Appeared in Nature, Oct.19-23, 1998.

Formation and Manipulation of a Metallic Wire of Single Gold Atoms. A. I. Yanson¹, G. Rubio Bollinger², H.E. van den Brom¹, N. Agraït², J.M. van Ruitenbeek¹

¹Kamerlingh Onnes Laboratorium, Leiden University, PO Box 9506, NL-2300 RA Leiden, The Netherlands e-mail: ruitenbe@phys.leidenuniv.nl tel: +31 71 527 5450 fax: +31 71 527 5404
²Laboratorio de Bajas Temperaturas, Dept. Física de la Materia Condensada C-III, Instituto Universitario de Ciencia de Materiales "Nicolás Cabrera'', Universidad Autónoma de Madrid, E-28049 Madrid, Spain e-mail: nicolas.agrait@uam.es tel. +34 91 397 4756 fax +34 91 397 3961

Abstract

The continuing miniaturization of microelectronics raises the prospect of nanometre-scale devices with mechanical and electrical properties that are qualitatively different from those at larger dimensions. The investigation of these properties, and particularly the increasing influence of quantum effects on electron transport, has therefore attracted much interest. Quantum properties of the   conductance can be observed when "breaking" a metallic contact: as two metal electrodes in contact with each other are slowly retracted, the contact area undergoes structural rearrangements until it consists in its final stages of only a few bridging atoms [1,2].Just before the abrupt transition to tunnelling occurs, the electrical conductance through a monovalent metal contact is always close to a value of 2e²/h»(12.9 kW)^-1, where e is the charge on an electron and h is Planck's constant [3,4,5]. This value corresponds to one quantum unit of conductance, thus indicating that the "neck" of the contact consists of a single atom [6]. In contrast to previous observations of only single-atom necks, here we describe the breaking of atomic-scale gold contacts, which leads to the formation of gold chains one atom thick and at least four atoms long. Once we start to pull out a chain, the conductance never exceeds 2e²/h, confirming that it acts as a one-dimensional quantized nanowire. Given their high stability and the ability to support ballistic electron transport, these structures seem well suited for the investigation of atomic-scale electronics.

Experimental techniques

Since atomic chains are very fragile objects, one has to isolate them from external disturbances in order to perform measurements. The destructive influence of thermal excitations, mechanical vibrations or chemical impurities has to be reduced as much as possible. The proper conditions for spontaneous formation of chains of gold atoms has been so far achieved using two different experimental approaches:

the Mechanically Controllable Break-junction (MCB) and

an ultra-stable low temperature Scanning Tunnelling Microscope (STM).

Mechanically Controllable Break-junction (MCB)

This method was invented by J. Moreland and J.W. Ekin [7] in 1985 and developed by C. Muller in 1988-1992 during his PhD in room 10 (kamer 10) of Kamerlingh Onnes Laboratory [8]. The basic principle of the method is to break a sample metal wire, glued on a flexible substrate, in a controlled way by bending the substrate and afterwards releasing the bending strain to re-gain contact. The core of every MCB-setup is shown in the figure below:

This is a schematic diagram of the MCB. It consists of the bending beam on which the sample (notched filament) is glued with epoxy adhesive. The substrate is then mounted in a 3-point bending configuration consisting of two counter supports and a piezoelectric actuator. The distance u (mechanical loop) is ~0.1-0.5 mm, the distance L is ~20 mm.

The whole system is then placed in cryogenic environment (P<10^-5 mbar, T=4.2 K), where the notched sample wire is initially broken by bending the substrate (which, in its turn is achieved by applying high voltage to the piezo element). Upon reducing the voltage on the piezo it is possible to bring the electrodes back into electrical contact and thus control the junction.

The major advantages of the MCB technique are:

clean contact surfaces	Since the notched wire is initially broken and kept at cryogenic temperatures in high vacuum, no contamination of the electrodes occur on the timescale of days. This allows the widest range of materials (from antimony to lithium).
exceptional mechanical stability	Since the mechanical loop (the distance between the anchoring points of the two electrodes) is extremely small (< 1 mm compared to > 1 cm for STM) and because of the bending-beam configuration the distance between the electrodes changes less than 0.1 Å per hour (without feedback).
ease of sub-angstrom adjustment	MCB with a bending beam gives a 20-50 times reduction between the expansion of the piezoelement and the distance between the electrodes, suppressing the effect of electric noise in high voltage and ensuring precise adjustment of the inter-electrode distance.
wide range of operation	MCB works in ambient conditions as well as in ultra high vacuum (< 10-5 mbar) and at very low temperatures (< 0.5 K).
ease of use and expandability	MCB can be employed to study various phenomena (shot noise, thermopower, Kondo effect, etc.) with the experimental success rate of 90%. Its flexibility in design allows for extensions such as MCB-STM, nanofabricated MCB and MCB-in-a-SQUID loop.

Here is another example of how a break-junction can be used to study materials that are considered "experimentally difficult":

This modification for studying alkali metals was developed by M. Krans [9] during his PhD in kamer 10, Kamerlingh Onnes laboratory, Leiden University

Scanning Tunnelling Microscope (STM)

STM has made a long way from a Nobel Prize winner [10] to a conventional tool in many physics laboratories in less than a decade. The basic scheme of an STM is shown in a picture below. For a more comprehensive description please refer to the STM page of the Technical University of Wien.

Experimental results

General information

In most experiments done on MCB the conductance of the sample is measured  while slowly breaking the contact. By bending the substrate in a controlled manner one can vary the distance between the electrodes which is related to the diameter of the contact and hence the conductance. While slowly pulling apart very large contacts  the "neck" (the narrowmost part of the wire) becomes thinner quasi-continuously due to the "chewing-gum" effect,  or goes through drastic changes due to avalanches of lattice defects and/or grain boundaries. In the first case the conductance will decrease proportionally to the stretching force in a smooth fasion, while in the second case it will show gigantic jumps. At the very last stages of contact stretching before the electric contact is broken, however, its area (and conductance) changes abruptly in a series jumps with plateaus in between associated with stress accumulation and relief on atomic level. These jumps are caused by atomic rearrangements inside the contact [11].

The discovery of gold chains formation

The first observations of "peculiar" behavior of ultra-small gold nanocontacts were made by M. Krans during his PhD in '92-'96 (see figure below). He was the first to notice the exceptional length of the last plateau on conductance scans [12].

He, among others, also showed that during the elongation of the contact the last value of contact resistance just before rupture (i.e. when the conductance changes changes its behavior from metallic to exponential tunnelling) for all monovalent metals is very close to the quantum unit of conductance, 1Gq=2e²/h»(12.9 kW)^-1. For this purpose a technique known as conductance histograms was employed [9]. It should be understood that since every time during the breaking of the contact thousands of atoms are involved in a very complex rearrangements, no two conductance scans reproduce themselves in full detail. Thus the method of histograms allows one to see the statistical preference of conductance to obtain certain values. The following figure illustrates the principle:

Here each conductance versus electrode separation (proportional to Vp) scan is projected onto the Y-axis and the number of datapoints fallen onto each interval on Y-axis is counted and added to the previous one. This way one obtains conductance histograms as shown below.

The histograms vary in behavior, although one feature is clear: the first peak is always centered near 1Gq and there are almost no points between this peak and zero conductance, indicating that the last metallic contact observed always has this conductance value, and since we naturally expect this last contact to be one atom wide, we conclude that the last plateau of conductance at 1Gq corresponds to a point contact with a single atom in diameter.

This deep conclusion (supported by the recent STM-TEM experiments) gives a new insight into the long last conductance plateaus for Au, like the one shown in the middle panel of the figure above. Plateaus like this one indicate that we are pulling chains of single gold atoms! The idea is illustrated in the figure below:

To justify this hypothesis we performed the following experiment: we pushed two golden electrodes into each other and pulled them out of contact (all the way into tunnelling) thousands of times, and each time we recorded the length of the last plateau. Afterwards we constructed a histogram of the lengths of this last plateau, shown in the figure below:

In accord with our hypothesis the histogram shows a well-pronounced peak structure with up to 5 equidistant peaks, which means that the structure formed during the last conductance plateau pull-out prefers to be of some certain specific lengths. The distance between the peaks is 3.6 Å, which is quite close to the interatomic distance for Au (2.8 Å). This number agrees well with the recent TEM observations of atomic chains formation by Takayanagi et. al. [13] (~ 3.5 Å). (One should keep in mind that the experiments of Takayanagi were performed at room temperature, while we are operating at very low temperatures T=4.2K).

A molecular dynamics simulation by the group of J. Nørskov in Lyngby, Denmark showed the formation of atomic chains on gold in an indentation-retraction numerical experiment at T=12K, supporting further the proposed interpretation of our experimental data. The last stage of elongation in the simulation is shown in the figure below:

More tests showed other interesting properties of these chains. We tried to analyse how far one has to return (push the electrodes back into contact) after the chain is broken in order to re-gain electrical contact. In other words, we tried to see if the two parts of the chain remain free-standing after the chain gets broken. The results of this experiment (return distance vs. plateau length) as well as the proposed model for explanation are presented in the figure above. We propose (and it is perfectly revealed in the experiments by Takayanagi) that the chain collapses completely onto the banks of the contact once broken.

Another thing we tried to do is to swing the chains sideways testing their lateral elasticity. For this we used an ultra-stable STM developed by N. Agrait's group in Madrid. The experiment and the results of it are depicted in the figure below:

The conductance trace was recorded, and at a certain point (indicated in the left panels) the retraction of the tip from the surface was stopped and lateral movement (in Y direction) was performed. For a longer chain in the lower panels the maximum "swinging" amplitude is much higher (of the order of the length of the chain itself), indicating that the chain becomes more flexible as it lengthens. In other words, the chains have higher lateral elasticity the longer they are.

Another test was to look at the plateaus at other conductance values. The graph below shows the histograms of plateau lengths for the last three (conductance intervals are indicated in the legend):

Only the last plateau shows a multi-peaked structure and has a non-zero tail at very high lengths.

Conclusions:

We conclude that what we observe is the formation of atomic chains wich have the following properties:

• we can pull chains up to 8 atoms long

• electron transport through the chains is purely ballistic
  while the system itself is purely one-dimensional

• since there is no scattering in the chains they can
  withstand voltages up to 1 V or currents up to 80 uA

• chains of single atoms are flexible and elastic

• these exotic one-dimensional structures are very stable
  with lifetime of hours at cryogenic temperatures

As any other scientific discovery this one raises many questions, among which are:

• is there a limit to the length of these 1D chains?
• can one observe purely 1D effects (such as
  Peierls transition, Luttinger liquid behavior, etc.)
  in such chains?

Further reading:

For a review of the properties of atomic size point contacts see ref. 1 of the list. For further questions, feel free to contact Nicolas Agrait or Jan van Ruitenbeek.

Our work reflected in press:

This is the article which appeared in NRC Handelsblad, a Dutch nation-wide daily newspaper, on Sat. Oct. 24, 1998.

Literature (references):

1. van Ruitenbeek, J.M., Quantum point contacts between metals, in Mesoscopic Electron Transport (eds Sohn, L.L., Kouwenhoven, L.P. & Schön, G.) 549-579 (NATO ASI Ser. E., Vol. 345, Kluwer Academic, Dordrecht, 1997).

2. Sutton, A.P. and Pethica, J.B. Inelastic electron flow processes in nanometre volumes of solids, J.Phys:Cond. Matter 2, 5317-5326 (1990); Landman, U., Luedtke, W.D., Burnham, N.A. & Colton, R.J. Atomistic mechanisms and dynamics of adhesion, nanoindentation and fracture, Science 248, 454-461 (1990).

3. Agraït, N., Rodrigo, J.G. & Vieira S. Conductance steps and quantization in atomic-size contacts, Phys. Rev. B 47, 12345-12348 (1993).

4. Pascual, J.I. et al. Quantum contact in gold nanostructures by scanning tunneling microscopy, Phys. Rev. Lett. 71, 1852-1855 (1993).

5. Krans, J.M. et al. One-atom point contacts, Phys. Rev. B, 48, 14721-14724 (1993).

6. Scheer, E. et al. The signature of chemical valence in the electrical conduction through a single-atom contact, Nature 394,  154-157 (1998).

7. Moreland, J. and Ekin, J.W., J.Appl.Phys. 58, 3888, (1985).

 8. Muller, C.J. PhD thesis, Leiden (1992); Muller, C.J., van Ruitenbeek, J.M. and de Jongh, L.J., Physica C 191, 485 (1992).

9. J.M. Krans, J.M. van Ruitenbeek, V.V. Fisun, I.K. Yanson and L.J. de Jongh, The signature of conductance quantization in metallic point contacts, Nature 375, 767 (1995).

10. Binnig, G., Rohrer, H., Gerber, Ch. and Weibel, E. Phys.Rev.Lett. 50, 120 (1983).

11. Ruibo, G., Agrait, N.   Viera, S. Atomic-sized metallic contacts: mechanical properties and electronic transport. Phys.Rev.Lett. 76, 2302 (1996).