Thermodynamics

Estimation of Pure Compound Properties

Using Group-Interaction Contributions
Jorge Marrero-Morejon
and Eladio Pardillo-Fontdevila
Centro de Qumica
Farmaceutica,
Ave. 200 y 21 Atabey, POB 16042, 11600-Habana, Cuba

A family of models proposed here estimate the critical constants and normal boiling
points of pure organic compounds from chemical structure by using a new structural
approach called group-interaction contributions, which considers the contributions of
interactions between bonding groups in the molecule instead of the contributions of
simple groups. Compared to the conentional group-contribution techniques, the proposed models demonstrate significant improements in accuracy, as well as the ability
to distinguish among isomers.

Introduction
The basis for the design of chemical processing units is the
set of physical and thermodynamic properties of the compounds that undergo transformation. However, it is not always possible to find experimental values of properties for
the compounds of interest in the literature. Since it is not
practical either to measure them as the need arises, estimation methods are profusely employed.
For the estimation of properties of pure compounds,
group-contribution methods Joback and Reid, 1987; Lyderson, 1955; Ambrose, 1978; Klincewicz and Reid, 1984; Lyman
et al., 1990. have been widely used. In these methods, the
property of a compound is a function of structurally dependent parameters S ., which are determined by summing the
number frequency of each group occurring in the molecule
multiplied by its contribution. These methods have the advantage of supplying quick estimates without requiring substantial computational resources. However, many of them are
of questionable accuracy and are unable to distinguish among
isomers due to the oversimplification of the molecular structure, which the simple group-contribution approach can.
To overcome this limitation, several attempts have been
reported in the literature. Constantinou et al. 1993, 1994.
have proposed a quite complex estimation technique, which
is based on conjugate forms alternative formal arrangements
of valence electrons.. This technique provides accurate estimations of several properties of pure compounds and allows

Correspondence concerning this article should be addressed to J. Marrero at his

current address: Dept. of Chemical Engineering, CAPEC, Technical Univ. of Denmark, Building 228, DK-2800 Lyngby, Denmark.

AIChE Journal

the capture of the differences among isomers. However, the

generation of conjugate forms is a nontrivial issue and requires a symbolic computing environment Prickett et al.,
1993..
A less complex method has been proposed by Constantinou and Gani 1994.. Estimation is performed at two levels:
the basic level uses the contribution from simple first-order
groups, while the second level uses a small set of second-order
groups that have the first-order ones as building blocks. The
role of the second-order groups is to consider the proximity
effects and to distinguish among isomers. For a proper determination of which of them occur in a given molecule, however, it is necessary to apply supplementary and nonreported
rules Abildskov, personal communication, 1995.. Otherwise,
the same molecule may be described in different ways.
1997. have rePardillo-Fontdevila and Gonzalez-Rubio

cently proposed a new structural approach wcalled groupinteraction contribution GIC.x, that considers the contributions of interactions between bonding groups instead of the
contributions of simple groups. In this work, we propose new
models, based on this new approach, for the estimation of
normal boiling points and critical constants of pure organic
compounds.

Development of the Models

In this work, the properties of a compound are considered
to be functions of structurally dependent parameters S .,
which are thereby determined by summing the number fre-

March 1999 Vol. 45, No. 3

615

Table 1. Estimation Equations Used in this WorkU

Property

Symbol

Unit

Normal boiling pointUU

Normal boiling point
Critical temperature
Critical pressure
Critical volume

Tb
Tb
Tc
Pc
Vc

K
K
K
bar
cm3rmol

Equation
Tbs M a Sq b
Tbs aq S
TcsTbrw aq bSy S 2 x
Pcs w aq bn A y S xy2
Vcs aq S

1.
2.
3.
4.
5.

Ss sum of the products each simple group or interaction between

groups . appears in the molecule and their contributions
Ms molecular weight

n As total number of atoms in the molecule

UU

for estimating the normal boiling point, which makes a significant improvement in accuracy as shown in the statistical results.
The contributions of simple groups and group-interactions
are presented in Tables 4 and 5, respectively. Notice that
there are group interactions that have not been calculated
because of the lack of property values of the compounds involving these interactions. We used the singular-value decomposition procedure Forsythe et al., 1977. as the optimization
algorithm for linear regression. For nonlinear regression, we
employed the well-known Levenberg]Marquardt procedure.
In both cases, the objective function was to minimize the sum
of squares of the differences between experimental and estimated values.

quency of each group-interaction occurring in the molecule

times its contribution. We selected 39 simple groups to generate a consistent set of group-interactions that allows one to
treat a wide variety of organic compounds. These groups are
the same as those used by Joback and Reid 1987., and earlier by Lydersen 1955., with the omission of s NH and s N]
nonring.. We compiled property values of 507 pure compounds, obtained exclusively from Reid et al. 1987., for use
in the development of the models.
With the object of comparing the group-contribution approach to the GIC approach, a least-square analysis was carried out to determine the contributions of both simple groups
and group interactions. The estimation equations for each
property are shown in Table 1. The fitted parameters of these
equations and the statistical summary of regression analysis
are presented in Tables 2 and 3, respectively. The equations
are the same as those used by Joback and Reid 1987.. In
addition, we propose a new alternative nonlinear equation

Table 4. Contributions of Simple Groups

Nonring groups
]CH 3
]CH 2 ]
)CH]
)C sCH 2
sCH]
sC sC s
sCH
'C]
Ring groups
]CH 2 ]
)CH]
)C sCH]
sC Halogen groups
]F
]Cl
]Br
]I

Table 2. Fitted Parameters of Estimation Equations

Group-Contribution
Approach

Group-Interaction
Contribution Approach

Eq.

1.
2.
3.
4.
5.

y0.366
218.13
0.5881
0.1218
26.3

149.84
}
y0.9305
0.4609
}

y0.404
204.66
0.5851
0.1285
25.1

156.00
}
y0.9286
y0.0059

Table 3. Statistical Summary of Regression Analysis

Eq.

Property

1.
2.
3.
4.
5.

Tb
Tb
Tc
Pc
Vc

Simple
Group-Interaction
Group-Contribution
Contribution
Data
U
UU

Points AAE STD

APE AAE STD APE
507
507
491
445
289

10.11
12.66
4.71
2.03
7.32

13.94
17.01
6.32
3.09
10.59

2.86
3.58
0.82
5.30
2.33

4.99
6.48
2.79
1.06
4.56

7.35
9.53
4.39
1.70
6.68

1.34
1.73
0.48
2.92
1.45

AAE s average absolute error

UU
STDs standard deviation of regression

APE s average percent error

Equation 1 is the one proposed for estimating Tb.

Note: AAE and STD have units of K for Tb and Tc, bar for Pc, and
cm3rmol for Vc.

616

Tb

TbU

Tc

Pc

Vc

11.41
22.90
34.09
48.08
2.82
27.76
31.63
28.10
y10.06
34.38

98.48
198.22
264.13
328.01
67.90
194.17
258.41
223.18
61.28
221.38

y0.0144
y0.0185
y0.0161
y0.0102
y0.0076
y0.0136
y0.0119
y0.0079
y0.0115
0.0032

1.8345
1.3731
0.9124
0.4525
1.3750
0.9144
0.4527
0.4604
0.9183
0.4602

58.6
56.6
48.3
42.2
50.5
47.8
38.0
35.5
42.1
37.7

22.50
30.48
46.39
23.24
38.42

174.48
247.01
324.31
179.86
290.21

y0.0092
y0.0141
y0.0046
y0.0088
y0.0103

1.3768
0.9125
0.4567
0.9180
0.4566

47.3
42.3
33.0
40.7
34.0

y14.79 y45.91 y0.0108 0.4560

25.63 267.42 y0.0084 0.4577
43.53 508.17 y0.0074 0.4562
89.39 950.06 y0.0017 0.4605

19.9
51.0
73.5
87.4

Oxygen groups
]OH alcohol.
84.51
]OH phenol.
59.95
]O] nonring.
25.83
]O] ring.
25.78
)CO nonring.
72.21
)CO ring.
96.98
]CHO aldehyde .
68.99
]COOH acid.
146.99
]COO] ester.
80.61
sO other.
y17.05
Nitrogen groups
]NH 2
58.67
) NH nonring.
56.91
) NH ring.
64.96
) N] nonring.
49.72
s N] ring.
44.74
]CN
111.45
]NO2
144.77

474.98
350.99
203.21
181.91
522.16
605.02
451.84
948.08
610.80
31.79

y0.0678
y0.0273
y0.0174
y0.0092
y0.0327
y0.0183
y0.0306
y0.0843
y0.0469
y0.0092

0.9309 20.5
0.9352 y7.6
0.4577 16.1
0.4637 11.1
0.9165 66.5
0.9203 52.4
1.3837 76.6
1.8407 85.4
1.3716 88.3
0.4669 32.7

325.69
362.92
369.36
347.77
240.89
639.92
912.64

y0.0210
y0.0291
y0.0154
y0.0158
y0.0108
y0.0525
y0.0409

1.3870 31.1
0.9204 44.4
0.9305 28.6
0.4565 34.6
0.4692 y24.1
0.9107 86.3
1.3878 88.3

Sulfur groups
]SH
]S] nonring.
]S] ring.

428.90
0.0048 0.9313
519.86 y0.0100 0.4618
444.31
0.0008 0.4652

52.67
72.12
66.09

62.8
59.5
30.1

Contributions to structurally dependent parameter S . of estimation

equation Eq. 1 ..

March 1999 Vol. 45, No. 3

AIChE Journal

Table 5. Contributions of Group Interactions

Tb

TbU

Interactions with CH3 ] (ia single bond)

CH 3 ] and CH 3 ]
y20.82 61.28
CH 3 ] and ]CH 2 ]
33.19 194.25
CH 3 ] and )CH]
26.94 194.27
CH 3 ] and )C 22.71 186.41
CH 3 ] and sCH]
18.17 137.18
CH 3 ] and sC 23.91 182.20
CH 3 ] and 'C]
23.04 194.40
CH 3 ] and )CH] r.
25.68 176.16
CH 3 ] and )C - r.
20.25 180.60
CH 3 ] and sC - r.
19.61 145.56
CH 3 ] and ]F
y9.96 160.83
CH 3 ] and ]Cl
44.44 453.70
CH 3 ] and ]Br
71.94 758.44
CH 3 ] and ]I
111.04 1181.44

Tc

Pc

Vc

Tb
and ]OH a.
and ]O]
and )CO
and ]CHO
and ]COOH
and ]COO] o.
and ]COO] c.
and NH 2 ]
and ) NH
and ) N]
and ]CN
and ]NO 2
and ]SH
and ]S]

TbU

Tc

Pc

Vc

y0.0213
y0.0227
y0.0223
y0.0189
0.8526
0.1792
0.3818
y0.0214
0.1117
0.0987
y0.0370
y0.9141
y0.9166
y0.9146

y0.0576 123.2
y0.0430 88.6
y0.0376 78.4
y0.0354 69.8
0.0654 81.5
0.0851 57.7
y0.2320 65.8
y0.0396 58.3
y0.0597 49.0
y0.0746 71.7
y0.0345 88.1
y0.0231 113.8
y0.0239 }
y0.0241 }

CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]
CH 3 ]

Interactions with nonring ] CH2 ] (ia single bond)

]CH 2 ] and ]CH 2 ]
22.00 244.88 y0.0206
]CH 2 ] and )CH]
19.78 244.14 y0.0134
]CH 2 ] and )C 22.17 273.26 y0.0098
]CH 2 ] and sCH]
25.30 201.80
0.8636
]CH 2 ] and sC 27.34 242.47
0.1874
]CH 2 ] and 'C]
26.20 207.49
0.4160
]CH 2 ] and )CH] r. 14.60 238.81 y0.0149
]CH 2 ] and )C - r. 19.06 260.00
0.1193
]CH 2 ] and sC - r.
4.10 167.85
0.1012
]CH 2 ] and ]F
6.27 166.59 y0.0255
]CH 2 ] and ]Cl
62.72 517.62 y0.0162
]CH 2 ] and ]Br
84.49 875.85 y0.0205
]CH 2 ] and ]I
107.75 1262.80 y0.0210

y0.0272 56.6
y0.0219 40.2
y0.0162 32.0
0.0818 50.7
0.1010 24.0
y0.2199 33.9
y0.0265 31.9
y0.0423 52.1
y0.0626 }
y0.0161 49.3
y0.0150 80.8
y0.0140 101.3
y0.0214 }

]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]

Interactions with nonring )CH ] (ia single bond)

)CH] and )CH]
21.94 291.41 y0.0039
)CH] and )C 31.03 344.06
0.0025
)CH] and sCH]
14.44 179.96
0.8547
)CH] and sC 33.24 249.10
0.1969
)CH] and )CH] r. 21.15 295.33
0.0025
)CH] and sC - r. y5.51 132.66
0.1187
)CH] and ]F
y2.06 68.80 y0.0200
)CH] and ]Cl
47.08 438.47 y0.0142

y0.0137
y0.0085
0.0816
0.1080
y0.0168
y0.0556
y0.0147
y0.0131

18.3
8.6
48.9
4.3
}
}
37.7
68.6

)CH]
)CH]
)CH]
)CH]
)CH]
)CH]
)CH]
)CH]

and
and
and
and
and
and
and
and

]OH a.
]O]
)CO
]CHO
]COOH
]COO] c.
]NH 2
) NH

84.70 585.99 y0.0757 y0.0093 45.6

14.40 215.94 y0.0162 y0.0155 23.7
45.66 434.45 y0.0194 y0.0112 39.3
78.46 630.07 y0.0406 y0.0280 92.2
170.37 1270.16 y0.1054 y0.0358 110.2
44.23 497.23 y0.0918 y0.2098 72.3
47.06 388.44 y0.0286 y0.0212 39.2
22.34 260.32 y0.0158 y0.0162 }

Interactions with nonring )C - (ia single bond)

)C - and )C 46.38 411.56
0.0084
0.0002
)C - and sCH]
23.36 286.30
0.8767
0.0953
)C - and sC 41.20 286.42
0.2061
0.1109
)C - and )CH] r.
25.89 340.00
0.0049 y0.0111
)C - and sC - r.
2.61 188.99
0.1249 y0.0510
)C - and ]F
y7.70 y16.64 y0.0176 y0.0161

22.7
23.4
8.8
}
}
30.0

)C )C )C )C )C )C -

and
and
and
and
and
and

]Cl
]Br
]OH a.
]O]
)CO
]COOH

33.83 360.79
50.42 610.26
76.39 540.38
23.46 267.26
38.63 373.71
164.43 1336.54

]OH a. 108.85 673.24 y0.0786 y0.0119 45.2

]O]
25.03 243.37 y0.0205 y0.0184 34.5
)CO
50.77 451.27 y0.0256 y0.0204 62.3
]CHO
88.49 648.70 y0.0267 y0.0210 106.1
]COOH 156.34 1180.39 y0.0932 y0.0253 114.0
]COO] o. 49.04 475.65
0.0276
0.1561 69.9
]COO] c. 53.64 541.29 y0.0993 y0.2150 79.1
]NH 2 ]
70.84 452.30 y0.0301 y0.0214 63.3
) NH]
35.62 314.71 y0.0248 y0.0203 49.4
) N]
18.11 240.08 y0.0161 y0.0170 32.7
]CN
130.85 869.18 y0.0654 y0.0329 113.5
]SH
70.35 612.31 y0.0137 y0.0163 93.3
]S]
47.45 451.03 y0.0192 y0.0173 57.9

and
and
and
and
and
and
and
and
and
and
and
and
and

Interactions with sCH2 (ia double bond)

sCH 2 and sCH 2
y35.36 51.13 y0.9129 y0.0476 105.3 sCH 2 and sC sCH 2 and sCH]
28.66 205.73 y0.8933 y0.1378 77.4 sCH 2 and sC s
Interactions with nonring sCH ] (ia double bond)
sCH] and sCH]
35.33 334.64 y1.7660
sCH] and sC 36.03 354.41 y1.2909
sCH] and sC s
40.99 316.46 y0.8945
sCH] and sCH]
20.63 174.18
1.7377
sCH] and sC 36.87 228.38
1.0731
sCH] and 'C]
22.58 174.39
1.2865
sCH] and sC - r.
18.08 184.20
0.9929
sCH] and ]F
y32.32
5.57
0.8623

and ]Cl
and ]O]
and yCHO
and ]COOH
and ]COO] o.
and ]COO] c.
and ]CN

y0.2291
y0.3613
y0.1202
0.1944
0.2146
y0.1087
0.0533
0.0929

47.8
73.6
43.6
42.1
16.6
26.3
}
41.4

sCH]
sCH]
sCH]
sCH]
sCH]
sCH]
sCH]

Interactions with nonring sC - (ia double bond)

sC - and sC 45.70 399.58 y0.8155 y0.4920

93.7

sC - and sC s

Contributions to structurally dependent parameter S . of estimation equation Eq.1 ..

a . ]OH in a nonaromatic compound.
p . ]OH in an aromatic compound.
c . Interaction with ester group via carbon atom.
o. Interaction with ester group via oxygen atom.
r . Ring group.
rr . Ring group interacting with a group occurring in a different ring heterocyclic compounds ..
} . Contribution that has not been calculated.

AIChE Journal

133.04 736.93 y0.0876 y0.0180 92.9

31.94 228.01 y0.0205 y0.0321 66.0
64.46 445.61 y0.0362 y0.0363 88.9
89.34 636.49 y0.0606 y0.0466 128.9
186.44 1228.84 y0.0890 y0.0499 145.9
58.87 456.92
0.0267
0.1462 93.3
65.95 510.65 y0.0974 y0.2290 108.2
62.14 443.76 y0.0397
}
}
41.60 293.86 y0.0313 y0.0317 }
23.78 207.75 y0.0199 y0.0348 76.3
150.14 891.15 y0.0766 y0.0507 147.9
169.64 1148.58 y0.0591 y0.0385 148.1
74.44 588.31 y0.9192 y0.0244 119.7
53.24 409.85 y0.0181 y0.0305 87.9

March 1999 Vol. 45, No. 3

13.11
17.02

y0.0129
y0.0121
y0.0094
y0.0103
y0.0085
y0.0455

63.7
85.7
40.6
40.8
62.1
89.0

215.27 y0.4158 y0.2709

183.55 y0.0123 y0.0239

99.2
68.4

42.28 370.60
14.95 204.81
92.68 658.53
180.68 1245.86
44.27 423.86
59.38 525.35
117.18 761.36

44.51

y0.0133
y0.0084
y0.0780
y0.0156
y0.0114
y0.1008

0.8613
0.0919 68.7
0.8565
0.0947 36.4
0.8246
0.0801 }
0.7862
0.0806 107.4
0.8818
0.2743 55.2
0.7780 y0.1007 64.1
0.8122
0.0771 107.4

321.02 y0.4009 y0.2502

58.1

(Table continued )

617

Table 5. Contributions of Group Interactions (Continued)

Tb

TbU

Tc

Pc

Interactions with nonring sC - (ia simple bond)

sC - and sC - r.
29.92 220.88
0.3043
sC - and ]F
y13.78 y37.99
0.1868

Vc

Tb

0.0705 }
0.1064 14.6

TbU

Tc

Pc

Vc

sC - and ]Cl

36.54 367.05

0.1886

0.1102 43.3

'CH and 'C]

22.20 222.40 y0.4222

0.1860 73.1

]CH 2 ]
]CH 2 ]
]CH 2 ]
]CH 2 ]

29.60
61.01
39.47
56.34

Interactions with nonring sC s (ia triple bond)

sC s and sO
10.32 160.42 y0.0159 y0.0010 51.4
Interactions with nonring 'CH (ia triple bond)
'CH and 'CH
y16.26 120.85 y0.0288 y0.0226 87.6
Interactions with nonring 'C ] (ia triple bond)
'C] and 'C]
49.36 333.26 y0.7958

0.3933 64.3

Interactions with ring ] CH2 ] (ia single bond)

]CH 2 ] r. and ]CH 2 ] r.
25.62 201.89 y0.0098 y0.0221 47.2
]CH 2 ] r. and )CH] r.
21.77 209.40 y0.0093 y0.0181 47.5
]CH 2 ] r. and )C - r.
20.34 182.74 y0.1386
0.0081 49.9
]CH 2 ] r. and sCH] r.
31.27 218.07
0.0976 y0.1034 42.5
]CH 2 ] r. and sC - r.
9.91 106.21
0.1089 y0.0527 }
Interactions with ring )CH ] (ia single bond)
)CH] r. and )CH] r.
19.23 210.66
)CH] r. and )C - r.
22.71 220.24
)CH] r. and ]O] r.
22.17 169.17
)CH] r. and )CH] rr.
20.17 242.01
)CH] r. and ]OH a.
81.38 597.82

y0.0055
y0.1341
y0.0218
y0.0059
y0.0737

y0.0088
0.0162
y0.0091
y0.0071
y0.0220

33.9
}
19.2
}
}

sCH] r. and s N] r.

0.2089 y0.1324 39.3

0.2190
}
29.8
0.0990 y0.0935 40.3

Interactions with ring sC - (ia double bond)

sC - r. and sC - r.
66.09 477.77 y0.4848

sCH]
sCH]
sCH]

225.52
451.74
283.55
424.13

y0.0092 y0.0119 29.2

y0.0148 y0.0177 50.7
y0.0139 y0.0127 38.8
y0.0071
}
}

sC sC sC sC sC sC sC sC sC -

30.44 221.55

0.3913

0.1356

8.3

r. and ) NH r. 68.48 420.54

0.0947
}
}
r. and s N] r. 49.83 321.44 y0.4067 y0.1491 65.9
r. and ]S] r.
45.58 348.00
0.1027 y0.0916 40.8

sC - r. and s N] r.

0.1432 37.8

Interactions with ring sC - (ia single bond)

sC - r. and sC - r.
1.99 180.07
0.2318 y0.0809 20.6
sC - r. and ]O] r.
14.56 134.23
0.1104 y0.0374 y0.3
sC - r. and s N] r.
16.03 174.31 y0.3972 y0.0971 y93.2
sC - r. and sC - rr. y32.07 153.05
0.2424 y0.0792 51.7
sC - r. and ]F
y8.96 y48.79
0.1069 y0.0504 23.7
sC - r. and ]Cl
30.76 347.33
0.1028 y0.0512 60.3
sC - r. and ]Br
51.77 716.23
0.1060 y0.0548 83.2
sC - r. and ]I
90.04 1294.98
0.1075 y0.0514 110.2
sC - r. and ]OH p.
64.74 456.25
0.0931 y0.0388 8.5
Interactions with ] Cl (ia single bond)
]Cl and )CO
54.79

]O] r.
)CO r.
) NH r.
]S] r.

Interactions with ring )C - (ia single bond)

)C - r. and )C - r. 20.52 348.23
0.0329
0.0057 36.2
)C - r. and sC - r. y3.11 y25.81 y0.0038
0.0072 }
)C - r. and )C - rr. 33.67 550.72
0.0662 y0.0509 }
)C - r. and sC - rr. 16.16 408.64
0.1615
0.1542 53.9
)C - r. and ]F
y0.85 y41.35 y0.0314 y0.0119 18.4

Interactions with ring sCH ] (ia double bond)

sCH] r. and sCH] r.
8.41 112.0 y0.2246
0.1490 36.5
sCH] r. and sC - r.
36.01 291.15 y0.3586 y0.1822 34.4
Interactions with ring sCH ] (ia single bond)
sCH] r. and sCH] r.
44.57 285.07
sCH] r. and sC - r.
24.95 237.22
sCH] r. and ]O] r.
22.15 171.59

r. and
r. and
r. and
r. and

r. and
r. and
r. and
r. and
r. and
r. and
r. and
r. and
r. and

43.35 334.09

0.2541

]O]
3.66 199.70
)CO
38.88 437.51
]CHO
92.60 700.06
]COOH 151.44 1232.55
]COO] c. 23.85 437.78
]NH 2
77.47 517.75
) NH
40.53 411.29
) N]
48.18 422.51
]CN
92.74 682.19

0.0997
0.1112
0.0919
0.0313
0.0241
0.0830
0.0978
0.0938
0.0768

y0.0523
y0.0528
y0.0597
y0.0684
y0.2573
y0.0579
y0.0471
y0.0462
y0.0625

}
46.3
}
100.2
55.2
33.2
}
}
}

532.24 y0.0191 y0.0125 84.0

Interactions with ] O ] (ia single bond)

]O] and )CO
42.16 367.83

0.3209

Interactions with nonring )CO (ia single bond)

)CO and )CO
83.64 734.86 y0.4957 y0.6898

Interactions with ] H (forming formaldehyde, formic acid, . . . )

]H and ]CHO
49.34 387.17 y0.0422 y0.0123
]H and ]COOH
169.14 1022.45 y0.0690
}

}
}

Interactions with ] NH2 (ia single bond)

]NH 2 and ) NH
115.75 673.59 y0.0301

0.1987

]O] r. and s N] r.

57.78 382.25

]H and ]COO] c.

44.47 298.12 y0.0781 y0.1878 51.2

Interactions with nonring ] S ] (ia single bond)

]S] and ]S]
61.17 597.59 y0.0124

Contributions to structurally dependent parameter S . of estimation equation Eq.1 ..

a . ]OH in a nonaromatic compound.
p . ]OH in an aromatic compound.
c . Interaction with ester group via carbon atom.
o. Interaction with ester group via oxygen atom.
r . Ring group.
rr . Ring group interacting with a group occurring in a different ring heterocyclic compounds ..
} . Contribution that has not been calculated.

618

March 1999 Vol. 45, No. 3

AIChE Journal

Table 6. Absolute Average Errors: Training Set vs.

Testing Set
Training Set

Testing Set

Eq. Property

Interactions
U

Data AAE
AAE
Data AAE
AAE
Points GCUU . GIC . Points GCUU . GIC .

TB
Tb
Tc
Pc
Vc

1.
2.
3.
4.
5.

Table 7b. Summations of Group-Interaction

Contributions for Ethyl Acrylate

407
407
391
345
189

10.07
12.43
4.54
1.97
7.13

4.87
6.36
2.61
0.98
4.33

98
99
90
87
83

11.01
13.12
5.02
2.12
7.88

5.22
7.01
2.73
1.00
4.69

AAE s average absolute error.

The notation GC stands for group contribution.

The notation GIC stands for group-interaction contribution.

Equation 1 is the one proposed for estimating Tb.

Note: AAE has units of K for Tb and Tc, bar for Pc, and cm 3rmol for Vc.
UU

No.

Tb

TbU

Tc

Pc

Vc

Interaction with sCH2 (ia double bond)

sCH 2 sCH]
1
28.66 205.73 y0.8933 y0.1378

77.4

Interaction with ] CH2 ]

]CH 3 ]CH 2 ]
1

194.25 y0.0227 y0.0430

88.6

475.65

0.0276

0.1561

69.9

525.35

0.7780 y0.1007

64.1

33.19

Interactions with ester group

via oxygen atom
]CH 2 ] ]COO] 1
49.04
via carbon atom
sCH] ]COO] 1
59.38

Structurally dependent parameter S

170.27 1400.98 y0.1104 y0.1254 300.0
U

Contributions to structurally dependent parameter S . of estimation

equation Eq. 1 ..

Reliability of the Models

In order to test the reliability of the models, we generated
two disjointed subsets by randomly excluding 100 compounds
from the entire data set of 507 compounds. The largest subset was used as a training set for fitting the equations presented in Table 1 and determining the contributions of both
simple groups and group interactions. The excluded compounds were used as a testing set by estimating their property values through the contributions and equations previously obtained using the training-set compounds.
The estimation errors for the testing-set compounds are
comparable with those for the training-set compounds, as
shown in Table 6, confirming the reliability of the proposed
models. Notice that some compounds were removed from the
testing set because they involved interactions whose contributions could not be obtained from the regression in which the
randomly selected training set was used. Since the models
presented perform well for compounds not included in the
training set, it is more suitable to employ the contributions
calculated by using the entire data set because of its larger
size. These contributions are reported in Table 5.

Table 7c. Summations of Group-Interaction

Contributions for Isoquinoline
Interactions

No.

Tb

TbU

Tc

Interactions within the rings (ia double bonds)

sCH] sCH] 3
8.41 112.00 y0.2246
sC sC - 1
66.09 477.77 y0.4848
sCH] s N]
1
30.44 221.55
0.3913

Pc

Vc

0.1490
0.1432
0.1356

36.5
37.8
8.3

Interactions within the ring (ia single bond)

sCH] s N]
1
49.83 321.44 y0.4067 y0.1491
sCH] sCH] 1
44.57 285.07
0.2089 y0.1324
sCH] sC - 4
24.95 237.22
0.2190
n.d.

65.9
39.3
29.8

Structurally dependent parameter S

315.96 2590.71 y0.0891

0.4443 380.0

Contributions to structurally dependent parameter S . of estimation

equation Eq. 1 ..
n.d.s interaction that could not be determined because of lack of data.

To illustrate our proposed application of the GIC approach, we have estimated the values of the normal boiling

point and critical constants using 1,3,5-trichlorotrifluorobenzene, ethyl acrylate, isoquinoline, and m-terphenyl as example compounds. In Tables 7a, 7b, 7c, and 7d, we show the
GIC summations for each property using the values from
Table 5. In Tables 8a, 8b, 8c, and 8d, we summarize the results and compare them with experimental values.

Table 7a. Summations of Group-Interaction

Contributions for 1,3,5-Trichlorotrifluorobenzene

Table 7d. Summations of Group-Interaction

Contributions for m-Terphenyl

Examples and Discussion

Interactions

No.

Tb

TbU

] Cl and ] F and ring sC sC - ]Cl

3
30.76 347.33
sC - ]F
3 y8.96 y48.79
Interactions within the ring
via double bond
sC - sC - 3
66.09
via single bond
sC - sC - 3
1.99

Tc

0.1028 y0.0512
0.1069 y0.0504

477.77 y0.4848
180.07

Pc

Vc
60.3
23.7

0.1432

37.8

0.2318 y0.0809

20.6

Structurally dependent parameter S

269.64 2869.14 y0.1299 y0.1179 427.2
U

Contributions to structurally dependent parameter S . of estimation

equation Eq. 1 ..

AIChE Journal

Interactions

No.

Tb

TbU

Tc

Pc

Vc

Interactions within the rings (ia double bonds)

sCH] sCH] 5
8.41 112.00 y0.2246 0.1490 36.5
sCH] sC - 4
36.01 291.15 y0.3586 y0.1822 34.4
Interactions within the ring (ia single bonds)
sC sC - 2 y32.07 153.05 0.2424 y0.0792 51.7
sCH] sCH] 5
44.57 285.07 0.2089 y0.1324 39.3
sCH] sC - 4
24.95 237.22 0.2190 n.d.
29.8
Structurally dependent parameter S
444.60 4404.93 y0.1521 y0.8042 739.20
U

Contributions to structurally dependent parameter S . of estimation

equation Eq. 1 ..
n.d.s interaction that could not be determined because of lack of data.

March 1999 Vol. 45, No. 3

619

Table 8a. Results for 1,3,5-Trichlorotrifluorobenzene

Table 8c. Results for Isoquinoline

Property

Eq.
Used

Est. Value
GIC.U

Est. Value
GC.UU

Exp.
Value

Property

Eq.
Used

Est. Value
GIC.U

Est. Value
GC.UU

Exp.
Value

Tb K.
Tb K.
Tc K.
Pc bar.
Vc cm3rmol.

1.
2.
3.
4.
5.

471.9
474.3
684.5
32.4
452.3

475.8
481.2
672.6
33.8
443.0

471.5
471.5
684.9
32.7
448.0

Tb K.
Tb K.
Tc K.
Pc bar.
Vc cm3rmol.

1.
2.
3.
4.
5.

591.5
520.6
782.5
n.e.
405.1

501.0
502.4
775.4
45.2
355.1

516.4
516.4
803.0
n.a.
n.a.

The notation GIC stands for group-interaction contribution.

The notation GC stands for group contribution.

The molecular weight M . was 235.419 grmol.

The experimental value of Tb was employed in Eq. 4. If the value

of Tb 471.9 K . estimated from Eq. 1 and group-interaction contributions had been used, a value of 685.0 K for Tc would have been
obtained.

n A, total number of atoms, is 12.

Note: Summation of contributions from Table 7a. Experimental values
were obtained from Reid et al. 1987 ..
UU

We have used a common set of experimental data to obtain the contribution values of both simple groups and group
interactions. It is worth mentioning that it would not have
been a fair comparison between both structural techniques if
we had not used a common set of data. As shown in Table 3,
the results achieved by using group interactions are significantly better than those obtained with the typical groupcontribution technique.
Another important feature of the GIC-based models presented is the ability to capture the fine differences among
isomers. Consider, for example, the estimation of the normal
boiling points of octane isomers. Table 9 shows the estimations through the GIC and group-contribution approaches,
using Eq. 1 as the estimation equation, and the group-contribution technique proposed by Constantinou and Gani 1994..
Clearly, by using the GIC approach, a higher number of isomers is distinguished and the estimates are significantly more
accurate.
The proposed GIC-based models were found to be quite
satisfactory for most of the chemical classes alkanes, alkenes,
cycloalkanes, cycloalkenes, aromatic hydrocarbons, ethers,
aldehydes, ketones, acids, esters, as well as nitrogen, sulfur,
and halogenated compounds., and relatively poor for alco-

The notation GIC stands for group-interaction contribution.

The notation GC stands for group contribution.
The molecular weight M . was 129.162 grmol.

The experimental value of Tb was employed in Eq. 4. If the value

of Tb 519.5 K . estimated from Eq. 1 and group-interaction contributions had been used, a value of 787.2 K for Tc would have been
obtained.

n A, total number of atoms, is 17.

Note: Summation of contributions from Table 7c. n.e.s value that cannot be estimated because of lack of contribution. n.a.s
experimental data not available. Experimental values were obtained from Reid et al. 1987 ..
UU

hols, phenols, and large heterocyclic compounds. All the

models yield large errors for polyhydroxy alcohols. It is remarkable that the GIC-based models perform better than the
group-contribution ones for all the classes.

Conclusion
A family of models for the estimation of normal boiling
points and critical constants of pure organic compound has
been developed by using the group-interaction contribution
GIC. approach. Compared to the classic group-contribution
techniques, the proposed models exhibit a drastically superior accuracy and are better able to distinguish among isomers. In addition, a nonlinear equation for estimating the
normal boiling point, making a significant improvement in accuracy, was reported.

Acknowledgment
CQF.,
Financial support of the Centro de Qumica
Farmaceutica

Havana, is gratefully acknowledged.

Table 8d. Results for m-Terphenyl

Table 8b. Results for Ethyl Acrylate

Property

Eq.
Used

Est. Value
GIC.U

Est. Value
GC.UU

Exp.
Value

Tb K.
Tb K.
Tc K.
Pc bar.
Vc cm3rmol.

1.
2.
3.
4.
5.

373.9
374.9
552.2
36.6
325.1

366.5
363.6
555.2
36.0
328.1

373.0
373.0
552.0
37.4
320.0

The notation GIC stands for group-interaction contribution.

UU
The notation GC stands for group contribution.

The molecular weight M . was 100.119 grmol.

The experimental value of Tb was employed in Eq. 4. If the value

of Tb 373.9 K . estimated from Eq. 1 and group-interaction contributions had been used, a value of 553.6 K for Tc would have been
obtained.

n A, total number of atoms, is 15.

Note: Summation of contributions from Table 7b. Experimental values
were obtained from Reid et al. 1987 ..

620

Property

Eq.
Used

Est. Value
GIC.U

Est. Value
GC.UU

Exp.
Value

Tb K.
Tb K.
Tc K.
Pc bar.
Vc cm3rmol.

1.
2.
3.
4.
5.

645.3
649.3
907.3
n.e.
764.3

652.3
697.2
893.5
27.1
732.1

638.0
638.0
924.9
35.1
768.0

The notation GIC stands for group-interaction contribution.

The notation GC stands for group contribution.
The molecular weight M . was 230.31 grmol.

The experimental value of Tb was employed in Eq. 4. If the value

of Tb 645.3 K . estimated from Eq. 1 and group-interaction contributions had been used, a value of 917.7 K for Tc would have been
obtained.

n A, total number of atoms, is 32.

Note: Summation of contributions from Table 7d. n.e.s value that cannot be estimated because of lack of contribution. Experimental
values were obtained from Reid et al. 1987 ..
UU

March 1999 Vol. 45, No. 3

AIChE Journal

Table 9. Normal Boiling Point Estimates of Octane Isomers

Isomeric Compounds

Exp. Values

GICU Estimates
Using Eq. 1

GCUU Estimates
Using Eq. 1

GCUU Estimates
Constantinou
and Gani, 1994.

n-Octane
2-Methylheptane
3-Methylheptane
4-Methylheptane

398.8
390.8
392.1
390.9

394.0
386.4
386.3
386.3

4.8.
4.4.
5.8.
4.6.

394.5
388.5
388.5
388.5

4.3.
2.3.
3.6.
2.4.

406.6
393.1
396.5
396.5

7.8.
2.3.
4.4.
5.6.

2,2-Dimethylhexane
2,3-Dimethylhexane
2,4-Dimethylhexane
2,5-Dimethylhexane

380.0
388.8
382.6
382.3

379.8
385.8
378.7
378.9

0.2.
3.0.
3.9.
3.4.

382.2
382.5
382.5
382.5

2.2.
6.3.
0.1.
0.2.

384.5
391.4
382.3
378.6

4.5.
2.6.
0.3.
3.7.

3,3-Dimethylhexane
3,4-Dimethylhexane
3-Ethylhexane

385.1
390.9
391.7

385.1
385.7
386.2

0.0.
5.2.
5.5.

382.2
382.5
388.5

2.9.
8.4.
3.2.

386.0
391.4
396.5

0.9.
0.5.
4.8.

2,2,3-Trimethylpentane
2,2,4-Trimethylpentane
2,3,3-Trimethylpentane
2,3,4-Trimethylpentane

383.0
372.4
387.9
386.6

382.6
372.2
388.1
385.3

0.4.
0.2.
0.2.
1.3.

376.2
376.2
376.2
376.6

6.8.
3.8.
11.7.
10.0.

383.5
369.5
385.0
386.1

0.5.
3.0.
2.9.
0.5.

2-Methyl-3-ethylpentane
3-Methyl-3-ethylpentane
2,2,3,3-Tetramethylbutane

388.8
391.4
379.6

385.7
390.5
381.8
AAE

3.1.
0.9.
2.2.
2.7

382.5
382.2
369.9
AAE

6.3.
9.2.
9.7.
5.2

382.3
386.0
383.8
AAE

6.5.
5.4.
4.2.
3.3

GIC s Group-interaction contribution.

GC s Group contribution.
Isomeric compounds that cannot be distinguished by the GIC approach.

Isomeric compounds that cannot be distinguished by the GC approach.

Isomeric compounds undistinguished by the technique proposed by Constantinou and Gani.

AAE s Absolute average error.

Note: Experimental values, estimates, and errors have units of K. Experimental values were obtained from Reid et al. 1987 ..
UU

Literature Cited
Ambrose, D., Correlation and Estimation of Vapor-Liquid Critical
Properties: I. Critical Temperatures of Organic Compounds, NPL
Rep. Chem., Nat. Physical Lab., Teddington, UK 1978..
Constantinou, L., S. E. Prickett, and M. L. Mavrovouniotis, Estimation of Thermodynamic and Physical Properties of Acyclic Hydrocarbons Using the ABC Approach and Conjugation Operators,
Ind. Eng. Chem. Res., 32, 1734 1993..
Constantinou, L., S. E. Prickett, and M. L. Mavrovouniotis, Estimation of Properties of Acyclic Organic Compounds Using Conjugation Operators, Ind. Eng. Chem. Res., 39, 395 1994..
Constantinou, L., and R. Gani, New Group Contribution Method
for Estimating Properties of Pure Compounds, AIChE J., 40, 1697
1994..
Forsythe, G. E., M. A. Malcolm, and C. B. Moler, Computer Methods
for Mathematical Computations, Prentice Hall, Englewood Cliffs,
NJ 1977..
Joback, K. G., and R. C. Reid, Estimation of Pure-Component
Properties from Group-Contributions, Chem. Eng. Commun., 57,
233 1987..

AIChE Journal

Klincewicz, K. M., and R. C. Reid, Estimation of Critical Properties

with Group-Contribution Methods, AIChE J., 30, 137 1984..
Lydersen, A. L., Estimation of Critical Properties of Organic Compounds, Rep., Univ. Cool. Exp. Stn., Madison, WI 1955..
Lyman, W. J., W. F. Reehl, and D. H. Rosenblatt, Handbook of
Chemical Property Estimation Methods, American Chemical Society,
Washington, DC 1990..
Pardillo-Fontdevila, E., and R. Gonzalez-Rubio,
A Group-Interac
tion Contribution Approach. A New Strategy for the Estimation of
Physico-Chemical Properties of Branched Isomers, Chem. Eng.
Commun., 163, 245 1997..
Prickett, S. E., L. Constantinou, and M. L. Mavrovouniotis, Computational Identification of Conjugate Paths for Estimation of Properties of Organic Compounds, Mol. Simulation, 11, 205 1993..
Reid, R. C., J. M. Prausnitz, and B. E. Poling, The Properties of Gases
and Liquids, 4th ed., McGraw-Hill, New York 1987..

Manuscript receied May 15, 1998, and reision receied No. 17, 1998.

March 1999 Vol. 45, No. 3

621