Secondary ion mass spectrometry (SIMS) spectra of Pd isotopes 104, 105, 106, 108, and 110 for the ND and PD layers showed that on PD layers, Pd content is about two orders of magnitude larger than in the ND layer. In Figure , SEM pictures of the PD and ND layers, respectively, prepared under the same conditions (66 V, 4 h), are presented. The density of the particles on the PD sample was larger while, in the case of the ND layer, small separated aggregates on the InP surface were observed. The PD sample exhibited charging in SEM; this is the reason we believe there are more nonconductive surfactant molecules which cover the majority of the InP surface. It suggests that surfactant aggregates without Pd particles are negatively charged in the solution, but the charging mechanism of nonpolar colloids is still far from clear. Because SEM observations of the particles from the colloid solution dropped on the TEM grid showed that the particles in the solution are not aggregated, it can be concluded that the particles form aggregates on InP during the electrophoretic deposition process.
SEM images. (a) SEM image of PD nanolayers and (b) ND nanolayers
An AFM profile measurement of the PD layer showed that nanoclusters on the surface are separated, and the distance between them is about 100 nm. This fact explains a significant lateral resistance when the distance between clusters is too large to enable quantum-mechanical tunneling. The size of aggregates grows with deposition time. The particles probably settle selectively in the vicinity of the previously deposited particles because of the higher gradient of the electric field. When the deposition time is long enough, e.g., 18 h, the layer turns to be laterally conductive.
A colloid graphite paint used for contacts makes a good Schottky barrier on an InP without a Pd layer, in contrast to formerly used colloid silver paint. I-V characteristics (Figure ) of the ND Pd layer with graphite contact exhibit a large rectification ratio and linear part gives Schottky barrier height 0.85 eV. Although an SIMS revealed less Pd isotopes on the ND layers, they have better rectifying ratios and sensing capabilities. We suppose that this fact is due to less coverage of the InP surface in combination with the porosity and high Schottky barrier of the graphite colloid paint on the InP.
I-V characteristics of ND Pd layers. Higher curve is for forward bias and the lower for reverse bias.
Formerly, we assumed that the PD layers were better for sensing because of their higher content of Pd. The response of these layers, equipped with silver contacts, to the presence of hydrogen was very quick, but the current change was not very high because of the impenetrability of the silver contact for the hydrogen molecules. The current change was more than two orders of magnitude. Now, we use graphite contacts, which are porous, and ND layers with lower density of coverage, so at the places, where are no Pd nanoparticles, the graphite contact touches the surface. The response of the structure on 0.1% H2
mixture can be seen in Figure . The current change was of 55,000 × when the current increased from its initial value 4.2 × 10-10
A to 2.3 × 10-5
A. The response and recovery times are of the order of 10 s, but the full recovery to its original value, probably due to the slow release of hydrogen diffused into the crystal lattice of the Pd nanoparticles, takes about 10 h [4
Current-voltage characteristic. Dependence of current on time of the ND Pd layer on InP interface in the presence of the 0.1% H2/N2 mixture. A forward bias of 0.1 V was applied.