noi and noj is the initial amount of the pure substance Aj used to generate the solution. The tpj's are related to the concentration variables xpoj, VOJ = nojlu^,
The dependence of any cpj on the composition of the phase can be represented by power-series expansion in ipoi and ^03 as introduced in (1.84), or
Bq>0 = 95* + = Ay,i^f
= ^03 ' + Xs'NR'lvvi.oi
+ Any*)
(21) (22)
+ Cx -
1
^0*1"1 ^2^0*2^1.01,
Unauthenticated Download Date | 9/24/15 11:21 PM
(23)
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems. Ill
Bv02 = Ay,2{q% + Xs lxF*i
1AV91.01
+ A21cp*)
+ Avl[(Vy>11.io-2X21X20K*20) — Xs
1
1431
{YJ^IANCP*
-
A2i/105 m _ 1
17
18
19
20
21
22
257.1 407.9 412.3 459.8 502.5 528.0 577.5 591.1 619.6 624.2
225.1 355.5 358.8 398.9 433.4 453.5 491.7 502.1 523.7 527.1
166.0 262.3 265.3 294.8 320.7 335.9 364.9 372.8 389.3 392.0
105.3 167.3 169.1 189.1 206.4 217.3 237.7 243.2 255.2 256.9
c^/mol m - 3 KoVm2 mo]-1 7.7372 29.413 30.721 53.033 91.446 129.79 270.14 339.53 589.28 654.65
191.5 306.2 309.4 347.4 383.2 405.1 448.4 460.7 486.4 490.0
Number of coefficients assumed unequal zero ^cKi/m-1 mol -2
t(Be Ki)
^4ci/m3 mol -1
t(Ae 1)
i?cK2/ni2 mol -3 t(B c K2) Act/lO-3 mKl — Ay>l Kq3 + B m 2^02 — Ay,2 Kq3 t^Q2 + By,K3(Vo2)2 -Ay,
24.1
K 0 3 (^O 2) + +
+
2
---.
(26)
Considering K03/^02 a s a random variable in dependence on KQ3 , If Q2 ' K03 V02 > (Vcfe)2* K0+3(^02)2, ... , analysis of variance (F- and t-test) and multiple regression lead to the number of coefficients significantly different from zero and to estimators for the coefficients a and AW0L. As concentration variables xpQ2 were used Cq2, Wq2, Xq2 and r ^ (compare Table 1 [36]); an example of the results using y>+ = c+ is shown in Table 2. From the results it may be recognized that the first four and only the first four coefficients of (26) are significantly different from zero with a statistical significance better than 95%. Multiple regression according to (26) assumes a random distribution of errors of Kq3/^02, but accurate values not only of ipQ2 but also of K^ and the latter assumption disagrees with experimental conditions. Hence the final values of the coefficients Aw\, Aw2, -BvK1 and BvK2 were determined by iterative non-linear Gauss-Newton approximation using the values of the coefficients ob-
4 66.98 382.4 0.1511 236.6 0.497 79.8 0.722 76.8
£cK3/10-5 m5 mol-4
t(Bc K3) F
3
5897
± 0.35 ±0.0013 ± 0.012 ± 0.019
5 66.87 ± 315.1 0.1504 ± 155.3 0.482 ± 29.2 0.698 ± 25.5 -0.20 ± 0.94 0.9
Unauthenticated Download Date | 9/24/15 11:21 PM
0.42 0.0019 0.033 0.054 0.41
Table 2. Results of multiple regression according to (26), Fand t-test for absorption measurements at
v = 18 • 105 m-1.
1433
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems. Ill
Table 3. Best estimators for the coefficients A v i and A w2 at 298.15 K. V>J
Xfß
Avi y e
Ay,2 (V0)2
cj
1 mol m - 3
0.15098 ± 0.00032
Wj Xj rj
1 1 1
(7.335 ± 0.072) • 10~4 71290 ± 690 79420 ± 770 80980 ± 770
1476.3 ± 3.2 1557.6 ± 3.3 1559.6 ± 3.3
tained from (26) as starting values and assuming a random distribution of errors of K^ and accurate values of y>Q2. The values of the estimators Aw\ and Av2 at different wavenumbers are equal within deviations as expected from the errors of the measured quantities. Averaging over all investigated wavenumbers leads to the best estimators as listed in Table 3. With the estimated values for Aw\ and A w 2 and the values of K^ and ip^, the left-hand side of the following equation can be calculated: Koa(l + Ay, i yjfe +
= JBvKi
Ayzjip^)2)
V>02 + BwK^+.
(27)
Considering the left-hand side of (27) as a random variable in dependence on ipQ2, linear regression leads to the best estimators for B v K l and B v K 2 ; some examples choosing ipo~2 = Cq2 are listed in Table 4. With the best estimators for Aw\, Aw which describes according to (7) the first-order activity effect on the 1:1 complex formation. Assuming 77^11.10 = 0, it becomes
assuming K* 21 = 0 it will be rj w u. 10 = A ^ / A y i , the values are listed in Table 5. If the assumption r)vllaia2 = 0 for ai, cn.2 = 0, 1 , 2 , . . . , which is equivalent to ^ 1 1 = 0, is met for one concentration variable xp^j* it usually will not be true for another concentration variable ytp as has been discussed at (IV.44). The relations between the coefficients 77^11.10 are rjcii.io =
M
F*
Vwi 1.10 + M1 Ffo - i f 2 V^i
= FQI^zii.IO + F*2 — F*i = V*lVrll.io+V*2.
(28)
The values of v . aia2 , which are different from zero for all but at least one concentration variable y)(j\ cause contributions to the estimators of Ay,\ and Ay,2 as listed in Table 3. Hence the values of Kfn and Kf2l, for example, obtained from Ac 1 and ^4C2, show some deviations from the values obtained from K * n and K* 2l [if =t= c) using (IV.43) as may be seen from the data in Table 5, the deviations of Kf 21 are even larger if the contributions due to the term K*nr)y,ii.io are taken into account (sixth row of Table 5). The problem of the choice of the most appropriate concentration variable for the evaluation of data was investigated by many authors
Be K2/IO-5 m 2 • mol -3
^*/m 2 • mol -1
4l/m2 • mol -1
39486 ± 50367 ± 49667 ± 41588 ± 30986 ± 20474 ±
341.9 ± 0 . 8 442.6 ± 1.0 461.3 ± 1.0 405.2 ± 0.9 298.7 ± 0.7 189.2 ± 0.5
538 ± 5 687 ± 7 677 ± 1 567 ± 6 422 ± 4 279 ± 3
25 28 23 12 19 17
M2
Table 4. Best estimators for the coefficients BcKI and BCK2 and molar absorption coefficients of the complexes HMB-TCNE and (HMB)2-TCNE in CC14 at ,298.15 K.
Unauthenticated Download Date | 9/24/15 11:21 PM
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems.Ill1434
1434
[20,24,29,30,49—51]; the best choice would be that one, where the terms r]viK(Xl(X2 iffy ^ol i*1 ( 7 ) g i y e the smallest contribution to An analysis of variance (F-test and t-test) of the data according to (26) including the term A V3 K^ (y^) 2 on the righthand side, does not lead to a statistically significant assertion even if the term seems to contribute the least for ip^ = c02I this gives some preference to this concentration variable and will therefore usually be chosen for the further evaluations in this study. The imperfect knowledge of the best choice of the concentration variable causes errors of the equilibrium constants Kfn and Kf2l of the order of the difference of the different values listed in Table 5 (first column and fourth to sixth row), i.e. 0.3% for Kfn and 4 % for K*n, and hence larger than the errors listed in Table 5, which are, as usual in this study, 95% confidence limits based on the Student's t-distribution ( ± ti-aS, where s is the standard deviation of the mean and a = 0.025). For optical absorption measurements, where the density D0 was identified with the absorption coefficient a, the MMQ's 99* occurring in (14) and (15) must, according to Sect. 9.4 of paper I, be identified with the molar absorption coefficient x* of substance A j . Since in the considered wavenumber interval it is x* = x* = x* = 0, one can assume Kxj.a2a3 = °>
for
« 7 = 1, 2, 3; a 2 = 0 , 1 , 2 , . . . ; oc3 =
0, 1 , 2 , . . . , as has been discussed in paper I. Hence from (23) and (24), it follows that the molar absorption coefficients of the 1:1 complex are x*t = Table 5. Equilibrium constants K*n CC14 at 298.15 K. rpj ipe
??vii.io V e
Kf2ll 10-4 Kt21/IO-4 ^vii-io V e
^cK2 = AC2(X*i — 2xt) + A c i Acxll.io' hexli.io describes according to Eq. (14) the solvent dependence of the molar absorption coefficient x\\ of the 1:1 complex in first-order relating to Co2 • Assuming Acxii.io = 0, then the molar absorption co. efficients x2l= ^cK2Mc2 of the 2:1 complex can be calculated. The values are listed in Table 4 and drawn in Figure 1. Assuming Ä"*0 = 0 and K%i = 0, then BcK2 — Aci (HiiVcll.lO + Xcxll.io) and the values of XCxii.io can be calculated, as they are represented in Figure 1. According to the above results the dependence of the optical absorption coefficients of solutions of HMB and TCNE in CC14, on their concentrations, can consistently be explained by two different models. According to model I, there exist two complexes HMB-TCNE and (HMB) 2 -TCNE, each one having an absorption band in the same wavenumber interval as shown in Figure 1. According to model II there exists just one complex, namely HMB-TCNE, and the further concentration dependences of the optical absorptions are caused by (1) the coefficient ^cii.io, which represents the activity effects on the quantity Kc\i of the complex formation, and (2) the coefficient Ac*ii.io> which represents the solvent dependence of the molar absorption coefficient x\\ of
and K*e>l for the formation of the complexes HMB-TCNE and (HMB)2-TCNE in
Remark
cj 1 mol m~3
1 2 3 4 4 5 6
0.15098 ± 0.00032 (7.335 ± 0.072) • lO"4 (4.858 ± 0.049) • 10"3
0
BcKiIAcI ; the values are listed in Table 4 and drawn in Figure 1. Assuming as above K%0 = 0 and ^cii.io — 0 it is
wj 3346.6 ± 251900 ± 48.29 ± 0.15119 ± 7.477 ± 7.592 ± -0.7414 ±
1 6.8 2400 0.48 0.00032 0.071 0.072 0.0009
xj
1557.6 ± 79420 ± 50.99 ± 0.15121 ± 7.485 ± 7.608 ± -0.8371 ±
1 3.3 770 0.51 0.00032 0.073 0.073 0.0009
Remarks: 1. 2. 3. 4. 5.
rj 1 1559.6 ± 3.3 80980 ± 770 51.92 ± 0.51 0.15140 ± 0.00032 7.632 ± 0.073 7.902 ± 0.073 -1.8371 ±0.0009
From the values Awx, Table 3. From the values A v 2 , Table 3, assuming K* 20 = 0 and = 0. = From the values A v i , AV2, Table 3, assuming K*2o — 0 a n ( l Calculated from K * n or if* 2 1 , respectively, using (IV. 43) (assuming r?vn. 10 = 0). Calculated from K*21 using (21) and (IV. 43) assuming ffcn.io = 0 but values of rjv 11.10 (v 4= c) as calculated from (28) and listed in the last row of Table 5. 6. Calcidated using (28) assuming r?cn.io = 0.
Unauthenticated Download Date | 9/24/15 11:21 PM
W . Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems. V
1435
the complex. A decision between these models is only possible if other data are known; for example, if the coefficients ^eii.io and ACxii.io c a n be determined independently. According to (7), (4) and (1.134) it is (ZFC n \ * / e i n (Jt»fUlfeii)\* ^ = ( ä > O c 02 \ OC02 /WO1,CO3,0. \ /woi.cos,«.
r)cii.io=
(29)
where / c / is the activity coefficient of the substance A j defined by (1.132) and (1.133). From (1.136) follows ein/c7\* C
0 O2
1
/ « O l , C03, » ,
RT
I 99mi \* \
+ noi
/«01.C03 5 / fymi \
„.(*=«)
+
S c 02
8C02 \
/n/>d(
- 1 1 -
Fn01
noi, co3,
hcvl. 10 F° ^01
(30)
where g m i is a contribution to the model molar Gibbs energy gi of the substance A / in the solution defined by (1.120), R is the gas constant, FQI is the molar volume of the pure solvent (CCI4), vi is the model molar volume of the substance A/ and XCvi. 10 is one coefficient of the power-series (14) for cpI = vi. Application of (1.17) and introducing (30) into (29) leads to ^cll.10 =
1 RT
/ dffm2 \ *
/ fym3 \
\ ÖC02 /«oi, Cos,«.
\ 9C02 J noi, Cos,'», \ 9C02 Jnoi, Cos,»,
+ Kl - v t — vf) (1 -
_
/ fymii \
(31)
+ (l/^0l) Uwll.10 - hv 2.10 - ^cv 3.10) .
The quantities gmi are dependent on the composition of the solution, a main contribution being caused by the electrostatic interactions of a solute
molecule with the surrounding molecules of the solution [52]. Neglecting the other composition dependent contributions to gmi, which is at least a rough approximation for the considered dilute solutions, g m i can be represented by [52] Na _ 9ml = 9ml
{[1
2
—
tr (£r) Og/]"! h (eT) (32)
-[l-f7(£r*)ag/]-if7(£*)}p.g/, where Nx is the Avogadro constant, (x g j and a g / are the electric dipole moment and the polarizability of a molecule A/ (in the electronic ground state) and e r is the relative permittivity of the solution with the value ef of the pure solvent. The tensor function f/(e r ) is defined by (11.35). According to (32) g m i can be considered as a function of e r , and hence
(HMB)2-TCNE J =21)
/ j w y \ OC02
J
^
J noi, Cos,», \
X Ö£r
l
p
X
,
]&, \ OC02 )noi, Cos, »,
where because of (32) 1
1 /dgmA
(33)
(34) ~
/ 9f/\
RT\ deT
C/10'm" Fig. 1. Molar absorption coefficients xfi and the complexes HMB-TCNE and (HMB)rTCNE in CC14 at 298.15 K and coefficients XCx 11.10, estimated from the data, and (2c*11.10) calc » calculated according to (37). For (^cxll.io)calc the scale factor on the right-hand side of the figure has to be multiplied by 10"3.
and because of Eq. (11.35) 9//A\
X=
fix
=
Ö£r M
(er — l ) [ e r — «/A(er — 1)] '
x,y,z.
Unauthenticated Download Date | 9/24/15 11:21 PM
(35)
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems.Ill1436
1436
In Sect. 4.2 the limits v*, v* and of the model molar volumes are estimated and it is shown that one has to assume Xcvi. 10 = 0, I = 2, 3, 11. In Section 4.3 the quantities p. g /, agi, fix and xi\ and the derivative (öer/Sco2)*0i a r e estimated. With these data, listed in Tables 12 and 10, Eqs. (31) to (35) lead to (^di.io)caic = 1.28 • 10~5 mol- 1 m 3 . The model molar absorption coefficient x\\ of HMB-TCNE depends on the relative permittivity £r and the square n 2 of the refractive index of the solution, which may be represented by an equation similar to (33) of a previous paper [53] xi{er, n22W4r= )\v — i3S « / ( v ( e * n*2) — Avi{eT, n2; ef, n*2)) {Agai{ef, n*2) — aga/[JSRM/(er, n2) -Enm(ef,n*2)]\\2,
(36)
where S =
where f/(e r ) is defined by (11.35) and (11.36) and f / = fi{n 2 ). p,a/ is the electric dipole moment of the molecule A/ in the considered excited electronic state. According to (22) [55] it is again with a a/ = Bau BcZ2, Bro, BcTl and BcT2 as given in column "3" of Table 9.
Unauthenticated Download Date | 9/24/15 11:21 PM
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems.Ill1440
1440
Fig. 3. Values of the limits Z ^ and ZQJ" of partial molar electric susceptibility and refraction of TCNE for solutions of HMB and TCNE in CC14 at 298.15 K. The curves are calculated using (19) and the data in Tables 2 and 9. The crosses ( + ) represent values of Z ^ according to the data of Briegleb, Czekalla, and Reuss [16] ; for these values the scale factor on the left-hand side of the figure has to be multiplied by 2.
for the further evaluation of the quantities BZO,..., BC1>2 according to (22) to (24) some of the coefficients Acjj.ai0t2 and AC£'j.aia£2 are needed, which are defined by (14). The further evaluation of the quantities £* and which may lead to the permanent electric dipole moment p. g j or the static polarizability a g j and the polarizability A g j at the wavenumber v used for the investigation (v = 9.259 • 105 m _ 1 ) of the molecule A j in the electronic ground state, must be based on a suitable model, for which we choose the extended Onsager model in ellipsoidal approximation [37]. In accordance with this model the quantities £* and as represented by (11.41) and (11.51), are dependent on £r and n 2 , respectively, at least at fixed values of the temperature T and the pressure p. Hence A, C?/'10
(—\*
-
8Cl\* I Ößr V
(—VI
\ \ 0£r )», [\OCO2/co3,0,
\0CO2 /Co3,a,
and The limit Z* 2 of the partial molar electric susceptibility is related to the limit £* of the model molar electric susceptibility of HMB according to (1.78) and (1.83) by Cj =
(49)
— AC£i.io/Foi
A similar equation holds for where Z*j is to be substituted by Z'0* and A^i.io by ^cs'i.io- Similarly Table 9. Results of multiple regression according to (19), F- and t-test for refractometric and permittivity measurements of solutions of HMB and TCNE in CC14 at 298.15 K. Number of coefficients assumed unequal zero £zo/10- 5 m3 mol"1
t(Bzo) -BeZl/10-7 m6 mol"2 t(BcZi) £CZ2/10-9 m9 mol- 3
t(BcZ2) Ä-Z3/10-11 m^2 mol-* t(BcZi) F JSz'o/10-5 m3 mol"1
t{Bro)
BcZ'i/10-7 m6 mol- 2
t(BcZ'i)
ÄZ'2/IO-9 m9 mol- 3
t(BcZ'2)
Bc Z'3/IO-11 m3 mol"1
t(Bc Z'3) F
2 -215 5.6 1677 27.2
3
—
16.5 ± 4 . 8 7.0 1081 ± 10 210.9 224.8 ± 3 . 4 134.6
—
—
—
—
18000 -110 14.30 ± 0.34 85.8 5.3 550 231.17 ±0.74 16.7 639.7 — 120.64 ± 0.24 — 1022.9 —
—
—
—
—
—
1000000
bsi.
lpC/r
(51)
2 \ 0CQ2 / C03 , 3CjV
02er
0£r M
0Cn 02 /Co3,
02C/\* / 8er \*2
+
0e2 )&i \0co2/Co8,ol
Similar equations hold true for ACfj.oi and AC£/.o2> where the indices " 2 " and " 3 " have to be exchanged. The coefficients AC£/.n and AC£/.21 can similarly be expressed. Representations for Acf/>ai(Xa result from similar equations, where eT is substituted by n2. The derivatives (öer/0Co2)CO3>^, (0er/3co3)co«, follow from (47), (48), (45) and (46), for example,
4 13.6 3.8 1091 103.1 217.3 29.3 1.23 1.0 1.1 13.96 56.9 232.31 322.0 119.57 236.7 1.44 1.8 3.2
( £ l >
z
- -
(
£
f
-
i
)
F
-
(52)
Similar equations hold for (0^2/0Co2)*3).1,
(dn2ldcos)t
where Z*j and ef have to be substituted by ZQ* and n*2. The data necessary for the estimation of the derivatives have been determined above and in Sect. 4.2, and the results are collected in Table 10. The derivatives (0£j/0e r )*, (02ti/0£?)a,, ••• follow from (11.41), and an example is given by (11.42). Similarly the derivatives (0£//0w2)* , ... follow from (11.51). For the derivatives with 1=1 the interaction radius a w i and the traces of the polariz-
Unauthenticated Download Date | 9/24/15 11:21 PM
1441
"W. Liptay et al. • Molecular Quantities from Measurements on Macroscopic Systems. Ill
Table 10. Derivatives of er and n2 with respect to C02 and co3 for solutions of HMB and TCNE in CC14 at 298.15 K.
culation of AC£/o\ 1
lim /
c0J£a2->0\
1
(55)
/no/,*. 8c£0J
\ I nor, 9.
Z^ficAr \
(56) (57)
are introduced to describe the field dependence of the molar absorption coefficients XEJ, the initial concentrations CEOJ of the substances used to generate the solution, and the equilibrium constants KECIV of the occurring chemical reactions. With (54) to (57) follows from (53) Yot =
[1
+ K*nC+
(ce)-i + K?21 (C0+2)2 ( c e ) - 2 ] - i
+ (P?i + &*)