1: \documentclass[amsmat,amssymb,amsfonts,aps,prb,twocolumn,showpacs]{revtex4}
2: %\documentclass[amsmath,amsfonts,aps,prb,preprint,showpacs]{revtex4}
3:
4: \usepackage{graphicx}
5: \usepackage{dcolumn}
6: \usepackage{bm}
7: \begin{document}
8:
9:
10: \title{Orbital eigenchannel analysis for ab-initio quantum transport calculations}
11:
12: \author{ David Jacob }
13: \email{david.jacob@ua.es}
14: \author{ J. J. Palacios}
15: \affiliation{Departamento de F\'isica Aplicada and Instituto Universitario de Materiales
16: de Alicante (IUMA), Universidad de Alicante, 03690 San Vicente del Raspeig, SPAIN }
17:
18: \date{\today}
19:
20: \pacs{73.63.-b,73.63.Rt,75.47.Jn}
21:
22: \begin{abstract}
23: We show how to extract the orbital contribution to the transport eigenchannels from
24: a first-principles quantum transport calculation in a nanoscopic conductor.
25: This is achieved by calculating and diagonalizing the first-principles transmission matrix
26: reduced to selected scattering cross-sections.
27: As an example, the orbital nature of the eigenchannels in the case of Ni nanocontacts
28: is explored, stressing the difficulties inherent to the use of non-orthogonal basis sets
29: %and first-principles Hamiltonians.
30: \end{abstract}
31:
32: \maketitle
33:
34: \section{Introduction}
35: %% Context
36: \textit{Ab initio} quantum transport calculations for nanoscopic conductors like
37: molecular junctions and nanocontacts have become standard\cite{mode-matching,NEGF},
38: %%\cite{Lang:prb:95,Hirose:prb:95,Di-Ventra:prl:01,Taraschi:prb:98,Kobayashi:prb:00,Damle:prb:01,Taylor:prb:01:63:2,Palacios:prb:01,Palacios:prb:02,Brandbyge:prb:02,Xue:prb:03:I,Thygesen:prl:03,Bagrets:05,Garcia-Suarez:prb:05},
39: and replace, to a large extent,
40: %%have taken over
41: earlier phenomenological or parametrized procedures for quantum transport
42: \cite{models,Cuevas:prl:98:80}.
43: %%\cite{Torres:prl:96,Brandbyge:prb:99,Cuevas:prl:98:80}.
44: %become a widely accepted standard
45: %% Landauer approach
46: A widely accepted starting point is the Landauer approach which describes the electron
47: transport as an elastic (and thus phase-coherent)
48: scattering process of non-interacting quasi-particles.
49: In this approach the conductance of a nanoscopic region is determined by the
50: quantum mechanical transmission probabilities of the incoming transport channels\cite{Datta:book:95}.
51: The transmission matrix is calculated by mode-matching of the incoming waves in one lead
52: with the outgoing waves on the other lead via the intermediate scattering region\cite{mode-matching}
53: %%\cite{Lang:prb:95,Hirose:prb:95,Di-Ventra:prl:01,Taraschi:prb:98,Kobayashi:prb:00,Khomyakov:prb:05}
54: %%\cite{Ando:prb:91,Lang:prb:95,Hirose:prb:95,Di-Ventra:prl:01,Taraschi:prb:98,Kobayashi:prb:00,Khomyakov:prb:05,Smogunov:ss:04}
55: or, equivalently\cite{Khomyakov:prb:05}, from the one-body Green's function of the scattering region connected to the metallic leads\cite{NEGF}.
56: %%\cite{Damle:prb:01,Taylor:prb:01:63:2,Palacios:prb:01,Palacios:prb:02,Brandbyge:prb:02,Xue:prb:03:I,Thygesen:prl:03,Bagrets:05,Garcia-Suarez:prb:05}.
57: %%
58: The latter is known as the non-equilibrium Green's function (NEGF) approach since
59: a finite bias voltage can also be taken into account when calculating the electronic
60: structure of the junction.
61: One of the advantages of the NEGF approach is that it does not need to be
62: implemented from scratch since one can make use of standard \textit{ab initio}
63: codes for the computation of the electronic structure.
64: %% Conductance and eigen channels
65:
66: It is often useful to decompose the total conductance into the contributions
67: of the transport eigenchannels, first introduced by B\"uttiker\cite{Buettiker:ibmjrd:88}.
68: These are defined as the linear combinations of the incoming
69: modes in a lead that do not mix upon reflection on the scattering region
70: and present a unique transmission value\cite{Brandbyge:prb:97}.
71: The decomposition of the measurable total transmission
72: in terms of the transmissions of these
73: eigenchannels simplifies considerably the interpretation of the
74: results. Knowledge and analysis of the eigenchannel wavefunctions would, in turn,
75: allow one to make predictions regarding the behavior of the
76: conductance upon distortions of the geometry or other perturbations of
77: the scattering region\cite{Jacob:prb:05}.
78: Unfortunately, in the NEGF approach it is not straightforward to extract the
79: orbital composition of the transport eigenchannels.
80: Only the eigenchannel transmissions can be obtained easily in the NEGF approach
81: from the non-negligible eigenvalues of the transmission matrix.
82: However, the associated eigenvectors turn out to be useless as obtained.
83: The reason is that these eigenvectors contain the contributions to the
84: eigenchannel wavefunctions of the atomic orbitals at one of the borders
85: of the scattering region immediately connected to the leads, but not inside.
86: %%which is immediately connected to the lead where the
87: %%eigenchannels (being linear combinations of the incoming modes of one lead)
88: %%are defined.
89:
90: %%The reason is that these eigenvectors only have weight at one of surface layers
91: %%of the scattering region (the ones immediately connected to the leads),
92: %%but not inside, and present the eigenchannels at the surface of the
93: %%scattering region connected to the lead where the
94:
95: %% on the physical conduction protagonists, i.e., the
96: %%atoms and orbitals involved in conduction. Some insight can be gained
97: %%from the analysis of the usually small number of non-negligible
98: %%eigenvalues of the transmission matrix. However, the associated
99: %%transport eigenchannels turn out to be useless as obtained.
100: %%about the orbital nature of the transport eigen-channels inside the scattering region
101: %%which is of fundamental interest for understanding the physical mechanisms
102: %%behind the transport properties of a particular device.
103: %%The reason is that these eigenchannels only have weight at the borders of the
104: %%scattering region (the ones immediately connected to the leads), but not inside.
105: % (i.e. the atomic constriction in a nanocontact or molecule-metal contact
106: %in a molecular junction device).
107:
108: Here we present a method for analyzing the orbital contributions to the transport eigenchannels
109: at an arbitrary cross-section of a nanoscopic conductor
110: by calculating the transmission matrix projected to that cross-section.
111: % from the Green's
112: %function of the whole device.
113: Our approach generalizes previous work by Cuevas et al.\cite{Cuevas:prl:98:80} for tight-binding-type
114: Hamiltonians to %%accounting for the nuances inherent to \textit{ab initio} Hamiltonians and
115: non-orthogonal atomic orbitals basis sets as those commonly used in quantum chemistry packages.
116: An alternative approach to investigate the contributions of certain atomic or molecular orbitals
117: to the conductance consists in directly removing the respective orbitals from the basis set
118: \cite{Thygesen:prl:05:94:3}.
119:
120: %% Eigenchannels
121: %%Consequently, the linear conductance (at zero bias) is given
122: %%by the trace of the transmission matrix:
123: %%
124: %%\begin{equation}
125: %% \label{eq:Landauer}
126: %% G = \frac{e^2}{h} {\rm Tr}[T].
127: %%\end{equation}
128: %%
129: %%The transmission matrix can be expressed as the absolute square of
130: %%the transmsission amplitude matrix $t$ which relates the incoming modes
131: %%of one lead to the outgoing modes in the other lead:
132: %%%%
133: %%\begin{equation}
134: %% \label{eq:T}
135: %% T = t^\dagger t.
136: %%\end{equation}
137: %%%%
138: %%Thus the probability that an incoming mode $m$ on one lead is scattered
139: %%into an outgoing mode $n$ on the other lead is given by $\|t_{mn}\|^2$.
140:
141: %%% Eigen channels
142: %%Diagonalizing the transmission matrix one obtains the so-called eigenchannels
143: %%(eigenvectors of the transmission matrix) and their corresponding transmissions
144: %%(eigenvalues).
145:
146:
147:
148: %In recent years ab-initio transport calculations of molecular electronics devices based
149: %on density functional theory (DFT) have become a widely accepted standard, and a wide
150: %range of codes is available nowadays [CITES].
151: %The electronic structure of the device connected to semi-infinite leads
152: %is calculated self-consistenly within the Kohn-Sham approximation of DFT
153: %taking into account the atomic structure of the device.
154: %The Landauer Formalism then describes the electron transport via the device as an elastic
155: %scattering problem in an effective single-particle picture.
156:
157:
158: \section{Method}
159: First we divide the system under study into 3 parts: The left lead (L),
160: the right lead (R), and the intermediate region called device (D) from now on,
161: where only elastic scattering takes place.
162: Figure \ref{fig:constriction}, shows a sketch of a
163: nano-constriction connecting two bulk leads.
164: %%
165: \begin{figure}
166: \includegraphics[scale=0.5]{LDR.eps}
167: \caption{Sketch of the scattering problem.
168: L: Left lead. D: device. R: right Lead.
169: 0: cross-section of interest.}
170: \label{fig:constriction}
171: \end{figure}
172: %%
173: We assume that the leads are coupled only to the constriction but not to each other.
174: The Hamiltonian describing this situation is then given by the matrix
175: %%
176: \begin{equation}
177: \label{eq:HLDR}
178: {\bf H} = \left(
179: \begin{array}{ccc}
180: {\bf H}_{\rm L} & {\bf H}_{\rm LD} & {\bf 0} \\
181: {\bf H}_{\rm DL} & {\bf H}_{\rm D} & {\bf H}_{\rm DR} \\
182: {\bf 0} & {\bf H}_{\rm RD} & {\bf H}_{\rm R}
183: \end{array}
184: \right).
185: \end{equation}
186: %%
187: Since many density functional theory (DFT) codes work in non-orthogonal basis sets,
188: we also allow explicitly for overlap between atomic-orbitals
189: given by the following overlap matrix:
190: %%
191: \begin{equation}
192: \label{eq:SLDR}
193: {\bf S} = \left(
194: \begin{array}{ccc}
195: {\bf S}_{\rm L} & {\bf S}_{\rm LD} & {\bf 0} \\
196: {\bf S}_{\rm DL}& {\bf S}_{\rm D} & {\bf S}_{\rm DR} \\
197: {\bf 0} & {\bf S}_{\rm RD} & {\bf S}_{\rm R}
198: \end{array}
199: \right).
200: \end{equation}
201:
202: The standard approach to calculate the conductance is to calculate the
203: self-energies of the leads from the Green's functions of the isolated leads,
204: i.e., for the left lead ${\bf\Sigma}_{\rm L}(E) = ({\bf H}_{\rm DL}-E {\bf S}_{\rm DL}) {\bf g}_{\rm L}(E) ({\bf H}_{\rm LD}-E {\bf S}_{\rm LD})$
205: where ${\bf g}_{\rm L}(E) = (E {\bf S}_{\rm L} - {\bf H}_{\rm L})^{-1}$
206: is the Green's function of the isolated left lead and analogously for the right lead.
207: From this we can calculate the Green's function of the device:
208: %%
209: \begin{equation}
210: \label{eq:GD}
211: {\bf G}_{\rm D}(E) = (E {\bf S}_{\rm D} - {\bf H}_{\rm D} - {\bf \Sigma}_{\rm L}(E) - {\bf \Sigma}_{\rm R}(E))^{-1},
212: \end{equation}
213: %%
214: which, in turn, allows us to calculate the (hermitian) transmission matrix
215: %%
216: \begin{equation}
217: \label{eq:TCarolis}
218: {\bf T}(E) = {\bf \Gamma}_{\rm L}(E)^{1/2} {\bf G}_{\rm D}^\dagger(E) {\bf \Gamma}_{\rm R}(E) {\bf G}_{\rm D}(E) {\bf \Gamma}_{\rm L}(E)^{1/2},
219: \end{equation}
220: %%
221: where ${\bf\Gamma}_{\rm L}=i({\bf\Sigma}_{\rm L}-{\bf\Sigma}_{\rm L}^\dagger)$ and
222: ${\bf\Gamma}_{\rm R}=i({\bf\Sigma}_{\rm R}-{\bf\Sigma}_{\rm R}^\dagger)$.
223: Typically the leads are only connected to the left and right borders of
224: the device and are sufficiently far away from the scattering region so that they
225: can be described by a bulk electronic structure.
226: From the structure of eq. (\ref{eq:TCarolis}) it follows that only
227: the submatrix of ${\bf T}$ representing the subspace of the device
228: immediately connected to one of the leads are non-zero.
229: Thus the eigenvectors obtained by diagonalizing ${\bf T}$ only contain
230: the atomic orbital contributions to the eigenchannels at one border of
231: the device region but not at the center where the resistance is ultimately
232: determined.
233:
234: %%Diagonalization of the ${\bf T}$ yields the eigenchannel transmissions.
235: %%Typically the leads are only connected to the left and right borders of the device
236: %%and are sufficiently far away from the scattering region so that they
237: %%can be described by a bulk electronic structure. The eigenvectors obtained by
238: %%diagonalizing ${\bf T}$ thus only contain the atomic orbital contributions to
239: %%the eigenchannels at the left and right borders of the device region but not
240: %%at the center where the resistance is ultimately determined.
241:
242: To investigate the orbital nature of the eigenchannels at an arbitrary part of the device
243: we can simply calculate the transmission matrix associated to this part.
244: By choosing this region to be a cross-section, like that indicated in Fig. \ref{fig:constriction},
245: current conservation guarantees that the so-calculated conductance is \emph{approximately} equal
246: to the conductance calculated from the transmission matrix of the whole device.
247: We want to emphasize here that this is really only approximately true for a Hamiltonian beyond
248: the tight-binding approximation since hoppings between atoms on both sides beyond the selected
249: region are neglected. Of course this approximation becomes better the thicker the chosen cross-section is.
250: We proceed by further subdividing the device region. The cross-section of interest will be referred to
251: as 0 while the regions on either side will be denoted as l and r, respectively:
252: %%
253: \begin{eqnarray}
254: \label{eq:HD}
255: {\bf H}_{\rm D} &=& \left(
256: \begin{array}{ccc}
257: {\bf h}_{\rm l} & {\bf h}_{\rm l0} & {\bf h}_{\rm lr} \\
258: {\bf h}_{\rm 0l} & {\bf h}_{\rm 0} & {\bf h}_{\rm 0r} \\
259: {\bf h}_{\rm rl} & {\bf h}_{\rm r0} & {\bf h}_{\rm r}
260: \end{array}
261: \right)
262: \\
263: \nonumber \\
264: \label{eq:SD}
265: {\bf S}_{\rm D} &=& \left(
266: \begin{array}{ccc}
267: {\bf s}_{\rm l } & {\bf s}_{\rm l0} & {\bf s}_{\rm lr} \\
268: {\bf s}_{\rm 0l} & {\bf s}_{\rm 0} & {\bf s}_{\rm 0r} \\
269: {\bf s}_{\rm rl} & {\bf s}_{\rm r0} & {\bf s}_{\rm r}
270: \end{array}.
271: \right)
272: \end{eqnarray}
273: %%
274: As mentioned above we will neglect the hoppings (and overlaps) between the
275: left and right layers outside the region of interest so we set
276: ${\bf h}_{\rm lr} = {\bf h}_{\rm rl} = {\bf s}_{\rm lr} = {\bf s}_{\rm rl} = 0$.
277: With this approximation the Green's function matrix of the cross-section $0$
278: can be written as
279: %%
280: \begin{equation}
281: \label{eq:G0}
282: {\bf G}_{\rm 0}(E) =( E {\bf s}_{\rm 0} - {\bf h}_{\rm 0} - {\bf\Sigma}^\prime_{\rm l}(E) - {\bf\Sigma}^\prime_{\rm r}(E) )^{-1}.
283: \end{equation}
284: %%
285: The self-energy matrices representing the coupling to the left and right lead,
286: ${\bf\Sigma}^\prime_{\rm l}(E) = ({\bf h}_{\rm 0l}-E {\bf s}_{\rm 0l}) {\bf g}^\prime_{\rm l}(E) ({\bf h}_{\rm l0}-E {\bf s}_{\rm l0})$ and
287: ${\bf\Sigma}^\prime_{\bf r}(E) = ({\bf h}_{\rm 0r}-E {\bf s}_{\rm 0r}) {\bf g}^\prime_{\rm r}(E) ({\bf h}_{\rm r0}-E {\bf s}_{\rm r0})$,
288: are given by the Green's function of the left layer $\rm l$ connected only to the left
289: lead $\rm L$ and the right layer $\rm r$ connected only to the right lead $\rm R$, respectively:
290: %%
291: \begin{eqnarray}
292: {\bf g}^\prime_{\rm l}(E) &=& ( E {\bf s}_{\rm l} - {\bf h}_{\rm l} - {\bf \Sigma}_{\rm L}(E) )^{-1} \\
293: {\bf g}^\prime_{\rm r}(E) &=& ( E {\bf s}_{\rm r} - {\bf h}_{\rm r} - {\bf \Sigma}_{\rm R}(E) )^{-1}.
294: \end{eqnarray}
295: %%
296: The \emph{reduced transmission matrix} (RTM) with respect to the chosen cross-section is now given by
297: %%
298: \begin{equation}
299: {\bf T}^\prime(E) = {\bf\Gamma}_{\rm l}^\prime(E)^{1/2} {\bf G}_0^\dagger(E) {\bf\Gamma}_{\rm r}^\prime(E) {\bf G}_0(E) {\bf\Gamma}_{\rm l}^\prime(E)^{1/2}
300: \end{equation}
301: with ${\bf\Gamma}_{\rm l}^\prime=i({\bf\Sigma}_{\rm l}^\prime-{{\bf\Sigma}_{\rm l}^\prime}^\dagger)$ and ${\bf\Gamma}_{\rm r}^\prime=i({\bf\Sigma}_{\rm r}^\prime-{{\bf\Sigma}_{\rm r}^\prime}^\dagger)$.
302: Diagonalizing ${\bf T}^\prime(E)$ now yields the contribution of the atomic orbitals within the cross-section $0$
303: to the eigenchannels.
304: %%As we will show in the next section the approximation made in the method is reasonable when the choosen
305: %%cross-section located in the bottle-neck of the system, e.g the contact atom of an atomic point-contact.
306: %%Nevertheless the method should also work well for a cross-section which is not in the bottle-neck
307: %%but some wider part as long as we choose the length of the cross-section sufficiently big, so that
308: %%hopping between atoms in the left and right outside layer become negligible.
309:
310:
311: \section{Results and Discussion}
312: In the following we apply the above described method to analyze the orbital nature of the conducting channels
313: of Ni nanocontacts which have recently attracted a lot of interest because of their apparently high magnetoresistive properties
314: \cite{Ni-experiments}.
315: %%\cite{Garcia:prl:99,Viret:prb:02}.
316: We consider the nanocontact to consist of two ideal pyramids facing each other along the (001) direction
317: and with the two tip atoms being 2.6 \r A apart. Bulk atomic distances (2.49 \r A) and perfect crystalline order are assumed for each pyramid.
318: Just as in our previous work on Ni nanocontacts\cite{Jacob:prb:05} we
319: perform \textit{ab initio} quantum transport calculations for this idealized geometry. To this end
320: we use our code ALACANT (ALicante Ab initio Computation Applied to NanoTransport). The electronic structure is computed at the
321: DFT local spin density approximation
322: level with a minimal basis set and the electrodes are described by means of
323: a semi-empirical tight-binding Bethe lattice model.
324:
325: %%
326: \begin{figure}
327: \includegraphics[width=0.65\linewidth]{Ni28.eps}
328: \caption{Sketch of Ni nanocontact consisting
329: of two pyramids facing each other along the (001)
330: direction with the two tip atoms forming a dimer bridge.
331: The device region (grey circles) consists of 28 Ni atoms
332: and the left and right electrodes (empty circles) are modelled by Bethe lattices
333: with appropriate tight-binding parameters to reproduce Ni Bulk DOS.}
334: \label{fig:Ni28}
335: \end{figure}
336: %%
337:
338: As indicated in Fig. \ref{fig:Ni28} we calculate the RTM ${\bf T}^\prime(E)$
339: for one of the tip atoms of the contact (labelled with 0) and diagonalize it
340: to obtain the eigenchannels and the corresponding transmissions projected
341: on the tip atom.
342: In Fig. \ref{fig:Ni28-channels} we compare the individual channel transmissions
343: calculated on the one hand from the full transmission matrix (FTM) ${\bf T}(E)$
344: and on the other hand from the RTM ${\bf T}^\prime(E)$.
345: Though the electron hopping between regions l and r of the contact has been neglected
346: in calculating the RTM the so calculated channel transmissions approximate very well
347: those calculated using the FTM so that it is very easy to relate the RTM channel
348: transmissions with the FTM channel transmission.
349: This shows that the hopping between the regions l and r on both sides of the tip atom
350: is almost negligible. Only for the one majority (M) channel we see a small deviation
351: near the Fermi energy indicating that here 2nd neighbour hopping contributes
352: to the transmission of that channel.
353: As the eigenvectors of the RTM (see Table \ref{tab:1}) reveal, this channel is mainly s-type.
354: Since s-electrons are strongly delocalized there is a small but finite contribution
355: from second-neighbour hopping explaining the deviation between the FTM and RTM transmission
356: in that channel.
357: The first minority (m) channel is also mainly s-type but now it is hybridized with
358: $\rm d_{3z^2-r^2}$ and $\rm p_z$ orbitals.
359: The other two m channels are degenerate and mainly $\rm d_{xz}$- and
360: $\rm d_{yz}$-type strongly hybridized with $\rm p_x$- and $\rm p_y$-orbitals, respectively.
361:
362:
363: %%
364: \begin{figure}
365: \includegraphics[width=\linewidth]{Ni28-channels.eps}
366: \caption{
367: Transmission functions of open transport channels for the Ni nanocontact
368: sketched in Fig. \ref{fig:Ni28} as calculated from the FTM
369: ${\bf T}(E)$ (solid line) and from the RTM ${\bf T}^\prime(E)$ (dashed lines).
370: (a) shows the only contributing M channel and (b)-(c) the three m
371: channels. See text for further discussion.
372: }
373: \label{fig:Ni28-channels}
374: \end{figure}
375: %%
376:
377: %%
378: \begin{table}
379: \begin{tabular}{l|c|c|c|c}
380: AO & majority & minority 1 & minority 2 & minority 3 \\
381: \hline
382: s & 97\% & 62\% & 0 & 0 \\
383: $\rm p_x$ & 0 & 0 & 28\% & 0 \\
384: $\rm p_y$ & 0 & 0 & 0 & 28\% \\
385: $\rm p_z$ & 3\% & 23\% & 0 & 0 \\
386: $\rm d_{3z^2-r^2}$ & 0 & 15\% & 0 & 0 \\
387: $\rm d_{xz}$ & 0 & 0 & 72\% & 0 \\
388: $\rm d_{yz}$ & 0 & 0 & 0 & 72\% \\
389: $\rm d_{x^2-y^2}$ & 0 & 0 & 0 & 0 \\
390: $\rm d_{xy}$ & 0 & 0 & 0 & 0 \\
391: \end{tabular}
392: \caption{Eigenvectors of the RTM at the Fermi level
393: for the contact sketched in Fig. \ref{fig:Ni28}.
394: Each column gives the weights of the atomic orbitals
395: (AO) given in the left column on the tip atom in each
396: eigenchannel shown in Fig. \ref{fig:Ni28-channels}.
397: }
398: \label{tab:1}
399: \end{table}
400:
401: %% Discussion
402: As discussed in our previous work\cite{Jacob:prb:05} the five
403: d-type transport channels for the m electrons available in the perfect Ni chain\cite{Smogunov:ss:02}
404: are easily blocked in a contact with a realistic geometry like that in Fig. \ref{fig:Ni28}
405: because the d-orbitals are very sensitive to geometry. We have refered to this
406: as \emph{orbital blocking}.
407: It is not so surprising that the $\rm d_{x^2-y^2}$- and $\rm d_{xy}$-channel
408: which are very flat bands just touching the Fermi level in the perfect chain
409: are easily blocked in a realistic contact geometry. These bands
410: represent strongly localized electrons which are easily scattered in geometries with low symmetry.
411: Interestingly, even the $\rm d_{3z^2-r^2}$-channel, which for the perfect chain is a very broad band
412: crossing the Fermi level at half band width, does not contribute to the conduction as our eigenchannel analysis shows.
413: This channel is blocked because the $\rm d_{3z^2-r^2}$-orbital lying along
414: the symmetry axis of the contact is not ``compatible'' with the geometry
415: of the two pyramids.
416: On the other hand the $\rm d_{xz}$- and $\rm d_{yz}$-channels are both open
417: in that geometry because their shape is compatible with the pyramid geometry of the
418: contacts.
419: This illustrates how the geometry of a contact can effectively block (or open) channels
420: composed of very directional orbitals. Of course, for different geometries we can expect
421: different channels to be blocked or opened.
422:
423: %%
424: \begin{figure}
425: \includegraphics[width=0.6\linewidth]{Ni27.eps}
426: \caption{Sketch of Ni nanocontact (27 atoms). As in Fig. \ref{fig:Ni28}
427: the contact consists of two pyramids along the (001) direction but
428: now both pyramids share the same atom at the tip.
429: The device region (grey circles) consists of 27 Ni atoms
430: and the left and right electrodes (empty circles) are modelled by Bethe lattices
431: with appropriate tight-binding parameters to reproduce Ni bulk DOS.
432: }
433: \label{fig:Ni27}
434: \end{figure}
435: %%
436:
437: Obviously, the approximation made in the calculation of the RTM
438: becomes worse the bigger the hopping between the regions l and r is.
439: For example, in the contact geometry shown in Fig. \ref{fig:Ni27}
440: electron hopping from the layers immediately connected to the central atom
441: (labeled 0) is certainly bigger than in the geometry of Fig. \ref{fig:Ni28}.
442: Indeed, Fig. \ref{fig:Ni27-channels} shows that
443: for almost all channels the RTM transmissions differ appreciably from FTM transmissions, making it difficult in some cases to relate them to each other.
444: Fortunately, we can judge by exclusion which
445: RTM transmission relates to which FTM transmission since for the
446: other channels at least the RTM transmission function mimics the overall
447: behaviour of the FTM transmission function.
448: However, for more complicated situations it might be impossible
449: to match the RTM transmission with the FTM transmission for all channels.
450: The cure to this problem is obvious: One has to choose a bigger
451: cross-section, i.e., add an atomic layer to the cross-section
452: so that the hopping between l and r becomes small again.
453: If we choose, e.g., the cross-section labeled with 0$^\prime$ in
454: Fig. \ref{fig:Ni27} (including the atomic layer to the right of the central atom)
455: the so calculated RTM transmissions now approximate very well the FTM transmissions as
456: can be seen in Fig. \ref{fig:Ni27-channels}.
457:
458: %%
459: \begin{figure}
460: %% \begin{tabular}{c}
461: \includegraphics[width=\linewidth]{Ni27-channels.eps} \\
462: \includegraphics[width=\linewidth]{Ni27-channels2.eps}
463: %% \end{tabular}
464: \caption{
465: Transmission functions of open transport channels for the Ni nanocontact
466: sketched in Fig. \ref{fig:Ni27} as calculated from the FTM ${\bf T}(E)$
467: (solid lines) and from the RTM ${\bf T}^\prime(E)$ (dashed lines).
468: The RTM transmissions calculated for the cross-section labeled with 0$^\prime$
469: in Fig. \ref{fig:Ni27} are given by the thin solid curves (labeled RTM$^\prime$).
470: }
471: \label{fig:Ni27-channels}
472: \end{figure}
473: %%
474:
475: %%Again, the channels that yield the highest deviation between FTM and RTM
476: %%transmission (M channel 1 and m channel 1) are mostly s-type describing
477: %%very delocalized electrons which thus have a lot of contributions from
478: %%second-neighbour hopping.
479:
480:
481: \section{Conclusions}
482: In summary, we have shown how
483: to obtain the orbital contributions to the eigenchannels at an arbitrary
484: cross-section of a nanoscopic conductor.
485: The method has been implemented into our \textit{ab initio} quantum
486: transport program ALACANT
487: and we have illustrated the method by exploring the orbital nature
488: of the eigenchannels of a Ni nanocontact.
489: The method works very well when the chosen cross-section is thick
490: enough so that hopping from the layers left and right to the
491: cross-section becomes negligible.
492: Hence in some cases an additional atomic layer has to be included
493: to the cross-section we are actually interested in.
494: Taking this into account the method has no limitations and can be readily applied
495: to \textit{ab initio} transport calculations in all types of
496: nanocontacts\cite{Fernandez:prb:05} and molecular junctions.
497:
498:
499: \section*{Acknowledgements}
500: We thank J. Fern\'andez-Rossier and C. Untiedt for fruitful discussions.
501: DJ acknowledges financial support from MECD under grant UAC-2004-0052.
502: JJP acknowledges financial support from Grant No. MAT2002-04429-C03 (MCyT) and from
503: University of Alicante.
504:
505:
506: %%%%%%%%%%%%%%%%%%%%%
507:
508: \bibliography{matcon}
509: %%\bibliographystyle{prsty}
510: \bibliographystyle{apsrev}
511: \end{document}
512: