Current status of the Functional-Segment Activity ... · UNIFAC COSMO-RS JCOSMO program COSMO-RS { Surface contacting theory For each surface pair contact, there is an energy change

IntroductionF-SAC

Conclusions

Current status of the Functional-Segment ActivityCoefficient Model: F-SAC

Prof. Rafael de Pelegrini Soares

UNIVERSIDADE FEDERAL DO RIO GRANDE DO SULESCOLA DE ENGENHARIADEPARTAMENTO DE ENGENHARIA QUIMICALAB. VIRTUAL DE PREDICAO DE PROPRIEDADEShttp://www.enq.ufrgs.br/labs/lvpp

July 14, 2013

Prof. Rafael de Pelegrini Soares Current Status of the F-SAC model

http://www.enq.ufrgs.br/labs/lvpp

IntroductionF-SAC

Conclusions

Summary

1 Introduction

2 F-SAC

3 Conclusions


Two families of predictive activity coefficient (γi ) models

Group contribution models,for instance:

COSMO-RS like models, forinstance:

UNIFAC (Do) – Group Contribution

According to revision 5 – Antje Jakob et al. In: Ind. Eng. Chem.Res. 45.23 (2006), pp. 7924–7933.

DOI:10.1021/ie060355c

http://dx.doi.org/10.1021/ie060355c

COSMO-RS – Surface contacting teory

First, for each insulated molecule theCOSMOa method is applied and theapparent surface charges are determinedin a cavity

Several approximations are consideredhere:

The cavity consists of spheres of fixedradiiMolecules surrounded by a perfectconductorThe methods used contains otherapproximations: AM1, RM1, DFT, MP2,etc.The escaping charges need to becorrected. . .

aA Klamt and G Schuurmann. In: J. Chem. Soc., PerkinTrans. 2 (1993), pp. 799–805

COSMO-RS – Surface contacting teory

First, for each insulated molecule theCOSMOa method is applied and theapparent surface charges are determinedin a cavity

Several approximations are consideredhere:

The cavity consists of spheres of fixedradiiMolecules surrounded by a perfectconductorThe methods used contains otherapproximations: AM1, RM1, DFT, MP2,etc.The escaping charges need to becorrected. . .

aA Klamt and G Schuurmann. In: J. Chem. Soc., PerkinTrans. 2 (1993), pp. 799–805

COSMO-RS – Surface contacting theory

In the COSMO-RSa surface contactingtheory, the molecules are sequentially putin contact

For each contact, the conductor ispartially excluded

If each molecule is completely surroundedby others, the real solution would beobtained

aAndreas Klamt. In: The J. of Phys. Chem. 99.7 (1995),pp. 2224–2235


In the COSMO-RSa surface contactingtheory, the molecules are sequentially putin contact

For each contact, the conductor ispartially excluded

If each molecule is completely surroundedby others, the real solution would beobtained

aAndreas Klamt. In: The J. of Phys. Chem. 99.7 (1995),pp. 2224–2235

IntroductionF-SAC

Conclusions

IntroductionUNIFACCOSMO-RSJCOSMO program


For each surface pair contact, there is anenergy change

This results in different behavior fordifferent substances in solution

Clearly, there are many different possiblecontacting arranges and statisticalthermodynamics is necessary

For the F-SAC model, the COSMO-RSvariant known as COSMO-SACa is used

aShiang-Tai Lin and Stanley I. Sandler. In: Ind. Eng.Chem. Res. 41.5 (2002), pp. 899–913


IntroductionF-SAC

Conclusions


Sigma profile – p(σ)

For the statisticalthermodynamics treatment,the 3D apparent surfacecharges are projected into asimple histogram

These pure compounddistributions, known assigma profiles – p(σ), arethe basis for computing theactivity coefficients inmixture

“It is always desirable to express the properties of a solution in terms that canbe calculated completely from the properties of the pure components.” –J. M. Prausnitz. Molecular thermodynamics of fluid-phase equilibria. Third.Prentice-Hall, 1999.


IntroductionF-SAC

Conclusions


Sigma profile – p(σ)

For the statisticalthermodynamics treatment,the 3D apparent surfacecharges are projected into asimple histogram

These pure compounddistributions, known assigma profiles – p(σ), arethe basis for computing theactivity coefficients inmixture

“It is always desirable to express the properties of a solution in terms that canbe calculated completely from the properties of the pure components.” –J. M. Prausnitz. Molecular thermodynamics of fluid-phase equilibria. Third.Prentice-Hall, 1999.


IntroductionF-SAC

Conclusions


JCOSMO program: sigma profile of pure substances

code.google.com/p/jcosmo/


http://code.google.com/p/jcosmo/

IntroductionF-SAC

Conclusions


JCOSMO program: activity coefficient and VLE predictions

code.google.com/p/jcosmo/



IntroductionF-SAC

Conclusions


JCOSMO program: surface charges visualization


F-SAC: Motivation

Although COSMO-RS models have exceptional theoreticalcharacteristics, in general quantitative experimentalagreement is usually obtained by empirical modifications

The F-SAC1 model combines group contribution with theCOSMO-RS theory:

DOI:10.1021/ie400170a

1R.P. Soares and R.P. Gerber. “Functional-Segment Activity Coefficientmodel. 1. Model formulation”. In: Ind. Eng. Chem. Res. (2013). doi:10.1021/ie400170a.

http://dx.doi.org/10.1021/ie400170a


IntroductionF-SAC

Conclusions

A new modelCurrent Status

Similarities/differences between COSMO-SAC and F-SAC

The model equations areidentical, the difference is inthe sigma profile

In the F-SAC model there isa neutral and two chargedpeaks per group


IntroductionF-SAC

Conclusions


Similarities/differences between COSMO-SAC and F-SAC

The model equations areidentical, the difference is inthe sigma profile

In the F-SAC model there isa neutral and two chargedpeaks per group


IntroductionF-SAC

Conclusions


Sigma profiles from group contribution

pi (σ) =∑

k ν(i)k pk (σ)

DOI:10.1021/ie400170a



F-SAC parameters

Parameters to be calibrated based on experimental data foreach group k :

Qk total group surface area, initial value fromCOSMO computations2

Q+k group positive area

Q−k group negative areaσ+

k surface charge density of the positive areaportion

Other parameters:

Q◦k = Qk − Q+

k − Q−k group neutral area

σ−k = σ+k Q+

k /Q−k surface charge density of the

negative area portion

2R.P. Gerber and R.P. Soares. In: Braz. J. of Chem. Eng. 30 (1 2013),pp. 1–11.

IntroductionF-SAC

Conclusions


Parameter fit with experimental data

Currently there are parameters for 24 functional groupsdivided in 48 subgroups3.

The parameter estimation was accomplished using InfiniteDilution Activity Coefficient (IDAC) for more than 2000mixtures at different temperatures.

Additionally, VLE data for the ethanol/water system was alsoconsidered.

3R.P. Soares et al. “Functional-Segment Activity Coefficient model. 2.Associating Mixtures”. In: Ind. Eng. Chem. Res. (2013). doi:10.1021/ie4013979.


http://dx.doi.org/10.1021/ie4013979

IntroductionF-SAC

Conclusions


F-SAC parameters

For the 24 grupos considered, only152 parameters were calibrated.

Only the hydrogen-bonding?

energies are pairwise

All other parameters are for puregroups alone, reducing the totalnumber of parameters

In order to represent the samemolecules in mixtures, 778parameters are used inUNIFAC (Do).


F-SAC parameters for Hydrogen-Bonding mixturesHB-acceptor is red (positive), HB-donor is blue (negative).

DOI:10.1021/ie4013979


F-SAC parameters for Hydrogen-Bonding mixtures

Only the hydrogen-bondingenergies are pairwisea

All other parameters are forpure groups alone

All F-SAC parameters have aphysical meaning, e.g. theestimated F-SAC water HBenergy is 21.84 kJ/mol and itshould beb in the range of 20.4to 23.3 kJ/mol.

aR.P. Soares et al. “Functional-Segment Ac-tivity Coefficient model. 2. Associating Mix-tures”. In: Ind. Eng. Chem. Res. (2013).doi: 10.1021/ie4013979

bK. Wendler et al. In: The J. of Phys. Chem.A 114.35 (2010), pp. 9529–9536


IntroductionF-SAC

Conclusions


IDAC comparison for non-associating mixtures

-2 -1 0 1 2 3 4 5 6Logarithm of Experimental IDAC

-2

-1

0

1

2

3

4

5

6

Log

arith

m o

f M

odel

ID

AC

(e) F-SAC


-2

-1

0

1

2

3

4

5

6

Log

arith

m o

f M

odel

ID

AC

(f) UNIFAC (Do)


IntroductionF-SAC

Conclusions


IDAC comparison for associating mixtures (water excluded)


-2

-1

0

1

2

3

4

5

6

Log

arith

m o

f M

odel

ID

AC

(g) F-SAC


-2

-1

0

1

2

3

4

5

6

Log

arith

m o

f M

odel

ID

AC

(h) UNIFAC (Do)


IntroductionF-SAC

Conclusions


IDAC comparison for aqueous mixtures

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Logarithm of Experimental IDAC

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

Log

arith

m o

f M

odel

ID

AC

(i) F-SAC

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Logarithm of Experimental IDAC

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

Log

arith

m o

f M

odel

ID

AC

(j) UNIFAC (Do)


IntroductionF-SAC

Conclusions


VLE predictions for non-associating mixtures

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0.075

0.100

0.125

0.150

0.175

0.200

0.225

0.250

Pre

ssu

re[b

ar]

Exp. 298.2 K

F-SAC

UNIFAC(Do)

COSMO-SAC

(k) Chloroform/n-heptane

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

Pre

ssu

re[b

ar]

Exp. 323.15 K

F-SAC

UNIFAC(Do)

COSMO-SAC

(l) Acetone/cyclohexane


IntroductionF-SAC

Conclusions


VLE predictions for non-associating mixtures

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

2.00

2.25

2.50

2.75

3.00

3.25

3.50

3.75

Pre

ssu

re[b

ar]

Exp. 283.6 K

F-SAC

UNIFAC(Do)

COSMO-SAC

(m) Dimethyl ether/1-butene

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Pre

ssu

re[b

ar]

Exp. 323.15 K

F-SAC

UNIFAC(Do)

COSMO-SAC

(n) Methyl acetate/1-hexene


IntroductionF-SAC

Conclusions


VLE predictions for associating mixtures

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0.25

0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

Pre

ssu

re[b

ar]

Exp. 298.15 K

F-SAC

UNIFAC(Do)

(o) diethyl-ether/chloroform

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0.25

0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

Pre

ssu

re[b

ar]

Exp. 380.15 K

Exp. 368.15 K

F-SAC

UNIFAC(Do)

(p) toluene/3-methyl,1-butanol


Predicting high-pressure VLE with SCMR

SRK with Mathias-Copeman α function and theSelf-Consistent Mixing Rule4 (SCMR) for propane-benzene,no parameter is adjusted for the mixture effects:

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0x1, y1

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

Pre

ssu

re[b

ar]

Exp. 310.93 K

Exp. 344.26 K

Exp. 377.59 K

Exp. 477.59 K

SRK(MC)+SCMR(FSAC)

4Paula B. Staudt and Rafael de P. Soares. In: Fluid Phase Equilibria 334(2012), pp. 76–88.

Possible representations for IL

According to the literature5, one of the following:

Meta-file: the IL is considered a single compound, theσ-profile is given by the sum of ions ⇐

Ion-pair: COSMO computations with the ion-pairElectroneutral mixture: independent ions in mixture in

equimolar proportion

5M. Diedenhofen and A. Klamt. In: Fluid Phase Equilib. 294.1-2 (2010),pp. 31–38.

IntroductionF-SAC

Conclusions


F-SAC: formulation for IL

We have selected the meta-file approach for F-SAC: cationsand anions are the functional groups and the IL σ-profile isgiven by the groups addition:


IntroductionF-SAC

Conclusions


F-SAC: parameters for charged groups

The number of independent parameters per group is the same:

Qak first charged peak

Qbk second charged peakσa

k charge density of the first peak

The other parameters are:

z group charge (+1, -1, etc.)Q◦

k = Qk − Qak − Qb

k : neutral areaσb

k = (z/Qk − σak Qa

k )/Qbk : charge density of the

second peak


IntroductionF-SAC

Conclusions


Cations selected for the first study


IntroductionF-SAC

Conclusions


Anions selected for the first study


IntroductionF-SAC

Conclusions


IL tested

A databank was assembled with 1031 mixtures with 9different IL:

[EMIM]+[BF4]−

[BMIM]+[BF4]−

[HMIM]+[BF4]−

[OMIM]+[BF4]−

[EMIM]+[TCB]−

[HMIM]+[TCB]−

[OMIM]+[CL]−

[EMIM]+[MESO3]−

[EMIM]+[CF3SO3]−


IntroductionF-SAC

Conclusions


IDAC of 1031 mixtures (substances dissolved in IL)

The average deviation was 0,17 ln units, similar to theobserved for the neutral mixtures


IntroductionF-SAC

Conclusions

ConclusionsOngoing researchLinks e more information

Conclusions

F-SAC is a new group contribution model

Its mathematical formulation is more complex (COSMO-RSbased), but it requires less parameters (and experimentaldata) than UNIFAC

When correlating a diverse IDAC database for non associatingmixtures, the F-SAC correlation deviated from experimentaldata by 0.073 ln units against 0.14 for UNIFAC (Do)predictions

For aqueous mixtures the F-SAC correlation deviated by0.338 while the UNIFAC (Do) predictions deviated by 1.23

Good predictions were also obtained for VLE data, notincluded in the parameter fitting process


IntroductionF-SAC

Conclusions


Ongoing research

The F-SAC model is currently being revised and extended

The number of groups is being extended

Now we are also considering VLE and LLE data in theparameter estimation process for improved parameter quality

Improved formulations for associating mixtures (hydrogenbonds) are being tested

Applications for polymer systems are being investigated

Parameters for light gases are being calibrated by means ofCEOs/GE mixing rules


IntroductionF-SAC

Conclusions


Thank you!

Download F-SAC demonstration code athttp://www.enq.ufrgs.br/labs/lvpp

Download the JCOSMO program, including source code, athttp://code.google.com/p/jcosmo/

DOI:10.1021/ie400170a DOI:10.1021/ie4013979


http://www.enq.ufrgs.br/labs/lvpp




Documents

Current status of the Functional-Segment Activity ... · UNIFAC COSMO-RS JCOSMO program COSMO-RS { Surface contacting theory For each surface pair contact, there is an energy change