Real time electron microscopy inspection of high temperature processes in W free standing wires J. G6mez-Herrero, J. I. Pascual, J. Mhndez, and A. M. Bar6 Departamento de Fisica de la Materia Condensada, hive&dad Authoma de Madrid, E-28049, Madrid, Spain (Received 29 July 1992; accepted for publication 4 January 1993) High temperature structural changes in a W wire are studied in situ by scanning electron microscopy. This is made possible by the microscopic size of the wire. Grain-boundary grooving and faceting are observed in real time. The constriction exhibits a strong thermal gradient (- 100 K pm-‘) which is the driving force for the migration mechanisms responsible for the observed phenomena. Free standing wires (FSW) have been fabricated in order to use them as mesoscopic system models to investi- gate quantum mechanics phenomena.* They also show in- teresting mechanical, thermal, and electrical properties.’ The free standing character of the wires is important in order to have them separated from the bulk substrate. FSW are currently produced by means of lithographic pro- cesses.2 In this letter we describe a method to fabricate tung- sten FSW by electrochemical etching. Due to the small volume of those devices, they can be heated at high tem- peratures with small power. The photon and electron emis- sion is so low that the device can be observed in situ at high temperatures by scanning electron microscopy (SEM). This allows a real time study of important morphological changes that occur when heating. Some of them are asso- ciated with the strong thermal gradient developed in the constriction. This is an interesting property of these de- vices. We have been able to observe processes such as grain-boundary grooving, recrystallization, and diffusion. Finally the wire breaks up in the form of a conical shape with a final diameter in the nanometer scale. This indicates that the device can also have interesting applications as a mesoscopic system. In practice we have fabricated a tungsten (W) neck in the following way: since we want to place it in the focus of a SEM beam, we start by spark welding a 75pm-diam tungsten wire to the two ends of a SEM filament support. Ordinary cold-drawn W is used. The wire, previously an- nealed to’ release stress and increase grain size, has a V-shape so that only the vertex is etched. The etching process is driven by means of a computer in order to produce pulses of voltage and to control the thickness of the neck by measuring its electrical resistance. The next step is to place the wire loop in the SEM (Jeol 840A) and to heat it up. Typically the dimensions are 10 pm in length and 1 pm in thickness. The small surface involved produces an electron and photon emission which is much lower than the secondary electron emission. Thanks to that, the SEM measurement is not perturbed in spite of the high temperature reached in the experiment. Another important property related to the small size of the systems is that the heat dissipation is exclusively re- lated with thermal conductivity and not with radiation or sublimation processes. Thus, it is easy to obtain the tem- perature distribution by solving the Fourier equation3 dT pi” --K--y. dx S(x) (1) Here p is the electrical resistivity, K the thermal con- ductivity, S(X) is the section of the wire, and I is the electrical current flowing through the constriction. We have solved that equation for two different microscopical shapes. The first one [labeled as (b) in Fig. ( 1)] represents a regular cylindrical constriction with thickness about 1 pm. With a power value of 50 mW, the Fourier equation gives a temperature at the center of the constriction of 2400 K. As we explain later, for long time heated constrictions an irregular shape appears. Those constrictions have been modeled assuming a sine wave for its profile. The resulting temperature distribution [labeled as (c) in Fig. ( I)] shows an oscillating behavior in the gradient with a maximum temperature of 2600 K. This increase can be interpreted as a decrease in the effective thickness of the wire. These cal- culations allow us to know the temperature gradients of the constrictions. Once we have discussed the heating process, we would like to discuss the structural changes observed by SEM. A noticeable fact is that constrictions obtained with the elec- trochemical etching show a characteristic prismatic shape with well-defined faces as can be seen in Fig. 2(a). This is a result of the annealing process and the electrochemical etching conditions. In particular, the prismatic shape is obtained by reducing the pulse length ( - 1 ms) and in- creasing the delay time ( - 1s) between etching pulses. We interpret these shapes as a result of low temperature diffu- sional currents while electrochemical etching. The thermal treatment makes all grains be oriented in the same way, and so regular faces appear along the wire during the etch- ing. If the voltage pulses are longer than 5-10 ms, the removal of tungsten from the constriction is strong enough to annihilate the effect of these diffusional currents and the cylindrical shape remains. The first feature observed by the initial heating is the appearance of a stairlike structure [Figs. 2(a)-2(c)]. This has been described in the literature as a grain-boundary grooving which is due to a strong reorganization of mate- rial by volume diffusion, although surface diffusion could also be important in this case due to the microscopic size of 1077 Appl. Phys. Lett. 62 (IO), 8 March 1993 0003-6951/93/l 01077-02$06.00 @I 1993 American Institute of Physics 1077 Downloaded 23 May 2005 to 161.111.20.5. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp