It seems to be "common sense" that the equation should have the g* on the the other side, since by depositing A on B, we are replacing the (free) energy g_B by g_A. If we take into account the interface energy g*, thinking of it as a positive but small energy, then indeed we should have g_A + g* < g_B; the free energy must go down to be consistent with layer growth. This statement occurs in our review article (Venables et al. 1984) on page 405 in equation 1.3, correctly. These statements are traceable directly to the review by Kern et al. (1979), where they discuss this topic in much more detail, considering several sub-cases and possible mechanisms. You may however note in passing that figure 1 in our review was garbled, with inconsistencies between the text, figure captions and the figure.

Figure 5.3 derives from my more recent review (Venables 1994, figure 1) where I was wishing to compare compactly both the layer and the SK growth mode in what is here figure 5.3(a). The point here is that layer growth is definitely favored if g_A < g_B. However if the interface energy builds up as the layers are deposited, then at some stage a switch to islands can occur; this is the SK mode. But figure 5.3(b) illustrates the growth of islands of B on A, not A on B. So in this case the equation on page 146 describes island growth of B on A: g_B + g* > g_A; the interest in considering A on B, and then B on A, arises from the role of these two different interfaces in multilayers.

Just to prove that this topic has a jinx on it, there was also an error in the 1994 review text description of what is here figure 5.3(a). But the bottom line is that the statement on page 146 is incorrect for describing layer growth of A on B; it should be g_A + g* < g_B. Thankyou for pointing out this error.

New references for section 5.1