Environmental Engineering Reference
In-Depth Information
terms of Eq. (2.22) with E ( D )/ E function in terms of Eq. (2.30) where S b ≈ 12 R is assumed
except for Sb and Bi [42], a similar size dependence is found (see figure 4). This agreement
indicates that the size dependence of γ sl ( D ) is originated from the size dependence of bond
energy. In terms of Eq. (2.28) or Eq. (2.29), γ sl ( D n , T n )/γ sl0 ( T m ) is within 10% of the value of
E ( D n )/ E b , which also confirms the validity of Eq. (2.28) or Eq. (2.29).
Another linear relationship between γ sl ( D n , T n ) or γ′ sl0 and T m with large scatter among
different groups of elements [11] has been rejected as the basis for an empirical rule in favor
of the correlation with H m [98]. However, direct calculations for γ′ sl0 values of transition fcc
metals with hard-sphere systems have shown a strong correlation with T m [88,98], which can
also be extended to γ sl0 ( T m ) for all fcc metals. Eq. (2.28-b) can confirm this correlation since
H m = T m S m and c 4 S m = 1.59 ± 7% nm J/mol K for all fcc metals listed in table 5. For other
elements, c 4 S m values show large scatter in a range of 1.04 to 5.44 nm J/mol K due to the
scatter of S m values. This indicates a disagreement for a linear relationship between T m and
H m . Thus, H m values, but not T m values, in general characterize γ sl0 ( T m ) value better.
Recently, it is thought [99] that Skapsky′s assumption, e.g. γ sl0 ( T m ) = γ sv0 ( T m )-γ lv0 ( T m )
[100], has been used in the derivation of Eq. (2.13) where subscript v denotes vapor phase.
Because there is no nucleation barrier for surface melting, solids can be hardly superheated
above T m , which leads to γ sl0 ( T m ) ≤ γ sv0 ( T m )-γ lv0 ( T m ). Thus, γ sl0 ( T m ) in terms of Eq. (2.13) is
the upper limit of γ sl . However, it is not the case. As a monotone function of T and D in terms
of Eq. (2.26), γ sl ( D , T ) takes its minimum at D n and T n ( D n and T n are the smallest values
possibly to be taken) and its maximum at D →∞ and T T m . Combining the values of D n / h
and θ listed in table 5, Eq. (2.26) gives the both limits of γ sl for elements as,
0.62 ≤ 3 V m R γ sl ( D , T )/(2 hS vib H m ) ≤ 1.
(2.31)
Eq. (2.31) shows the reason why γ sl0 ( T m ) in terms of Eq. (2.13) is the upper limit of γ sl
and overestimates γ sl ( D n , T n ) 50% to 100% [46].
γ sl (D n ,T n ) for Alkali Halides
In terms of Eqs. (2.26-b) and (2.27-b), the γ sl ( D n , T n ) values for alkali halides (ionic
crystals) are also predicted and shown in table 6. For comparison, the corresponding γ CNT
values [101] are also listed.
As shown in table 6, the difference between γ sl ( D n , T n ) and γ CNT is smaller than 20%
except for CsF where the difference reaches to 33%. Considering that both γ sl ( D n , T n ) and γ CNT
values are very smaller, tiny difference in absolute values will make the deviation larger.
Thus, it can also claim that an agreement between γ sl ( D n , T n ) and γ CNT is found.
Moreover, for alkali halides, most of the D n / h values are smaller while the corresponding
θ n values are larger than those of metallic and semiconductors elements, which leads to that
γ sl ( D n , T n )/γ sl0 values of alkali halides will be smaller than those of elements in terms of Eq.
(2.26).
Search WWH ::




Custom Search