0810.1170/AMR.tex
1: \documentclass[aps,prl,showpacs,twocolumn,a4paper,unsortedaddress,amsmath,amssymb,byrevtex]{revtex4}
2: 
3: \usepackage[latin1]{inputenc}
4: \usepackage{graphicx}
5: \usepackage{color}
6: \usepackage{dcolumn}
7: \usepackage{bm}
8: \usepackage{hyperref}
9: \usepackage{times}
10: 
11: \begin{document}
12: 
13: \title{Huge anisotropic magneto-resistance in iridium atomic chains}
14: 
15: \author{V. M. Garc\'{\i}a-Su\'arez$^2$}
16: \author{D. Zs. Manrique$^{1,2}$}
17: \author{C. J. Lambert$^2$}
18: \author{J. Ferrer$^{1,*}$}
19: 
20: \affiliation{$^1$  Departamento de F\'{\i}sica, Universidad de Oviedo, 33007 Oviedo, Spain}
21: \affiliation{$^2$ Departament of Physics, Lancaster University, Lancaster LA1
22: 4YB, UK}
23: \affiliation{$^*$ e-mail: ferrer@condmat.uniovi.es.}
24: \date{\today}
25: 
26: \begin{abstract}
27: We analyze in this article the magneto-resistance ratio of finite
28: and infinite iridium and platinum chains. Our calculations, that
29: are based on a combination of non equilibrium Green function
30: techniques and density functional theory, include a fully
31: self-consistent treatment of non-collinear magnetism and of the
32: spin-orbit interaction. They  indicate that, in addition to having
33: an extremely large magnetic anisotropy that may overcome the super-paramagnetic
34: limit, infinite and also
35: realistic finite-length iridium chains show sizeable anisotropic
36: magnetoresistance ratios. We therefore propose iridium
37: nanostructures as promising candidates for nanospintronics logic
38: devices.
39: \end{abstract}
40: 
41: \pacs{85.75.-d,73.63.Nm,75.30.Gw}
42: 
43: \maketitle
44: 
45: % BODY
46: The ability to enhance and tailor at the same time the magnetic anisotropy and
47: magneto-resistance of atomic-sized magnetic bits and junctions
48: will determine whether nanospintronics will be a viable
49: technology. Cobalt clusters and chains with magnetic
50: anisotropy barriers of the order of 1 meV were fabricated a few 
51: years ago by deposition on substrates of 5d elements\cite{gambardella1,gambardella2}.
52: Furthermore, several
53: recent theoretical predictions have pointed out that atomic
54: clusters and chains made of 4d elements should have even higher
55: magnetic anisotropy barriers than Co nanostructures\cite{blugel06,so-clusters,Carlo}. 
56: In the case of 5d
57: elements like platinum and iridium, calculations predict that the
58: anisotropies of iridium and platinum atomic clusters and chains
59: can be even larger than room-temperature thermal energies, so that
60: these nanostructures can overcome the technologically relevant
61: superparamagnetic limit\cite{so-clusters,predictions,Tos08}. 
62: 
63: Suspended chains of gold atoms connecting two gold electrodes were 
64: fabricated using the scanning tunneling microscope
65: and mechanically controllable break junction (MCBJ)
66: techniques\cite{ohn,yan}, where a quantized conductance close to
67: $G_0=2\,e^2/h$ was measured. Subsequent experiments found that
68: platinum and iridium also form suspended
69: chains, but that other elements like Ni, Co, Pd or Rh  do not
70: \cite{smit, Smit-thesis}, unless doped with oxygen or other
71: elements\cite{thijssen}. It was also demonstrated that longer
72: atomic chains could be made by deposition on stepped
73: surfaces\cite{gambardella1}, or by encapsulation in carbon
74: nanotubes\cite{mildred}.
75: 
76: The reduced dimensionality of nanometric objects increases their
77: tendency towards magnetism. Ugarte and coworkers proposed that
78: their experimental conductance data for platinum chains could be
79: explained provided that these were magnetic\cite{ugarte}.
80: Subsequent theoretical work indeed found that 4d and 5d infinite
81: linear chains of bulk-paramagnetic materials do become
82: magnetic\cite{Delin1,Delin2}. Very recently, several groups have
83: found that the magnetic anisotropy energy (MAE) per atom of
84: infinite zigzag platinum and iridium chains\cite{predictions}, and
85: of linear platinum chains\cite{Tos08} show values as large as 50
86: meV to 100 meV when stretched, which correspond to equivalent
87: temperatures of 500K to 1000 K. This suggests that the blocking
88: temperature $T_B$ of such nanostructures will attain a similar
89: value and therefore they should overcome the
90: superparamagnetic limit, so that their magnetization would be
91: finite at room temperature.
92: 
93: We draw the attention in this article to iridium and
94: platinum atomic structures as promising candidates for
95: nanospintronics. We show that iridium atomic chains not only
96: have huge magnetic anisotropies, but also meet the
97: requirement of having a significant magneto-resistance ratio.
98: The simulations presented here are based on density functional
99: theory\cite{Koh65} as implemented in the SIESTA code\cite{SIESTA},
100: and include a fully self-consistent and spin non-collinear
101: implementation of the spin-orbit interaction\cite{Lucas}. This
102: spin-orbit implementation has been successfully used to compute
103: the magnetic anisotropy of infinite Ir and Pt atomic
104: chains\cite{predictions} and of transition metal atomic
105: clusters\cite{so-clusters}, in agreement with the results of other
106: simulations\cite{blugel06,Carlo,Tos08}. The conductance
107: calculations were carried out with our spin non-collinear code
108: SMEAGOL\cite{Roc06}, which includes the fully self-consistent 
109: implementation of the spin-orbit interaction referred above.
110: We note that SMEAGOL has been
111: shown to provide conductance data that compare rather accurately
112: with the available experimental data on platinum
113: chains\cite{Gar05,Smit-thesis,smit}. The simulations have used the
114: Local Density Approximation\cite{Per81} and norm-conserving
115: pseudopotentials\cite{Tro91}. The atomic parameters that were fed
116: in the pseudopotential generator, as well as the basis set used
117: are similar to those reported in references
118: \cite{Gar05,predictions,so-clusters}.
119: 
120: Figure 1 (c) shows a schematic view of the experimental setup of a
121: MCBJ experiment, where at the last stages before breakage of a Pt
122: or Ir strip, a short suspended atomic chain is formed. In these
123: experiments, an electric current is applied to the circuit, and
124: the conductance $G$ is measured many times as the chain elongates.
125: The whole setup can also be understood as a magnetic junction that connects 
126: two paramagnetic electrodes. Recent work has shown that theory can  
127: reproduce accurately the experimental conductance data of finite-length platinum
128: chains\cite{Gar05}, but has not found significant differences
129: between the paramagnetic and the magnetic cases\cite{Pal05}. 
130: 
131: We show the energy of linear platinum and iridium chains as a function of $d_z$
132: in Figs. 2(a) and 3(a). 
133: It is important to notice that these chains are metastable. Indeed, allowing for 
134: lateral displacements of the atoms yields the energy curves in Figs. 2(b) and 3(b),
135: that only have minima for ladder or zigzag arrangements similar to those shown 
136: in Figs. 1(a) and (b). Furthermore, relaxation of the forces always finishes in 
137: zigzag or ladder chains, unless the atoms are constrained to lie in a linear array.
138: Actually, ladders are energetically more stable than zigzag geometries. But the relaxation
139: of the forces in a MCBJ geometry, like the one shown in Fig. 1(c) only leads to zigzag
140: arrangements, that gradually straighten as the electrodes are pulled apart.
141: Related to this effect, we note that infinite platinum (iridium) zigzag chains become linear only  
142: for distances $d_z$ longer than 2.55 \AA~ (2.40 \AA), since at these
143: distances, the zigzag angles fall abruptly to zero\cite{predictions}. These values of
144: $d_z$ are marked by a dotted vertical line in the right panels of
145: Figures 2 and 3. We will therefore only present results for the conductance of
146: infinite zigzag and linear, as well as finite zigzag, chains in this article. 
