Diffusivity of Cu

Copper introduces new problems in the fabrication of interconnects on chips, the most important of which is the diffusion of Cu into Si, Si02, and other dielectrics (4),... 

Figure 6.12 Comparison of NEB results for exchange (squares) and direct adatom hopping (circles) mechanisms for self diffusion of Cu on Cu(100).

The importance of the measurements that we have presented so far for the diffusion of embedded tracer atoms becomes evident when we now use these measurements and the model discussed in Section 3 to evaluate the invisible mobility of the Cu atoms in a Cu(00 1) terrace. The results presented in Section 2 imply that not just the tracer atom, but all atoms in the surface are continuously moving. From the tracer diffusion measurements of In/Cu(0 0 1) we have established that the sum of the vacancy formation energy and the vacancy diffusion barrier in the clean Cu(0 01) surface is equal to 717 meV. For the case of self-diffusion in the Cu(0 01) surface we can use this number with the simplest model that we discussed in Section 3.2, i.e. all atoms are equal and no interaction between the vacancy and the tracer atom. In doing so we find a room temperature hop rate for the self-diffusion of Cu atoms in a Cu(00 1) terrace of v = 0.48 s-1. In other words, every terrace Cu atom is displaced by a vacancy, on average, about once per two seconds at room temperature and about 200times/sec at 100 °C. We illustrate this motion by plotting the calculated average displacement rate of Cu terrace atoms vs. 1 /kT in Fig. 14. 

Size Effect. With increase in cluster size the occurrence of RA is significantly suppressed. In Au clusters of approximately 10 nm in mean size, the rapid diffusion of Cu atoms takes place only at the shell-shaped region beneath the free surface of an individual cluster, while pure gold is retained at the center of the cluster. In Au clusters of approximately 30 nm in the mean size, the RA does not take place. It should be emphasized that the critical size of the RA increases with the negative heat of solution and temperature. 

Copper introduces new problems in the fabrication of interconnects on chips. The most important one is the diffusion of Cu into Si, SiC>2, and other dielectrics [92], and reaction of Cu with Si forming silicides [109]. Diffusion of Cu through Si results in the poisoning of devices (transistors) and diffusion through SiC>2, dielectrics, leads to the degradation of dielectrics, which can result in... 

To achieve successful Cu interconnects, it is essential to deposit effective barrier layers to avoid the oxidation and diffusion of Cu into interlevel dielectrics such as Si02. Both oxidation and diffusion of... 

A mathematical model for the regeneration of Cu -saturated Dowex XFS 4195.02 ion exchange resin was developed, assuming that intraparticle diffusion of Cu" through the resin is a controlling step in the presence of 0.5 N aqueous ammonia as a regenerant, as shown in Equation 5. 

The effects of the initial Cu concentration upon the removal fraction are shown in Figure 4. In Figure 4, the symbols are the data, and the solid lines are the best-fit prediction from the removal model. Removal fractions increase with decreasing initial Cu concentration. These observations indicate that a controlling step for the removal of Cu from aqueous solution is diffusion of Cu through the resin. 

The effects of varying the relative amounts of solution and resin on the removal of Cu were examined by treating varying quantities of 80 ppm Cu solution with 0.2 g fixed quantities of the Dowex XFS 4195.02 resin at 25 C. The removal-fraction versus removal-duration data were applied to the removal model in order to identify effects of varying the proportions of solution and resin on the removal of Cu. As is shown in Figure 5, the intraparticle diffusivities of Cu" through the resin increase with the amounts of solution. This may indicate that increased mass ratios of solution to resin result in increased Cu removal. 

Figure 5. Effects of initial volume on intraparticle diffusivity of Cu through 16-50 mesh Dowex XFS 4195.02 ion exchange resin at 25°C and pH 1.5.

The following conclusions were drawn on the basis of experimental data for the removal of heavy metals from aqueous solutions in the presence of Dowex 4195.02 ion exchange resin, and the regeneration of spent Dowex XFS 4195.02 in the oresence of aqueous ammonia as a regenerant. ) The removal of Cu from aqueous solutions is controlled by intraparticle diffusion of Cu through the resin. Removal rates of... 

Kever et al. [10] have proposed that the main reason for the switching is the field-assisted diffusion of Cu-ions into the oxide layer that is commonly found at the Cu(TCNQ)/metal interface in devices with aluminium top-contacts, leading to the formation of conductive filaments through the oxide. We have... 

QCM measurements directly give in-situ kinetic information of reactions inside the cell. Figure 8 shows the tight correlation of dissolution with the square root of time. The resonance frequency change was converted to that of surface mass using eq.(3). This linear dependency indicates a strong possibility of diffusion-controlled dissolution. The rate-limiting step, in this case, seems to be diffusion of Cu(acac)2 to the ambient fluid. 

We can estimate the diffusivity of Cu(acac)2 in CO2 from the slope in the released-mass line with respect to Vt if is known. In this experiment, d>o is the equilibrium solubility of Cu(acac)2 in CO2 at the given condition. Table I summarizes the obtained diffusivity and data used for estimation. The solubility of Cu(acac)2 in liquid CO2 was calculated by the equation of the state suggested by Cross, et al.[2l ]. The difilusivities obtained in the liquid and the supercritical states are in the range of the value of the self-diffusion coefficient of CO2 [1,22,23]. 

Table 1. Diffusivity of Cu(acac)2 in CO and data used for the estimation.

Barrier metal such as Ta or TaN prevents the diffusion of Cu because Cu s solubilify and diffusivity are high when wire fhaf uses Cu is required. And also because Cu and low-k maferial are very soff, fhere is difficulf poinf fhaf CMP process must perform under low shear sfress and pressure. 

The oxidation of copper metal in a low partial pressure of oxygen produces cuprous oxide, CU2O, by a mechanism involving diffusion of Cu cations and electrons. The reaction is described by the chemical equation ... 

SIMS measurements on the precursor confirm intermixing of the layers occurs at room temperature (Figure 1.25). In particular, the Cu signal is clearly seen in both the Sn and the Zn phases. The majority of the Zn is still confined to the top of the precursor, while the Sn is distributed towards the bottom, suggesting it is mainly diffusion of Cu that is responsible for alloy formation. 

In the case of Cu deposition, the initial persistence of the PET pealcs is believed to be related to Cu penetration into the polymer. A confirmation of the Cu atom incorporation underneath the PET surface has been obtained by Ion Scattering measurements. For Al deposited on PET, the ISS spectra show a well-defined single scattering peak due to collisions with Al surface atoms, whereas in the case of Cu deposition, only a broad distribution with a high energy edge at Che Cu kinematic factor is observed. This indicates that essentially no Cu atoms are present in the uppermost monolayer since mainly inelastic and multiple collisions sequences are detected. Diffusion of Cu into the polymer bulk has also been mentioned in the literature for polyimide B substrates. 

Laz] examined the volume and boundary diffusion of Cu in flie Fe-Mo alloy with 0.7 mass% Mo in the (yFe) region. It was established that Mo did not effect the mobility of Cu atoms in the (yFe) phase volume, but increased diffusion mobility of Cu atoms along the grain boundaries. 

Che] Vibrating sample magnetometer Coercivity change diffusion of Cu into PtFe thin film. 

Table 1 contains the elemental compositional results obtained by ESCA for Cu coated polyimide films before and after T H exposure. As shown, the metal diffuses into the polymer matrix after the T H exposure. The same result was obtained for Ni. The diffusion of Cu into PI-5878 surfaces detected in our ESCA results correlates with the 4X increase in peel strength (after 500 hours at T H - 85 C 80%) measured (4) for thick Cu films deposited onto cured PI-5878... 

---