1: \documentclass[preprint,aps,showpacs]{revtex4}
2: %\documentclass[aps,prl,preprint,groupedaddress]{revtex4}
3: %\documentclass[pre,twocolumn,showpacs,preprintnumbers]{revtex4}
4:
5:
6: %\newcommand{\mwcd}[1]{\marginpar{\protect\scriptsize \protect\setlength
7: % {\baselineskip}{8pt} to editor- MS revision: #1}}
8: %\setlength {\marginparwidth}{7.0cm}
9:
10: \usepackage{graphicx}% Include figure files
11:
12: \begin{document}
13:
14: \title
15: { Pair-distribution functions
16: of two-temperature two-mass systems:
17: Comparison of MD, HNC, CHNC, QMC and
18: Kohn-Sham calculations for dense hydrogen
19: }
20:
21: %
22: \author
23: {
24: M.W.C. Dharma-wardana}
25: %
26: \affiliation{
27: National Research Council of Canada, Ottawa, Canada, K1A 0R6
28: }
29: \email[Email address:\ ]{chandre.dharma-wardana@nrc-cnrc.gc.ca}
30:
31: \author
32: {Michael S. Murillo}
33: %
34: \affiliation{Physics Division, MS D410, Los Alamos National Laboratory, Los Alamos, New Mexico, 87545}
35: \email[Email address:\ ]{murillo@lanl.gov}
36: %
37: %\date{13-feb-2002}
38: \date{\today}
39: %
40:
41: \begin{abstract}
42: Two-temperature, two-mass quasi-equilibrium plasmas may occur in electron-ion plasmas,
43: nuclear-matter, as well as in electron-hole condensed-matter systems.
44: Dense two-temperature hydrogen plasmas straddle the difficult partially-degenerate
45: regime
46: of electron densities and temperatures which are important in astrophysics, in
47: inertial-confinement
48: fusion research, and other areas of warm dense matter physics. Results from
49: Kohn-Sham calculations and QMC are used to benchmark the procedures used in
50: classical molecular-dynamics simulations, HNC and CHNC methods to
51: derive electron-electron and electron-proton pair-distribution functions.
52: Then, nonequilibrium
53: molecular dynamics for two-temperature, two-mass plasmas are used to
54: obtain the pair distribution.
55: Using these results, the correct HNC and
56: CHNC procedures for the evaluation of pair-distribution functions in
57: two-temperature two-mass two-component
58: charged fluids are established. Results for a mass ratio of 1:5, typical of
59: electron-hole fluids, as well as for compressed hydrogen are presented.
60: \end{abstract}
61: \pacs{PACS Numbers: 52.25.Kn, 52.25Gj, 71.10.-w, 52.27.Gr, 26.30.+k}
62: %\vspace{0.5in}
63: %
64: \maketitle
65: %
66: \section{Introduction}
67: The study of hot strongly-coupled dense charged
68: fluids is a difficult task, especially near the
69: %Kris T. Delaney, Carlo Pierleoni, D. M. Ceperley. Phys. Rev. Lett. 97, 235702 (2006).
70: %Carlo Pierleoni, David M. Ceperley, Markus Holzmann. Phys. Rev. Letts. 95, 146402 (2004)
71: regime of or molecular and atomic species\cite{ceperleyH}, or excitons in electron-hole plasmas.
72: The system is better understood
73: for fully ionized systems, such as hydrogen, in the form of free electrons
74: and protons, and fully-ionized electron-hole condensates.
75: In fact, considerable headway
76: has been made using methods based on density-functional theory (DFT), even
77: for plasmas with multiple states of ionization.
78: DFT methods have been used with molecular-dynamics based
79: approaches\cite{kwon,dejarlais,mazevet}, as well as within multi-component
80: integral-equation approaches \cite{pdw95,ilciacco}. Equilibrium properties
81: of plasmas, as well as their linear transport properties, have been
82: successfully studied in these papers, and excellent agreement
83: between the molecular-dynamics based DFT and integral-equation
84: based DFT has been found \cite{lvm}.
85:
86: On the other hand, laser-produced plasmas are initially formed as
87: two-temperature plasmas, where the electrons have absorbed the laser
88: energy and have self-equilibrated to some ``electron temperature''
89: $T_e$, while the ions remain cool, at some temperature $T_i$, with
90: $T_i<T_e$. The opposite
91: situation arises in shock-wave generated plasmas, where the ions absorb the
92: shock energy and $T_i>T_e$. Such two-temperature plasmas also occur in
93: astrophysical settings, affecting the time of termination
94: of synthesis of light-nuclei
95: to occur at different stages of cooling of the electrons\cite{astrop}, and
96: influencing the Coulomb nuclear-tunneling rates\cite{dewitt}.
97: The possibility of such well-defined
98: two-temperature plasmas is largely a result of the extreme mass ratio
99: $m_i/m_e\ge 1836$ between ions and electrons. Similar, but less well
100: defined situations can arise
101: in electron-hole plasmas, where the masses are of the same order of
102: magnitude (e.g, the electron and hole masses in GaAs are 0.067$m_e$ and
103: 0.34$m_e$ respectively, with an electron-hole mass ratio of $\sim 5$).
104: GaAs is a direct bandgap material, and electron-hole plasmas are more
105: easily studied in indirect-gap systems like Si where the density-of states
106: mass ratio is $\sim 3$.
107: The simulation of such systems at two temperatures, using quantum Monte-Carlo
108: methods is at present unavailable, even in regimes of densities
109: and temperatures where bound states (or exciton formation in electron-hole
110: systems) do not exist. Thus it is natural to look for analytical methods based on
111: integral-equation techniques which are computationally simple
112: and physically insightful. However, although $T_e,\, T_i$ define
113: the temperatures of each subsystem and the pair-distribution
114: functions (PDFs) $g_{ee}$ and $g_{ii}$, the `temperature' $T_{ei}$
115: entering into
116: the cross-correlations $g_{ei}$, as well as the the effects
117: of electron spin, exchange
118: etc., relevant to two-temperature systems need to be clarified. In this context
119: we use $T_{ee}=T_e$, $T_{ii}=T_i$, and $T_{ei}$ to refer to the
120: electron-, ion-, and electron-ion temperatures as they enter independently
121: into the Ornstein-Zernike (OZ) and hypernetted chain (HNC) equations. In fact,
122: some authors\cite{seuf} have proposed to modify
123: the well-established
124: OZ equations in dealing with
125: two-temperature (2T) two-mass (2M) systems.
126:
127: The objective of this paper is to study such 2T-2M plasmas
128: using results from molecular-dynamics (MD)\cite{mc}, HNC\cite{hncref},
129: classical-map HNC (CHNC)\cite{prl1,prb3d}, quantum Monte Carlo (QMC)\cite{mc}
130: and Kohn-Sham (KS)\cite{ilciacco} methods to establish the
131: proper implementation of quantum effects and 2T, 2M situations in
132: simulation studies.
133: One of our main interests would be uniform hydrogenic plasmas free of
134: bound states, in the regime of warm-dense matter.
135: %
136: \section{Theoretical Methods}
137: A system of classical particles, e.g., hard-spheres or Lennard-Jones fluids,
138: or classical
139: ions in a uniform neutralizing background, can
140: be studied completely using the method of molecular dynamics (MD) where
141: the classical equations
142: of motion are integrated numerically, for a sufficiently large number of
143: particles contained
144: in a simulation box. It has been found that the particle distribution functions,
145: e.g., the pair-distribution function (PDF) $g_{ij}(r)$ (where $i,j$
146: specify the species, spin etc.),
147: obtained from MD simulations for charged classical ions
148: can also be accurately reproduced via suitable integral-equations which are
149: computationally very economical and efficient.
150: The pair-potentials, quantum corrections etc.,
151: needed to simulate systems with ions and electrons, or purely electron systems
152: (with a uniform neutralizing background) will be discussed in this section. First, we compare
153: the usual HNC approximation with the CHNC method. Then, these results are compared with a
154: full quantum (Kohn-Sham) calculation; these results are then used to determine the effective
155: diffractive interaction used in the CHNC method. Finally, we discuss how the CHNC can be extended to
156: ``classical map molecular dynamics'' (CMMD).
157: %
158: \subsection{HNC and CHNC methods}
159: The HNC equation and
160: its straight-forward generalizations, coupled with the OZ equation
161: have lead
162: to very accurate results for classical charged-particle interactions.
163: The exact equations for the PDFs are of the form:
164: \begin{equation}
165: g_{ij}(r)=e^{-\beta_{cf} \phi_{ij}(r)
166: +h_{ij}(r)-c_{ij}(r) + B_{ij}(r)}
167: \label{hnc}
168: \end{equation}
169: Here $\phi_{ij}(r)$ is the pair potential between the
170: species $i,j$.
171: If the bridge function $B_{ij}(r)$ is set to zero we have the HNC approximation.
172: Then, given the temperature $T=1/\beta$, the particle densities $n_i$, and
173: the pair-potentials $\phi_{ij}(r)$, the pair-correlation function $h_{ij}(r)$ and
174: the direct correlation function $c_{ij}(r)$ can be self-consistently obtained via the
175: HNC and OZ equations, which have the form
176: \begin{equation}
177: \label{oz1}
178: h_{ij}(r) = c_{ij}(r)+
179: \sum_ s n_s\int d{\bf r}'h_{is}
180: (|{\bf r}-{\bf r}'|)c_{sj}({\bf r}').
181: \end{equation}
182: This method already fails for strictly attractive potentials. Thus,
183: in simulations of electron-proton systems, $\phi_{ep}(r)$ has to
184: be replaced by effective-potentials which attempt to incorporate
185: quantum diffraction effects \cite{Jones}.
186: Even with purely repulsive potentials, classical simulations fail to
187: incorporate Fermi or Bose statistics which begin to
188: manifest as the density is increased and
189: the temperature is lowered. A well established means of
190: incorporating quantum effects is to derive
191: integral equations from correlated-determinantal
192: wavefunctions, as done in the Feenberg
193: approach\cite{feenberg,lantto}. The resulting integral equations
194: are very daunting, and
195: in fact, not easy to use. Quantum Monte-Carlo (QMC) itself may be considered
196: as an adaptation of the Feenberg-functional to generate a statistical measure
197: for the stochastic algorithms used in MD. An alternative approach, using
198: Feynman paths instead of classical trajectories, provides another class
199: of simulation techniques. However, these
200: quantum simulation methods become computationally extremely
201: heavy. Such methods are
202: best suited for the establishment of bench-mark results, and for the
203: ``calibration'' of
204: other methods which contain approximation schemes. In fact, the QMC
205: techniques have been most useful in providing the ``exchange-correlation''
206: potentials $V_{xc}(r)$ needed in the Kohn-Sham
207: density-functional theory (DFT) equations. While DFT is itself exact,
208: one has to use results from QMC and such microscopic methods to model
209: the unknown $V_{xc}(r)$. Given the $V_{xc}$, the inhomogeneous density distribution
210: around a given particle can be calculated, and the pair-distribution is deduced from
211: it.
212:
213: The method followed here is to exploit the well established Kohn-Sham
214: equations as the reference calculation, and determine the effective
215: potentials to be used in the classical simulations of 2T and 2M systems.
216: To this end we present comparisons of Kohn-Sham calculations of $g_{ep}(r)$
217: for
218: H-plasmas with available QMC results to mutually validate these methods.
219: Another theoretical tool we use is the ``classical-map
220: HNC'', i.e., CHNC equations which incorporate quantum effects including
221: Fermion statistics via effective potentials and effective temperatures.
222: The CHNC has been extensively tested via comparisons
223: with QMC results in 2-D and 3-D electron systems, and shown to provide excellent
224: agreement, even at the extreme quantum limit of zero temperature\cite{prl1, prl2}.
225: CHNC uses
226: a ``quantum temperature'' $T_q$, which depends on the Fermion density. If the
227: physical temperature of the quantum fluid is $T$, the distribution functions
228: are obtained\cite{prl1} from a
229: classical fluid at the temperature $T_{cf}$ such that:
230: \begin{equation}
231: \label{tcf}
232: T_{cf}=\sqrt{T^2+T_q^2}.
233: \end{equation}
234: The temperature $T_q$ is defined to be such that the classical Coulomb fluid has the same
235: exchange-correlation
236: energy as the
237: quantum fluid [\onlinecite{prl1}]. DFT assures us that only
238: the true density
239: distribution possesses the true exchange-correlation energy.
240: Thus, the charge distributions, i.e., the
241: PDFs obtained from CHNC are found to be in excellent agreement with those
242: from Monte-Carlo simulations of 2D and 3D systems.
243: This agreement is obtained by including the exchange-hole of parallel-spin
244: electrons as an
245: effective potential, called in CHNC the Pauli exclusion potential $P_{ij}(r)$.
246: Clearly, this is zero if $i\neq j$. For $i=j=\parallel$ spins, the potential $P_{ij}(r)$ is such that the
247: non-interacting PDFs, i.e.,
248: $g_{ij}^0(r)$ are correctly recovered from the integral equations\cite{lado}.
249: Thus, using atomic units where $|e|=\hbar=m_e=1$, the effective
250: pair potentials $\phi_{ij}(r)$ are of the form:
251: \begin{eqnarray}
252: \label{pots}
253: \phi_{ij}(r)&=&P_{ij}(r)+ V^c_{ij}(r)\\
254: V^c_{ij}(r)&=& z_iz_j(1-e^{-k_{ij}r})/r
255: \end{eqnarray}
256: Here $z_e=-1,\,z_p=1$ and $k_{ij}$ is a cut-off 'momentum' defining a
257: diffraction correction allowing for
258: quantum effects. In the simplest formulation $k_{ij}$ is the thermal de
259: Broglie momentum given by:
260: \begin{eqnarray}
261: \label{kdifr}
262: k_{ij}&=&k^{dB}_{ij}=(2\pi\, m_{ij} T_{ij})^{1/2}\\
263: m_{ij}&=&m_im_j/(m_i+m_j)
264: \end{eqnarray}
265: The temperatures $T_{ij}$ entering into the HNC equations for the $g_{ij}(r)$
266: are given by:
267: \begin{equation}
268: \label{crosstemp}
269: T_{ij}/m_{ij}=T_i/m_i+T_j/m_j.
270: \end{equation}
271: The large mass of the proton ensures that the diffraction
272: correction, as well as the $T_q$, is negligible
273: for the proton-proton scattering process. A more complete approach to
274: determining $k_{ij}$ is to
275: solve the corresponding Kohn-Sham equation for the two particles in the
276: Kohn-Sham potential of the
277: medium, and matching the $k_{ij}$ so that the quantum and classical values
278: of the PDF at contact are in agreement.
279:
280: The Pauli exclusion potential is usually determined by inverting the
281: HNC equations applied to the exactly known
282: non-interacting quantum PDFs $g^0_{ii}(r)$ of the uniform electron fluid.
283: Here $i$ runs through $e\uparrow, e\downarrow$,
284: i.e., a spin-resolved, two-component electron system is used. If $i\ne j$, $P_{ij}=0$.
285: In the absence of strong magnetic fields,
286: the spin-resolution is not needed in warm dense systems.
287: Treating the electrons as a spin-averaged
288: one-component subsystem simplifies the
289: simulations.
290: Due to the non-linearity of the inversion of the HNC relations given
291: in Eqs.~\ref{paupot},
292: the Pauli exclusion potential
293: for paramagnetic electrons is not a simple average of the Pauli exclusion
294: potentials of the spin-resolved cases.
295: The corresponding Pauli potential
296: $P_e(r)$ has to be extracted directly from the averaged $g^0_{ee}(r)$.
297: Thus we have, for the spin-resolved and -unresolved cases:
298: \begin{eqnarray}
299: \label{paupot}
300: \beta P_{ii}(r)&=&-\ln[g_{ii}^0(r)]+N_{ii}^0(r), \\
301: \beta P_e(r) &=&-\ln[\{g_{ii}^0(r)+1\}/2]+N^0_{ii}/2 \\
302: N_{ii}^0(r) &=& h_{ii}^0(r)-c_{ii}^0(r).
303: \end{eqnarray}
304: A bridge term $B_{ij}(r)$ is used to correct the HNC for multi-particle
305: effects poorly rendered by HNC. Such bridge corrections are found to be
306: very significant in the 2D electron fluid\cite{prl2}, but not
307: for 3D electrons at densities and temperatures considered in this study.
