Search

The volume diffusion of Ni was studied in a nearly equiatomic $B2$ NiTi compound at 1285 K in the pressure range of 0\char21{}0.91 GPa. Our measurements, carried out by the tracer technique, proved that the volume diffusion coefficient is independent of the external pressure, and the evaluated activation volume is close to zero. Results may refer either to the interstitial diffusion mechanism or vacancy-mediated diffusion, indicating that in the latter case the vacancy concentration is pressure independent in the investigated pressure range.

X-ray scattering experiments on coherent CuNi multilayers are performed at various annealing temperatures. First, we show that the classical thermoelasticity theory can be applied in such nanosamples to link composition and strain fields at intermediate temperature. Second, when interdiffusion takes place at higher temperature, the satellite peaks measured at different annealing times indicate the presence of a layer-by-layer interdiffusion mode controlled by the asymmetry of the atomic mobilities in this system.

In our paper, we offer a natural resolution for a long-standing paradox in diffusion. We show that the growth rate of the diffusion zone (reaction layer) should not go to infinity with decreasing time (as $1∕\sqrt{t}$), just because the diffusion permeability of the interface is finite. Expression for the changeover thickness ${X}^{*}$ between the linear and parabolic regimes of the interface shift in phase separating binary $A(B)$ systems is derived in the framework of a deterministic atomistic model for diffusion. ${X}^{*}$ lies typically between $0.01$ and $300\phantom{\rule{0.3em}{0ex}}\mathrm{nm}$, depending on the composition dependence of the diffusion coefficient and the phase separation tendency of the alloy. While in ideal binary alloys with composition independent diffusivity, the deviation from the parabolic law practically cannot be observed, in real systems (where the diffusion coefficient can change several orders of magnitude with the composition), measurable deviations are expected as it was experimentally observed very recently in the $\mathrm{Ni}∕\mathrm{Cu}$ and $\mathrm{Au}∕\mathrm{Ni}$ systems. We also offer an atomistic explanation for the phenomenological interface transfer coefficient $K$. It measures the finite interface permeability (proportional to the jump frequency across the interface) and thus it controls the shift of the interface at short times (diffusion distances). Although it is almost exclusively accepted in the literature that linear growth kinetics are the result of interface reaction control, our results suggest that the linear or nonparabolic growth of a reaction layer on the nanoscale cannot be automatically interpreted by an interface reaction.

We report experimental evidence for the violation of the ``diffusional parabolic law'' ($x\ensuremath{\propto}{t}^{0.5}$, where $x$ is the interface shift and $t$ is time) as predicted recently by computer simulation [Z. Erd\'elyi et al., Phys. Rev. B 69, 113407 (2004)] for a binary system with restricted mutual solubility. Using x-ray photoelectron spectroscopy we investigated the shift of the interface between a thin Ni layer (3 nm) deposited under UHV conditions on top of a Au(111) single crystal during its thermally driven dissolution into the Au substrate. From the temporal evolution of the $\mathrm{Ni}\text{\penalty1000-\hskip0pt}2p$ and $\mathrm{Au}\text{\penalty1000-\hskip0pt}4f$ core level intensities at various fixed temperatures, a power law could be extracted for the time dependence of the interface shift $(\ensuremath{\propto}{t}^{{k}_{c}})$ with exponents in the range of 0.6--0.7. Thus, clear experimental evidence is provided that the kinetics of such a shift might differ from the well known parabolic law if restricted to the nanometer scale.

Proper transition rates in kinetic mean-field and Monte Carlo simulations of dynamic processes are very important to obtain realistic results. We show that by proper choice of the transition rates one can unify the advantages and eliminate the disadvantages of different models used in the literature [see, e.g., Senhaji et al., Surf. Sci. 274, 297 (1992 )( kinetic tight binding Ising model), Martin, Phys. Rev. B 41, 2279 (1990) and Cserhati et al., Surf. Sci. 290, 345 (1993)]. Furthermore, we also show that this choice cannot be considered simply as an extension of the previous ones. It contains an additional parameter controlling the ratio of the transition rates in the bulk and close to the surface sGsurf / Gbulkd, which influences the kinetics of surface segregation. We show that some results obtained previously need reconsideration. We illustrate, e.g., how the composition dependence of the transition rates (diffusion coefficient) influences the ‘surfactant formation and dissolution mode’. For example, the dissolution kinetics can deviate from the parabolic law on the nanoscale in accordance with our recent results [Erdelyi et al., Phys. Rev. B 69, 113407 (2004)]. Furthermore, we also present how the sGsurf / Gbulkd ratio affects the kinetic segregation isotherm.

Computer simulations were used to study the interplay of the diffusion asymmetry ~composition dependence of diffusion coefficient! and the phase-separation tendency ~chemical effect! in the kinetics of the interface shift during dissolution in a binary system with restricted solubility. We have found that—on nanoscale, taking into account only the diffusion asymmetry—the shift of the chemically sharp interface is not proportional to the square root of the time as would be expected from Fick’s laws but to t kc, where 0.25,k c,1 ~deviations from the parabolic law!. In ideal systems 0.5

As an extension of our previous paper [Phys. Rev. Lett. 89, 168901 (2002)], we demonstrate, using computer simulations based on a general continuum model, that the diffusional interface sharpening in ideal binary alloys (found recently from deterministic kinetic equations in discrete atomic approach) is still present even if the stress development and relaxations during interdiffusion or if the ``built-in'' stresses (caused by lattice mismatch or thermal expansion) are also taken into account. Thus we conclude that this diffusional sharpening, in systems with large diffusion asymmetry, is really a quite general phenomenon.