Science & Technology

A Study of the Affinity of Dyes for Cellulose Fiber within the Framework of a Fragment Approach in QSPR

Description
A Study of the Affinity of Dyes for Cellulose Fiber within the Framework of a Fragment Approach in QSPR
Published
of 5
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  1070-4272/05/7806-1013    2005 Pleiades Publishing, Inc.  Russian Journal of Applied Chemistry, Vol. 78, No. 6, 2005, pp. 1013   1017. Translated from Zhurnal Prikladnoi Khimii, Vol. 78, No. 6, 2005, pp. 1034   1037.Original Russian Text Copyright     2005 by Zhokhova, Baskin, Palyulin, A.Zefirov, N.Zefirov. MACROMOLECULAR CHEMISTRY   AND POLYMERIC MATERIALS A Study of the Affinity of Dyes for Cellulose Fiberwithin the Framework of a Fragment Approach in QSPR N. I. Zhokhova, I. I. Baskin, V. A. Palyulin, A. N. Zefirov, and N. S. Zefirov  Lomonosov State University, Moscow, Russia Received October 20, 2004; in final form, March 2005 Abstract    The affinity of azo and anthraquinone dyes for the cellulose fiber was first studied by the fragmentapproach in terms of the QSPR method. Linear-regression models that can prognosticate this parameter onthe basis of descriptors taking into account the fragment composition of a molecule were suggested. The interaction of various dyes with a cotton fiberis a complex physicochemical process governed byspecific features of the structure of the textile poly-mer and by the nature of a dye molecule. This inter-action is affected by various factors (electrostaticfields, Van der Waals forces, formation of hydrogenbonds, hydrophobicity, etc.), and the available ex-perimental data are rather diverse. That is why theo-retical methods for calculation of quantitative struc-ture    property/activity relationships (QSPR/QSAR and3D-QSAR) are widely used to study these interac-tions [1].One of the main parameters characterizing the in-teraction of a dye with a fiber is the chemical affinity   0 (kJ mol   1 ). It has been shown that, in the fixa-tion of a dye molecule on a cellulose fiber, the keyinfluence on the affinity of anionic and neutral azo[2, 3], heterocyclic monoazo [4], symmetric diazo [5],and anthraquinone [6] dyes for the cellulose fiber isexerted by electrostatic interactions. At the same time,it is commonly assumed thatthe binding of dye mol-ecules may occur with specific binding centers in-volved, and the existence of these centers is due to thesupramolecular structure of cellulose [7]. When study-ing dye    fiber intermolecular interactions, the authorsof [8] concluded that steric factors are less importantin this case than for the classical ligand    receptor inter-actions. As for the spatial structure of a dye moleculeand presence of various kinds of substructures, theyshould be regarded as additional factors that affect thedistribution of electrostatic potentials around a mol-ecule, which are of primary importance for the dye   fiber interaction, too. In this context, it seems to beof interest to study how the affinity of dye moleculesfor cellulose depends on their fragment composition,which remains virtually unstudied. In the study,the affinity of a number of azo and anthraquinonedyes for cellulose was examined using the fragmentapproach the framework of the QSPR methodology.Previously, descriptors characterizing the fragmentcomposition of a molecule have been successfullyused for QSPR-prognostication of physicochemicalproperties of organic compounds of various classes,including the chromatographic retention indices [9],boiling point [9], flash point [10], sublimation en-thalpy [11], polarizability [12], and magnetic suscep-tibility [13].The study was performed using the NASAWINQSAR/QSPR software package [14, 15] and the in-tegrated FRAGMENT software. Databases were com-piled using MEOW and BASTET software speciallydeveloped for creating and refining structural data-bases [12]. Three samples of compounds, containing30 anionic sulfur-containing azo dyes [3] were studied(Database 1). The affinity of a dye for the fabric waschosen as the property to be modeled. It was calcu-lated as the difference of the standard chemical poten-tials of a dye molecule fixed on the fabric surfaceand that in an aqueous solution. These thermodynamicdata were experimentally obtained in aqueous solu-tions at three temperatures and were described indetail in [16]. Further, data for 49 anthraquinone dyes[8] were collected (Database 2). Finally, a combinedsample was obtained, which contained both sets of structures (Database 3). Based on a preliminary veri-fication of all the databases by the BASTED software,the following corrections were introduced, by agree-  RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 78 No. 6 2005 1014 ZHOKHOVA  et al. Statistical parameters of QSPR models based on fragment descriptors for the affinity of azo and anthraquinone dyesfor cellulose fiber   Model *  Database   R 2learn   s   R 2**prog   F    N  ***   1   Database 1 (azo dyes)   0.949   0.87   0.896   88.6   4/62   0.957   0.81   0.850   81.7   5/63   0.958   0.83   0.839   64.9   6/64   0.949   0.87   0.900   88.6   4/105   0.957   0.81   0.850   81.7   5/106   0.971   0.95   0.908   161.7   4/157   Database 2 (anthraquinone dyes)   0.918   0.56   0.860   51.1   7/158   0.924   0.55   0.866   47.5   8/159   0.931   0.53   0.866   44.9   9/1510   Database 3 (azo and anthraquinone dyes)   0.940   1.66   0.765   94.3   9/1511   0.942   1.65   0.772   86.6   10/1512   0.945   1.63   0.785   80.7   11/15   * Model nos. 1    6: Fr1/=CR    C, Fr2/RC Ar    C Ar    NH 2 , Fr3/    =               (   is an arbitrary atom), Fr4/      N;   )  model nos. 2, 5:Fr5/C    (C    ) 2 2C    C=O; model no. 3: Fr5/     O, )   C Ar Fr6/      O; )    model no. 6: Fr4/C    C    (C    ) 3 C    N=N    (C    ) 5 C    N; model no. 7:Fr1/RC Ar , Fr2/C sp 3    NHR, Fr3/C Alk     NR    C, Fr4/C Alk     RC Ar    C Ar , Fr5/C Alk     C    NH    (C Ar    ) 2 C Ar , Fr6/>N    (C Ar    ) 3 C Ar    N<,Fr7/C sp 3    O    (C Ar    ) 7 C Ar    N<; model no. 8: Fr1    Fr4 as in model nos. 7, Fr5/(C Ar    ) 3 C Ar    OH, Fr6, Fr7, Fr8 as Fr5, Fr6 and Fr7model nos. 7, respectively; model no. 9: Fr1    Fr7 as in model nos. 8, Fr8/(C Ar    ) 5 C Ar    Cl, Fr9 as Fr8 model nos. 8; modelno. 10: Fr1/C=, Fr2/     OH, Fr3/C sp 3    N sp 3    C, Fr4/RC Ar    C Ar    C Ar , Fr5/(C    ) 3 C    N, Fr6/(C Ar    ) 3 C Ar    C=O, Fr7/      (     ) 4         =  (   is an arbitrary atom), Fr8/=C    N sp 3    (C Ar    ) 7 C Ar    OH, Fr9/Cl    C Ar (    C Ar H) 2 ; model no. 11: Fr1    Fr6 as in model nos. 10,Fr7/C    C    (C Ar    ) 3 C    N, Fr8, Fr9, Fr10 as Fr7, Fr8, Fr9 model nos. 10, respectively; model no. 12: Fr1    Fr4 as in model nos. 11,Fr5/N sp 3    C Ar    C Ar    N=, Fr6, Fr7, Fr8, Fr9, Fr10, Fr11 as Fr5, Fr6, Fr7, Fr8, Fr9, Fr10 model nos. 1 1. ** Control sample includes every fifth compound of a database. *** Number of fragment descriptors/Maximum number of atoms in a fragment. ment with the authors of [8], into Database 2: sub-stituent 1-NHCOCH 3  in structure no. 12 changed wschanged to 1-NHCOCH 2 Cl, and the place of additionof the substituent    [4-OCH 3 C 6 H 4 CONH    ] in structureno. 41, from position 5 to position 4.In the first stage of the study, linear-regressionmodels based on descriptors characterizing fragmentswith the maximum chain lengths of 6, 10, and 15atoms and an external control sample including everyfifth compound of a database were constructed forthe samples under study using the NASAWIN soft-ware package. These models (see table, model nos. 1   12) have good descriptive and prognosticating prop-erties. The best prognosis quality was achieved withfragments having chains up to 15 atoms long includedin the model.The quality of fragment-based and published mod-els obtained for the same samples by means of a com-parative molecular field analysis and a multiple linearregression with the use of quantum-chemical descrip-tors (azo compounds) [3], and by the method of a comparative molecular surface analysis (CoMSA,azo and anthraquinone compounds) [8] was comparedby constructing a number of models using a slid-ing control with exclusion of a single compound.The models obtained are comparable in their proper-ties with those cited in the literature, and in somecase, even surpass them. For example, the regressionmodel constructed for Database 1 with four fragmentdescriptors (chain in a fragment up to 15 atoms long)has parameters (  R 2learn  = 0.967; standard deviation  s  =0.66;  F   = 181.6; squared correlation coefficient witha sliding control,  Q 2 = 0.949; root-mean-square error  RMS  sl  = 0.74; standard deviation with a sliding con-trol,  s sl  = 0.80) exceeding those of the best regressionmodel constructed for the same sample with the ener-gies of the upper occupied and lower unoccupiedmolecular orbitals  E  HOMO ,  E  2HOMO , and  E  LUMO , cal-culated by the AM1 method, used as descriptors(three descriptors,  R 2learn  = 0.92,  s  = 1.02,  F   = 94.9, Q 2 = 0.89, standard deviation with a slidingcontrol 1.19) [3]. For the PLS CoMFA [3] andCoMSA [8] models obtained for the given samples  RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 78 No. 6 2005 A STUDY OF THE AFFINITY OF DYES FOR CELLULOSE FIBER 1015 the values of   Q 2 fall within the ranges 0.63    0.75and 0.829    0.970, respectively. The model construct-ed includes descriptors that describe the follow-ing fragments of azo compound molecules:    RC    C,RC Ar    C Ar    NH 2 ,                  (   is an arbitraryatom) and C    C    C    (C    ) 2 C    N    N    C    (C    ) 4 C    N.These fragment chains are more important for charac-terizing the affinity of the azo dyes. The largest differ-ence between the calculated and experimental affin-ities is observed for structure  1  (1.3 kJ mol   1 ) and  7 (1.2 kJ mol   1 ):        S      N    N      SOH      OO 1          N    N      SOH      OO 7   OH   Combination of fragment descriptors with quantum-chemical descriptors of the energy of molecular or-bitals,  E  HOMO  and  E  ILUMO  (data of [3] are used), failsto improve the quality of the models. Possibly, thesedescriptors are approximated by fragment descriptors.The fragment models constructed for a sample of anthraquinone dyes (Database 2) also compare wellin their prognosticating properties with publishedmodels [8]. Squared correlation coefficient witha sliding control for a model including eight de-scriptors (chains of 15 atoms;  R 2learn  = 0.942,  s  =0.46,  F   = 81.5,  Q 2 = 0.915,  RMS  sl  = 0.50,  s sl  = 0.55)exceeds the maximum value of   Q 2 (0.88) forthe Co MSA model [8]. The largest contributionto the model RC Ar , and comes from the follow-ing components: C Ar    N sp 3    C, HC Ar    C Ar    NHR, andC sp 3    O    (C Ar    ) 7 C Ar    N sp 3 . The maximum deviationfrom the correlation is observed for structure  8 (0.98 kcal mol   1 ) and  19  (0.94 kcal mol   1 ):     S    ONH    N    N     SOH      OO    NH 2 HO        OOSHO 8        N    N    HO     SOH      OO        OOSOH   19 The most important fragments included in the mod-els for azo and anthraquinone compounds are distin-guished by the existence of long chains of conjugatedsystems, which reflects their importance for the char-acterization of the property being modeled.An attempt was made to construct a more versatilemodel for describing the dye    cellulose affinity bymerging narrow samples of azo and anthraquinonedyes (Database 3). The prognosticating ability of this model was assessed by using the sliding controlmethod. To make Database 2 uniform, the affinitiesfrom this database (kcal mol   1 ) were expressed inkJ mol   1 . The model obtained includes 10 fragmentdescriptors (chains of 15 atoms) and has the followingparameters:  R 2learn  = 0.933,  s  = 1.72,  F   = 94.7,  Q 2 =0.901,  s sl  = 2.09,  RMS  sl  = 1.94. The largest deviationfrom the correlation is observed for compound no.68 (5.5 kJ mol   1 ):       ONH  OHOH    NH Cl  O   Cl Below, the model obtained with this structure ex-cluded from the learning sample is presented:   0 =    4.36 + 1.87Fr1    2.63Fr2    1.31Fr3   1.38Fr4 + 0.67Fr5 + 1.56Fr6 + 1.14Fr7+ 1.78Fr8    0.82Fr9 + 1.38Fr10, (1) where  R 2learn  = 0.947,  s  = 1.47,  F   = 119.4,  Q 2 =0.909,  s m  = 1.93,  RMS  m  = 1.79; Fr1/C=, Fr2/     OH,Fr3/C sp 3    N sp 3    C, Fr4/C Ar    (C=)    C, Fr5/RC Ar    C Ar    C Ar ,Fr6/N    C    C    N, Fr7/                   ; Fr8/C Ar    (C=)   NR    (C Ar    ) 7 C Ar    OH, Fr9/C    N    (C    )7C    N    C=O,Fr10/Cl    C Ar (    C Ar H) 2 .In contrast to the case of models constructed fornarrow samples (Databases 1 and 2), the largest con-  RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 78 No. 6 2005 1016 ZHOKHOVA  et al. Diagram of scatter of the calculated and experimentalvalues of the affinity of dyes for cellulose,    0 , fora learning sample of Database 3, obtained in terms of thelinear-regression model [Eq. (1)]. tribution to the combined model comes from descrip-tors of rather simple fragments: Fr1/C= and Fr2/     OH.The correlation plot for the calculated and experi-mental values of     0 for the combined model isshown in the figure.Of interest is assessment of the prognosticatingcapacity of combined fragment models for the azoand anthraquinone dye samples under study. For thispurpose, models were obtained using the combinedDatabase 3 and the sliding control and were testedon two independent external samples including, re-spectively, every fifth compound from Databases 1and 2. With these samples excluded from Database 3,the number of azo compounds and anthraquinones inthe combined learning sample was 24 and 40, respec-tively. It was found that the prognosticating capacityof such a model (10 fragment descriptors, chains upto 15 atoms long) for an independent control sampleof azo compounds considerably exceeds that for a sam-ple of anthraquinones. For example, squared correla-tion coefficients between the experimental and pre-dicted values,  R 2prog , for the appropriate control sam-ples are 0.941 and 0.573, respectively. These dataare in an agreement with the results of the study [8],in which the CoMSA models obtained for a sample of azo compounds surpass in prognosticating propertiesthe models for a sample of anthraquinones. This con-firms the efficiency of application of the fragmentapproach to modeling of the properties under study.To improve the quality of the models, it seemedappropriate to include, in addition to the fragmentdescriptors, those characterizing the geometry (shapeand size) of a molecule and making it possible toevaluate the degree of its compactness, i.e., topolog-ical indices. Descriptors of this kind were used tomodel the affinity of anthraquinone dyes for cellulosein [17]. More than 30 topological indices calculatedusing the EMMA software package [18, 19] were ana-lyzed. However, the desired improvement of the qual-ity of the models could not be achieved. This meansthat the set of fragment descriptors used in the presentstudy enables a rather complete description of the re-lationship between the structural features of moleculesof the dye groups studied and their affinity for cel-lulose.CONCLUSIONLinear-regression QSPR models making it possibleto prognosticate the affinity of azo and anthra-quinone dyes for the cellulose fiber on the basis of descriptors that take into account the fragment com-position of a dye molecule were suggested for the firsttime.REFERENCES 1. Timofei, S., Schmidt, W., Kurunczi, L., and Simon, Z.,  Dyes Pigm. , 2000, vol. 47, nos. 1    2, pp. 5    16.2. Funar-Timofei, S. and Schr uurmann, G.,  J. Chem. Inf. Compt. Sci. , 2002, vol. 42, no. 4, pp. 788    795.3. Schr uurmann, G. and Funar-Timofei, S.,  J. Chem. Inf. Compt. Sci. , 2003, vol. 43, no. 5, pp. 1502    1512.4. Timofei, S. and Fabian, W.M.F.,  J. Chem. Inf. Compt.Sci. , 1998, vol. 38, no. 6, pp. 1218    1222.5. Fabian, W.M.F. and Timofei, S.,  J. Mol. Struct.,Theochem , 1996, no. 362, pp. 155    162.6. Fabian, W.M.F., Timofei, S., and Kurunczi, L.,  J. Mol.Struct., Theochem , 1995, no. 340, pp. 73    81.7. French, A.D., Battisla, O.A., Cuculo, J.A., andGray, D.G.,  Kirk-Othmer Encyclopedia of ChemicalTechnology, 4th ed. , New York: Wiley, 1993, vol. 5,p. 476.8. Polanski, J., Gieleciak, R., and Wyszomirski, M.,  J. Chem. Inf. Comp. Sci. , 2003, vol. 43, no. 6,pp. 1754    1762.9. Zefirov, N.S. and Palyulin, V.A.,  J.Chem. Inf. Compt.Sci. , 2001, vol. 41, no. 4, pp. 1022    1027.10. Zhokhova, N.I., Palyulin, V.A., Baskin, I.I.,  et al. ,  Izv. Akad. Nauk. Ser. Khim. , 2003, no. 9,pp. 1787    1793.11. Zhokhova, N.I., Palyulin, V.A., Baskin, I.I.,  et al. ,  Zh. Prikl. Khim. , 2003, vol. 76, no. 12, pp. 1914    1919.12. Zhokhova, N.I., Palyulin, V.A., Baskin, I.I.,  et al. ,  RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 78 No. 6 2005 A STUDY OF THE AFFINITY OF DYES FOR CELLULOSE FIBER 1017  Izv. Akad. Nauk. Ser. Khim. , 2003, no. 5,pp. 1005    1009.13. Zhokhova, N.I., Palyulin, V.A., Baskin, I.I.,  et al. ,  Zh. Strukt. Khim. , 2004, vol. 45, no. 4, pp. 659    668.14. Baskin, I.I., Palyulin, V.A., and Zefirov, N.S.,  QSARand Molecular Modelling: Concepts, ComputationalTools and Biological Applications , Sanz, F., Giraldo, J.,and Manaut, F., Eds., Barcelona: Prous Sci. Publishers,1995, p. 30.15. Baskin, I.I., Palyulin, V.A., and Zefirov, N.S.,  J. Chem. Inf. Comput. Sci. , 1997, vol. 37, no. 4,pp. 715    721.16. Alberti, G., Cerniani, A., de Giori, M.R., and Seu, G., Tinctoria , 1980, no. 5, pp. 141    143.17. Timofei, S., Schmidt, W., Kurunczi, L.,  et al. ,  DyesPigm. , 1994, vol. 24, no. 4, pp. 267    279.18. Petelin, D.E., Palyulin, V.A., and Zefirov, N.S.,  Dokl. Akad. Nauk  , 1992, vol. 324, no. 5, pp. 1019    1022.19. Zefirov, N.S. and Palyulin, V.A.,  J. Chem. Inf. Compt.Sci. , 2002, vol. 42, no. 5, pp. 1112    1122.
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks