Nanowire synthesis has generally involved "bottom-up" assembly from atomic or molecular precursors. The alternative, "top down", approach involves starting with a big wire and reducing its diameter into the nanometer range. The problem with the latter approach is that it is difficult to make the removal of material from a large wire precisely uniform over its surface, thereby insuring that the roughness of this wire is preserved as it becomes smaller. The usual fate of the wire is shown in the figure below. Basically, since the electrochemical etch rate, dr/dt, is larger for smaller diameter sections of the nanowire, as it is etched it rapidly decomposes into an ensemble of particles.

In this paper we describe a method for circumventing this problem - a method for carrying out etching while retaining the intrinsic uniformity of the wire at the beginning of the etching process. During ESED nanowire electrodeposition, a constant, convection-controlled current, igrowth, is generally observed. Under these conditions, the increase in wire radius, r(t), is approximately linear with the square root of the deposition time, t_growth^1/2. For a hemicylindrical wire of initial radius r_o, r(t) is given by Eq. (1):

Coupling nanowire growth according to Eq. 1 with nanowire oxidation according to Eq. 2s provides a means for preparing dimensionally uniform nanowires of diminutive diameter. Consider the growth of a "rough" silver (n=1, V_m = 10.26 cm³/mol) nanowire consisting of a linear ensemble of r = 20 nm hemispherical nanoparticles, shown below left.

Within the context of the ESED experiment, this nanowire is produced by the nucleation of silver, followed by growth until these particles are, on average, tangential to one another (at the threshold of coalescence). This nanowire has a mean radius, r_ave = 15.7 nm with St.Dev._r = 4.5 nm and is characterized by the distribution of radii shown above. Now Eq. 1 can be used to predict the profile of this nanowire as it is grown under conditions of constant total current. As shown in Fig. 1a and b, growth under conditions of i_growth = 10^-9 A/cm² for t_growth = 600 s produces a smoother nanowire with r_ave = 65.8 nm and a standard deviation of the nanowire radius, St.Dev._r = 0.9 nm - the wire radius has increased by a factor of 4 while its roughness has been reduced by 80%. Eq. 2 can then be used to propagate the wire radii distribution as this r_ave = 65.8 nm nanowire is oxidized (n=1) under conditions of kinetic control using J_ox = 10^-4 A cm^-2 for 500 s. The wire profile and the width of the radius distribution remain constant as r_ave is reduced to 10.1 nm. In principle, t_ox can be extended with further shrinkage of r_ave to approximately 3 nm before electrooxidation begins to introduce breaks. This numerical calculation shows that a relatively large, but "smoothed" nanowire resulting from ESED growth can be shrunk to 1/20th of its initial radius without inducing roughening and without introducing breaks.

Can a similar result be achieved experimentally? We have addressed this question for nanowires composed of three disparate materials: Gold (a noble metal), antimony (a base metal), and bismuth telluride (Bi₂Te₃, a compound). Typical results for a kinetically-controlled etching are shown in the SEM images below.