308:
309: The CHNC differs from HNC in the use of the Pauli potentials and the
310: quantum temperature $T_q$ when treating quantum fluids. Also, the pair-potentials
311: used in HNC have been constructed to agree with KS-charge profiles (see below).
312: Hence any insights obtained for the two-temperature two-mass HNC can be
313: easily transfered to the CHNC. The two-temperature electron-ion
314: plasma was discussed in a formal analysis by Boercker and More\cite{bm}, using
315: a product form for the partition function. However, no
316: comparisons of their results with actual simulations are available.
317: The more general two-temperature two-mass HNC type equations
318: have been discussed, most recently by Seuferling et
319: al.\cite{seuf}. Using an analysis based on the
320: Bogoliubov-Born-Green-Kirkwood-Yvon (BBGKY) hierarchy as
321: well as some factorization assumptions, the authors of
322: Ref.~[\onlinecite{seuf}] have proposed modified OZ type
323: equations for 2T-2M plasmas. While their formulae reduce to the usual
324: OZ equations, viz., Eq.~\ref{oz1}, for the $m_a>>m_b,\; T_a=T_b$, the
325: case $m_a=m_b,\; T_a=T_b$ is not correctly recovered. The results
326: presented in our study imply that the usual OZ equations hold in all cases,
327: as long as the correct mass-dependent $T_{ij}$, Eq.~\ref{crosstemp}
328: is used in 2T-2M systems.
329: \subsection{The Kohn-Sham reference calculation}
330: Kohn-Sham theory at finite temperatures\cite{ilciacco} states that
331: the true one-particle density distribution of the system subject to an
332: external potential is such that the free
333: energy of the system is minimized. This theorem holds rigorously for
334: a system in equilibrium and we use it to derive distribution functions
335: by considering the inhomogeneous electron distribution
336: around a proton in the plasma. Let $n(r)$ and $\rho(r)$ be the electron
337: and proton charge densities around the proton at the origin. These tend
338: to the average densities $\bar{n}=\bar{\rho}$ far
339: away($r\to \infty$) from the proton at the center; then,
340: \begin{equation}
341: g_{ep}(r)=n(r)/\bar{n}.
342: \end{equation}
343: Instead of using a two-component DFT procedure, we make the further
344: approximation, well established
345: in practice, where the proton subsystem is replaced by a uniform positive
346: background with a cavity, viz., a Wigner-Seitz sphere of radius
347: $r_s=\left[3/(4\pi\bar{n}\right]^{1/3}$. The positive charge scooped out
348: to form the cavity is placed as a point charge at the origin, and forms
349: the central proton. The finite-temperature Kohn-Sham equation is a
350: consequence of the Euler equation for the stationary property of the free
351: energy under functional derivation with respect to the
352: electron-density distribution, {\it viz}.,
353: \begin{equation}
354: \frac{\delta F([n(r)])}{\delta n(r)}=0.
355: \end{equation}
356: A standard Kohn-Sham type analysis now leads to the equation:
357: \begin{equation}
358: \label{kohnsham}
359: [-\nabla^2/2+Z/r-V_{ks}(r)]\psi_\nu(r)=
360: \epsilon_\nu\psi_\nu(r),
361: \end{equation}
362: where
363: \begin{eqnarray*}
364: V_{ks}(r)&=&V_p(r,n(r))+V_{xc}(r,n(r),T_e)\\
365: n(r)&=&\sum_\nu |\phi_\nu(r)|^2f(\epsilon_\nu/T_e).
366: \end{eqnarray*}
367: Here
368: $$V_P(r,n(r))=\int d{\bf r^\prime}n(r^\prime)/|{\bf r}-{\bf r^\prime}|$$
369: is the Poisson potential of the electron distribution $n(r)$. This
370: distribution is evaluated self-consistently from the Kohn-Sham
371: wavefunctions $\psi_\nu(r),\;\nu=n,l,m$,
372: energy $\epsilon_\nu$, with the occupation factor
373: given by the Fermi function $f(\epsilon/T_e)$.
374: The potential
375: due to the proton at the origin is $Z/r$, with $Z=1$,
376: and $V_{xc}(r,n(r),T_e)$ is the finite-temperature Kohn-Sham
377: exchange-correlation potential\cite{prb3d} which depends self-consistently on
378: the charge profile $n(r)$. This is evaluated using the local-density
379: approximation (LDA), unlike in CHNC where a fully non-local $V_{xc}(r)$
380: is evaluated via a coupling-constant integration over the $g_{ee}(r)$.
381: The Kohn-Sham procedure uses only $n(r)=\bar{n}g_{ep}(r)$, and does not
382: provide a $g_{ee}(r)$. Since this problem contains only one proton, there
383: is no proton temperature in the theory. However, due to the
384: large mass of the proton, and due to the exclusion of other protons by the
385: central proton (modeled by the Wigner-Seitz cavity), the value of
386: $g_{ep}(r)$ at $r\to 0$ given by the
387: Kohn-Sham calculation is expected to be a valid estimate for the full
388: electron-proton plasma. In fact, in the two-temperature electron-proton
389: plasma, $T_{ep}$ of Eq.~\ref{crosstemp}
390: reduces to $T_{ee}$, as in the Kohn-Sham calculation,
391: since $m_p>>m_e$. That this one-proton Kohn-Sham calculation correctly
392: reproduces the $g_{ep}(r)$ of the plasma, even at very low temperatures,
393: is seen from the comparisons given in Fig.~\ref{ksqmc}, where the path-integral Monte
394: Carlo (PIMC) PDFs for hydrogen from the work of Militzer and Ceperley\cite{milicep}
395: have been used.
396: %
397: \begin{figure}
398: \includegraphics*[width=8.8 cm, height=11.0 cm]{qmks.eps}
399: \caption
400: {(Online color)Panels (a),(b) present a comparison of the DFT, QMC,
401: CHNC, MD, and HNC calculations of the electron-proton PDF.
402: The last three use the effective potentials
403: of Eq.~\ref{effpots}. A quantum temperature $T_q$, a Pauli potential
404: and the effctive potentials with $f_{ep}$ different from unity are used in
405: CHNC. The lower panels compare the spin-resolved
406: electron-electron PDfs in the H-plasma, obtained from QMC and CHNC.
407: P-Dw(DFT) and CHNC calculations use the formulations of Dharma-wardana
408: and Perrot\cite{ilciacco, prl1}. The
409: QMC (PIMC) is from Militzer and Ceperley\cite{milicep},
410: while the H-M(HNC) follow the MD calculations of
411: Hanson and MacDonald\cite{hanmac} using the potentials of Eq.~\ref{zerothset}.
412: }
413: \label{ksqmc}
414: \end{figure}
415: %
416: \subsection{The effective electron-proton interaction
417: for classical simulations.}
418: In a classical simulation of an electron-proton plasma, or in a CHNC
419: calculation, the Coulomb interaction $V_{ep}(r)$ appears. This is the
420: attractive classical potential associated with the quantum-mechanical
421: operator $-1/r_{ep}$. It is expected to have the form:
422: \begin{eqnarray}
423: \label{effpots}
424: V_{ep}(r)&=&-\left[1-e^{-k_{ep}r}\right]/r\\
425: k_{ep}&=&k^{dB}_{ep}f_{ep}\\
426: k^{dB}_{ep}&=&\left[2\pi T_{ep}m_{ep}\right]^{1/2}
427: \end{eqnarray}
428: The thermal de Broglie momentum $k^{dB}$
429: provides a first approximation to $k_{ep}$.
430: But we choose $k_{ep}$
431: such that the $g_{ep}(r\to 0)$ generated by the classical procedure (e.g., MD
432: or CHNC) agrees with the $g_{ep}(r\to 0)$ obtained from the Kohn-Sham
433: calculation at the given $r_s$ and $T_e$.
434: It turns out that the correction factor $f_{ep}$ is
435: quite close to unity for sufficiently high temperatures.
436: Even at $T$=10.77 eV, $r_s=1$,
437: i.e., $T/E_F$=0.215, $f_{e}$=0.922 and we see from Fig.~\ref{ksqmc} that the
438: agreement between QMC, DFT, and CHNC is quite good. We have
439: determined the value of $f_{ep}$
440: as a function of $T_e, r_s$ by matching the CHNC calculation
441: and the Kohn-Sham calculation. In effect, $f_{ep}$ is
442: similar to a pseuodopotential or form factor for the electron-proton
443: interaction. When bound states begin to be formed ($r_s>1.8$), the form of
444: $f_{ep}$ becomes more critical, but this problem does not arise within
445: the densities studied here. However, classically, for attractive
446: potentials, dynamical instabilities could occur at any $r_s$, $T$
447: and these have to be controlled using close-approach cutoffs on
448: the potentials, as well as controls on the
449: velocity distribution functions, to maintain the meaning
450: of ``subsystem temperatures'' which are set to $T_e$ and $T_i$.
451: Investigation of such instabilities where the velocity
452: distributions do not conform to the two-temperature model
453: is outside the scope of this study.
454:
455: The electron-electron interaction used in
456: CHNC, and MD simulations,
457: is also a diffraction-corrected Coulomb potential,
458: $V_{ee}(r)$, with $k_{ee}$ being
459: $(2\pi m_{ee} T_{cf})^{1/2}$, with $m_{ee}=m_e/2$ and requiring no
460: additional correction factors.
461: This
462: diffraction correction
463: can be derived from the
464: Schr\"odinger equation describing electron-electron scattering\cite{prl3}.
465: %
466: \subsection{Classical-map Molecular dynamics}
467: The HNC method using all three items: (i) diffraction-corrected effective potentials,
468: (ii) the Pauli exclusion potential and (iii) the quantum temperature $T_q$,
469: is the CHNC scheme. If the same three items were included in classical
470: molecular dynamics simulations, we have a classical-map-MD scheme (CMMD).
471: The CMMD is superior to CHNC since it
472: automatically includes any bridge corrections, etc.
473: that are not in the HNC scheme. In CMMD the electron temperature would be $T_{cf}$,
474: Eq.~\ref{tcf}, as in CHNC. However, since bridge corrections are expected to be
475: negligible in the regime of densities and temperatures considered here,
476: we do not carry out CMMD simulations.
477: %
478: \section{Results}
479: We first provide comparisons between simple classical MD simulations
480: and HNC calculations of PDFs of 2T-2M systems using the simplest diffraction corrected
481: pair potentials, given by:
482: \begin{eqnarray}
483: \label{zerothset}
484: \phi^0_{ij}&=&z_iz_j(1-e^{-k^{dB}_{ij}r})/r\\
485: k^{dB}_{ij}&=&\left[2\pi m_{ij}T_{ij}\right]^{1/2}
486: \end{eqnarray}
487: The MD simulations only need the individual subsystem temperatures $T_{ii}$, $T_{jj}$, and
488: no cross-species temperature $T_{ij}, i\ne j$ is needed. This is achieved by employing
489: two velocity-scaling thermostats that adjust the electron and ion velocity distributions to
490: have the desired mean values. In contrast, the HNC needs a
491: specification for $T_{ij}$. Seuferling et al.\cite{seuf} have suggested that
492: the OZ equations also
493: need to be modified. These issues can be tested by comparison with the MD results.
494: \subsection{Two-mass two-temperature systems.}
495: Systems where the two masses $m_a$ and $m_b$ of the two components $a,b$
496: are equal cannot produce two-temperature
497: quasi-equilibrium systems unless $V_{ab}$ is, for some reason, extremely different from
498: $V_{bb}$ and $V_{aa}$. Thus two-temperature plasmas may exist for significant times,
499: even when the mass-ratio is
500: of the order of 3-10, as in some solid state electron-hole plasmas where band-structure
501: effects associated with the existence of indirect gaps introduce restrictions on
502: electron-hole recombination. Here we present HNC
503: calculations of the PDFs of plasmas with $m_b/m_a=5$,
504: and compare them with MD simulations,
505: to establish the correct implementation of HNC and CHNC procedures.
506: %
507:
508: In Fig.~\ref{tttm} we show the PDFs calculated for a two-component system with a mass
509: ratio of 5, using both the HNC with the standard OZ relations and MD. The MD simulations used 300
510: particles, 40,000 equilibration steps, a time step of 0.02 of the inverse electron-plasma
511: frequency, and data was then accumulated over 120,000 steps using the two thermostats described
512: above. In the HNC calculation, the pair-potentials are given by Eq.~\ref{zerothset}, and the
513: cross-species temperature is as in Eq.~\ref{crosstemp}.
514: This simple HNC-OZ procedure is in very good agreement with
515: MD, both for the equilibrium and non-equilibrium (two-temperature) cases, and we
516: conclude that the additional procedures proposed by Seuferling et al.\cite{seuf} in
517: their Eq. (37) are not needed. That is, our results show that a modified OZ equation is not
518: necessary.
519: The comparison between the HNC and the MD establishes the correctness
520: of the basic HNC procedures even in the quasi-equilibrium case where the formal
521: derivation of the HNC equations becomes an open question.
522: However, once the correct HNC procedure is established, the
523: calculations for the quantum two-temperature two-mass system can be carried out using
524: the CHNC, with the same temperature assignments $T_{ij}$ extended to include the
525: quantum temperatures $T_q$, and the Pauli potentials.
526: %
527: %
528: \begin{figure}
529: \includegraphics*[width=8.5 cm, height=11.0 cm]{tmgij.eps}
530: \caption
531: {(Online color)The upper panel shows the PDFs for a two-component ($a,b$)
532: system at a temperature $T_a=T_b=100$ eV, $r_s=1$, the masses $m_a,\,m_b$ being 1 and 5.
533: The component $a$ electron like, while $b$ is hole-like, i.e., positively charged.
534: Lower panel shows the two-temperature two-mass case, with $T_b$ lowered to 30 eV.
535: In both cases the PDFs $g_{ij}(r)$ calculated using the standard HNC and the
536: OZ equation, Eq.~\ref{oz1}
537: agree well with the MD results using the same input potentials as in HNC.}
538: \label{tttm}
539: \end{figure}
540: %
541: \subsection{Electron-proton systems in thermal equilibrium}
542: The simple diffraction corrected potentials, Eqs.~\ref{zerothset}
543: were used by Hanson and MacDonald in H-plasma
544: simulations\cite{hanmac}. Their PDFs can also be generated using the
545: simple HNC equations if the above $\phi_{ij}(r)$ are used. Hence, in
546: Fig.~\ref{ksqmc} we have labeled the corresponding $g_{ep}(r)$ as H-M(HNC,MD).
547: The PDFs obtained from the DFT calculation, using Eq.~\ref{kohnsham},
548: as implemented in the
549: codes by Perrot and Dharma-wardana\cite{dpcodes}, as well as the PIMC
550: results of Militzer and Ceperley are also shown, to establish that these
551: two first-principles methods are in excellent agreement. Here we note
552: that the CHNC results for $g_{ep}$ and also the spin-resolved $g_{ee}$ are in
553: excellent agreement with the QMC PDFs. To obtain this agreement,
554: the CHNC uses the slightly modified
555: diffraction parameter $k_{ep}=f_{ep}k^{dB}_{ep}$ with $f_{ep}$=0.922 and 0.965
556: at $r_s$=1 and 1.5 respectively, as obtained by matching the contact value
557: of the CHNC $g_{ep}$ to the value from the KS calculation.
558: The MD-HNC using the Hanson-MacDonald
559: approach leads to a large value of $g_{ep}$ at $r\to0$, while the $g_{ee}$
560: (not shown in the figure) are in strong disagreement.
561: The agreement between QMC and CHNC shown in Fig.~\ref{ksqmc} holds even
562: better at higher temperatures, and this justifies our use of the CHNC
563: and Kohn-Sham results as the reference calculations when QMC
564: results are not available.
565: %
566: \subsection{Two-temperature electron-proton systems}
567: In this sub-section we compare classical two-temperature H-plasmas and show that
568: the temperature $T_{ep}$ that appears in the cross-species HNC equation is
569: indeed the electron temperature $T_{ee}$, as in Eq.~\ref{kdifr},
570: for the limit $m_p>>m_e$. Thus, we use the same $T_{ep}$ in the CHNC,
571: to include
572: the quantum corrections and compute $g_{ij}(r)$.
573: % Results from the Kohn-Sham procedure
574: % and from CMMC are used to confirm these results.
575: In Fig.~\ref{mdhnc} we
576: show the cross-species (electron-proton) PDF for a hydrogen plasma with
577: $T_e=100$ eV, $r_s=1$, for the four cases: $T_i=100,\,60,\,30$ and 10.
578: From panel (a)
579: we see that the classical procedures (HNC and MD) using the simplest
580: set of $\phi_{ij}(r)$, Eq.~\ref{zerothset}, overestimate the $g_{ep}$
581: in comparison
582: to the Kohn-Sham (DFT) estimate. In panels (b-d) we have two-temperature
583: plasmas, and the MD calculation (which needs only $T_e,T_i$) is
584: closely reproduced by the
585: HNC if $T_{ep}$ is set to $T_e$. In panel (b) we show that the choice
586: $T_{ep}=(T_e+T_p)/2$ in the HNC is clearly inapplicable if the system
587: is entirely specified by $T_e,\,T_p,\,$ and $r_s$.
588: %
589: \begin{figure}
590: \includegraphics*[width=9 cm, height=12.0 cm]{mdhnc.eps}
591: \caption
592: {(Online color) Dense hydrogen: Panel (a) compares HNC and MD $g_{ep}(r)$ using the simplest set
593: of classical potentials (Eq.~\ref{zerothset}). The DFT PDF shows that the
594: classical potentials are an overestimate. In panels (b-d) we use the
595: same classical potentials to establish that the temperature $T_{ep}$
596: needed in the HNC is indeed $T_e$ if HNC and MD are to agree for
597: two-temperature electron-proton systems.}
598: \label{mdhnc}
599: \end{figure}
600: %
601: As seen from Fig.~\ref{mdhnc}, quantum effects may significantly modify the
602: PDFs even when the electrons are at 100 eV.
603:
604: Thus, in Fig.~\ref{chnc2t}, we present
605: CHNC calculations for a two-temperature plasma with $T_e=100, T_i=30$ at the
606: density $r_s=1$. The top panel shows that the proton-proton PDF calculated from
607: the quantum procedure (using CHNC) is more strongly coupled than in the classical
608: (using HNC) $g_{pp}$. The stronger e-p interaction in the classical system,
609: as shown in the enhanced $g_{ep}$, leads to greater screening,
610: weakening the ion-ion
611: interaction. The lower panel shows the spin-resolved e-e PDFs, labeled $g_{uu},g_{ud}$
612: obtained from CHNC, and the classical $g_{ee}$ obtained from HNC. The CHNC correctly
613: incorporates the exclusion effects via the Pauli potential, Eq.~\ref{paupot}.
614: %
615: \begin{figure}
616: \includegraphics*[width=9 cm, height=12.0 cm]{chnc2t.eps}
617: \caption
618: {(Online color)The upper panel shows the e-p and p-p PDFs for
619: an electron-proton plasma with $T_e=100$eV, $T_p=30$eV, $r_s=1$, calculated using
620: HNC and CHNC. The lower panel shows the e-e PDFs, where the HNC does not
621: incorporate the effects of the exclusion principle. The CHNC $g_{uu}$ and
622: $g_{ud}$ refer to spin parallel and antiparallel PDFs respectively.}
623: \label{chnc2t}
624: \end{figure}
625: %
626: %
627: \subsection{The electron-proton PDF and Pauli exclusion effects}
628: The electron-proton pair distribution function is mainly determined by the e-p interaction
629: which is spin-independent. However, once an electron is correlated with a proton, the
630: correlation of that electron with other electrons would be affected by Pauli exclusion
631: effects associated with the electron spin. In the CHNC and CMMD schemes, the
632: effect of the Pauli principle are incorporated as a potential, Eq.~\ref{paupot},
633: between parallel-spin electrons. This potential is not used in MD and in
634: the pure HNC scheme. Hence the
635: $g_{ee}(r)$ obtained from HNC,
636: shown in the lower panel of Fig.~\ref{chnc2t}(b) is identical for
637: parallel and anti-parallel PDFs. However, Fig.~\ref{chnc2t}(a) shows that the $g_{ep}$
638: obtained by the full CHNC, inclusive of the Pauli potential, $f_{ep}$, and $T_q$ features
639: is quite close to the pure-HNC result where $f_{ep}=1$ in the
640: diffraction potentials. At $r_s=1$, $T_q/E_F=0.768$, and hence, when $T_e=100$ eV, i.e.,
641: $T_e/E_F=1.9956$, then $T_q$ itself is substantial. Thus the larger value of $g_{ep}(r)$ at
642: $r\to0$ found in the HNC and MD is not due to the Pauli exclusion effects,
643: but due mainly to two reasons: (i)the overestimate contained in
644: the zeroth set of effective potentials where $f_{ep}=1$, and
645: (ii)the use of the physical temperature $T_e$ as the effective temperature
646: of the classical electron fluid,
647: while $T_{cf}>T_e$ is used in the CHNC.
648: To check these, we have run CHNC calculations where (i) the Pauli
649: potential was switched off while the $T_q$, $f_{ep}$ were included; (ii) only the Pauli
650: and $f_{ep}$ were included; (iii) only the Pauli and $T_q$ were included; and so forth.
651: Such ``numerical experiments'' enable us to conclude that the Pauli exclusion effect is
652: of relatively low importance for the $g_{ep}(r)$ when $T_e$ is 100 eV and $r_s=1$.
653: %
654:
655:
656: \section{concluding discussion}
657: The simplest classical rendering of
658: quantum plasmas, based on the use of diffraction corrected potentials
659: (Eq.~\ref{zerothset}) was used with HNC calculations and MD simulations to resolve
660: the ambiguities and difficulties in handling the two-temperature, two-mass system.
661: We conclude that the modifications to the OZ equations proposed by
662: Seuferling et al.\cite{seuf}., are not needed.
663: The classical
664: mapping of quantum systems to the HNC equations, as used in the CHNC was confirmed by
665: comparisons with Kohn-Sham DFT calculations as well as with available PIMC
666: results for compressed hydrogen plasmas at finite temperatures.
667: We conclude that the HNC and CHNC, together with the standard OZ
668: equations provide
669: excellent, accurate and simple analytical tools for the investigation
670: of many-particle quasi-equilibrium
671: systems for which direct quantum simulations continue to remain too prohibitive
672: or unfeasible.
673: %
674: \begin{thebibliography}{99}
675: %
676: \bibitem{ceperleyH}
677: K. T. Delaney, C. Pierleoni, D. M. Ceperley. Phys. Rev. Lett. {\bf 97}, 235702 (2006).
678: C. Pierleoni, D. M. Ceperley, Markus Holzmann. Phys. Rev. Lett. {\bf 95}, 146402 (2004)
679: \bibitem{kwon}
680: I. Kwon, L. Collins, J. Kress and N. Troullier, Phys. Rev. E. {\bf 54},
681: 2844 (1996)
682: \bibitem{dejarlais}
683: M. P. Desjarlais, Phys. Rev. B {\bf 68}, 64204 (2003);
684: \bibitem{mazevet}
685: S. Mazevet, M. P. Desjarlais, L. A. Collins, J. D. Kress and N. H. Magee,
686: Phys. Rev. E {\bf 71}, 016409 (2005)
687: \bibitem{pdw95}
688: F. Perrot and M.W.C. Dharma-wardana, Phys. Rev. E. {\bf 52 }, 5352 (1995)
689: \bibitem{ilciacco}
690: {\it Density Functional Theory}, Ed. E. K. U. Gross and Dreizler,
691: NATO ASI Series B: Physics 337 (Plenum, NY. 1993)
692: \bibitem{lvm}
693: M.W.C. Dharma-wardana, Phys. Rev. E. {\bf 73}, 036401 (2006)
694: \bibitem{astrop}
695: F. A. Agronyan and R. A. Syunyaev,
696: Astrophysics {\bf 27}, 413-422; Translated from Astrofizika; 27: No. 1, 131-145 ( 1987)
697: %http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=6894805
698: \bibitem{dewitt}
699: A.I. Chugunov, H. E. DeWitt, and D. G. Yakovlev, Phys. Rev. D {\bf 76}, 025028 (2007)
700: \bibitem{seuf}
701: P.Seuferling, J. Vogel, and C. Toepffer, Phys. Rev. A {\bf 40}, 323 (1989)
702: \bibitem{mc}
703: D. P. Landau and K. Binder {\it A guide to Monte Carlo simulations in statistical Physics}
704: (Cambridge University Press 2005)
705: \bibitem{hncref}
706: J. M. J. van Leeuwen, J. Gr\"oneveld, J. de Boer, Physica {\bf 25}, 792 (1959)
707: \bibitem{prl1}
708: M. W. C. Dharma-wardana and F. Perrot, Phys. Rev. Lett. {\bf 84}, 959 (2000)
709: \bibitem{prb3d}
710: F. Perrot and M.W.C. Dharma-wardana, Phys. Rev. B, {\bf 62}, 16536 (2000);
711: %\cite{feenberg,lantto,KP}
712: \bibitem{Jones} C. S. Jones and M. S. Murillo,
713: High Energy Density Phys. (doi:10.1016/j.hedp.2007.02.038)
714: (2007).
715: \bibitem{feenberg}
716: E.Feenberg, {\em Theory of Quantum Fluids} (Academic, New York 1969)
717: \bibitem{lantto}
718: L. J. Lantto, Phys. Rev. B {\bf 22}, 1380 (1980)
719: \bibitem{prl2}
720: F. Perrot and M. W. C. Dharma-wardana, Phys. Rev. Lett. {\bf 87}, 206404
721: (2001)
722: \bibitem{lado}
723: F. Lado, J. Chem. Phys. {\bf 47}, 5369 (1967)
724: \bibitem{bm}
725: D. B. Boercker and R. M. More, Phys. Rev A {\bf 33} 1859 (1986)
726: \bibitem{milicep}
727: B. Militzer and D. Ceperley , Phys. Rev. Lett. {\bf 85}, 1890 (2000);
728: B. Militzer, Thesis, (2000), see http://militzer.gl.ciw.edu/diss/diss\_militzer.pdf
729: \bibitem{dpcodes}
730: M. W. C. Dharma-wardana and Fran\c{c}ois Perrot;\\
731: http://athens.phy.nrc.ca/ims/qp/codes/chandre/D\_P/\\
732: access by password obtainable from: chandre.dharma-wardana@nrc-cnrc-gc.ca
733: %
734: \bibitem{hanmac}
735: J.-P. Hansen, I. R. MacDonald, Phys. Rev. Lett. {\bf 41} 1379 (1978)
736: \bibitem{prl3}
737: M.W.C. Dharma-wardana and F. Perrot, Phys. Rev. Lett. {\bf 90}, 136601 (2003)
738:
739:
740: \end{thebibliography}
741:
742:
743:
744: \end{document}
745:
746:
747:
748:
749:
750:
751:
752: %
753: %
754: