambda $ is constant by its 
167: definition, $\alpha $ is also assumed to be a constant, and $\kappa $ may 
168: be taken as constant provided that the Poisson ratio $\nu $ has a very 
169: weak pressure dependence. In fact, for an isotropic medium \cite{Gschn} 
170: \beq
171: \nu =\frac{1}{2}\;\!\frac{3B-2G}{3B+G}. 
172: \eeq
173: Although both $G$ and $B$ vary with pressure, the ratio in Eq.\ (7) varies 
174: only weakly. Consider, for example, Cu, for which $\nu \approx 0.34$ 
175: at $p=0.$ At $p=100$ GPa, we calculate the values of $G$ and $B$ with the 
176: help of Eqs.\ (13) and (14) below, with their pressure derivatives taken from 
177: ref.\ \cite{GS}, and find $\nu \approx 0.38.$ Therefore, in this case the 
178: corresponding values of $1/\kappa \approx 1-\nu /2$ \cite{prev2} 
179: are 0.83 and 0.81, respectively, so that the variation in the average value of 
180: $1/\kappa $ is $\approx 2$\%. Thus, the pressure dependence of $1/\kappa $ can 
181: be safely neglected. (There exists an upper bound on the change in the value 
182: of $1/\kappa $ with pressure. In the ultra-high-pressure limit, $p\propto 
183: \eta ^{5/3},$ in agreement with %the Thomas-Fermi theory \cite{Stein}. 
184: the theory of the free electron gas (Fermi gas). Therefore, $B\equiv -Vdp/dV=
185: \eta \;\!dp/d\eta \propto \eta ^{5/3}\gg G\propto \eta ^{4/3},$ and hence $\nu 
186: \rightarrow 1/2,$ in view of Eq.\ (7) (see also \cite{Kop}). Thus, in contrast 
187: to the Poisson ratio which changes by $\approx 50$\%, $1/\kappa \approx 1-\nu 
188: /2$ changes by $\approx 10$\%: $5/6\rightarrow 3/4.)$ 
189: 
190: We assume further that the mean interdislocation spacing at the melting point, 
191: $R\approx 1/\sqrt{\rho (T_m)},$ scales with $b,$ independent of pressure, and 
192: hence $b^2\rho (T_m)$ is a pressure-independent constant (with a numerical 
193: value of $0.64\pm 0.14,$ in view of Eq.\ (6)). It then follows from Eq.\ (1) 
194: that, provided the coordination number does not change with pressure 
195: (i.e., the element either remains in the same crystalline phase, or changes 
196: phase without changing the coordination number, e.g., a face-centered cubic 
197: structure $\leftrightarrow $ a hexagonal close-packed structure), the melting 
198: relation is given by 
199: \beq
200: \frac{G(p,T_m(p))v_{WS}(p,T_m(p))}{T_m(p)}={\rm const.} 
201: \eeq
202: 
203: The dependence of $v_{WS}$ on pressure and temperature is just the equation 
204: of state of the metal. Let us first focus on its temperature dependence. The 
205: fixed-pressure ratio of Wigner-Seitz volumes at $T_m$ and $T=0$ is equal to 
206: $1+\beta T_m,$ where $\beta $ is the volume expansivity. At $p=0,$ $\beta $
207: is typically of order $10^{-5}$ K$^{-1},$ and melting temperatures are at most 
208: about 4000 K, so $v_{WS}$ changes by only a few percent between $T=0$ and 
209: $T_m.$ Assuming that $\beta $ does not increase appreciably with compression, 
210: we can use room-temperature values for $v_{WS}.$ 
211: 
212: In contrast to $v_{WS},$ the dependence of $G$ on $T$ is not necessarily weak. 
213: Its $T$-dependence involves two characteristic temperatures, namely the Debye 
214: temperature, $T_D,$ and the melting temperature. $G$ is always monotonically 
215: decreasing with $T,$ and is nonlinear for $T\stackrel{<}{\sim }T_D$ and 
216: linear from $T_D$ to $T_m.$ However, there are no experimental data, no 
217: computer calculations, and no theoretical guidance that tells us how the 
218: temperature dependence of $G$ varies with pressure. In particular, how does 
219: the (negative) slope of the linear region vary with $p$? At this point we have 
220: no choice but to conjecture. We assume that $G(p,T_m(p))/G(p,0)$ is a slowly 
221: varying function of $p,$ so it can be considered constant up to moderate 
222: compressions, say, 20\% to 30\%. Thus, $G(p,T_m)$ is replaced by $G(p,0)$ 
223: in Eq.\ (8). In addition, data on the $p=0$ temperature dependence of shear 
224: moduli \cite{SW} clearly show that $G(p,300)\approx G(p,0),$ and therefore, 
225: we use the room temperature value of the shear modulus in our melting relation.
226: 
227: Subsequently, the explicit dependence of $G$ and $v_{WS}$ on $T$ will be 
228: dropped. It will be understood that $G$ and $v_{WS}$ are at room temperature. 
229: Our melting relation now reads 
230: \beq 
231: \frac{G(p)v_{WS}(p)}{T_m(p)}={\rm const.} 
232: \eeq
233: 
234: Differentiating Eq.\ (9) with respect to $p,$ one finds 
235: \beq 
236: \frac{1}{T_m}\;\!\frac{dT_m}{dp}=\frac{1}{G}\;\!\frac{dG}{dp}-\frac{1}{B}, 
237: \eeq
238: where we have used the definition of the bulk modulus, 
239: \beq 
240: B(p)\equiv -V\frac{dp}{dV}=-v_{WS}\frac{dp}{dv_{WS}}. 
241: \eeq
242: Thus, upon integration, Eq.\ (10) gives 
243: \beq
244: \frac{T_m(p)}{T_m(0)}=\frac{G(p)}{G(0)}\;\!\exp \;\!\left\{ -\int _0^p
245: \frac{dp'}{B(p')}\right\} . 
246: \eeq
247: To proceed further, we have to specify $G(p)$ and $B(p).$ 
248: 
249: \subsection{The shear modulus $G$ at finite pressure}
250: 
251: For the shear modulus at all pressures, we use the relation \cite{GS2} 
252: \beq
253: G=G_0+G'_0\;\!\frac{p}{\eta ^{1/3}}, 
254: \eeq
255: where $G_0'\equiv (dG/dp)_0.$ %and $\eta \equiv V_0/V.$ 
256: The subscript 0 refers to ambient conditions: $T\simeq 300$ K and $p=0.$ 
257: 
258: This equation satisfies the requirement that 
259: %the shear modulus should have an asymptotic compression dependence in 
260: %agreement with the Thomas-Fermi theory in which $p\propto \eta ^{5/3}$ and 
261: $G\propto \eta ^{4/3}$ as $\eta \rightarrow \infty ,$ since $p\propto \eta ^{
262: 5/3}.$ With the values of $G'_0$ for 32 elements tested in ref.\ \cite{GS2} 
263: Eq.\ (13) gives nearly the right value for the proportionality constant 
264: between $G$ and $\eta ^{4/3}$ at high compressions. Eq.\ (13) works 
265: well for a diverse selection of engineering metals covering many different 
266: crystal structures and nearly all groups of the Periodic Table \cite{GS2}. 
267: 
268: \subsection{Compression and the bulk modulus $B$ at finite pressure} 
269: 
270: Expanding the bulk modulus around $p=0$ we have 
271: \beq 
272: B(p)=B_0+B_0'p+\frac{1}{2}\;\!B_0''p^2+\ldots , 
273: \eeq 
274: where $B_0$ and $B_0'\equiv (dB/dp)_0,$ $B_0''\equiv (d^2B/dp^2)_0,\ldots $ 
275: can be extracted from equation of state data. Values of $B_0''$ are known 
276: for a few elements only (their determination is highly uncertain and involves 
277: an error of order 100\% \cite{FI}), and besides, $B_0''$ first appears in the 
278: $(p/B_0)^3$ term in the power series expansion of $\eta :$ 
279: $$\eta =\exp \;\!\left\{ \int _0^p\frac{dp'}{B(p')}\right\}=\left[ \frac{2B_0+
280: (B_0'+\sqrt{B_0'^2-2B_0B_0''})\;\!p}{2B_0+(B_0'-\sqrt{B_0'^2-2B_0B_0''})\;\!p}
281: \right] ^{1/\sqrt{B_0'^2-2B_0B_0''}}$$ 
282: \beq 
283: =1+\left( \frac{p}{B_0}\right) -\frac{B_0'-1}{2}\left( \frac{p}{B_0}\right) ^2
284: +\frac{(B_0'-1)(2B_0'-1)-B_0B_0''}{6}\left( \frac{p}{B_0}\right) ^3+\ldots \;. 
285: \eeq 
286: Since only $B_0$ and $B_0'$ are generally known (for almost all the elements, 
287: see ref.\ \cite{GS}), we restrict ourselves instead to the first two terms in 
288: Eq.\ (14). Then the compression simplifies to 
289: \beq
290: \eta =\left( 1+\frac{B_0'}{B_0}\;\!p\right) ^{1/B'_0}\!. 
291: \eeq 
292: Eqs.\ (15) and (16) are two different approximations to 
293: the Murnaghan equation of state \cite{Murn,HS}. 
294: 
295: It then follows from Eqs.\ (12), (13) and (16) that the equation of the 
296: melting curve is 
297: \beq
298: T_m(p)=T_m(0)\left( 1+\frac{B'_0}{B_0}\;\!p\right) ^{-1/B'_0}\left[ 1+\frac{
299: G'_0}{G_0}\;\!p\left( 1+\frac{B'_0}{B_0}\;\!p\right) 
300: ^{-1/3B'_0}\right] .
301: \eeq
302: As discussed in Section 4, this equation is only valid for pressures 
303: $p\stackrel{<}{\sim }2B.$ 
304: 
305: It follows from (17) that for $p\ll B_0$ 
306: \beq 
307: T_m(p)=T_m(0)\left[ 1+\left( \frac{B_0G_0'}{G_0}-1\right) \left( \frac{p}{B_0}
308: \right) -\left( \frac{4}{3}\;\!\frac{B_0G_0'}{G_0}-\frac{B_0'+1}{2}\right) 
309: \left( \frac{p}{B_0}\right) ^2+\ldots \right] . 
310: \eeq 
311: For the vast majority of the elements, $B_0'>5/3$ and $B'$ approaches 5/3 in 
312: the limit of large compressions. (In this limit $p\propto \eta ^{5/3},$ and 
313: therefore $B\equiv -Vdp/dV=\eta \;\!dp/d\eta =5p/3,$ i.e., $B'=5/3.)$ In fact, 
314: the average value of $B_0'$ for the 65 elements analyzed in \cite{GS}, except 
315: for Ce for which $B_0'<0,$ is $4.30\pm 1.40.$ Hence, if 
316: \beq 
317: \frac{G_0'}{G_0}>\frac{3}{8}\;\!\frac{B_0'+1}{B_0}, 
318: \eeq 
319: it follows from $B_0'>5/3$ that also $G_0'/G_0>1/B_0,$ i.e., Eq.\ (18) is of 
320: the form $T_m(p)=T_m(0)(1+ap-bp^2+\ldots ),$ $a,b>0,$ and describes melting 
321: curves for which melting temperatures increase with pressure \cite{Young}. 
322: If, however, 
323: \beq 
324: \frac{G_0'}{G_0}<\frac{1}{B_0} 
325: \eeq 
326: and $B_0'>5/3,$ then also  $G_0'/G_0<3/8\;\!(B_0'+1)/B_0,$ i.e., Eq.\ (18) is 
327: of the form $T_m(p)=T_m(0)(1-ap+bp^2-\ldots ),$ $a,b>0,$ and describes melting 
328: curves for which melting temperatures initially decrease with pressure 
329: \cite{Young}. For Si, for example, with the data from ref.\ \cite{GS} we find 
330: $G_0'/G_0<1/B_0$ and $B_0'=4.19,$ in agreement with the negative initial slope 
331: of the experimental melting curve. Eqs.\ (19) and (20) plus $B_0'>5/3$ 
332: should be considered our criteria for the two types of melting curves 
333: discussed above. 
334: 
335: \section{Melting curves: comparison with data} 
336: 
337: In this section we compare our melting curve, Eq.\ (17), to some experimental
338: melting curves, and predict a number of melting curves that can be compared 
339: with experiment in the not-so-distant future. 
340: 
341: We have found 5 elements for which melting curves have been measured to higher 
342: pressures, $p\sim O(100$ GPa): Al, Fe, Ni, Pb and U. We compare experimental 
343: data for these elements with our curves in Figs.\ 1-5. For Al, we also show 
344: the best fit to data in the form of the Simon equation, $T_m(p)=T_m(0)(1+ap)^
345: b$ \cite{Al}. For Fe, the experimental data are from ref.\ \cite{Fe}, and from 
346: ref.\ \cite{Ni} for Ni. For Pb, we combine the high-pressure data of ref.\ 
347: \cite{Pb} with the low-pressure data of ref.\ \cite{Pb2} as corrected in ref.\
348: \cite{Pb3}. For U, the high-pressure data of ref.\ \cite{U} are combined with 
349: the low-pressure data of ref.\ \cite{U2}. 
350: 
351: As claimed in ref.\ \cite{Al}, the Simon equation may not be the best 
352: functional form for a fit to data. In fact, the initial slope provided by this 
353: equation for the Al melting curve is 80 K/GPa, in contrast to 59 and 65 K/GPa 
354: from the two previous low-pressure measurements \cite{Al}. This accounts for 
355: the difference between the two curves in Fig.\ 1. 
356: %For comparison, our melting curve has an initial slope of 52 K/GPa. See, 
357: %however, our comment in the end of Section 3. 
358: 
359: In Fig.\ 4, in addition to Fe, we also plot melting curves for Ru and Os, 
360: elements in the same column of the Periodic Table. Those curves should be 
361: considered predictions for these metals. 
362: 
363: In Fig.\ 5, in addition to U for which there are high-pressure data, we also 
364: plot melting curves for the 5 actinides Am, Cm, Np, Pa and Th. For Np, we also 
365: show the low-pressure data of ref.\ \cite{Np}. We do not show the low-pressure 
366: data of ref.\ \cite{Am} for Am since they would overlay the low-pressure U 
367: data. We have checked that our melting curve is in agreement with the 
368: low-pressure Am data. For Cm, the values of $B_0$ and $B_0'$ are taken from 
369: ref.\ \cite{Cm}, and the value of $G_0$ is that estimated in ref.\ 
370: \cite{prev1}. For Pa, the values of $B_0$ and $G_0$ come from ref.\ 
371: \cite{Gschn}. We estimate the values of $G_0'$ for Cm and Pa from Th and U, 
372: their neighbors in the same row in the Periodic Table. Our earlier $G_0'$ 
373: estimates for Am and Np lead to $\gamma =1.05$ and 1.09, respectively, in 
374: Eq.\ (24), which implies that such estimates are reliable. The values of 
375: $B_0'$ for Np and Pa are also estimated by interpolating between Am, Cm, Th 
376: and U. (We note that this estimation of $B_0'$ is justified by the pronounced 
377: periodic behavior of $B_0'$ in $Z$ \cite{Stein}.) The value of $B_0'$ for Am 
378: is taken from \cite{Am}. We emphasize that the predicted melting curves assume 
379: constancy of coordination number along them. In the case of Am, e.g., there is 
380: still disagreement over the correct sequence of phases and their transition 
381: pressures \cite{Young}, so this assumption may well be incorrect. For Th, 
382: however, it is claimed that there is a transition from a face-centered cubic 
383: structure to a body-centered tetragonal structure that changes coordination 
384: number \cite{Vohra}. This transition occurs in the pressure range of $70-100$ 
385: GPa \cite{Vohra}, and thus our predictions for the Th melting curve up to 
386: 75 GPa should be quite reliable. 
387: 
388: Although we can account for a decrease in melting temperature with pressure 
389: in our theoretical framework (Eqs. (17),(20)), we do not consider such cases 
390: here, among which there are Pu and Ce. It has been established  \cite{Th} 
391: that for Th, which is in the same column as Ce, $\triangle V>0,$ and 
392: therefore, in view of Eq.\ (23), its melting temperature increases with 
393: pressure. 
394: 
395: % 
396: %\hskip 1.9in    
397: % \epsfysize=3in
398: \begin{center}
399: \vspace{2cm} 
400: %\parbox{6in}{abs } 
401: \epsfig{file=CurvesAl.eps,width=15cm,angle=0}
402: \end{center}
403: Fig.\ 1. Melting curve for Al. The dashed line is the Simon-fit to the data 
404: of ref.\ \cite{Al}, which are not shown explicitly. The diamonds are the 
405: low-pressure data from ref.\ \cite{LW}. The triangle is the shock-melting 
406: point at 125 GPa from ref.\ \cite{Al3}. The boxes are the points at 25, 69 
407: and 137 GPa calculated in ref.\ \cite{Urlin} from shock-melting data. 
408: They are assigned 20\% error bars \cite{Urlin}. 
409:  \\
410: 
411: %\hskip 1.9in    
412: % \epsfysize=3in
413: \begin{center}
414: \vspace{-0.3cm} 
415: %\parbox{6in}{abs } 
416: \epsfig{file=CurvesNi.eps,width=15cm,angle=0}
417: \end{center} 
418: Fig.\ 2. Melting curve for Ni. The diamonds (with small error bars) are the 
419: data of ref.\ \cite{Ni}. The boxes are the points at 79 and 250 GPa calculated 
420: in ref.\ \cite{Urlin} from shock-melting data. The corresponding error bars 
421: are not quoted in ref.\ \cite{Urlin}. 
422:  \\
423: 
424: %\hskip 1.9in    
425: % \epsfysize=3in
426: \begin{center}
427: %\vspace{0.5cm} 
428: %\parbox{6in}{abs } 
429: \epsfig{file=CurvesPb_1.eps,width=15cm,angle=0}
430: \end{center}
431: Fig.\ 3. Melting curve for Pb. The diamonds are the low-pressure data of 
432: ref.\ \cite{Pb2} corrected as in ref.\ \cite{Pb3}. The triangles are the data 
433: from ref.\ \cite{Pb}, and the dashed line is a best fit \cite{Pb} to the data.
434: The boxes are the points at 12, 34 and 68 GPa calculated in ref.\ \cite{Urlin} 
435: from shock-melting data. The corresponding error bars are not quoted 
436: in ref.\ \cite{Urlin}. The star is the point at 118 GPa calculated in ref.\ 
437: \cite{God} from shock-melting data. 
438:  \\
439: 
440: %\hskip 1.9in    
441: % \epsfysize=3in
442: \begin{center}
443: \vspace{2cm} 
444: %\parbox{6in}{abs } 
445: \epsfig{file=CurvesFeRuOs.eps,width=15cm,angle=0}
446: \end{center}
447: Fig.\ 4. Melting curves for the elements of the iron group (Fe, Ru, Os). The 
448: diamonds are the data of ref.\ \cite{Fe}, and the dashed line is a best fit 
449: \cite{Fe} to the data. The boxes are the shock-melting points at 235 GPa and 
450: 300 GPa \cite{Fe2}. The triangle is the shock-melting point at 240 GPa 
451: \cite{Fe3}. The star is the shock-melting point at 243 GPa \cite{Fe4}. 
452:  \\
453: 
454: %\hskip 1.9in    
455: % \epsfysize=3in
456: \begin{center}
457: \vspace{0.5cm} 
458: %\parbox{6in}{abs } 
459: \epsfig{file=CurvesActinides.eps,width=15cm,angle=0}
460: \end{center}
461: Fig.\ 5. Melting curves for the actinides Am, Cm, Np, Pa, Th and U. The data 
462: for U are from ref.\ \cite{U}, and for Np they come from ref.\ \cite{Np}. 
463:  \\
464: 
465: In Fig.\ 6 we plot the low-pressure data of ref.\ \cite{W} for W, and our 
466: melting curves for W, Mo and Cr. The initial slope of our melting curve of 
467: Mo, 26 K/GPa, is consistent with that predicted in ref.\ \cite{Mo}: 
468: $(34\pm 6)$ K/GPa. The same melting curve gives $T_m\simeq 9650$ K at $p=390$ 
469: GPa, in good agreement with the shock-melting temperature $\sim 10000$ K at 
470: the same pressure, found in ref.\ \cite{Mo2}. 
471: 
472: %
473: %\hskip 1.9in    
474: % \epsfysize=3in
475: \begin{center}
476: \vspace{2cm} 
477: %\parbox{6in}{abs } 
478: \epsfig{file=CurvesCrMoW.eps,width=15cm,angle=0}
479: \end{center}
480: Fig.\ 6. Melting curves for the elements of the chromium group (Cr, Mo, W). 
481: The data for W are from ref.\ \cite{W}. 
482:  \\
483: 
484: In Fig.\ 7 we compare the low-pressure data of ref.\ \cite{CuAgAu} for Cu, Ag 
485: and Au with our corresponding melting curves. Although the initial slopes of 
486: these curves are somewhat less than those of the data (the corresponding 
487: values of $\gamma $ in Fig.\ 1 are $\simeq 0.8),$ they are in good agreement 
488: with the best extrapolation of data to higher pressures made in ref.\ 
489: \cite{CuAgAu}, and with the calculation of ref.\ \cite{Urlin} in the case of 
490: Cu. 
491: 
492: %
493: %\hskip 1.9in    
494: % \epsfysize=3in
495: \begin{center}
496: \vspace{2cm} 
497: %\parbox{6in}{abs } 
498: \epsfig{file=CurvesCuAgAuCombo.eps,width=15cm,angle=0}
499: \end{center}
500: Fig.\ 7. Melting curves for the elements of the copper group (Cu, Ag, 
501: Au). The diamonds, the stars, and the triangles are the low-pressure data of 
502: ref.\ \cite{CuAgAu} in the inset, and the best extrapolations of these data 
503: to 20 and 30 GPa in the main plot for Cu, Ag and Au, respectively. The boxes 
504: are the points at 45 and 128 GPa calculated in ref.\ \cite{Urlin} from 
505: shock-melting data for Cu. The corresponding error bars are not quoted 
506: \cite{Urlin}. The gray diamond is the shock-melting point for Cu 
507: at 37 GPa \cite{Cu2}.
508:  \\ 
509: 
510: In Fig.\ 8 we compare the low-pressure data on the noble gases to our 
511: corresponding melting curves. The unknown values of $G'_0$ for Ne, Ar, Kr 
512: and Xe are calculated with the help of Eq.\ (25) below using the measured 
513: values of $B_0,$ $G_0$ and $\gamma _0$ (the Gr\"{u}neisen constant) 
514: \cite{prev2}. The values of $B_0'$ for Ne, Ar, Kr and Xe are taken from 
515: \cite{B'}. In the case of Rn, for which $B_0,$ $B_0',$ $G_0$ and $G_0'$ have 
516: not been measured, we first calculate $G_0$ using the (approximate) relation 
517: $GV/T_m={\rm const}$ for the noble-gas group, where $V=v_{WS}N_A$ is the 
518: molar volume. This relation follows from Eq.\ (1) provided that $\kappa ,$ 
519: $\lambda ,$ $\alpha $ and $z$ do not vary within this group. The value of the 
520: constant is determined by using the corresponding Ne, Ar, Kr and Xe data in 
521: this relation. We then calculate $B_0$ using Eq.\ (7) with the value of the 
522: Poisson ratio for Rn determined by extrapolating from the corresponding values 
523: for Ne, Ar, Kr and Xe. Finally, we determine both $B_0'$ and $G_0'$ by again 
524: extrapolating the Ne, Ar, Kr and Xe data. 
525: 
526: %
527: %\hskip 1.9in    
528: % \epsfysize=3in
529: \begin{center}
530: \vspace{2cm} 
531: %\parbox{6in}{abs } 
532: \epsfig{file=CurvesNobleGases.eps,width=15cm,angle=0}
533: \end{center}
534: Fig.\ 8. Melting curves for the noble gases Ne, Ar, Kr, Xe and Rn. 
535: The diamonds are the data of ref.\ \cite{Ne}. The stars are the data of 
536: ref.\ \cite{Ar}. The boxes and triangles come from ref.\ \cite{KrXe}. 
537:  \\ 
538: 
539: Finally, in Fig.\ 9 we show the experimental data and our theoretical 
540: melting curve for Mg. 
541: 
542: %
543: %\hskip 1.9in    
544: % \epsfysize=3in
545: \begin{center}
546: \vspace{-0.8cm} 
547: %\parbox{6in}{abs } 
548: \epsfig{file=CurvesMg.eps,width=15cm,angle=0}
549: \end{center}
550: Fig.\ 9. Melting curve for Mg. The low-pressure data are from ref.\ 
551: \cite{Mg_}. The high-pressure data are the shock-melting points of ref.\ 
552: \cite{Mg^}.
553:  \\ 
554: 
555: The 24 melting curves considered above constitute convincing evidence for the 
556: validity of our formula for melting temperature as a function of pressure, 
557: Eq.\ (17).  
558: 
559: \section{The range of validity of the new melting curve equation} 
560: 
561: In deriving our melting curve, Eq.\ (17), we have used both Eq.\ (13) for 
562: the pressure dependence of the shear modulus and the Murnaghan equation of 
563: state, Eq.\ (16). Since Eq.\ (13) has the correct zero-pressure limit (its 
564: Taylor series expansion in $p$ at $p=0$ is $G=G_0+G_0'p-(G_0'/3B_0)p^2+\ldots 
565: )$ and is claimed to have the correct ultra-high-pressure limit \cite{GS2}, 
566: we assume that this equation is valid over the entire pressure range. In any 
567: event, we do not have data to either confirm or invalidate this assumption. It 
568: then follows that the range of validity of Eq.\ (17) depends crucially on the 
569: range of validity of the Murnaghan equation of state, Eq.\ (16). 
570:  
571: The Murnaghan equation of state was examined in ref.\ \cite{HS}, together 
572: with a number of different equations of state, by comparing with the 
573: theoretical results calculated by the augmented-plane-wave method and the 
574: quantum-mechanical model proposed by Kalitkin and Kuz'mina \cite{KK} from 
575: low to ultra-high pressures. It was shown that the Murnaghan equation 
576: %with the real low-pressure values for $B_0$ and $B_0'$ 
577: is in good agreement with the theoretical results up to $V/V_0\simeq 0.7,$ 
578: i.e., up to compressions $\simeq 1.4-1.5.$ Since for the vast majority of the 
579: elements $B_0'\approx 5$ \cite{GS}, 
580: %it is seen in Eq.\ (16) , with $B_0'=4-5,$ $\eta =
581: %1.43-1.50$ for $p=B.$ With $B_0'=6-8,$ $\eta =1.42-1.53$ for $p=2B.$ Thus, 
582: we conclude, on the basis of Eq.\ (16), that the Murnaghan equation, and 
583: consequently, our equation for melting curve, Eq.\ (17), is valid up to 
584: pressures $p\approx 2B_0.$ The melting curves for Al and Pb in Figs.\ 1 and 3, 
585: respectively, and for Ne and Ar in Fig.\ 8 show that in some cases Eq.\ (17) 
586: is good to pressures even greater than $2B_0.$ 
587: 
588: For reliable predictions of melting curves to much higher pressures, 
589: $p\stackrel{>}{\sim }1$ TPa, one has to use a better equation of state than 
590: Murnaghan's. Hama and Suito \cite{HS} claim that the Vinet equation of state 
591: \cite{Vinet} is consistent with first-principles theoretical calculations to 
592: compressions $\eta \sim 5.$ Reference \cite{CGH} also finds the Vinet equation 
593: of state to be most accurate among various suggested equations of state. 
594: In fact, we have calculated that the melting temperatures for Fe and Pb at 
595: pressures $p\sim 50\;\!B_0$ as given by Eq.\ (17) are about two times higher 
596: than those given by a relation that derives from Eqs.\ (12), (13), and the 
597: Vinet equation of state. 
598: 
599: Another possible source of disagreement between the new melting curve, Eq.\ 
600: (17), and data may be inaccurate values of elastic constants and their pressure
601: derivatives in some cases. The Murnaghan equation and its frequently used 
602: partner -- the Birch equation \cite{Birch} -- are derived from the 
603: second-order Taylor series expansion of the bulk modulus [as in Eq.\ (14)] or 
604: the elastic strain energy with respect to pressure or strain, respectively. 
605: Thus their validities are, in principle, restricted to a narrow range of 
606: compression. Extending this range would entail the inclusion of higher-order 
607: terms. This could explain why the values of $B_0,$ and especially those of 
608: $B_0'$ and $B_0'',$ obtained from experiments which cover different ranges of 
609: compression by using a fitting method, are usually different. In many cases 
610: these differences between different experiments are small and can be safely 
611: neglected. In some cases, however, they are large, and so their use for 
612: predicting physical observables, such as melting temperature, is dubious. 
613: For example, in the case of Ni, we have used the value $B_0'=6.20$ given 
614: in \cite{GS}. Reference \cite{Ger}, however, quotes $B_0'\simeq 30$ (!). 
615: Similarly, for Mo we have used $B_0'=4.4$ of ref.\ \cite{GS}, while ref.\ 
616: \cite{Ger} gives $B_0'\simeq 20.$ (We note that the use of the values $B_0'=
617: 6.20$ for Ni and 4.4 for Mo is justified in view of the recent compilations 
618: of experimental data on $B_0'$ \cite{RMR}.) Although the numerical value of 
619: $B_0'$ does not matter at low $p,$ since it first appears in the $(p/B_0)^2$ 
620: term, in view of Eq.\ (15), it would strongly affect the predicted melting 
621: curve at pressures $p\sim O(B_0).$ 
622:  
623: There are also inconsistencies in the values of $G_0$ quoted in the 
624: literature. For example, for Pb we use the value $G_0=8.6$ GPa from ref.\ 
625: \cite{GS}, whereas ref.\ \cite{Gschn} quotes $G=5.5$ GPa. (Our own calculation 
626: \cite{prev1}, based on the values of the elastic constants $c_{11},$ $c_{12}$ 
627: and $c_{44},$ shows that 8.6 GPa is preferred over 5.5 GPa.) Likewise the 
628: values of $G_0$ for K and Na from ref.\ \cite{GS} are 0.9 and 1.98 GPa, 
629: whereas ref.\ \cite{Gschn} quotes 1.3 and 3.5 GPa, respectively. 
630: 
631: %\section{Apparent equivalence of defect and mechanical approaches to melting 
632: %curve}
633: \section{Relation of dislocation-based melting relation to the Lindemann 
634: criterion} 
635: 
636: The well-known Lindemann melting rule is based on the assumption that all 
637: elemental solids melt when the atomic vibrational amplitude is a fixed 
638: pressure-independent fraction of the interatomic distance. As shown by 
639: Lindemann \cite{Lin}, this implies the invariance of the Lindemann number 
640: \beq
641: \theta _D\left( \frac{M}{T_m}\right) ^{1/2}V^{1/3}=L
642: \eeq 
643: along the melting curve. Here $\theta _D$ is the density-dependent Debye 
644: temperature, $V$ is the molar volume, and $M$ is the molar mass. It is found 
645: that $L\approx 150$ \cite{Gschn}. 
646: 
647: There are compelling reasons to suppose that our dislocation-based melting 
648: relation is somehow equivalent to the Lindemann criterion. First of all, Eq.\ 
649: (21) gives melting curves that are typically very close to those predicted by 
650: our dislocation-based melting relation. (For example, the melting curve for Mg 
651: in Fig.\ 6.3 of ref.\ \cite{Young} is {\it very} similar to our curve in Fig.\ 
652: 9.) Furthermore, the Lindemann number, which is proportional to the ratio of 
653: atomic vibrational amplitude to the lattice constant at melt, is analogous to 
654: $b^2\rho (T_m),$ since both are presumed constant along the melting curve. 
655: In the dislocation-based approach, melting is associated with a critical 
656: configuration of dislocations, and for any such configuration there is a 
657: corresponding mean displacement of atoms from their equilibrium positions. 
658: Hence $L$ and $b^2\rho (T_m)$ are clearly related, and therefore the 
659: left-hand sides of Eqs.\ (9) and (21) are related as well. 
660: 
661: The mathematical equivalence of our melting relation and the Lindemann 
662: criterion would be established if it  could be determined that the left-hand 
663: side of Eq.\ (21) is a fixed fraction of the left-hand side of Eq.\ (9). A 
664: search of the literature has turned up two results which show that $L^2$ is 
665: {\it approximately} proportional to $Gv_{WS}/T_m.$ For a Debye solid the 
666: relation is \cite{Klein} $L^2=f(\nu (p,T))Gv_{WS}/T_m,$ where $f$ is a 
667: complicated function of $\nu .$ Thus the two melting relations are not 
668: rigorously equivalent. 
669: 
670: A second connection between the two melting formulas is provided by the 
671: following approximation for the Gr\"{u}neisen constant \cite{GS}, 
672: \beq 
673: \gamma (p)=\frac{2}{3}\;\!\gamma _S(p)+\frac{1}{3}\;\!\gamma _L(p), 
674: \eeq 
675: where
676: \beq 
677: \gamma _S(p)=\frac{G'(p)}{2}\;\!\frac{B^T(p)}{G(p)}-\frac{1}{6}, 
678: \eeq 
679: \beq 
680: \gamma _L(p)=\frac{1}{2}\;\!\frac{B^T(p)}{B^S(p)+\frac{4}{3}G(p)}\;\!\frac{
681: d(B^S(p)+\frac{4}{3}G(p))}{dp}-\frac{1}{6}, 
682: \eeq 
683: are the contributions of the shear (transverse) and longitudinal acoustic 
684: modes. Here $B^T$ is the isothermal bulk modulus, which is equivalent to 
685: $B$ that we are using in this paper. Eqs.\ (22)-(24) follow from the two 
686: assumptions that (i) the only appreciable contribution to the heat capacity 
687: of a crystal arises from lattice vibrations, and (ii) averaging over all modes 
688: is equivalent to averaging only over the low-frequency acoustic modes. (I.e., 
689: the contribution of the optical modes is equal to that of the acoustic modes.)
690: If in addition it is assumed that $B^S(p),$ the isentropic bulk modulus, is 
691: proportional to $G(p),$ then 
692: \beq 
693: \gamma _S(p)=\gamma _L(p)=\gamma (p)=\frac{G'(p)}{2}\;\!\frac{B(p)}{G(p)}-
694: \frac{1}{6}. 
695: \eeq 
696: However, there is no basis for this assumption, i.e., $\gamma _S(p)\neq \gamma 
697: _L(p)$ is to be expected. For example, in the ultra-high pressure limit, 
698: $B^S(p)\sim B^T(p)=5p/3$ and $G(p)\sim p^{4/5},$ quite different dependencies. 
699: 
700: Integration of Eq.\ (25), using $B(p)=-dp/d\ln V(p)$ and $\gamma (p)=-d\ln 
701: \theta _D(p)/d\ln V(p),$ gives 
702: \beq 
703: \frac{\theta _D^2(p)V^{2/3}(p)/T_m(p)}{G(p)V(p)/T_m(p)}={\rm const,} 
704: \eeq
705: that is, $L^2\propto Gv_{WS}/T_m.$ We emphasize that this proportionality 
706: is founded on a number of uncontrolled approximations. 
707:  
708: Equation (25), which would ensure a rigorous mathematical equivalence of 
709: the defect and mechanical (Lindemann's) approaches to melting, does {\it not} 
710: follow from first principles. %even under certain assumptions. 
711: This means that the defect and mechanical approaches to melting are basically 
712: {\it different.} Moreover, since the mechanical approach does not have a solid 
713: thermodynamic basis, it cannot, for example, predict the latent heat of 
714: fusion. In contrast, the defect approach 
715: %establishes the melting rule, in terms of Eqs.\ (1),(6), which can be 
716: %translated into the Lindemann rule, in view of Eq.\ (24), and, in addition, 
717: predicts the latent heat of fusion, Eq.\ (9), which is in good 
718: agreement with data for three-quarters of the Periodic Table \cite{prev2}. 
719: 
720: Finally, we wish to make the following comments on Eq.\ (25). We did not check 
721: extensively its validity at zero pressure, 
722: %i.e., the validity of the relation $\gamma _0=G'_0/2\;B_0/G_0-1/6,$ 
723: since that would go beyond the scope of this paper. We do, however, 
724: have some evidence that Eq.\ (25) is rather well satisfied: 
725: with the data from ref.\ \cite{GS}, we calculate from the above relation 
726: $\gamma _0=2.25$ vs.\ measured 2.40 for Ag, 3.08 vs.\ 2.99 for Au, 1.66 vs.\ 
727: 1.78 for Fe, 1.28 vs.\ 1.29 for K, and 1.18 vs.\ 1.19 for Na. We have actually 
728: used Eq.\ (25) in Section 3 to calculate $G_0'$ for noble gases in order to 
729: get their melting curves via Eq.\ (17) and to compare with experiment. Good 
730: agreement between the calculated and experimental curves is another hint 
731: on the approximate validity of this formula. Also, Eq.\ (25) has the correct 
732: ultra-high-pressure limit in which $\gamma \rightarrow 1/2$ \cite{Kop}, 
733: since in this limit $G\sim p^{4/5}$ and $B=5p/3.$ 
734: 
735: \section{Concluding remarks} 
736: 
737: We have extended the framework of melting as a string-mediated phase 
738: transition to non-zero pressure and derived a new equation for the 
739: melting curve, Eq.\ (17). As discussed above, with accurate experimental 
740: values of all the parameters involved, this equation reproduces the 
741: existing experimental melting data, and predicts unknown melting curves to 
742: pressures $p\stackrel{<}{\sim }2B_0.$ For higher pressures, a better equation 
743: of state than Murnaghan's should be used, e.g., the Vinet equation of state. 
744: 
745: We have addressed the apparent equivalence of defect and mechanical 
746: approaches to melting curve, and demonstrated that both approaches are 
747: basically different. We have shown that their would-be rigorous mathematical 
748: equivalence must manifest itself in a new relation, Eq.\ (25), 
749: %the validity of which has not been tested here. 
750: which we have not tested in detail. 
751: 
752: To summarize, we have calculated melting curves for 24 elements: Al, Mg, Ni, 
753: Pb, the iron group (Fe, Ru, Os), the chromium group (Cr, Mo, W), the copper 
754: group (Cu, Ag, Au), the noble gases Ne, Ar, Kr, Xe and Rn, and the six 
755: actinides Am, Cm, Np, Pa, Th and U. These calculated melting curves are 
756: in good agreement with existing data. 
757:  
758: \section*{Acknowledgements} 
759: 
760: We wish to thank T. Goldman and A.Z. Patashinski for valuable discussions 
761: during the preparation of this work. One of us (L.B.) wishes to thank 
762: B.K. Godwal for useful correspondence. 
763: 
764: \bigskip
765: \bigskip
766: \begin{thebibliography}{9} 
767: \bibitem{Mott} C. Mott, Proc. Roy. Soc. A {\bf 215} (1952) 1 
768: \bibitem{MO} S. Mizushima, J. Phys. Soc. Japan {\bf 15} (1960) 70 \\ 
769: A. Ookawa, J. Phys. Soc. Japan {\bf 15} (1960) 2191 
770: %\bibitem{Poirier} J.P. Poirier, Geophys. J. Roy. Astr. Soc. {\bf 85} (1986) 
771: %315 
772: \bibitem{MD} R.M.J. Cotterill, Phys. Rev. Lett. {\bf 42} (1979) 1541 
773: \bibitem{MC} W. Janke, Int. J. Theor. Phys. {\bf 29} (1990) 1251 \\ 
774: L. G\'{o}mez, A. Dobry and H.T. Diep, Phys. Rev. B {\bf 55} (1997) 6265 
775: \bibitem{Craw} R.K. Crawford, Bul. Am. Phys. Soc. {\bf 24} (1979) 385 \\ 
776: R.M.J. Cotterill and J.K. Kristensen, Phil. Mag. {\bf 36} (1977) 453 \\ 
777: Y. Wang and K. Kakimoto, J. Crystal Growth {\bf 208} (2000) 303 
778: \bibitem{prev1} L. Burakovsky and D.L. Preston, Analysis of dislocation 
779: mechanism for melting of elements, Los Alamos preprint LA-UR-99-4171 
780: [cond-mat/0003494], Solid State Comm., {\it in press}
781: \bibitem{prev2} L. Burakovsky, D.L. Preston and R.R. Silbar, Melting as a 
782: string-mediated phase transition, Los Alamos preprint LA-UR-99-5914 
783: [cond-mat/0004011], Phys. Rev. B, {\it in press}
784: %\bibitem{HL} J.P. Hirth and J. Lothe, {\it Theory of Dislocations,} 2nd ed., 
785: %(Krieger Publishing, Malabar, FL, 1992) 
786: %\bibitem{Sar} G.F. Sarafanov, Phys. Solid State {\bf 39} (1997) 1403 
787: \bibitem{Gschn} K.A. Gschneidner, Jr., in {\it Solid State Physics, Advances 
788: in Research and Applications,} Eds. F. Seitz and D. Turnbull, (Academic Press,
789: New York, 1965), Vol.\ 16, p.\ 275
790: \bibitem{GS} M.W. Guinan and D.J. Steinberg, J. Phys. Chem. Solids {\bf 35}
791: (1974) 1501
792: \bibitem{Kop} V.P. Kopyshev, Sov. Phys. Doklady {\bf 10} (1965) 338 
793: \bibitem{SW} G. Simmons and H. Wang, {\it Single Crystal Elastic Constants and
794: Calculated Aggregate Properties,} 2nd ed., (MIT Press, Cambridge, MA, 1971) 
795: \bibitem{GS2} M. Guinan and D. Steinberg, J. Phys. Chem. Solids {\bf 36} (1975)
796: 829 
797: \bibitem{FI} J. Freund and R. Inglass, J. Phys. Chem. Solids {\bf 50} (1989) 
798: 263 
799: \bibitem{Murn} F.D. Murnaghan, Proc. Nat. Acad. Sci. US {\bf 30} (1944) 244 
800: \bibitem{HS} J. Hama and K. Suito, J. Phys.: Cond. Mat. {\bf 8} (1996) 67 
801: %\bibitem{KKD} P. Kuchhal, R. Kumar and N. Dass, J. Phys.: Cond. Mat. {\bf 9} 
802: %(1997) 2987 
803: \bibitem{Young} D.A. Young, {\it Phase Diagrams of the Elements,} (University 
804: of California Press, Berkeley, 1991) 
805: \bibitem{Am} D.R. Stephens, H.D. Stromberg and E.M. Lilley, J. Phys. Chem. 
806: Solids {\bf 29} (1968) 815 
807: \bibitem{Np} D.R. Stephens, J. Phys. Chem. Solids {\bf 27} (1966) 1201 
808: \bibitem{Tonkov} E.Yu. Tonkov, {\it High Pressure Phase Transformations,} 
809: (Gordon and Breach, Philadelphia, 1992) 
810: \bibitem{Ubb} A.R. Ubbelohde, {\it The Molten State of Matter,} (Wiley, New 
811: York, 1978) 
812: \bibitem{Ir} G.R. Gathers, J.W. Shaner, R.S. Hixson and D.A. Young, High 
813: Temp.-High Press. {\bf 11} (1979) 653 
814: \bibitem{Mo} J.W. Shaner, G.R. Gathers and C. Minichino, High Temp.-High Press.
815: {\bf 9} (1977) 331 
816: \bibitem{Nb} W.M. Hodgson, Ph.D. thesis (Lawrence Livermore preprint UCRL-52493
817: (1978) 
818: \bibitem{Th} M. Boivineau, H. Colin, J.M. Vermeulen and Th. Th\'{e}venin, Int. 
819: J. Thermophys. {\bf 17} (1996) 1001  
820: \bibitem{W} A. Kloss, H. Hess, H. Schneidenbach and R. Grossjohann, Int. J. 
821: Thermophys. {\bf 20} (1999) 1199 
822: \bibitem{MYR} J.A. Moriarty, D.A. Young and M. Ross, Phys. Rev. B {\bf 30} 
823: (1984) 578 
824: \bibitem{Al} R. Boehler and M. Ross, Earth Plan. Sci. Lett. {\bf 153} 
825: (1997) 223 
826: \bibitem{Fe} Q. Williams, R. Jeanloz, J. Bass, B. Svedensen and T.J. Ahrens, 
827: Science {\bf 236} (1987) 181 
828: \bibitem{Ni} P. Lazor, G. Shen and S.K. Saxena, Phys. Chem. Min. {\bf 20} 
829: (1993) 86 
830: \bibitem{Pb} B.K. Godwal, C. Meade, R. Jeanloz, A. Garcia, A.Y. Liu and 
831: L. Cohen, Science {\bf 248} (1990) 462 
832: \bibitem{Pb2} P.W. Mirwald and G.C. Kennedy, J. Phys. Chem. Solids {\bf 37} 
833: (1976) 795 
834: \bibitem{Pb3} M. Akaishi, H. Kanda, N. Setaka and O. Fukunaga, Japan J. Appl. 
835: Phys. {\bf 16} (1977) 1077 
836: \bibitem{U} C.S. Yoo, J. Akella and J.A. Moriarty, Phys. Rev. B {\bf 48} 
837: (1993) 15529 
838: \bibitem{U2} J. Ganguly and G.C. Kennedy, J. Phys. Chem. Solids {\bf 34} 
839: (1973) 2272 
840: \bibitem{Cm} R.G. Haire, U. Benedict, J.R. Peterson, C. Dufour and 
841: J.P. Iti\'{e}, J. Less-Common Met. {\bf 109} (1985) 71 
842: \bibitem{Stein} D.J. Steinberg, J. Phys. Chem. Solids {\bf 43} (1982) 1173 
843: \bibitem{Vohra} Y.K. Vohra and J. Akella, Phys. Rev. Lett. {\bf 67} (1991) 
844: 3563 \\ K. Ghandehari and Y.K. Vohra, Scr. Metal. Mater. {\bf 27} (1992) 195 
845: \bibitem{LW} J. Lees and B.H.J. Williamson, Nature {\bf 208} (1965) 278 
846: \bibitem{Al3} J.M. Shaner, J.M. Brown and R.G. McQueen, in {\it High Pressure 
847: in Science and Technology,} Eds. C. Homan, R.K. MacCrone and E. Whalley, 
848: (North Holland, Amsterdam, 1984), p.\ 137 
849: \bibitem{Urlin} V.D. Urlin, Zh.E.T.F. {\bf 49} (1965) 485 
850: \bibitem{God} B.K. Godwal, S.K. Sikka and R. Chidambaram, Pramana 
851: {\bf 29} (1987) 93. The typo in the value of $T_m(118$ GPa) found 
852: by us was communicated with B.K. Godwal. 
853: \bibitem{Fe2} C.S. Yoo, N.C. Holmes, M. Ross, D.J. Webb and C. Pike, Phys. 
854: Rev. Lett. {\bf 70} (1993) 3931 
855: \bibitem{Fe3} J.M. Brown and R.G. McQueen, J. Geophys. Res. {\bf 91} (1986) 
856: 7485 
857: \bibitem{Fe4} J.D. Bass, in {\it High Pressure Research in Mineral Physics,} 
858: Eds. M.H. Manghnani and Y. Syono, (Terra Scientific, Tokyo, 1987), p.\ 393 
859: \bibitem{Mo2} R.S. Hixson, D.A. Boness, J.W. Shaner and J.A. Moriarty, Phys. 
860: Rev. Lett. {\bf 62} (1989) 637 
861: \bibitem{CuAgAu} P.W. Mirwald and G.C. Kennedy, J. Geophys. Res. {\bf 84}
862: (1979) 6750 
863: \bibitem{Cu2} A.I. Funtikov, Fiz. Goreniya Vzryva {\bf 5} (1969) 510 
864: \bibitem{B'} {\it Rare Gas Solids,} Eds. M.L. Klein and J.A. Venables, 
865: (Academic Press, London, 1977), Vol.\ II, Chapter 13 
866: \bibitem{Ne} W.L. Vos, J.A. Schouten, D.A. Young and M. Ross, J. Chem. Phys. 
867: {\bf 94} (1991) 3835 
868: \bibitem{Ar} C.-S. Zha, R. Boehler, D.A. Young and M. Ross, J. Chem. Phys. 
869: {\bf 85} (1986) 1034 
870: \bibitem{KrXe} P.H. Lahr and W.G. Eversole, J. Chem. Eng. Data {\bf 7} (1962) 
871: 42 
872: \bibitem{Mg_} G.C. Kennedy and R.C. Newton, in {\it Solids under Pressure,} 
873: Eds. W. Paul and D.M. Warschauer, (McGraw-Hill, New York, 1963), p.\ 163 
874: \bibitem{Mg^} P.A. Urtiew and R. Grover, J. Appl. Phys. {\bf 48} (1977) 1122 
875: \bibitem{KK} N.N. Kalitkin and L.V. Kuz'mina, Sov. Phys. Solid State {\bf 13} 
876: (1972) 1938 
877: \bibitem{Vinet} P. Vinet, J. Ferrante, J.R. Smith and J.H. Rose, J. Phys. C 
878: {\bf 19} (1986) L467 \\  
879: P. Vinet, J. Ferrante and J.H. Rose, Phys. Rev. B {\bf 35} (1987) 1945 
880: \bibitem{CGH} R.E. Cohen, O. G\"{u}lseren and R.J. Hemley, Amer. Mineral. 
881: {\bf 85} (2000) 338 
882: \bibitem{Birch} F. Birch, J. Geophys. Res. {\bf 57} (1952) 227, {\bf 83} 
883: (1978) 1257 
884: \bibitem{Ger} L. Gerward, J. Phys. Chem. Solids {\bf 46} (1985) 925 
885: \bibitem{RMR} Z.-H. Fang and L.-R. Chen, Phys. Stat. Sol. B {\bf 180} (1993) K5
886:  \\ S. Raju, E. Mohandas and V.S. Raghunathan, J. Phys. Chem. Solids {\bf 58} 
887: (1997) 1367 
888: \bibitem{Lin} F.A. Lindemann, Phys. Z. {\bf 11} (1910) 609 
889: \bibitem{Klein} H. Kleinert, {\it Gauge Fields in Condensed Matter,} 
890: (World Scientific, Singapore, 1989), Vol.\ II 
891: \end{thebibliography}
892: \end{document}
893: \end
894: