Chemistry Reference
In-Depth Information
can be used to predict static elasticity from the mass spectrum, if the sur-
face pressure (or surface tension) is known. This is illustrated in Figure 3
where we plot the predicted and measured H
o
vs.
3 curves for the three
samples of Figure 3 of Frew et al. (2006). The results suggest that the D1
scores and the resulting regression relations can be used to predict surface
elasticities at a given surface pressure reasonably accurately (to within 20
%) for the F1 fraction. Deviations of the predicted elasticities from the
measured values are mainly due to the fact that only D1 was used in the
prediction, thus ignoring information in lower ranked discriminants. In the
future, a more comprehensive approach using canonical correlation tech-
niques may improve this by incorporating multiple discriminant functions.
Table 2.
Linear fit coefficients a and b, respectively, and correlation coefficient r
at an elasticity (3) = a [D1 Score] + b
3
(mN m
-1
)
a
b
r (df)*
0.5
2.59
1.38
0.557 (28)
1.0
7.96
2.39
0.910 (28)
1.5
15.19
3.62
0.944 (28)
2.0
22.33
4.75
0.957 (28)
2.5
28.38
5.86
0.960 (27)
3.0
34.11
6.93
0.969 (27)
3.5
38.83
7.94
0.969 (27)
4.0
43.53
8.88
0.976 (27)
4.5
46.69
9.79
0.982 (27)
5.0
49.36
10.64
0.983 (27)
5.5
51.19
11.47
0.984 (26)
6.0
53.33
12.19
0.987 (26)
6.5
55.04
12.84
0.988 (26)
7.0
56.18
13.43
0.987 (26)
7.5
57.26
13.99
0.985 (25)
*Degrees of freedom = n-2
