There is a marked particle size effect on both R and K. Table V gives an example of increasingly finer grinds in flotation with a different sulfide ore blend. Within the times of grind indicated, there is a tendency for both the over-all R and K values to increase with finer grind [see also Table III of reference]. This trend continues until sufficient fine material is produced (e.g., note the 17 minute over-all R and K values in Table V). Table V demonstrates clearly the greatest loss of Cu recovery is in the very fine sizes (-325 mesh) and the coarse sizes (+70 mesh). This is often observed also in plant practice.

Finer grind, of course, implies more complete liberation of Cu bearing mineral from the gangue matrices thus allowing for potentially increased recovery. However, finer grind also increases the surface area (hence reagent usage) and hydrodynamic limitations of the float cell which are forces counteracting the liberation benefits of fineness of grind. Increasing fineness of grind will thus eventually lead to a decrease in the over-all R and K values, as is the case with 17 minutes of grind in Table V.

Another important result demonstrated in Table V is the rate effects due to individual particle sizes. A series of separate flotation experiments were performed using cell charges consisting only of the particle size under question. Thus a K value for each size i was determined with the values indicated. If there were no interactions between sizes in a flotation charge consisting of a mixture of sizes, it should be possible to estimate the over-all R and K of the mixture by taking a weighted average (based on mass fraction of feed sizes to the cell) of the individual values. It can be seen from the tables, that the estimated parameters do follow the same qualitative pattern of change that the parameters calculated directly from the mixture follow. However, the parameter values estimated do not agree exactly with the parameters calculated directly from the mixtures. This indicates that there are in the mixture of sizes, particle-particle interactions which limit the assumption of strict size independence.

The real value of such an individual rate analysis of size is to indicate that the potential changes in R and K strictly due to particle size are of the same magnitude as those due to changes in collector and/or frother concentracion, pH. etc. (see Tables III and IV). This fact is often overlooked in laboratory chemical screening because of the extra complexity it introduces to the testing. Plant experience seems to be indicating that collector-particle size effects on R and K are not as great, for example, as are frother-particle size effects. This area obviously needs more detailed work which is underway in cooperation with various mining companies. This frother sensitivity to particle size helps to explain the variety of trends in frother dosage effects on R and K that are observed in the lab and plant. It also appears from very preliminary work that correlations can be developed between frother type and effectiveness as a function of particle size. In this paper, the emphasis has been on the correlation of frother type with an “average” or “typical” mix of particle size.

Another interesting demonstration can be made from the data of Table V. It is well known from industrial grinding experience, that the incorporation of classification into a grinding operation (closed circuit) leads to a steeper product size distribution on a plot of log % less than size versus log size than using a mill without classification (open circuit). What is less known is the relationship between the influence of these two modes of grinding on over-all flotation rate and equilibrium recovery. Figure 2 shows the comparison on the R and K values between 5 different simulated grinding tests with each test having a common size control point. The over-all K values quoted are using the individual K values listed in Table V adjusted by the mass-size distribution make-up as indicated by each curve of Figure 2. The over-all R values were calculated in a similar manner except that the set of individual Ri values used were the result of an average of the Ri values over the three grinding times indicated.

Upon comparing the calculated flotation parameters from Figure 2, it is obvious that for both open and closed circuit operations, an optima exists in the R and K sets of values in the size region corresponding to approximately tests 1 and 2. Even more important is the observation that the R and K optima for the closed circuit operation is significantly higher in both rate of mass removal and equilibrium recovery. Such an analysis indicates the direct correlation of the shape of size distributions from grinding and classification on flotation behavior. This is often not fully appreciated because of the routine use of one point product size control of grinding circuit performance (e.g. x%< some mesh size).