147: 
148: \begin{figure}
149: \includegraphics[width=0.7\columnwidth]{Fig1.ps}
150: \caption{(Color online) The geometry of the chains simulated in this
151: article, referred to the three coordinate axes. (a) Ladder
152: geometry that corresponds to the absolute minimum of the infinite
153: chains, see the top right panel in Figures 2 and 3; (b) zigzag
154: geometry that corresponds to the second minimum of the infinite
155: chains, see the top right panels in Figures 2 and 3; (c) schematic
156: view of the geometry of the relaxed finite-length chains in
157: contact to two electrodes. }
158: \end{figure}
159: 
160: 
161: \begin{figure}
162: \includegraphics[width=\columnwidth]{Fig2.ps}
163: \caption{(color online) Results of the simulations of infinite platinum
164: chains as a function of the distance $d_z$ defined in Figure 1 for
165: orientations of the spins along the $=x$, $y$ and $z$ axes. 
166: The left panels correspond to the case of metastable linear
167: chains. The
168: right panels corresponds to the geometries (b) and (c) in Figure 1.
169: The dashed vertical line indicates the
170: distance $d_z$ at which the zigzag chains cease to be stable.
171: (a) Energies $E_{x,y,z}$ per atom; (b) magnetic anisotropy
172: energies per atom $\Delta_{x,y}$; (c) spin moments per atom; (d) conductance. }
173: \end{figure}
174: 
175: We have performed separate
176: simulations for orientations of the atomic spins along the chain
177: axis ($z$- or easy-axis), perpendicular to it, but still in the
178: zigzag plane ($x$-axis) and perpendicular to the zigzag plane
179: ($y$-axis), but the energies $E_{x,y,z}$  can not be discriminated in the
180: energy scale used in Figs. 2(a,b) and 3(a,b). We have therefore
181: computed\cite{predictions} the energy differences
182: $\Delta_{x,y}=E_z-E_{x,y}$ per atom, that we define as the MAEs, and
183: plotted them in Figs. 2(c,d) and 3(c,d) for platinum and iridium 
184: respectively. Our definition implies that when both $\Delta_{x,y}$ are 
185: negative, the magnetization lies parallel to the easy axis. On the contrary,
186: when $\Delta_{x,y}$ is positive, the magnetization is oriented perpendicular to the
187: easy axis, and aligns either parallel to the $x$- or the $y$-axes,
188: depending on whether $\Delta_x$ is larger or smaller than $\Delta_y$, respectively.
189: Notice that our results for linear
190: platinum chains (Fig. 2(c)) are similar to those presented in Ref.
191: \cite{Tos08}. More interestingly, the anisotropy of zigzag
192: platinum and iridium chains is $x$-axis for short $d_z$, but
193: shifts to easy axis if the chains are elongated, with the MAE
194: achieving values as large as 50-80 meV depending on the specific
195: value of $d_z$. The third row in Figs. 2 and 3 shows the spin
196: moment per atom as a function of $d_z$. Smogunov and coworkers\cite{Tos08}
197: found that linear platinum chains become magnetized at a shorter elongation
198: $d_z$ for spin lying in the axis of the chain than for spins
199: oriented perpendicular to it. This fact led them to predict a
200: colossal magnetic anisotropy for the window of elongations within
201: the onset of both magnetizations. However, we note here that the
202: magnetization of the more realistic zigzag chains is finite for
203: almost the full range of elongations regardless of the spin
204: orientation, and therefore that effect is not expected to be seen. Interestingly,
205: we find that the magnetization of iridium chains decreases when
206: the chain is pulled apart, in a window of elongations around the equilibrium $d_z$.
207: 
208: \begin{figure}
209: \includegraphics[width=\columnwidth]{Fig3.ps}
210: \caption{(Color online) Results of the simulations of infinite iridium
211: chains as a function of the distance $d_z$ defined in Figure 1 for
212: orientations of the spins along the $=x$, $y$ and $z$ axes. 
213: The left panels correspond to the case of metastable linear
214: chains. The
215: right panels corresponds to the geometries (b) and (c) in Figure 1.
216: The dashed vertical line indicates the
217: distance $d_z$ at which the zigzag chains cease to be stable.
218: )a) Energies $E_{x,y,z}$ per atom; (b) magnetic anisotropy
219: energies per atom $\Delta_{x,y}$; (c) spin moments per atom; (d) conductance. }
220: \end{figure}
221: 
222: Since infinite Ir and Pt chains show large MAEs, an obvious
223: question is whether they also show a strong anisotropic
224: magneto-resistance effect that could be activated by the presence
225: of a magnetic field. If so, this effect could open the way towards
226: room-temperature inorganic nanomagnetic memory bits. We have
227: computed the low-voltage conductance $G_{x,y,z}$ for orientations
228: of the spins along the three axes, that we show in
229: Figs. 2(g,h) and 3(g,h), again as a function of $d_z$.
230: We have found that the magneto-resistance ratio, defined as
231: \begin{equation}
232: AMR_{x,y} = 2\,\frac{G_{x,y}-G_z}{G_{x,y}+G_z}
233: \end{equation}
234: \noindent is finite but small over wide ranges of elongations
235: $d_z$. Furthermore, iridium chains show
236: larger magneto-resistance ratios than platinum ones for a wider
237: range of $d_z$ at and above the equilibrium elongation. We therefore
238: expect that the magneto-resistance of iridium chains should be
239: more robust than that of platinum chains.
240: 
241: Remarkably, for certain windows of the elongations $d_z$ in zigzag chains,
242: one of the conductances may fall exactly to zero, marking the
243: onset of a perfect magneto-resistance ratio. Notice though  that these
244: windows of perfect magneto-resistivity do not appear for linear chains.
245: Zigzag platinum chains show three of those windows,
246: which occur when the atomic spin moments are oriented in the $y$-axis
247: (i.e. perpendicular to the zigzag plane). Two of these appear for
248: unrealistically short elongations $d_z$, but the third develops
249: for distances slightly larger that the equilibrium $d_z$ and
250: therefore could have the potential to be realized and measured
251: experimentally. Iridium chains show only one of these windows,
252: where $G_{x}$ vanishes, which occurs for elongations about the
253: equilibrium $d_z$, where the magnetic anisotropy favors
254: orientations of the spins along the $x$-axis. The MAE in this
255: window of elongations is of the order of 25 meV. Interestingly,
256: the vanishing conductance develops precisely for orientations of
257: the spins along the $x$-axis. 
258: 
259: Since the above finite magneto-resistance ratios originate from
260: the different number of channels available for electron conduction
261: close to the Fermi energy for the different spin orientations, we
262: now study the band structure of infinite chains. Indeed, the
263: number of conductance channels can be found by counting the number
264: of bands crossing the Fermi energy. We note that each band moves
265: up or down in energy at a different pace as the chain elongates.
266: Furthermore, since the spin-orbit interaction couples spin and
267: orbital degrees of freedom, these shifts are different for the
268: different spin orientations. Since the band structure of
269: non-conducting chains must show a mini-gap at the Fermi energy, we
270: have plotted in Fig. 4 the band structure of a platinum chain at
271: $d_z=2.45$ \AA~ for paramagnetic as well as for spin-orbit
272: simulations. We indeed find that when the atomic spins point in
273: the $y$-axis there exists a small mini-gap $E_g=25$ meV at
274: $k=0.4\,\pi/d_z$. This value allows us to estimate the minimum
275: number of atoms that a finite-sized platinum chain must have to
276: develop such a mini-gaps. Using the relationship $N\simeq
277: \hbar\,v_F / E_g\,d_z$, where we estimate the Fermi velocity at
278: this $k$-point as $\hbar\,v_F\simeq1000$ meV $d_z$, we find a
279: critical length of about 40 atoms.  We therefore expect that it
280: will be difficult to measure a significant magneto-resistance
281: ratio in platinum chains. We have indeed simulated 5-atoms long
282: platinum chains contacted to (001) platinum electrodes, in a
283: geometry similar to Fig. 1 (c) and found that the
284: magneto-resistance ratio was essentially zero.
285: 
286: 
287: \begin{figure}
288: %\includegraphics{Fig4-alt.jpg}
289: \centerline{\includegraphics[angle=-90,width=\columnwidth]{Fig4.ps}}
290: \caption{Band structure of a zigzag platinum chain with an
291: elongation $d_z=2.45$ \AA. From left to right the panels
292: show the bands for a paramagnetic calculation, and for spin-orbit
293: simulations whereby the spins are oriented along the $z$-, $x$-
294: and $y$-axes, respectively. }
295: \end{figure}
296: 
297: In contrast, Fig. 5 shows the band structures of iridium chains,
298: obtained from paramagnetic and also for spin-orbit simulations for an
299: elongation $d_z$ such that $G_x$ falls to zero. We note that the
300: spin-orbit interaction generates much larger gaps for iridium than
301: for platinum. For this particular elongation, the gap when the
302: spins point along the $x$-axis is positioned exactly at the Fermi
303: energy and has a value $E_g\simeq 100$ meV at the $\Gamma$ point.
304: Using the estimated Fermi velocity $\hbar\,v_F\simeq1500$ meV
305: $d_z$, we predict that chains containing  about 15 atoms should
306: show large magneto-resistance ratios.
307: 
308: \begin{figure}
309: %\includegraphics{Fig4-alt.jpg}
310: \centerline{\includegraphics[angle=-90,width=\columnwidth]{Fig5.ps}}
311: \caption{Band structure of a zigzag iridium chain with an
312: elongation $d_z=1.96$ \AA. From left to right the panels
313: show the bands for a paramagnetic calculation, and for spin-orbit
314: simulations whereby the spins are oriented along the $z$-, $x$-
315: and $y$-axes, respectively. }
316: \end{figure}
317: 
318: 
319: To confirm this, we have simulated 2- and 21-atoms long iridium
320: chains in contact to (111) iridium electrodes terminated by a
321: pyramid, as in Fig. 1 (c), for a number of elongations of the
322: chain, which cover both the equilibrium distance and also stretched
323: situations. We have found that the magneto-resistive ratio of the 2-atom
324: chains was negligible for whatever of these elongation. In contrast, 
325: we have found that $G_{x}$ is indeed substantially smaller than $G_{y,z}$, 
326: as we show in Fig. 6, rendering a finite ratio $AMR_x$. We stress that
327: our previous analysis of Figs. 3 and 5 indeed led us to expect that $G_x$
328: had to be smaller than $G_{y,z}$.
329: 
330: \begin{figure}[t]
331: \centerline{\includegraphics[angle=-90,width=0.8\columnwidth]{Fig6.ps}}
332: \caption{(Color online) Conductance as a function of $d_z$ of iridium chains 
333: containing 21 atoms. }
334: \end{figure}
335: 
336: 
337: To summarize, we expect that short platinum chains should show a
338: negligible magneto-resistance ratio until the number of atoms in
339: the chain exceeds of order 40. Since such a length is very
340: difficult to attain experimentally, we believe that platinum is
341: not the optimum magneto-resistance material. We predict that a
342: better candidate element is iridium. Short iridium chains are
343: expected to display small but measurable magneto-resistance
344: ratios. Longer chains could develop perfect magneto-resistance for
345: certain elongations. These clear differences are experimentally
346: accessible and therefore we anticipate that future experimental
347: studies of iridium chains will prove to be extremely fruitful.
348: 
349: We acknowledge useful conversations with C. Untiedt, J. J.
350: Palacios and A. Halbritter. L. 
351: Fern\'andez-Seivane tested the gaps in platinum
352: chains using the code Quantum ESPRESSO (http://www.pwscf.org). 
353: We acknowledge financial support
354: from the European Union (MRTN-CT-2003-504574), the
355: Spanish MEC (MEC-FIS2006-12117) and the UK EPSRC.
356: 
357: 
358: \begin{thebibliography}{99}
359: 
360: \bibitem{gambardella1}
361: P. Gambardella, et al.,
362: %Ferromagnetism in one-dimensional monatomic metal chains,
363: Nature {\bf 416}, 301 (2002).
364: 
365: \bibitem{gambardella2}
366: P. Gambardella, et al.,
367: %Giant magnetic anisotropy of single cobalt atoms and nanoparticles,
368: Science {\bf 300}, 1130 (2003).
369: 
370: \bibitem{blugel06}
371: %Y. Mokrousov, G. Bihlmayer, S. Heinze and S. Blugel,
372: Y. Mokrousov, et al.,
373: %Giant magnetocrystalline anisotropies of 4d transition metal monowires,
374: Phys. Rev. Lett. {\bf 96}, 147201 (2006).
375: 
376: \bibitem{so-clusters}
377: L. Fern\'andez-Seivane and J. Ferrer,
378: %Magnetic anisotropies of late transition metal atomic clusters,
379: Phys. Rev. Lett. {\bf 99} 183401 (2007)
380: 
381: \bibitem{Carlo}
382: T. O. Strandberg, C. M. Canali and A. H. Macdonald,
383: %Transition metal dimers and physical limits on magnetic anisotropy,
384: Nature Materials {\bf 6}, 648 (2007)
385: 
386: 
387: \bibitem{predictions}
388: L. Fern\'andez-Seivane, V. M. Garc\'{\i}a-Su\'arez and J. Ferrer,
389: %Predictions for the formation of atomic chains in mechanically
390: %controllable break junctions, 
391: Phys. Rev. B {\bf 75}, 075415 (2007)
392: 
393: \bibitem{Tos08}
394: %A. Smogunov, A. dal Corso, R. Weht, A. Delin and E. Tosatti,
395: A. Smogunov, et al.,
396: %Colossal magnetic anisotropy of monatomic free and deposited
397: %platinum wires, 
398: Nature Nanotech. {\bf 3}, 22 (2008)
399: 
400: \bibitem{ohn}
401: H. Ohnishi, Y. Kondo and K. Takayanagi, K., 
402: %Quantized conductance through individual rows of suspended gold atoms, 
403: Nature {\bf 395}, 780 (1998).
404: 
405: \bibitem{yan}
406: %A. I. Yanson, G. Rubio Bollinger, H. E. van der Brom, N. Agrait
407: %and J. M. Van Ruitenbeek, 
408: A. I. Yanson, et al., 
409: %Formation and manipulation of a metallic wire of single gold atoms, 
410: Nature {\bf 395}, 783 (1998).
411: 
412: \bibitem{Smit-thesis}
413: R. H. M. Smit, 
414: %From quantum point contacts to monatomic chains:
415: %fabrication and characterization of the ultimate nanowire, 
416: Ph. D. Thesis, Leiden University (2003).
417: 
418: \bibitem{smit}
419: %R. H. M. Smit, C. Untiedt, G. Rubio-Bollinger, R. C. Segers and
420: %J. M. van Ruitenbeek, 
421: R. H. M. Smit, et al., 
422: %Observation of a parity oscillation in the conductance of atomic wires, 
423: Phys. Rev. Lett. {\bf 91}, 076805 (2003).
424: 
425: 
426: \bibitem{thijssen}
427: %W. H. A. Thijssen, D. Marjenburgh, R. H. Bremmer and J. M. van
428: %Ruitenbeek, 
429: W. H. A. Thijssen, et al., 
430: %Oxygen-enhanced atomic chain formation, 
431: Phys. Rev. Lett. {\bf 96}, 026806 (2006).
432: 
433: \bibitem{mildred}
434: %H. Muramatsu, T. Hayashi, Y. A. Kim, D. Shimamoto, M. Endo, 
435: %M. Terrones and M. Dresselhaus, 
436: H. Muramatsu, et al., 
437: %Synthesis and isolation of molybdenum atomic wires, 
438: Nano Letters {\bf 8}, 237 (2008).
439: 
440: \bibitem{ugarte}
441: %V. Rodr\'{\i}gues, J. Bettini, P. C. Silva and D. Ugarte,
442: V. Rodr\'{\i}gues, et al.,
443: %Evidence for spontaneous spin-polarized transport in magnetic nanowires, 
444: Phys. Rev. Lett. {\bf 91}, 096801 (2003).
445: 
446: \bibitem{Delin1}
447: A. Delin and E. Tosatti, 
448: %Magnetic phenomena in 5d transtion metal nanowires, 
449: Phys. Rev. B {\bf 68}, 144434 (2003).
450: 
451: 
452: \bibitem{Delin2}
453: A. Delin, E. Tosatti and R. Weht, 
454: %Magnetism in atomic-sized palladium contacts and nanowires, 
455: Phys. Rev. Lett. {\bf 92}, 057201 (2004).
456: 
457: \bibitem{Koh65}
458: W. Kohn, L. J. and Sham, 
459: %Self-consistent equations including exchange and correlation effects, 
460: Phys. Rev. {\bf 140}, A1133 (1965).
461: 
462: \bibitem{SIESTA}
463: J. M. Soler, et al., 
464: %The Siesta method for ab initio order-N materials simulation, 
465: J. Phys.: Condens. Matter {\bf 14}, 2745 (2002).
466: 
467: \bibitem{Lucas}
468: %L. Fern\'andez-Seivane, M. A. Oliveira, S. Sanvito and J. Ferrer,
469: L. Fern\'andez-Seivane, et al.,
470: %On-site approximation for spin-orbit coupling in linear combination 
471: %of atomic orbitals density functional methods, 
472: J. Phys.: Condens. Matt. {\bf 18}, 7999 (2006).
473: 
474: \bibitem{Roc06}
475: %A. R. Rocha, V. M. Garc\'{\i}a-Su\'arez, S. W. Bailey, C. J. Lambert,
476: %J. Ferrer and S. Sanvito, 
477: A. R. Rocha, et al., 
478: %Spin and molecular spintronics in atomically generated orbital landscapes, 
479: Phys. Rev. B {\bf 73}, 085414 (2006).
480: 
481: %\bibitem{Natmat}
482: %A. R. Rocha, V. M. Garc\'{\i}a-Su\'arez, S. W. Bailey, C. J. Lambert,
483: %J. Ferrer and S. Sanvito, 
484: %Towards Molecular Spintronics,
485: %Nature Materials {\bf 4}, 335 (2005).
486: 
487: %\bibitem{data}
488: %S. Datta,  
489: %Electronic transport in mesoscopic systems, Cambridge University Press (1995)
490: 
491: \bibitem{Gar05}
492: %V. M. Garc\'{\i}a-Su\'arez, A. R. Rocha, S. W. Bailey, C. J. Lambert,
493: %S. Sanvito and J. Ferrer, 
494: V. M. Garc\'{\i}a-Su\'arez, et al., 
495: %Conductance oscillations in zigzag platinum chains, 
496: Phys. Rev. Lett. {\bf 95}, 256804 (2005).
497: 
498: 
499: \bibitem{Per81}
500: J. P. Perdew and A. Zunger, 
501: %Self-interaction correction to density-functional approximations for 
502: %many-electron systems, 
503: Phys. Rev. B {\bf 23}, 5048 (1981).
504: 
505: \bibitem{Tro91}
506: N. Troullier and J. L. Martins, 
507: %efficient pseudopotentials for plane-wave calculations, 
508: Phys. Rev. B {\bf 43}, 1993 (191).
509: 
510: %\bibitem{carlos}
511: %M. Reyes Calvo, J. Fern\'andez-Rossier and C. Untiedt, 
512: %Kondo-like features in chemically pure magnetic atomic-size
513: %contacts, 
514: %2008 March Meeting of the American Physical Society,
515: %http://meetings.aps.org/Meeting/MAR08
516: 
517: \bibitem{Pal05}
518: %J. Fern\'andez-Rossier, D. Jacob, C.  Untiedt  and J. J. Palacios, 
519: J. Fern\'andez-Rossier, et al., 
520: %Transport in magnetically ordered Pt nanocontacts, 
521: Phys. Rev. B. {\bf 72}, 224418 (2005)
522: 
523: \end{thebibliography}
524: 
525: \end{document}
526: