1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % Kolmogorov-Calandra-Curtarolo
3: % metal-borides ternary paper
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5: %\documentstyle[multicol,aps,epsfig,prl]{revtex4}
6: \documentclass[twocolumn,prl,aps]{revtex4}
7: %\documentstyle[aps,amssymb,epsfig,preprint]{revtex}
8: \voffset 10mm
9:
10: %\documentstyle[multicol,aps,prl]{revtex4}
11: %\documentclass[twocolumn,prb,aps]{revtex4}
12:
13: \usepackage{epsfig}
14: %\usepackage{graphicx}
15:
16: \begin{document}
17: \textheight 226mm
18:
19: \title{\Large Thermodynamic stabilities of ternary metal borides:
20: An \textit{ab initio} guide for synthesizing layered superconductors}
21:
22: \author{Aleksey N. Kolmogorov$^1$, Matteo Calandra$^2$, and Stefano Curtarolo$^3$}
23:
24: \affiliation{
25: $^1$Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, United Kingdom \\
26: $^2$Institut de MinΩralogie et de Physique des Milieux condens\'es %,
27: case 115, 4 place Jussieu, 75252, Paris cedex 05, France\\
28: $^3$Department of Mechanical Engineering and Materials Science %,
29: Duke University, Durham, North Carolina 27708, USA
30: }
31:
32: \date{\today}
33:
34: \begin{abstract} {
35: Density functional theory calculations have been
36: used to identify stable layered Li-$M$-B crystal structure phases
37: derived from a recently proposed binary metal-sandwich (MS)
38: lithium monoboride superconductor. We show that the MS lithium
39: monoboride gains in stability when alloyed with electron-rich
40: metal diborides; the resulting ordered Li$_{2(1-x)}M_x$B$_2$
41: ternary phases may form under normal synthesis conditions in a
42: wide concentration range of $x$ for a number of group-III-V metals
43: $M$. In an effort to pre-select compounds with the strongest
44: electron-phonon coupling we examine the softening of the in-plane
45: boron phonon mode at $\Gamma$ in a large class of metal
46: borides. Our results reveal interesting general trends for the
47: frequency of the in-plane boron phonon modes as a function of the
48: boron-boron bond length and the valence of the metal. One of the
49: candidates with a promise to be an MgB$_2$-type superconductor,
50: Li$_2$AlB$_4$, has been examined in more detail: according to our
51: {\it ab initio} calculations of the phonon dispersion and the
52: electron-phonon coupling $\lambda$, the compound should have a
53: critical temperature of $\sim4$ K. }
54: \end{abstract}
55: \pacs{63.20.Kr, 63.20.Dj , 78.30.Er, 74.70.Ad}
56:
57: \maketitle
58:
59: %\draft
60:
61: \renewcommand{\topfraction}{0.85}
62: \renewcommand{\textfraction}{0.1}
63: \renewcommand{\floatpagefraction}{0.75}
64:
65: \subsection{I. Introduction}
66:
67: Observation and explanation of the superconducting transition in
68: MgB$_2$ at an unexpectedly high temperature of 39 K
69: \cite{origin,band_filling} have stimulated extensive research aimed at
70: developing related layered phonon-mediated superconductors
71: \cite{CaAlSi,TaB2,MB2,NbxB2,LiBC,LixBC,SCGIC1,SCGIC2,SCGIC3,US1,MgC}.
72: Interesting results have been recently obtained for carbon-based
73: layered materials: the efforts to adjust the properties of graphite
74: via intercalation with various metals have led to the discovery of a
75: CaC$_6$ superconductor with a critical temperature of 11.5 K
76: \cite{SCGIC1,SCGIC2} (15.1 K under hydrostatic pressure \cite{SCGIC3}).
77:
78: Tuning the properties of boron-based superconductors has proven to be
79: difficult because, despite the existence of over a dozen of stable
80: metal diborides with the C32 structure \cite{ICSD,PAULING,Oguchi},
81: only one of them, MgB$_2$, has the hole-doped boron $p\sigma$ states
82: that couple strongly to the in-plane boron phonon modes
83: \cite{band_filling}. For comparison, critical temperatures in other
84: C32-$M$B$_2$ superconductors, such as TaB$_2$ or NbB$_2$
85: \cite{TaB2,MB2,NbxB2}, do not exceed 10 K because these electron-rich
86: compounds have fully occupied boron $p\sigma$ bands and thus lack the
87: important nearly two-dimensional Fermi surfaces. The attempts to raise
88: $T_c$ in MgB$_2$ via substitutional doping have been unsuccessful for
89: various reasons: i) electron-doping led to the decrease of the
90: electron-phonon coupling due mainly to the filling of the $p\sigma$
91: band \cite{band_filling}; ii) hole-doping or substitution of Mg with
92: large metals turned out to be thermodynamically unfavorable
93: \cite{dope_review,MgB2_defects}; and iii) co-doping of MgB$_2$ with
94: Li-Al Li-C caused the reduction of the $T_c$ because of the concurrent
95: filling of the $p\sigma$ band and depletion of the $p\pi$ band
96: \cite{MgAlLiB1}. Recently, Palnichenko {\it et al.} have reported
97: interesting results on a possible onset of superconductivity in
98: MgB$_2$ after thermal treatment in the presence of Rb, Cs, and Ba; the
99: structure of the resultant materials and the mechanism of the $T_c$
100: enhancement have yet to be determined \cite{Palnichenko,Chepulskyy}.
101:
102: Identification and synthesis of new stable compounds will be a
103: critical step to overcome the limitations of the existing layered
104: metal borides and to have a chance of obtaining better
105: superconductors. Using a data-mining approach we have recently found a
106: previously unknown metal sandwich (MS) crystal structure and
107: demonstrated that lithium monoboride in this configuration is
108: marginally stable under ambient conditions and is likely to form under
109: pressure \cite{US1,US2}. Remarkably, our calculations indicated that,
110: relative to MgB$_2$, MS-LiB has a higher density of boron $p\sigma$
111: states at the Fermi level \cite{US1,US2}, a feature long sought for in
112: MgB$_2$-related materials. Subsequent theoretical studies of the
113: electron-phonon coupling predicted that the $T_c$ in MS-LiB would be
114: around 10 K \cite{Mazin,US3}. The relatively low value of $T_c$ has
115: been attributed to an accidental electronic structure feature present
116: in the pristine MS-LiB: the crossing of the $p\pi$ bands happens to be
117: exactly at the Fermi level \cite{Mazin,US3}. After examining the
118: compound's response to hydrostatic pressure and substitutional doping
119: we have concluded that the recovery of the $p\pi$ density of states
120: (DOS) at the Fermi level would be accompanied by an unfortunate drop
121: of the $p\sigma$ DOS under these conditions \cite{US3}. The question of
122: whether it could be possible to tune the properties of MS-LiB to
123: obtain higher $T_c$ remains open.
124:
125: In this work we expand the search for stable crystal structure phases
126: to ternary metal borides. Namely, we use density functional theory
127: calculations to explore the relative stability of different layered
128: configurations with composition Li$_{2(1-x)}M_x$B$_2$ for a library
129: of over 30 metals $M$. We show that MS lithium monoboride gains in
130: stability when alloyed with electron-rich metal diborides; the
131: resulting ordered Li$_{2(1-x)}M_x$B$_2$ ternary phases may form under
132: normal synthesis conditions in a large concentration range of $x$ for
133: a number of group-III-V metals $M$. For several promising candidates
134: we evaluate the softening of the in-plane boron phonon mode at $\Gamma$
135: in order to identify which of these compounds has large
136: coupling to the in-plane boron modes (similarly to what happens in
137: MgB$_2$). For one case, Li$_2$AlB$_4$, we also perform a full
138: calculation of the critical temperature. We believe that our results
139: will be of considerable help to experimental groups working on the
140: development of new boron-based layered superconductors.
141:
142: The paper is organized as follows. In Sec. II we describe the geometry
143: of the ternary MS structures. Section III is devoted to the
144: zero-temperature thermodynamic stability of Li$_{2(1-x)}M_x$B$_2$
145: phases and to the special case Li$_{2(1-x)}$Al$_x$B$_2$. Electronic
146: properties of two representatives, Li$_2$AlB$_4$ and Li$_2$TiB$_4$,
147: are addressed in Sec. IV. In Sec. V we examine the phonon softening in
148: a large set of metal borides. In Sec. VI we present the {\it ab initio}
149: superconductivity analysis of Li$_2$AlB$_4$. Conclusions are given in
150: Sec. VII.
151:
152: \subsection{II. Li$_{2(1-x)}M_x$B$_2$ metal sandwich structures}
153:
154: Consideration of ternary Li-$M$-B systems dramatically increases the
155: number of candidates to be screened because one should check both
156: different compositions and various populations of metal sites. We
157: reduce the search space by recalling the binding trends for the
158: binary MS metal borides demonstrated in our previous
159: study \cite{US2}. Namely, the MS lithium monoboride has plenty of
160: available bonding boron $p \sigma$ states and may additionally
161: stabilize when mixed with electron-rich metals. Because
162: straightforward substitutional doping generally causes strain and loss
163: of binding within the metal layers \cite{MgB2_defects, US4} we separate
164: regions with different metal species by boron layers and, thus, avoid
165: putting different metals in direct contact. This idea follows a
166: natural segregation tendency in layered borides, which has been
167: observed, for instance, in MgB$_2$ under heavy Al doping: the
168: resulting compound was a C32 superstructure with alternating Mg and Al
169: layers \cite{MgAlB4}.
170:
171: % here is another one on MgAlB4: PHYSICAL REVIEW B 71, 174506 (2005)
172:
173: \begin{figure}[t]
174: \begin{center}
175: % \vspace{-10mm}
176: \centerline{\epsfig{file=fig1.eps,width=80mm,clip=}}
177: % \vspace{-5mm}
178: \caption{ \small (Color online). Structure of the proposed ternary
179: metal sandwich phases and their location on a schematic ternary
180: (Li-$M$-B) phase diagram. Formation energies are shown along
181: the vertical axis. The relative stability is calculated with
182: respect to 11-LiB (MS2) and 0-MB$_2$ (C32) (notation explained in
183: the text).}
184: \label{fig1}
185: \end{center}
186: % \vspace{10mm}
187: \end{figure}
188: %\vspace{-5mm}
189: %
190: Our candidate segregated superstructures are built by combining MS-LiB
191: and C32-$M$B$_2$ unit cells. Since there are numerous possibilities to
192: stack metal and boron layers, it is convenient to adopt the following
193: compact notation: MS1-LiB and C32-$M$B$_2$ basic units will be denoted
194: as ``1''- and ``0''-unit, respectively. In this way one can uniquely
195: specify the structure with a string of ones and zeros. Note that the
196: presence of one ``1''-unit results in a stacking shift along the
197: $c$-axis which makes the unit cell rhombohedral \cite{US1,US2}. For
198: convenience, in candidate phases containing an even number of
199: ``1''-units we choose to reflect every second ``1''-unit in the
200: $(x,y)$-plane to obtain a hexagonal unit cell. This choice is expected
201: to not affect our results, since we have found no appreciable
202: differences in the properties of the rhombohedral and hexagonal
203: allotropes of MS-LiB \cite{US1}. Examples of two superstructures with
204: small unit cells are 10-Li$_2$$M$B$_4$ [Fig. \ref{fig1}(c)] and
205: 110-Li$_4$$M$B$_6$ [Fig. \ref{fig1}(b)], the latter being the
206: superposition of MS2-LiB and C32-$M$B$_2$.
207:
208: Figure \ref{fig1} shows that such compounds are located along the red
209: line on the ternary phase diagram. Note that MS-LiB is marginally
210: stable (the known competing phase LiB$_y$ has a completely different
211: structure \cite{US2}). This implies that if (i) the C32-MB$_2$ exists
212: and (ii) the relative stability of the MS ternary phases along this
213: line is negative, then one would have a better chance of synthesizing
214: the MS ternary compounds from elemental materials rather than the two
215: binaries MS2-LiB and $M$B$_2$ separately. Of course, one needs to make
216: sure that no other competing phase forms, which can be a challenging
217: task for ternary systems.
218:
219: \subsection{III. Thermodynamic stability}
220:
221: {\it Computational details.} In the evaluation of the low temperature
222: stability of ternary compounds, we rely on the ``{\it ab initio}
223: formation energy'' criterion, which has been shown to be a reliable
224: approach for binary systems \cite{Curtarolo_DMQC,Curtarolo_JMEAS}
225: (expected probability of predicting the correct experimental compounds
226: or immiscibilities is $\eta^{\star}\approx97.3\%$ as defined in
227: equation (5) of Ref. \cite{Curtarolo_CALPHAD}). {\it Ab initio} total
228: energy calculations are performed in the generalized gradient
229: approximation with projector augmented-wave pseudopotentials (PAW)
230: \cite{bloechl994} and Perdew, Burke, and Ernzerhof \cite{PBE}
231: exchange-correlation functional, as implemented in {\small VASP}
232: \cite{kresse1993,kresse1996b}. Because of a significant charge
233: transfer between metal and boron in most structures considered we use
234: PAW pseudopotentials in which semicore states are treated as
235: valence. This is especially important for the Li-B system as discussed
236: in Refs. \cite{US_PAW} and \cite{US1}. Simulations are carried out at
237: zero temperature and without zero-point motion; spin polarization is
238: used only for borides of Fe, Co, and Ni. We use an energy cutoff of 398
239: eV and at least 4000/(number of atoms in unit cell) ${\bf k}$-points
240: distributed on a Monkhorst-Pack mesh \cite{MONKHORST_PACK}. All
241: structures are fully relaxed and numerically converged to within 1-2
242: meV/atom.
243:
244: %
245: {\it Stability of Li$_2$$M$B$_4$ phases.} Figure
246: \ref{fig2} shows the relative formation energy $\Delta E_f^{10}$ of
247: 10-Li$_2M$B$_4$ metal borides (Fig. \ref{fig1}(c)) with respect to
248: phase separation into 11-LiB (MS2) and 0-MB$_2$ (C32):
249:
250: \begin{eqnarray}
251: \Delta E_f^{10}\equiv E_f^{10}-\frac{4}{7} E_f^{11}-\frac{3}{7} E_f^{0},
252: \end{eqnarray}
253: %
254: (all formation energies are per atom). We observe that there are many
255: metals that can stabilize MS2-LiB and a number of them do have stable
256: C32-MB$_2$ phases ($M$ = Al, Hf, Ti, V, Nb, Ta)
257: \cite{PAULING,US2_note1}. Considering that metals $M$ = Hf, Ti, V, Nb,
258: Ta have no reported stable compounds with Li or Li-B
259: \cite{PAULING,ICSD}, there is a good chance that the predicted layered
260: phases will form in the Li-$M$-B ($M$ = V, Nb, Ta) ternary
261: systems. Note that stable structures for several transition-metal
262: diborides have larger unit cells with buckled boron layers and
263: different location of metal sites [WB$_2$ ($hP12$, $hR18$), MoB$_2$
264: (hR18), RuB$_2$ and OsB$_2$ ($oP6$) (the $\delta$ phase
265: \cite{US2_note2} is basically a non-corrugated $oP6$)]. For example,
266: the oP6-RuB$_2$ phase is favored over C32-RuB$_2$ by about 0.4
267: eV/atom, which would make 10-Li$_2$RuB$_4$ unstable. However, one
268: could expand the library of possible ternary configurations by
269: creating more stable MS sequences with buckled boron layers. As for
270: the mono- and divalent metal diborides, it is not surprising that they
271: cannot improve the stability of MS-LiB in the corresponding ternary
272: alloy because they are underdoped themselves \cite{Oguchi}. Structural
273: and electronic properties for selected proposed 1010-Li$_2$$M$B$_4$
274: phases are summarized in Table I.
275:
276: \begin{figure}[t]
277: \begin{center}
278: % \vspace{-4mm}
279: \centerline{\epsfig{file=fig2.eps,width=80mm,clip=}}
280: % \vspace{-8mm}
281: \caption{ \small (Color online). Relative stability of the
282: 10-Li$_2M$B$_4$ metal borides (Fig. \ref{fig1}(c)) with respect
283: to phase separation into 11-LiB (MS2) and 0-$M$B$_2$ (C32) (see
284: Eq. 1) as a function of metal valence. The group-II-V metals
285: highlighted in blue form stable C32-type diborides.}
286: \label{fig2}
287: \end{center}
288: \end{figure}
289: %
290: %
291: \begin{table*}[bht]
292: \begin{center}
293: % \vspace{-7mm}
294: \caption{ \small Structural and electronic properties of the
295: proposed 1010-Li$_2$$M$B$_4$ phases for metals $M$ that have stable
296: C32-$M$B$_2$ phases. We list fully relaxed lattice vectors and
297: Wyckoff positions for the $P6_3/mmc$ (no. 194) unit cells: B1 ($4f$)
298: (1/3,2/3,$z_{B1}$); B2 ($4e$) (0,0,$z_{B2}$); Li ($4f$)
299: (1/3,2/3,$z_{Li}$); and $M$ ($2c$) (1/3,2/3,1/4).
300: $E^{\sigma,1}_{\Gamma}$ and $E^{\sigma,2}_{\Gamma}$ and positions of
301: the boron $\sigma$ states above the Fermi level at the $\Gamma$
302: point and $E_f^{1010}$ is a relative formation energy per atom with
303: respect to MS2-LiB and C32-$M$B$_2$. Band structures for
304: 1010-Li$_2$AlB$_4$ and 1010-Li$_2$TiB$_4$ are shown in Figs. 5 and
305: 6, respectively.}
306: \begin{tabular}{c|@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}c@{\hspace{7mm}}}
307: \hline\hline
308: $M$ in & $a$ & $c$ & $z_{B1}$ & $z_{B2}$ & $z_{Li}$ & $E^{\sigma,1}_{\Gamma}$ & $E^{\sigma,2}_{\Gamma}$ & $E_f^{1010}$ \\
309: 1010-Li$_2$$M$B$_4$& (\AA) & (\AA) & & & & eV & eV & eV/atom \\
310: \hline
311: Mg & 3.048 & 18.94 & 0.6560 & 0.6560 & 0.4214 & 0.93 & 0.75 & 0.014 \\
312: Al & 3.012 & 18.46 & 0.6643 & 0.6644 & 0.4198 & 0.62 & 0.14 & -0.052 \\
313: Sc & 3.116 & 17.77 & 0.6524 & 0.6526 & 0.4263 & 0.78 & -0.21 & 0.009 \\
314: Y & 3.225 & 16.14 & 0.6298 & 0.6302 & 0.4482 & 0.55 & -0.13 & 0.033 \\
315: Ti & 3.040 & 17.99 & 0.6626 & 0.6627 & 0.4197 & 0.28 & -1.54 & -0.019 \\
316: Zr & 3.138 & 17.85 & 0.6514 & 0.6515 & 0.4265 & 0.16 & -1.25 & 0.008 \\
317: Hf & 3.119 & 18.20 & 0.6558 & 0.6559 & 0.4229 & 0.05 & -1.40 & -0.010 \\
318: V & 2.968 & 18.48 & 0.6670 & 0.6670 & 0.4181 & -0.04 & -2.43 & -0.044 \\
319: Nb & 3.044 & 19.17 & 0.6610 & 0.6610 & 0.4188 & -0.28 & -2.45 & -0.074 \\
320: Ta & 3.036 & 19.12 & 0.6615 & 0.6615 & 0.4194 & -0.48 & -2.61 & -0.108 \\
321: \hline\hline
322: \end{tabular}
323: %\vspace{-0.4cm}
324: \label{SUM}
325: \end{center}
326: \end{table*}
327:
328: {\it Li-Al-B system}. Aluminum is a special case because it is the
329: least electron-rich metal on the list that provides the desired
330: stabilization for the ternary MS configurations. The sizeable
331: 50-meV/atom energy gain for 10-Li$_2$AlB$_4$ indicates that this
332: compound would be stable with respect to C32-AlB$_2$ and the known
333: off-stoichiometry LiB$_y$ as well \cite{US2}. For the analysis of the
334: thermodynamic stability of Li$_{2(x-1)}$Al$_x$B$_2$ we also need to
335: consider the known stable Li-Al binary phases B32-LiAl,
336: hR15-Li$_3$Al$_2$, and mS26-Li$_9$Al$_4$ \cite{PAULING,LiAl1,LiAl2}
337: and the only reported ternary LiAlB$_{14}$ compound \cite{LiAlB14}
338: which was observed in experiments on doping AlB$_2$ with
339: Li \cite{LiAlB1}.
340:
341: \begin{figure}[t]
342: \begin{center}
343: % \vspace{-5mm}
344: \centerline{\epsfig{file=fig3.eps,width=80mm,clip=}}
345: % \vspace{-8mm}
346: \caption{ \small (Color online). Top: location of known and predicted phases on
347: the Li-Al-B phase diagram (the boron-rich Li-B and Al-B phases are
348: not shown). Bottom: Relative stability of the proposed
349: Li$_2$AlB$_4$ with respect to the decomposition along the lines
350: shown on the top panel.}
351: \label{fig3}
352: \end{center}
353: \end{figure}
354: %\vspace{-2mm}
355: %
356: The location of these phases on the Li-Al-B phase diagram is shown in
357: Fig. 3. Because 10-Li$_2$AlB$_4$ lies very close to the line
358: connecting LiAlB$_{14}$ and mS26-Li$_9$Al$_4$, its relative stability
359: will be defined primarily by the formation energies of these two
360: compounds. We find that 10-Li$_2$AlB$_4$ is virtually degenerate in
361: energy with respect to the mixture of LiAlB$_{14}$, mS26-Li$_9$Al$_4$,
362: and Li at zero temperature and pressure and without zero-point
363: corrections; the finite temperature contributions must have a negative
364: effect on the 10-Li$_2$AlB$_4$ relative stability, as there are no
365: reports on the formation of this compound.
366:
367: In order to stabilize the proposed phase one could use high pressures:
368: as we have demonstrated previously the MS phases are unusually
369: soft \cite{US1,US2}. Indeed, the calculated relative formation enthalpy
370: as a function of pressure for 10-Li$_2$AlB$_4$ in Fig. 3 either
371: becomes negative (with respect to LiAlB$_{14}$, mS26-Li$_9$Al$_4$, and
372: Li) or remains negative (with respect to 11-LiB and 0-AlB$_2$). The
373: considerable change in slope at about 6 GPa is related to a sudden
374: $\sim10$\% decrease in the 10-Li$_2$AlB$_4$ atomic volume at that
375: pressure (Fig. 3). Overall, 10-Li$_2$AlB$_4$ compresses by over 25\%
376: when the pressure is increased from 0 GPa ($v$ = 10.4 \AA$^{3}$/atom,
377: $H_f=-0.163$ eV/atom) to 12 GPa ($v$ = 7.7 \AA$^{3}$/atom,
378: $H_f=-0.302$ eV/atom). For comparison, the LiAlB$_{14}$ phase is much
379: more compact at $P=0$ GPa ($v$ = 7.7 \AA$^{3}$/atom, $H_f=-0.168$
380: eV/atom) but compresses by only 5\% at $P=12$ GPa ($v$ = 7.3
381: \AA$^{3}$/atom, $H_f=-0.217$ eV/atom). These results suggest that
382: this ternary and other binary boron-rich Li-B and Al-B
383: phases \cite{PAULING,POLY}, also comprised of rigid boron polyhedra
384: (e.g., Li$_3$B$_{14}$), should not prohibit the formation of the
385: predicted layered compounds under the pressures considered.
386:
387: \begin{figure}[t]
388: \begin{center}
389: % \vspace{-5mm}
390: \centerline{\epsfig{file=fig4.eps,width=80mm,clip=}}
391: % \vspace{-8mm}
392: \caption{ \small Relative stability of Li$_{2(1-x)}$Al$_x$B$_2$
393: ternary metal borides with respect to the phase separation into
394: 11-LiB (MS2) and 0-AlB$_2$ (C32). The dotted lines connect
395: members of the 110$\cdots$0 and 11110$\cdots$0 series.}
396: \label{fig4}
397: \end{center}
398: \end{figure}
399: %\vspace{-2mm}
400: %
401:
402: There is strong evidence that another stable phase has formed in
403: different experiments involving Li, Al, and
404: B \cite{LiAlB1,LiAlB2}. Monni {\it et al.} suggested that
405: their unidentified low-angle spurious peaks could belong to a binary
406: Li-B phase \cite{LiAlB2}. The authors purposely prepared a sample at the LiB$_2$
407: composition and indeed observed a matching set of the low-angle
408: reflections (see Fig. 7 in Ref. \cite{LiAlB2}). Their conclusion that
409: the unknown phase is a binary Li-B compound agrees well with the
410: observation of 12.2$^{\circ}$ and 20.9$^{\circ}$ peaks at 40\% - 50\%
411: Li compositions by Wang {\it et al.} in
412: 1978 \cite{LiB_Wang78}. However, in a series of detailed studies of the
413: Li-B system W\"{o}rle {\it et al.} have conclusively determined that
414: such low-angle peaks correspond to Li$_6$B$_{18}$(Li$_2$O)$_x$ phases
415: (0$<x<$1) \cite{LiB_B6}. The authors' latest analysis of the X-ray
416: patterns and electron density distributions has indicated that the
417: zeolite-like structure of Li$_6$B$_{18}$(Li$_2$O)$_x$ consists of
418: interconnected boron octahedra with large tunnels filled with neutral
419: Li$_2$O template \cite{LiB_B6}.
420: Therefore, it appears that the stable oxygen-containing phase also
421: formed in the earlier experiments involving Li and B at about 1:2
422: composition. This knowledge is important for future experiments on
423: the ternary metal borides: in order to obtain the predicted layered
424: compounds (whose stability can be affected by a few meV/atom) one
425: needs to reduce the amount of oxygen in the system.
426:
427: In order to determine whether there are more stable ternary MS
428: configurations we calculate a number of Li-Al-B phases with various
429: stacking sequences and compare their relative stability, $\Delta E_f$,
430: with respect to the phase separation in 11-LiB and 0-MB$_2$. The
431: results in Fig. \ref{fig4} suggest that 10-Li$_2$AlB$_4$ would be the
432: most LiB-rich phase to form, while one could observe several
433: AlB$_2$-rich phases at different concentrations (a similar phase
434: diagram is obtained for the Li-Ti-B system). The effect of permutation
435: of the ``1'' and ``0'' units on the energy at given composition is
436: illustrated with the 10$\cdots$0, 110$\cdots$0, 11110$\cdots$0 series
437: shown in Fig. \ref{fig4}. These results clearly demonstrate the
438: benefit of having stackings with alternating electron-poor ``1''- and
439: electron-rich ``0''-units. For instance, 1010 is 20 meV/atom below
440: 1100, which implies that the boron layer in the 11 block of 1100,
441: surrounded by two Li layers on each side, does not take full advantage
442: of the available charge provided by Al. This behavior is consistent
443: with our previous observation that substitutional doping leads to the
444: filling of boron states only in the layer closest to the dopant
445: \cite{US3}.
446:
447: Note that the boundary of the convex hull
448: MS2-LiB$\leftrightarrow$C32-AlB$_2$ is determined by the 10$\cdots$0
449: series, which implies that there is a thermodynamic driving force for
450: the formation of ``1''-units in the ``0'' matrix. Indeed, doing
451: extrapolation over the 10$\cdots$0 series we find that replacement of
452: one Al layer via $n$AlB$_2$ + 2Li = Li$_2$Al$_{n-1}$B$_{2n}$ + Al is
453: an exothermic reaction with $\Delta E$ = 0.6 eV/(Li atom)
454: ($n\rightarrow\infty$). There is an alternative exothermic reaction
455: with an even bigger $\Delta E$ = 0.8 eV/(Li atom)
456: ($n\rightarrow\infty$) that preserves the C32 structure: $n$AlB$_2$ +
457: Li = LiAl$_{n-1}$B$_{2n}$ + Al. Therefore, one would expect to see
458: substitution of Al layers with Li first, and only at high Li
459: concentration should the ``1''-units start forming: in the limiting
460: case of 1:1 Li:Al composition the reaction LiAlB$_4$ + Li =
461: Li$_2$AlB$_4$ would be energetically favorable by $\Delta E$ = 0.4
462: eV/(Li atom). However, these reactions have apparently not been
463: observed due to the formation of the stable ternary LiAlB$_{14}$ and
464: the oxygen-containing phases \cite{LiAlB2}.
465:
466: In summary, based on our formation enthalpy calculations,
467: Li$_2$AlB$_4$ is marginally stable with respect to the considered
468: known compounds under normal conditions but can be stabilized by
469: hydrostatic pressure. Considering the stabilizing effect of Al we
470: expect a lower pressure threshold for the formation of the ternary
471: MS-Li-Al-B phases compared to the case of the binary MS-Li-B phases.
472:
473: More detailed studies should be carried out in the future to account
474: for the finite-temperature contributions in the Gibbs free energy;
475: this will be a difficult task because the boron-rich metal borides
476: have large unit cells with fractional occupancies. In addition, one
477: should consider not only compounds reported for a given ternary
478: system, but also crystal structure phases observed in similar
479: systems. For example, the absence of any stable Li-$M$-B phases ($M$ =
480: V, Nb, Ta) in the ICSD database \cite{ICSD} could simply be an
481: indication that these ternary systems have not been fully explored
482: experimentally yet. It would not be surprising then if attempts to
483: synthesize the predicted layered phases would lead to the formation of
484: phases not considered here, such as the metal-rich Li$_2$Pd$_3$B
485: superconductor comprised of linear chains of boron
486: \cite{Li2Pd3B}. High-throughput simulation of selected ternary systems
487: identified in our present work is subject of future studies.
488:
489: \subsection{IV. Electronic properties}
490:
491: %
492: \begin{figure}[t]
493: \begin{center}
494: % \vspace{-5mm}
495: \centerline{\epsfig{file=fig5.eps,width=80mm,clip=}}
496: % \vspace{-8mm}
497: \caption{ \small (Color online). Band structure and
498: partial density of states (PDOS) in 1010-Li$_2$AlB$_4$. The top
499: panels show the region near the Fermi level (0 eV).}
500: \label{fig5}
501: \end{center}
502: \end{figure}
503: %
504:
505: It is interesting to see what changes in properties the combination of
506: the non-superconducting $M$B$_2$ and potentially superconducting
507: MS-LiB structures would induce. Electronic properties of the proposed
508: MS ternary phases are examined for two representative
509: 1010-Li$_2$AlB$_4$ and 1010-Li$_2$TiB$_4$ compounds with hexagonal
510: unit cells (1010 and 10 phases are nearly degenerate in energy). Boron
511: layers in the 1010 structure are indistinguishable, but there are two
512: different sets of degenerate boron states derived from those in the
513: original ``0'' and ``1'' metal borides. For example, in MS2-LiB the
514: practically dispersionless $p\sigma$ states of boron are 1.0 eV above
515: the Fermi level \cite{US1,US2} while in C32-AlB$_2$ they are
516: completely filled [$E^{\sigma}_{\Gamma}$=-1.6 eV, $E^{\sigma}_A$=-1.0
517: eV]; in 1010-Li$_2$AlB$_4$ the two sets of boron $p\sigma$ states are
518: not completely filled, being 0.6 and 0.1 eV above the Fermi
519: level. This is a very satisfying outcome, because they still
520: contribute a considerable 0.042
521: states/(eV$\cdot$spin$\cdot$boron\,atom) to the density of states
522: (DOS) at the Fermi level (to be compared to 0.059 and 0.049
523: states/(eV$\cdot$spin$\cdot$boron\,atom) in MS2-LiB and C32-MgB$_2$,
524: respectively\cite{US2}). Because boron orbitals do not overlap across the
525: Li-filled portion of 1010-Li$_2$AlB$_4$, the $p\sigma$ states are not
526: dispersed and give rise to the desired, nearly cylindrical
527: two-dimensional Fermi surfaces similar to those of MS-LiB
528: \cite{US1,US2,Mazin}. In addition, the compound has $p\pi$ Fermi
529: surfaces, as the $p\pi$-bands crossing now happens at about 0.7 eV. The
530: $p\pi$ states DOS of 0.019 states/(eV$\cdot$spin$\cdot$ boron atom) is
531: a significant improvement with respect to MS2-LiB, which lacks these
532: states altogether. However, this contribution is still below the value
533: in MgB$_2$ (0.032 states/(eV$\cdot$spin$\cdot$ boron atom)) and,
534: unfortunately, comes at the expense of losing some $p\sigma$ DOS.
535:
536: \begin{figure}[t]
537: \begin{center}
538: % \vspace{-5mm} \
539: \centerline{\epsfig{file=fig6.eps,width=80mm,clip=}}
540: % \vspace{-8mm}
541: \caption{ \small (Color online). Band structure and partial
542: density of states (PDOS) in 1010-Li$_2$TiB$_4$. The top panels
543: show the region near the Fermi level (0 eV).}
544: \label{fig6}
545: \end{center}
546: \end{figure}
547: %
548:
549: \begin{figure}[htb]
550: \begin{center}
551: % \vspace{-4mm}
552: \centerline{\epsfig{file=fig7.eps,width=80mm,clip=}}
553: % \vspace{-8mm}
554: \caption{ \small (Color online). Frequency of the in-plane boron
555: phonons at M (open symbols) and $\Gamma$ (solid symbols) points
556: as a function of boron-boron bond length. The known C32-$M$B$_2$
557: diborides are shown as blue squares, the proposed MS2-LiB is
558: shown as red circles. The solid line shows a linear fit of the
559: M-point frequencies for the known diborides with metal valence
560: from III to V.}
561: % \vspace{-6mm}
562: \label{fig7}
563: \end{center}
564: \end{figure}
565: %\vspace{-5mm}
566: %
567:
568: \begin{figure}[htb]
569: \begin{center}
570: % \vspace{-6mm}
571: \centerline{\epsfig{file=fig8.eps,width=80mm,clip=}}
572: % \vspace{-8mm}
573: \caption{ \small (Color online). Frequency of the in-plane boron
574: phonons at M (open symbols) and $\Gamma$ (solid symbols) for
575: C32-$M$B$_2$ diborides of group I, II, and III-V metals. The
576: blue squares, black triangles, and red circles correspond to the
577: known, unstable, and proposed (11-LiB and 1010-Li$_2$$M$B$_4$)
578: compounds, respectively. The linear fits to the sets of M point
579: frequencies are labeled according to the valence of the metal.}
580: % \vspace{-7mm}
581: \label{fig8} \end{center}
582: \end{figure}
583: %\vspace{-5mm}
584: %
585:
586: Figure \ref{fig6} shows that boron in 1010-Li$_2$TiB$_4$ is too
587: overdoped to have any strong MgB$_2$-type electron-phonon
588: coupling. The bottom $p\sigma$ band at -1.5 eV is completely filled,
589: and even though the top band is still 0.3 eV above the Fermi level the
590: steep $\partial E/\partial k$ derivative on the narrow cylinder along
591: ${\Gamma}$-A results in a relatively low DOS of 0.010
592: states/(eV$\cdot$spin$\cdot$boron\,atom). 1010-Li$_2$HfB$_4$ exhibits
593: similar properties but has a slightly higher $p\sigma$ DOS of 0.017
594: states/(eV$\cdot$spin$\cdot$boron\,atom). 1010-Li$_2$$M$B$_4$
595: compounds with transition metals of valence V and higher present cases
596: of borides with completely filled boron $p\sigma$ states (see Table I).
597:
598: \subsection{V. Phonon softening}
599:
600: As previously discussed \cite{US3}, the next step in the analysis of
601: superconducting properties of the proposed materials should be an {\it
602: ab initio} calculation of the electron-phonon coupling
603: $\lambda$. However, the full scale calculations become prohibitively
604: expensive for compounds with large unit cells, and more so if they
605: have nested Fermi surfaces. In the following we attempt to obtain
606: information on the strength of the electron-phonon coupling from the
607: compounds' vibrational properties.
608:
609: Indeed, it has been previously argued that the unexpected decrease of
610: $\lambda$ in MS-LiB with respect to that in MgB$_2$ can be related to
611: the increase of the in-plane boron frequency and to the smaller
612: softening of this mode. \cite{US3,Mazin} The considerations were based
613: on the fact that for a zone-center optical phonon mode the softening
614: due to screening by metallic electrons, $\Delta\omega$, is connected
615: to the Fermi-surface average of the square of the electron-phonon
616: matrix element, $g$, through $\Delta\omega^2=-4\omega\langle
617: g^2\rangle N(0)$ ($N(0)$ is the electronic DOS at the Fermi level)
618: \cite{Mazin,Mazin2,Mazin18}.
619:
620: Unfortunately, there is no simple way to evaluate the unscreened
621: frequency in real materials. However, in MgB$_2$-type superconductors
622: the M-point phonons have wavevectors larger than $2k_F$ and are thus
623: screened much less than the ${\Gamma}$-point phonons. For this
624: reason we will use the $\omega_M - \omega_{\Gamma}$ difference to
625: reveal how the in-plane boron vibrations soften across a set of binary
626: and ternary layered metal borides. It should be stressed that the
627: above frequency difference cannot be treated as the real softening for
628: quantitative estimates of the electron-phonon coupling; nevertheless,
629: we expect it to be useful for selecting candidates with the strongest
630: electron-phonon coupling.
631:
632: For the C32-$M$B$_2$ diborides and MS2-LiB monoboride, we calculate
633: the frequencies of the in-plane boron mode at ${\Gamma}$ and the TO
634: mode at M points in the frozen-phonon approach using the fourth-order
635: corrections and the same settings as described in Refs.
636: \cite{Kunc,US1}. We neglect the anharmonic and non-adiabatic effects
637: since we are interested in general trends. Full phonon spectra in
638: existing metal diborides have been calculated previously
639: \cite{MgB2_Dispersion_1,MgB2_Dispersion_2}, and the general inverse
640: dependence of the $E_{2g}$ frequency at ${\Gamma}$ on the boron-boron
641: bond length has been pointed out \cite{MgB2_Dispersion_2}. In
642: Fig. \ref{fig7} we plot, as functions of the boron-boron bond length
643: $a_{B-B}$, the ${\Gamma}$- and M-point sets of calculated
644: frequencies. The latter exhibits a much more consistent linear
645: decrease with the bond length for all known metal diborides of groups
646: III-V. One can further discern two subsets (Al, Hf, Zr, Y and the
647: rest) which have ${\Gamma}$-point frequencies following the linear
648: dependence with similar negative slopes (we see no apparent factor
649: which could be responsible for this behavior). Note that for the
650: electron-rich metal diborides $\omega_M$ is below $\omega_{\Gamma}$,
651: which illustrates why the former cannot be considered as the
652: unscreened frequency. However, the calculated values of ${\Gamma}$-
653: and M-point frequencies will be useful as a reference when we examine
654: the vibrational properties of the ternary MS phases derived from the
655: corresponding metal diborides.
656:
657: As expected, MgB$_2$ stands out as a compound with a large softening
658: of the frequency at ${\Gamma}$. In addition, Fig. \ref{fig7} shows
659: that the M-point frequency does not follow the general trend of
660: group-III-V metal diborides either; this behavior cannot be directly
661: related to the electron-phonon coupling with the $p\sigma$ boron
662: states. The phenomenon is explained in Figure \ref{fig8} where we plot
663: calculated frequencies for a large set of group-I-II metal
664: diborides. The M-point frequency $\omega_M$ for divalent (Be, Mg, Ca)
665: and monovalent (Li, Cu, Au, Ag, Na, K) metals shows similar and nearly
666: perfect linear dependences on the boron-boron bond length. One can
667: expect to have the same trends for the unscreened frequency, only it
668: would be shifted up by a constant offset.
669:
670: These results give a new perspective on the relation between the
671: vibration properties and the strength of the electron-phonon coupling
672: in LiB discussed in Refs. \cite{Mazin,US3}. As a boride isovalent to
673: MgB$_2$, LiB follows the $\omega_M(a_{B-B})$ dependence obtained for
674: the Be-Mg-Ca series of diborides shown in Fig. \ref{fig8}.
675: Therefore, one of the key factors in the LiB hardening of the in-plane
676: frequency [with respect to that in MgB$_2$] is the shortening of the
677: boron-boron bond length, which depends on the metal and the particular
678: filling of the $p\pi$ and $p\sigma$ states of boron. Hence, it follows
679: that the frequency softenings in LiB and MgB$_2$ are, in fact,
680: comparable.
681:
682: Finally, we can comment on the superconducting properties of the
683: proposed ternary compounds. The introduction of electron-rich metals
684: (group IV and higher) into 1010-Li$_2$$M$B$_4$ overdopes the boron
685: layers, and causes the vibrational properties to become closer to
686: those of the corresponding diborides: $\omega_{\Gamma}$ is still above
687: $\omega_M$ for Ti and Hf (Fig. \ref{fig8}). 1010-Li$_2$AlB$_4$ is a
688: more promising MgB$_2$-type superconducting material, judging by the
689: pronounced softening of the in-plane boron mode (note that in AlB$_2$
690: $\omega_{\Gamma}$ is harder than $\omega_M$). Unfortunately, aluminum
691: induces the shortening of the bond length and a substantial increase
692: of the in-plane boron frequency, which will likely weaken the
693: electron-phonon coupling \cite{Mazin,US3} (the next section is devoted
694: to a more accurate description of this compound). As for the proposed
695: MS phases at other compositions, we find that all ternary systems with
696: negative relative stability for 1010-Li$_2$$M$B$_4$ (Fig. \ref{fig2})
697: have negative relative stability for 110-Li$_4$$M$B$_6$ as well (the
698: latter are typically 2-7 meV/atom above the line connecting 11-LiB
699: and 1010-Li$_2$$M$B$_4$ and are only metastable). Therefore, if
700: 110-Li$_4$$M$B$_6$ could still be synthesized with large in size
701: transition metals, such as Hf or Ta, the superconducting properties of
702: the boron layer in the ``11'' portion (which should be well isolated
703: by the double layers of lithium from the electron-rich ``0''-unit)
704: would be enhanced due to the stretching of the overall boron bond
705: length in the compound. A similar phenomenon has been recently
706: observed in stretched MgB$_2$ films \cite{MgB2_stretched_films}. A
707: recent Raman study of the vibrational and superconducting properties
708: of doped MgB$_2$ has revealed that the adjustment of the frequency
709: alone might not be enough to achieve a higher $T_c$ \cite{Raman_MgB2}.
710:
711: The trend shown in Figs. \ref{fig7}-\ref{fig8} is corroborated by the
712: well known correlation between the increase of the electron-phonon
713: coupling and the inevitable dynamical destabilization of the structure
714: \cite{Correlation_Testardi,Pickett_review,Correlation_Cordero,Correlation_BaKBiO3_1,Correlation_BaKBiO3_2}
715: (a too strong renormalization can cause a Peierls-type distortion, a
716: band Jahn-Teller transformation or a structural transition
717: \cite{Pickett_review}). In our considered metal boride prototypes, the
718: specific relation between stability and superconductivity originates
719: from the subtle effects associated with the filling of binding states
720: of boron.
721:
722: The focus of our study has been the MgB$_2$-type
723: superconductors, while there have been reports on Nb-deficient
724: superconductors based on C32-NbB$_2$ with $T_c$ of nearly 10
725: K \cite{MB2,NbxB2}. Indeed, it would be interesting to explore the
726: superconducting potential of Li$_2$$M$B$_4$ compounds ($M$ =
727: V,Nb,Ta) which would have a non-MgB$_2$-type electron-phonon
728: coupling mechanism.
729:
730: \subsection{VI. Phonon spectrum and electron-phonon coupling.}
731:
732: In the previous sections we have shown that Li$_2$AlB$_4$ has a
733: significant amount of boron $p\sigma$ and $p\pi$ DOS at the Fermi
734: level and exhibits a characteristic softening of the in-plane boron
735: phonon mode at ${\Gamma}$. All these quantities are lower compared to
736: those in MgB$_2$ so that one can expect the superconducting properties
737: of Li$_2$AlB$_4$ to be less appealing. However, a full calculation of
738: the electron-phonon coupling (for all the modes and all {\bf k}-points
739: in the Brillouin zone) is needed to understand the superconducting
740: properties of Li$_2$AlB$_4$. For example, in MgB$_2$ the intercalant
741: modes are weakly coupled; based on our results for LiB \cite{US3} the
742: situation could be different in Li$_2$AlB$_4$. In addition, due to
743: symmetry constraints in calculations on periodic systems, energy
744: minimization might not guarantee convergence to an equilibrium free of
745: dynamical instabilities (imaginary phonon frequency at ${\bf
746: q}\ne{\Gamma}$). Thus, dynamical stability has to be checked by
747: explicit calculation of the phonon frequencies in the whole BZ. In
748: this section we calculate vibrational phonon frequencies and the
749: electron-phonon coupling of Li$_2$AlB$_4$ using density functional
750: theory in the linear response approach \cite{BaroniRMP2001}.
751:
752: We find \cite{Phonon_Details} that the theoretically devised structure
753: of Li$_2$AlB$_4$ is dynamically stable with no imaginary
754: phonon-frequencies. The obtained phonon density of states (PHDOS) is
755: plotted in Fig. \ref{fig9}. As it can be seen from the decomposition
756: along selected cartesian vibrations, B modes are dominant at energies
757: larger than 50 meV. In the low energy region ($< 50$ meV), two clear
758: peaks are seen, one at $\approx 40$ meV due to Li vibrations and the
759: other one at $30 $ meV due to Al vibrations. As expected from the
760: large interlayer spacing between the B-layers, B-modes are flat in the
761: direction perpendicular to the B-layers.
762:
763: The electron-phonon coupling
764: $\lambda_{{\bf q}\nu}$ for a phonon mode
765: $\nu$ with momentum ${\bf q}$ is:
766: \begin{equation}\label{eq:elph}
767: \lambda_{{\bf q}\nu} = \frac{4}{\omega_{{\bf q}\nu}N(0) N_{k}} \sum_{{\bf k},n,m}
768: |g_{{\bf k}n,{\bf k+q}m}^{\nu}|^2 \delta(\epsilon_{{\bf k}n}) \delta(\epsilon_{{\bf k+q}m})
769: \end{equation}
770: where the sum is over the Brillouin Zone.
771: The matrix element is
772: $g_{{\bf k}n,{\bf k+q}m}^{\nu}= \langle {\bf k}n|\delta V/\delta u_{{\bf q}\nu} |{\bf k+q} m\rangle /\sqrt{2 \omega_{{\bf q}\nu}}$,
773: where $u_{{\bf q}\nu}$ is the amplitude of the displacement of the phonon,
774: $V$ is the Kohn-Sham potential and $N(0)$ is the electronic density of states
775: at the Fermi level.
776: The calculated average electron-phonon coupling is
777: $\lambda=\sum_{{\bf q}\nu} \lambda_{{\bf q}\nu}/N_q\approx 0.41$
778: ($N_{k}$ and $N_{q}$ are the ${\bf k}$-space and ${\bf q}$-space mesh dimensions, respectively \cite{Phonon_Details}).
779:
780: \begin{figure}[htb]
781: \begin{center}
782: \vspace{-2mm}
783: \centerline{\epsfig{file=fig9.eps,width=80mm,clip=}}
784: % \vspace{-5mm}
785: \caption{ \small (Color online). Phonon Density of States (PHDOS) decomposed
786: over selected vibrations, Eliashberg function $\alpha^2F(\omega)$
787: and integrated Eliashberg function $\lambda(\omega)$ in Li$_2$AlB$_4$.}
788: \label{fig9}
789: % \vspace{-5mm}
790: \end{center}
791: \end{figure}
792: \vspace{-7mm}
793:
794: The Eliashberg function
795: \begin{equation}
796: \alpha^2F(\omega)=\frac{1}{2 N_q}\sum_{{\bf q}\nu} \lambda_{{\bf q}\nu} \omega_{{\bf q}\nu} \delta(\omega-\omega_{{\bf q}\nu} )
797: \end{equation}
798: and the integral $\lambda(\omega)=2 \int_{0}^{\omega} d\omega^{\prime}
799: \alpha^2F(\omega^{\prime})/\omega^{\prime}$
800: are shown in Fig. \ref{fig9}.
801: As can be seen, the contribution of the coupling due to in-plane vibration is
802: substantially reduced with respect to MgB$_2$. This confirm what could be qualitatively
803: inferred from the softening calculated in Fig. \ref{fig8}.
804: On the contrary the coupling to intercalant modes is not negligible.
805:
806: The critical superconducting temperature is estimated using the McMillan
807: formula \cite{mcmillan}:
808: \begin{equation}
809: T_c = \frac{\langle \omega \rangle_{log}}{1.2}\, \exp \left[{ - \frac{1.04 (1+\lambda)}{\lambda-\mu^* (1+0.62\lambda)}}\right]
810: \label{eq:mcmillan}
811: \end{equation}
812: where $\mu^*$ is the screened Coulomb pseudopotential and
813: \begin{equation}
814: \langle\omega\rangle_{log} = e^{\frac{2}{\lambda}\int_{0}^{+\infty}
815: \alpha^2F(\omega)\log(\omega)/\omega\,d\omega }
816: \end{equation}
817: the phonon frequencies logarithmic average. We obtain $\langle\omega\rangle_{log}=59.4$ meV
818: leading to $T_c$ of approximately 3.6 K for $\mu^{*}=0.1$.
819: This value might be enhanced by multiband effects.
820:
821: \subsection{VII. Summary}
822:
823: We have demonstrated that there is a number of Li-$M$-B systems, in
824: which MS ternary borides gain in stability with respect to MS2-LiB and
825: C32-$M$B$_2$. The finding suggests that these potential
826: superconductors could be synthesized under normal conditions or grown
827: on metal diboride substrates with a matching lattice constant. We show
828: that some boron states in the MS ternary compounds [derived from those
829: in MS2-LiB and C32-$M$B$_2$] are still hole-doped and could give rise
830: to the MgB$_2$-type superconductivity. In order to pre-select
831: compounds with the strongest electron-phonon coupling we examine the
832: softening of the in-plane boron phonon mode in a large class of metal
833: borides. We find a very well defined correlation between the frequency
834: of this mode and both the boron-boron bond length and the valence of
835: the metal. This observation allows us to give a new interpretation of
836: the hardening of the in-plane boron phonon mode in MS-LiB and identify
837: one compound, Li$_2$AlB$_4$, that should be a superconductor with
838: $T_c$ of about 3.6 K.
839:
840: We thank R. Margine, L. Boeri, W. Setyawan, M. Mehl, A. Liu and
841: I. Mazin for valuable discussions. We acknowledge the Teragrid-TACC
842: center, the Pennsylvania State supercomputer center, and the IDRIS
843: supercomputing center (project 081202) for computational
844: support. Research supported by ONR (N00014-07-1-0878) and NSF
845: (DMR-0639822).
846:
847: \vspace{-3mm}
848: \begin{thebibliography}{99}
849: \vspace{-3mm}
850:
851: \bibitem{origin}
852: J. Nagamatsu, N. Nakagawa, T. Muranaka, Y. Zenitani, and J. Akimitsu,
853: Nature {\bf 410}, 63 (2001).
854:
855: \bibitem{band_filling}
856: J. Kortus, O. V. Dolgov, R. K. Kremer, and A. A. Golubov,
857: Phys. Rev. Lett. {\bf 94} 027002 (2005) and references therein.
858:
859: \bibitem{CaAlSi}
860: %Superconducting properties of single-crystalline Ca(Al-0.5,Si-0.5)(2): A ternary silicide with the AlB2-type structure
861: M. Imai, E. Abe, J. Ye, K. Nishida, T. Kimura, K. Honma, H.
862: Abe, and H. Kitazawa, Phys. Rev. Lett. {\bf 87}, 077003 (2001).
863:
864: \bibitem{TaB2}
865: H. Rosner, W. E. Pickett, S.-L. Drechsler, A. Handstein, G. Behr,
866: G. Fuchs, K. Nenkov, K.-H. Muller, and H. Eschrig, PRB {\bf 64},
867: 144516 (2001)
868:
869: \bibitem{MB2}
870: A. Yamamoto, C. Takao, T. Masui, M. Izumi, and S. Tajima, Physica C {\bf 383}, 197 (2002).
871:
872: \bibitem{NbxB2}
873: Z.-A. Ren, S. Kuroiwa, Y. Tomita and J. Akimitsu, Physica C: Superconductivity {\bf 468}, 411 (2008)
874:
875: \bibitem{LiBC}
876: H. Rosner, A. Kitaigorodsky, and W.E. Pickett, Phys. Rev. Lett. {\bf 88}, 127001 (2002).
877:
878: \bibitem{LixBC}
879: A. M. Fogg, J. B. Claridge, G. R. Darling, and M. J. Rosseinsky, Chem. Commun. (Cambridge) {\bf 12} 1348 (2003);
880: A. M. Fogg, J. Meldrum, G. R. Darling, J. B. Claridge, and M. J. Rosseinsky, J. Am. Chem. Soc. {\bf 128}, 10043 (2006).
881:
882: \bibitem{SCGIC1}
883: T. E. Weller, M. Ellerby, S. S. Saxena, R. P. Smith, and N. T. Skipper, Nature Phys. {\bf 1}, 39 (2005).
884:
885: \bibitem{SCGIC2}
886: N. Emery, C. Herold, M. d'Astuto, V. Garcia, C. Bellin, J. F. Mareche, P. Lagrange, and G. Loupias,
887: Phys. Rev. Lett. {\bf 95}, 087003 (2005).
888:
889: \bibitem{SCGIC3}
890: A. Gauzzi, {\it et al.}, Phys. Rev. Lett. {\bf 98}, 067002 (2007)
891:
892: \bibitem{US1}
893: A. N. Kolmogorov and S. Curtarolo, Phys. Rev. B {\bf 73}, 180501(R) (2006).
894:
895: \bibitem{MgC}
896: P. Zhang, S. Saito, S. G. Louie, and M. L. Cohen, Phys. Rev. B {\bf 77}, 052501 (2008).
897:
898: \bibitem{ICSD}
899: Inorganic crystal structure database, http://icsd.ill.fr/icsd/index.html
900:
901: \bibitem{PAULING}
902: P. Villars, K. Cenzual, J. L. C. Daams, F. Hulliger, T. B. Massalski, H. Okamoto, K. Osaki, A. Prince, and S. Iwata,
903: Crystal Impact, {\it Pauling File. Inorganic Materials Database and Design System},
904: Binaries Edition, ASM International, Metal Park, OH (2003).
905:
906: \bibitem{Oguchi}
907: T. Oguchi, J. Phys. Soc. Jpn. {\bf 71}, 1495 (2002).
908:
909: \bibitem{dope_review}
910: R.J. Cava, H.W. Zandbergen, and K. Inumaru, Physica C {\bf 385}, 8 (2003).
911:
912: \bibitem{MgB2_defects}
913: F. Bernardini and S. Massidda, Europhys. Lett. {\bf 76}, 491 (2006).
914:
915: \bibitem{MgAlLiB1}
916: F. Bernardini and S. Massidda,
917: Phys. Rev. B {\bf 74}, 014513 (2006).
918:
919: %\bibitem{MgBLiC}
920: %{\sf we should put this one, see email, if you agree let`s find the real citation}
921: %http://arxiv.org/abs/0704.3526
922:
923: \bibitem{Palnichenko}
924: A. V. Palnichenko, O. M. Vyaselev, and N. S. Sidorov, JETF Lett. {\bf 86}, 272 (2007).
925:
926: \bibitem{Chepulskyy}
927: R. V. Chepulskii, I. I. Mazin, and S. Curtarolo,
928: {\it First-principles search for potential high temperature superconductors
929: in the Mg-B-A (A=alkali and alkaline earth metals) system with high boron content}, (2008).
930:
931: \bibitem{US2}
932: A. N. Kolmogorov and S. Curtarolo,
933: Phys. Rev. B {\bf 74}, 224507 (2006).
934:
935: \bibitem{Mazin}
936: A. Y. Liu and I. I. Mazin,
937: Phys. Rev. B {\bf 75}, 064510 (2007).
938:
939: \bibitem{US3}
940: M. Calandra, A. N. Kolmogorov, and S. Curtarolo,
941: Phys. Rev. B {\bf 75}, 144506 (2007).
942:
943: \bibitem{US4}
944: A. N. Kolmogorov, R. Drautz, D. G. Pettifor, Phys. Rev. B {\bf 76}, 184102 (2007),
945:
946: \bibitem{MgAlB4}
947: J. Q. Li, L. Li, F. M. Liu, C. Dong, J. Y. Xiang, and Z. X. Zhao,
948: Phys. Rev. B {\bf 65}, 132505 (2002).
949:
950: \bibitem{Curtarolo_DMQC}
951: S. Curtarolo, D. Morgan, K. Persson, J. Rodgers, and G. Ceder, Phys. Rev. Lett. {\bf 91}, 135503 (2003).
952:
953: \bibitem{Curtarolo_JMEAS}
954: D. Morgan, G. Ceder, and S. Curtarolo, Meas. Sci. Technol. {\bf 16}, 296-301 (2005).
955:
956: \bibitem{Curtarolo_CALPHAD}
957: S. Curtarolo, D. Morgan, and G. Ceder, Calphad {\bf 29}, 163-211 (2005).
958:
959: \bibitem{bloechl994}
960: P. E. Blochl,
961: Phys. Rev. B {\bf 50}, 17953 (1994).
962:
963: \bibitem{PBE}
964: J. P. Perdew, K. Burke, and M. Ernzerhof,
965: Phys. Rev. Lett. {\bf 77} 3865 (1996).
966:
967: \bibitem{kresse1993}
968: G. Kresse and J. Hafner, Phys. Rev. B {\bf 47}, 558 (1993).
969:
970: \bibitem{kresse1996b}
971: G. Kresse and J. Furthmuller,
972: Phys. Rev. B {\bf 54}, 11169 (1996).
973:
974: \bibitem{US_PAW}
975: G. Kresse and D. Joubert,
976: Phys. Rev. B {\bf 59}, 1758 (1999).
977:
978: \bibitem{MONKHORST_PACK}
979: J. D. Pack and H. J. Monkhorst, {Phys. Rev. B} {\bf 13}, 5188 (1976);
980: {\bf 16}, 1748 (1977).
981:
982: \bibitem{US2_note1}
983: Figure 3 of Ref \cite{US2}.
984:
985: \bibitem{US2_note2}
986: Figure 2 of Ref \cite{US2}.
987:
988: \bibitem{LiAl1}
989: M. Sluiter, D. de Fontaine, X. Q. Guo, R. Podloucky, and A. J. Freeman,
990: Phys. Rev. B {\bf 42}, 10460 (1990).
991:
992: \bibitem{LiAl2}
993: M.H.F. Sluiter, Y. Watanabe, D. de Fontaine, and Y. Kawazoe,
994: Phys. Rev. B {\bf 53}, 6137 (1996).
995:
996: \bibitem{LiAlB14}
997: T. Ito and I. Higashi, Acta Cryst. B {\bf 39}, 239 (1983).
998:
999: \bibitem{LiAlB1}
1000: T. Sun {\it et al. }
1001: J. Superconductivity {\bf 17}, 473 (2004).
1002:
1003: \bibitem{POLY}
1004: Consideration of these binary Li-B and Al-B compounds with large unit
1005: cells and fractional occupancies of atomic sites is beyond the scope
1006: of the present study.
1007:
1008: \bibitem{LiAlB2}
1009: M. Monni {\it et al. }
1010: Phys. Rev. B {\bf 73}, 214508 (2006).
1011:
1012: \bibitem{LiB_Wang78}
1013: F.E. Wang, M.A. Mitchell, R.A. Sutula, and J.R. Holden, J. Less-Common Metals {\bf 61}, 237 (1978).
1014:
1015: \bibitem{LiB_B6}
1016: M. W\"{o}rle, R. Nesper, G. Mair, and H.G. von Schnering, Solid State Science {\bf 9}, 459 (2007) and references therein.
1017:
1018: \bibitem{Li2Pd3B}
1019: %Superconductivity in the metal rich Li-Pd-B ternary boride
1020: K. Togano, P. Badica, Y. Nakamori, S. Orimo, H. Takeya, and K. Hirata, Phys. Rev. Lett. {\bf 93}, 247004 (2004).
1021:
1022: \bibitem{Mazin2}
1023: I. I. Mazin and V. P. Antropov,
1024: Physica C {\bf 385}, 49 (2003).
1025:
1026: \bibitem{Mazin18}
1027: C. O. Rodriguez {\it et al. }
1028: Phys. Rev. B {\bf 42}, 2692 (1990).
1029:
1030: \bibitem{Kunc}
1031: K. Kunc, I. Loa, K. Syassen, R. K. Kremer, and K. Ahn, J. Phys.: Condens. Matt. {\bf 13}, 9945-9962 (2001).
1032:
1033: \bibitem{MgB2_Dispersion_1}
1034: K.-P. Bohnen, R. Heid, and B. Renker, Phys. Rev. Lett. {\bf 86}, 5771 (2001).
1035:
1036: \bibitem{MgB2_Dispersion_2}
1037: R. Heid, B. Renker, H. Schober, P. Adelmann, D. Ernst, and K. -P. Bohnen, Phys. Rev. B {\bf 67}, 180510(R) (2003).
1038:
1039: \bibitem{MgB2_stretched_films}
1040: A. V. Pogrebnyakov {\it et al. }
1041: Phys. Rev. Lett. {\bf 93}, 147006 (2004).
1042:
1043: \bibitem{Raman_MgB2}
1044: %Raman study of element doping effects on the superconductivity of MgB2
1045: W. X. Li, Y. Li, R. H. Chen, R. Zeng, S. X. Dou, M. Y. Zhu, and H. M. Jin,
1046: Phys. Rev. B {\bf 77}, 094517 (2008).
1047:
1048: \bibitem{Correlation_Testardi}
1049: L.R. Testardi,
1050: in {\it Physical Acoustics}, edited by W.P. Mason and R.N. Thurston (Academic, New York, 1973).
1051:
1052: \bibitem{Pickett_review}
1053: W.E. Pickett,
1054: % The next breakthrough in phonon-mediated superconductivity
1055: Physica C {\bf 468}. 126 (2008).
1056:
1057: \bibitem{Correlation_Cordero}
1058: F. Cordero, R. Cantelli, G. Giunchi and S. Ceresara
1059: %Search for incipient lattice instabilities in MgB2 by anelastic spectroscopy
1060: Phys. Rev. B {\bf 64}, 132503 (2001).
1061:
1062: \bibitem{Correlation_BaKBiO3_1}
1063: S. Zherlitsyn {\it et al. }
1064: %, , B. L<FC>thi, V. Gusakov, B. Wolf, F. Ritter, D. Wichert, S. Barilo, S. Shiryaev, C. Escribe-Filippini, and J.L. Tholence,
1065: %Structural instability and superconductivity in Ba/sub 1-x/K/sub x/BiO/sub 3/
1066: Eur. Phys. J. B {\bf 16}, 59 (2000).
1067:
1068: \bibitem{Correlation_BaKBiO3_2}
1069: M. Braden {\it et al. }
1070: %, ,, W. Reichardt, E. Elkaim, J.P. Lauriat, S. Shiryaev, and S.N. Barilo,
1071: % Structural distortion in superconducting Ba1-xKxBiO3
1072: Phys. Rev. B {\bf 62}, 6708 (2000).
1073:
1074: \bibitem{BaroniRMP2001}
1075: S. Baroni, S. de Gironcoli, A. Dal Corso, and P. Giannozzi,
1076: %{\it Phonons and related crystal properties from density-functional perturbation theory},
1077: Rev. Mod. Phys. {\bf 73}, 515 (2001).
1078:
1079: \bibitem{Phonon_Details}
1080: We use the Quantum-espresso code (www.pwscf.org),
1081: norm-conserving pseudopotentials, the generalized gradient approximation (PBE) and a 60 Rydberg cutoff for the
1082: kinetic energy. We perform the electronic integration using
1083: $8\times8\times8$ {\bf k}-point mesh. The dynamical matrices were computed
1084: on a $N_q=4\times4\times4$ {\bf q}-point mesh centered at ${\Gamma}$. The electron-phonon
1085: coupling was computed using a $N_k=30\times30\times30$ {\bf k}-point mesh.
1086:
1087: \bibitem{mcmillan}
1088: W. L. McMillan, Phys. Rev. {\bf 167}, 331 (1968).
1089:
1090: \end{thebibliography}
1091:
1092: \end{document}
1093:
1094: