SURFACE PHASES OF SILVER AND TININ CRYSTALLINE AND LIQUID COPPER-BASED ALLOYS
G.P. Vyatkin, T.P. Privalova, T.O. Alekseeva,
A.E. Chudakov, S.I. Morozov, A.V. Gusev
Southern Ural State University, Lenin str, 76, Chelyabinsk, Russia, 454080
There is a strong surface segregation of Ag and Sn atoms in Cu-Ag and Cu-Sn alloys. Surface segregation changes physical and chemical properties of these alloy surfaces. Enrichment of the surface in these alloys is found by means of spectroscopic experiments and computer simulation, but in a solid state and for low Ag and Sn bulk concentration only. Our spectroscopic and computer experiments confirm these results and show that surface segregation exists in a liquid state too. Moreover, this effect can be found up to about 15 at.% of Ag and Sn in bulk. Our experimental method is temperature-programmed desorption (TPD); computer simulation is based on Monte Carlo method with Embedded-Atom Method (EAM).
INTRODUCTION
It was found many years ago there is significant surface segregation of Sn and Ag atoms in copper based alloys. This phenomenon is studied by means different experimental methods during the past few years (1). Now the theoretical study of this problem is growing extensively, for Cu–Ag solid alloys especially. The theoretical investigations in this area are often provided by means mean-field calculations (2) or approaches based on density functional theory, e.g. Embedded Atom Method (EAM) (3). We used the Temperature-Programmed Desorption method (TPD) (4) as the main experimental method and EAM in conjunction with Monte Carlo modeling as a simulation method (5). These two methods are appropriate both for solid and liquid alloys under consideration. Moreover, they give comparable results on structure and surface stuff.
EXPERIMENTAL
The measurements were executed on equipment, which is created on the basis of a magnetic mass spectrometer (4). The alloys are prepared from pure metals, such as oxygen-free copper (99,98 %), silver (99,99 %), tin (99,99 %) by melting in atmosphere of helium during 10 min at 1500K and with a annealing within 2 hours at 1100K. Investigated metal was placed in the holder of tungsten. The cleaning of an investigated surface from impurities and molecules, which one are adsorbed from a gas phase, was made by desorption in repeated heatings with melting of metal. Experiments were provided in conditions of vacuum 10-5Pa at partial pressure of oxygen less than 10-7Pa. Intensity of desorption flux I at continuous heating and consequent cooling with constant speed (from 5 up to 20Kps) in temperature range from 400 up to 1500K was registered for following spectral lines of masses: 53,5 and 54,5 (Ag ++), 107 and 109 (Ag), 63 and 65 (Cu), 118 and 119 (Sn).
The parameters of desorption of atoms Ag, Sn and Cu were instituted from experimental TPD-spectra of these fractions, which can be described by next equation:
I =a T -1/2 N x Cexp(-E / kT),
where I is the flux of desorption of fragments; a is the equipment coefficient; Т is the temperature of metal; N is the surface concentration or number of fractions per a unit area; x is the order of desorption (х=1); С is the exponential factor; Е is the activation energy of desorption; k is constant Boltzmann’s constant.
Value of desorption activation energy of Ag, Sn and Cu atoms is instituted on linear lease of ln(I T 1/2) – 1/T dependence. Method, which one is basis on the first equation, are used for definition of surface concentrations. This method are used a measurement of relative variations of flux of desorption of each component of alloy at phase change of melting (or crystallization) from TPD-spectra:
а1=I1liq/I1sol; а2=I2liq/I2sol .
COMPUTER SIMULATION
Our computer simulation is based on Monte Carlo modeling in conjunction with EAM (3). The transition probability for atomic structure changes is evaluated from total energy. The total energy is a sum over all individual contributions:
,
where the 1st term is embedding energy, or energy to embed atom i into the electron density r, the 2nd term is short-range electrostatic pair potential. Electron density ri is approximated by the superposition of atomic densities.
Computer simulation which is based on Monte Carlo modeling of (N,V,T) ensemble in conjunction with EAM allows to explore different metal properties. For alloys it is possible to evaluate the surface segregation degree, the microscopic surface structure and concentration profile. In order to provide such simulation it is enough to fit a few elastic and vacation properties of pure metals, and to get a set of parameters for description of these elements in EAM. These parameters are appropriate both for pure metals and alloys.
In order to verify the evaluated parameters and to check the program complex, the surface relaxation and surface energy of silver and copper were calculated:
s=(Е–NЕsub)/S,
were E is the energy of ensemble with surface S, and this energy is evaluated according to EAM, and NЕsub is the energy of the same number of N atoms in bulk.
These properties simulation shows good results, which are similar to experimental ones (Table 1).
Table 1
Structural and energy properties of pure surfaces
(Monte Carlo with EAM and experimental)
Metal / Surface energy, arb.u. / Surface relaxation, 10-9 mEAM, (111) / Experimental / EAM, (111) / Experimental
Cu / 1240±200 / 1790 / –5,0±2,0 / –2,6
Ag / 675±120 / 1240 / –6,0±2,0 / –3,8
The next step of parameters’ verification was simulation of pure silver and copper melting. The atom distribution analysis in solid and liquid state shows the presence of melting phase transition in slab under simulation. Also good agreement with experiment was found for melting thermal effect calculations.
RESULTS AND DISCUSSION
TPD experiments with Cu1-xAgx (x=0.005 .. 0.13) alloys
During the heating and cooling of polycrystalline alloys Cu1-xAgx (x=0.005 .. 0.13) TPD experiment shows stimulating of surface segregation and desorption of silver. This is confirmed by a sharp increase of desorption rate during crystallization and melting. This increase is due to intensive segregation of silver. Evaluation of silver surface concentration from TPD spectra shows that it is 60±10 at.% in solid polycrystalline state and it is 23–35 at.% in liquid state.
Surface concentration of silver atoms in liquid and solid state for different temperature ranges are listed in Table 2.
Table 2
Surface concentration of silver in Cu–Ag alloys
Ag bulk concentration, at.% / Parametersa1=(Isol/Iliq)Cu
a2=(Isol/Iliq)Ag / Surface concentration of Ag, at. %, and corresponding temperature ranges
Liquid state / Crystalline state
0,3 / a1=0,47±0,05 / 1350…1450 К / 970…1100К
a2=2,32±0,10 / 24±5 / 60±10
0,86 / a1=0,69±0,02 / 1370…1420 К / 950…1100 К
a2=1,56±0,05 / 30±5 / 49±7
1,3 / a1=0,51±0,03 / 1370…1450 К / 1060…1100 К
a2=1,96±0,07 / 28±5 / 61±5
3,0 / a1=0,51±0,08 / 1300…1350 К / 900…1050 К
a2=2,72±0,15 / 18±7 / 55±13
3,6 / a1=0,57±0,05 / 1350…1450 К / 950…1050 К
a2=1,71±0,05 / 32±5 / 59±9
12,8 / a1=0,56±0,02 / 1250…1280 К / 850…1000 К
a2=1,68±0,12 / 33±8 / 60±10
Considering these experimental data with respect to width of each temperature range, the increase of silver bulk concentration finally gives an increase of surface concentration in a liquid state, and a decrease of surface concentration in a solid state.
Simulations of Cu1-xAgx alloys
In a liquid alloy the increase of silver bulk concentration during simulation also gives an increase of surface concentration of silver atoms, and the decrease of surface concentration in a solid state too. Of course, values obtained during simulation differ from experimental ones but ranges of calculated surface concentration and the measured one are overriding for each bulk concentration values. The surface concentration of silver is higher than the bulk one both for solid and for liquid state. Possibly, the reason of such surface segregation is the size mismatch between Ag and Cu atoms, which doesn’t depend on the temperature and structure. Also there is a dependence of surface concentration on surface index. In Cu–3 at.% Ag alloy surface concentration is 11 at.% on (110)Cu, 25 at.% on (100)Cu, 60 at.% on (111)Cu. If to remember that (111) fcc surface has the minimal surface energy, it is good agreement with TPD experiment (see table 2).
In the second atomic layer concentration of silver atoms is less about 10 times than in the surface layer. The third layer contains approximately the same quantity of silver atoms as any bulk layer. Thus, it is possible to approve, that in computing experiment the monotonous structure of concentration is observed. An exit of separate atoms of silver in ad layer also is take place in simulation.
It is observed the change of silver contents in volume on 300 % corresponds the change on 10 % in a surface only. It can be explained by the segregation of silver occurs due to formation of specific surface phases. In a surface of a crystal (111) Cu (Ag) silver and copper join in separate surface phases; the Ag atoms are forming p(1х1) structures, the placement of separate Ag atoms in ad layer above a surface monolayer also is observed. In liquid alloy the surface consists of copper atoms mainly, It is smoother, without precisely expressed ad layer. The significant part of Ag atoms stays separately from other silver atoms, i.e. in an environment of atoms Cu. In a solid state there are several surface phases, which correspond to different indexes of monocrystal. In the liquid state surface structure p(1x1)Ag is preferable. Also this structure exists on (111)Cu surface of monocrystal. If to compare the surface concentrations of silver in (100), (110), (111) copper surface, then it is possible to look following law: than more densely packing of a surface, then segregation degree of Ag is more. By the way, the structure of a surface liquid Cu has appeared very similar to a side (111)Cu according to our simulation .
In copper monocrystals with low silver contents the analysis of concentration shows that the gradient of concentration of silver atoms practically disappears since the 3rd layer. This allows to consider layers below as a volume layers. In more concentrated solutions the volumetric stratification on phases enriched with silver and copper is observed. In the liquid alloy (fig. 1) the top monolayer contains almost all Ag atoms of slab. After that sharp jump silver enrichment is absent almost.
Fig. 1. Concentration profile of liquid Cu–3 at.% Sn alloy
T=1500 K; Sn is black; Cu is white
TPD experiments with alloys Cu1-xSnx (x=0.005.. 0.17)
In alloys Cu1-xSnx (x=0.005.. 0.17) the experiment hold by TPD method gives that nature of segregation of tin is about the same as segregation of silver in a system Cu–Ag with the small differences. The surface monolayer contains approximately twice less quantity of 2nd component than in previous system both in solid and in liquid state. Then the computer simulation demonstrates that in second, third and even in the fourth layers there is a large surplus of Sn atoms respectively to bulk, i.e. the segregation encompasses several layers. Moreover, the ad-layers replacement of tin in a surface is confirmed by experimental values of an activation energy of desorption Sn, which are close to a dissociation energy of dimers Sn2.
Simulation of alloys Cu1-xSnx
Results of surface modeling of alloys Cu-Sn of different composition (from 0.5 up to 17 at. % Sn) in solid and liquid states demonstrate a high scale of surface segregation of tin. The coverage of a surface by atoms Sn in solid state reaches about half of monolayer. For melts the surface concentration Sn is less, but if a bulk concentration of tin increases then a surface concentration increase too. Formation of phases on the basis of an edge (111)Cu, and (100)Cu is equally favorite for a monocrystals. The obtained concentration values are in the quantitative agreement with accounts of a surface concentration of tin in alloys of the same compositions executed on the basis of dates TPD-experiment.
Computer simulation gives the snapshots of alloy surface. It makes possible to indicate structures, which are specific for different indexes of a surface: for example, р(1х2)Sn structure for an edge (111)Cu and for liquid surface, and phase with (2х2) Sn structure for less packed edge (100) Cu.
In a surface (100)Cu the structure c(2х2)Sn is the most stable. In a surface (111)Cu, as well as in liquid state, the structure р(1х2)Sn is the most stable. These surface structures are approximately equivalent on the formation energy and their combination is possible. In surface layer it is possible to find the chains from Sn atoms, which are placed in one layer with atoms of copper. Their existence can be explained by following. In a weak solution the Sn atoms substitute atoms of copper, but the formation of two- or three-dimensional fields from atoms of tin appears energetically unprofitable because positive cores are repulsing and these cores would appear too close in solid area. The realization of zigzag line-ups in Cu–Sn alloys allows to reach a maximum of electron density in the location of each of atoms of tin, and consequently, minimum of the contribution from energy of attraction between atom core and electron density in this point.
The small tendency to oscillating allocation of concentration of tin is revealed during analysis of concentration profiles. It is visible from tab.3 for two melts and from a fig.2 and tab.3 for two mono-crystalline compositions. The formation in the bulk of structure, which is similar to a-phase, is appropriate to all compositions, that corresponds to the phase diagram of a system. The typical scale of oscillations is 3-4 layers.