www.elsevier.com/locate/jphotobiol

Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251

Study on the Interaction between the chiral drug of propranololand a1-acid glycoprotein by fluorescence spectrophotometry

Feng Zhang a,b, Yingxiang Du a,b,*, Baofen Ye b, Ping Li a

a Key Laboratory of Modern Chinese Medicines, Ministry of Education, China Pharmaceutical University, Nanjing 210009, Chinab Department of Analytical Chemistry, China Pharmaceutical University, Nanjing 210009, China

Received 12 August 2006; received in revised form 3 November 2006; accepted 16 November 2006Available online 5 January 2007

Abstract

The interaction between the chiral drug of propranolol (PPL) and a1-acid glycoprotein (AGP, orosomucoid) has been first studied byfluorescence spectrophotometry. The fluorescence intensity of PPL increased due to the addition of AGP into PPL. The equation of Scat-chard was employed to calculate the association constant and binding site number of the two enantiomers with AGP. The associationconstant is 2.62 · 105 M�1 for R-PPL and 8.57 · 105 M�1 for S-PPL and the binding site number is 0.41 for R-PPL and 1.17 for S-PPL at17 �C respectively. The method of thermodynamics was applied to determine the binding type of S-PPL with AGP. The results suggestedthat the binding type is mainly van der waals force or hydrogen bond. At last the effect of three metal cations on the association constantand the binding site number of S-PPL with AGP was examined.� 2006 Elsevier B.V. All rights reserved.

Keywords: Propranolol; a1-acid glycoprotein; Chiral drug; Fluorescence spectrophotometry; Interaction

1. Introduction

An extensive attention has been paid to study on theinteraction between drug and biology macromolecule (suchas protein, DNA, etc.) in the scope of life science, chemistryand clinic medicine [1]. The interaction of drug with proteinin serum can influence bioavailability of drug and affect thefunction of several biomolecules. Clinical effect of the plas-matic level of a drug is a very important pharmacologicalparameter determined by absorption, distribution andelimination. Chiral drug plays an important role in ourdaily administration of medicine. The different isomershave different clinical effects. Probably one isomer can curedisease while the other isomer can result in side effect, evenmake disease worse, and maybe it is fatal [2–4].

1011-1344/$ - see front matter � 2006 Elsevier B.V. All rights reserved.

doi:10.1016/j.jphotobiol.2006.11.002

* Corresponding author. Address: Department of Analytical Chemistry,China Pharmaceutical University, No. 24, Tongjiaxiang, Nanjing, Jiangsu210009, China. Tel./fax: +86 25 83221790.

E-mail address: yingxiangd@yahoo.com.cn (Y. Du).

The human a1-acid glycoprotein (AGP, orosomucoid) isa small acute-phase glycoprotein (Mr � 44,000) that is neg-atively charged at physiological pH. It consists of a chainof 181 amino acids, and contains 40% carbohydrate byweight and has 16 sialic acid residues (10–14% by weight).Five heteropolysaccharide groups are linked via an N-gly-cosylic bond to asparaginyl residues of the protein. Theprotein contains tetrantennary as well as di- and trianten-nary glycans [5–8]. In the acute phase (e.g. cancer orinflammation and so on), its content increases evidently.It is a very important protein in the blood, which bindsto basic drugs. Propranolol (PPL, Fig. 1) is one of mostwidely used and well-know racemic drugs, a blocking drugfor treating certain cardiovascular via restraining sympa-thetic nerve from undue excitement or excreting catechol-amine. So it can be used as anti-arrhythmia drug. Itsmain carrier protein is a1-acid glycoprotein [9,10]. PPLhas two isomers, and its racemate is commonly used fortreatment. But R-PPL can bring about side effect.

Presently, the methods employed to study the interac-tion of drug with protein mainly include: ultraviolet

O

OH

NH

CH3

CH3

Fig. 1. The structure of PPL.

F. Zhang et al. / Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251 247

spectrophotometry, fluorescence spectrophotometry, chro-matography (GC, LC, CE, etc.), equilibrium dialysis[5,11–15] and so on. Previously, there were several methodsto study the reaction mechanism of PPL with AGP, forexample capillary electrophoresis [5], equilibrium dialysis[14] and so on. However, those methods were laboriousand time consuming.

In this paper, fluorescence spectrophotometry was firstemployed to study the interaction of PPL and AGP. Theresults can explain why S-PPL is more effective thanR-PPL. The proposed method is very rapid and highly sen-sitive because sample preparation and fluorescence deter-mination are simple and facility.

2. Experimental

2.1. Materials and apparatus

AGP (purity >99%), and R- and S- enantiomers of PPLwere purchased from Sigma Chemical Co. F-5301PC fluo-rescence spectrophotometer (Shimidzu Company, Jap.),pH 25 pH meter (Shanghai, China). The other reagentswere all analytical purity. Water used in the experimentwas doubly distilled water.

100150200250300350

F

2.2. Methods

2.2.1. Spectroscopic measurement

Aqueous stock solution (1.0 ml) of R-, S- or DL-PPLwas accurately transferred into a 5 ml volumetric flask,and then an appropriate amount of 2.0 · 10�5 mol/LAGP solution and 1 ml of 0.05 mol/L Tris–HCl buffer(pH 7.0, containing 0.1 mol/L sodium chloride) were addedsuccessively. The mixed solution was diluted to the markwith 0.1 mol/L sodium chloride solution, shaken thor-oughly and then equilibrated for 10 min. Fluorescencespectra were recorded or fluorescence value was measuredat the excitation wavelength of 305 nm and emission wave-length of 337 nm.

00 2 4 6 8 10 12

50

pH

Fig. 2. Effect of pH on the fluorescence intensity of PPL CDL-PPL =2.24 · 10�6 mol/L.

2.2.2. Fluorescence quenching by KI

S-PPL (1.0 ml of 1.6 · 10�5 mol/L), 2.0 ml of 2.0 ·10�5 mol/L AGP solution (the AGP solution was notadded in the experiment of the fluorescence quenching of

S-PPL by KI), 1 ml of 0.05 mol/L Tris–HCl buffer (pH7.0, containing 0.1 mol/L sodium chloride) and a differentvolume of 0.1 mol/L KI solution were accurately addedinto a 5 ml volumetric flask successively. The followingprocedure was the same as Section 2.2.1. Fluorescencemeasurement was also performed as Section 2.2.1.

2.2.3. Effect of pH on the fluorescence of PPL

DL-PPL stock solution (1.0 ml) was accurately addedinto a series of 5 ml volumetric flasks and diluted to themark with 0.01 mol/L Tris–HCl buffer of different pHvalues containing 0.1 mol/L sodium chloride, and thentheir fluorescence values were measured at the excitationwavelength of 305 nm and emission wavelength of337 nm.

3. Results and discussion

3.1. Effect of pH on the fluorescence of PPL

The effect of pH on the fluorescence of DL-PPL was firstexamined. The results suggested that the fluorescenceintensity of PPL decreased in the hard acidic or alkalinesolution. Moreover, the red shift of the maximum emissionwavelength of PPL was observed in above conditions. Ascan be seen from Fig. 2, the fluorescence intensity of PPLdecreased when pH was above 9.5 or below 2.3. But thefluorescence intensity change of PPL almost did not takeplace between pH 2.3 and 9.5. So pH 7.0 was chosen in thisstudy.

It is known that there is a –OH group in PPL molecule(Fig. 1). In the neutral, soft basic or acidic condition, the–OH group can form hydrogen bond with H of naphtha-lene ring, so that the rigidity of PPL molecule becomesstronger and the fluorescence intensity of PPL increases.But in hard acidic condition, H+ in solvent can formhydrogen bond with the –OH group in PPL competi-tively. Similarly in hard alkaline condition OH� in solventcan form hydrogen bond with H of naphthalene ring andthe –OH group in PPL. Therefore, the hydrogen bondbetween the –OH group in PPL and H of naphthalenering becomes weak and the fluorescence intensity ofPPL decreases.

380

390

400

410

420

430

0 50 100 150reaction time (min)

F

Fig. 3. Dependence of the fluorescence intensity of PPL on reaction timeCDL-PPL = 2.64 · 10�6 mol/L, CAGP = 2.03 · 10�6 mol/L. t = 17 �C.

248 F. Zhang et al. / Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251

3.2. Choice of reaction time, excitation and emission

wavelength

Dependence of the fluorescence intensity of DL-PPL onreaction time was investigated. The result was showed inFig. 3. Obviously, before 10 min the fluorescence intensityof PPL increased first and then decreased, but it was steadyafter 10 min. Therefore, the reaction time was chosen as10 min in this paper.

In addition, it was found that AGP could emit fluores-cence and disturb the fluorescence measurement of PPLwhen excitation wavelength was below 300 nm. But whenexcitation wavelength was more than 300 nm, the fluores-cence of AGP disappeared. So fluorescence measurementwas performed at the excitation wavelength of 305 nmand emission wavelength of 337 nm (the maximum emis-sion wavelength of PPL) throughout all experiment.

3.3. Mechanism of fluorescence enhancement

Fig. 4 shows the phenomenon of fluorescence enhance-ment. When the concentration of DL-PPL was fixed, thefluorescence of DL-PPL solution was enhanced with theincrease of AGP concentration. As AGP contains threetryptophan (Trp) residues, one of which is on the surfaceand the other two embedded in the protein matrix [12], it

Fig. 4. Effect of the concentration of AGP on the fluorescence of PPL pH7.00, CDL-PPL = 2.64 · 10�6 mol/L, the concentration of AGP(·10�6 mol/L) was 0, 0.41, 2.03, 3.24, 4.86 (From 1 to 5). t = 17 �C.

can emit fluorescence at the excitation wavelength rangingfrom 270 nm to 300 nm. But above the excitation wave-length of 300 nm, the fluorescence disappeared, as seen inSection 3.2. As a result, it is evident that the fluorescenceenhancement of PPL results from the formation of a newcomplex of PPL and AGP through hydrogen bond.Because the rigidity and coplanarity of PPL Æ AGP com-plex is stronger than that of PPL, the fluorescence intensityof PPL is increased.

In order to explain the formation mechanism ofPPL Æ AGP complex further, we investigated the effect ofKI on the fluorescence quenching of S-PPL and S-PPL Æ AGP complex. Only S-PPL was examined becauseR-PPL can lead to side effect. The explanation is basedon the Stern–Volmer equation, as follows [16–18]:

F 0=F ¼ 1þ KsC ð1Þ

where F0 is the fluorescence value of PPL or PPL Æ AGPwithout KI, F is the fluorescence value with addition ofKI into the solution of AGP or PPL Æ AGP, Ks is quench-ing constant, C is the concentration of KI.

Fig. 5 shows the Stern–Volmer plots. It was clear thatthe quenching degree of PPL by KI was more than thatof PPL Æ AGP. It is indicated that PPL binds into the innerof AGP because if PPL binds on the surface of AGP, thequenching degree of PPL and PPL Æ AGP by KI shouldbe equivalent [19].

3.4. Calculation of association constant (KA) and bindingsite number (n)

3.4.1. Calculation of the concentration of bound PPL (Cb)

and free PPL (Cf)

As the fluorescence of PPL can be enhanced by AGP, anexperiential formula lg F 0�F

F ¼ lg KA þ n lg½Q� [16] cannot bedirectly used to estimate KA and n. Formula (8) and (9) inSection 3.4.2 were applied to calculate these two parame-ters in our work. But the concentration of free drug (Cf)and bound drug (Cb) must be first obtained. In a mixedsolution of PPL and AGP the total fluorescence value (F)can be expressed as follows:

F ¼ K fCf þ KcCc ð2Þ

Fig. 5. The Stern–Volmer curve of fluorescence quenching by KICS-PPL = 3.33 · 10�6 mol/L, CAGP = 8.56 · 10�6 mol/L.

F. Zhang et al. / Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251 249

where Kf is the fluorescence constant of free drug, Cf is theconcentration of free drug, Kc is the fluorescence constantof complex, Cc is the concentration of complex.

Cc ¼1

a:Cb ð3Þ

where a is the mole number of PPL (aPPL + bAGP =PPLa Æ AGPb). So formula (2) can be expressed as

F ¼ K f Cf þ1

aKcCb ð4Þ

Because Kc/a is also a constant, we can use Kb to replaceKc/a. Then formula (2) can be represented further as

F ¼ K f Cf þ KbCb ð5ÞCt ¼ Cf þ Cb ð6Þ

where Ct is the total concentration of drug. When Cb = 0,Kf = F/Ct.

Kb value can be acquired based on a simple experimentwithout separation of free PPL from a mixture solution, asfollows. S-PPL (2.5 ml) (C = 6.76 · 10�7 mol/L) or R-PPL(2.5 ml) (C = 6.97 · 10�7 mol/L) solution (the solvent ispH 7.0, 0.01 mol/L Tris–HCl buffer containing 0.1 mol/Lsodium chloride) was exactly transferred into a cuvettewith 1 cm light path and then an appropriate amount(<30 ll) of 2.5 · 10�4 mol/L AGP solution was added.The mixed solution was shaken thoroughly and equili-brated for 10 min. Fluorescence value was measured atthe excitation wavelength of 305 nm and emission wave-length of 337 nm. The results were shown in Fig. 6.

The fluorescence intensity of R-PPL Æ AGP and S-PPL Æ AGP tended to increase as AGP concentrationincreased. For S-PPL fluorescence intensity reached tothe maximum of 130.8 at AGP concentration of

Fig. 6. Calculation of KSb and KR

b (a) S-PPL, C = 6.76 · 10�7 mol/L; (b)R-PPL, C = 6.97 · 10�7 mol/L. t = 17 �C.

1.48 · 10�6 mol/L and did not change almost above thisconcentration. The similar tendency was observed forR-PPL. It was indicated that all of R-PPL and S-PPLenantiomers bound to AGP nearly above the maximumconcentration of AGP. That was, Cf in formula (5) and for-mula (6) could be considered 0 in this condition. Accordingto these two equations, Kb can be obtained. They are1.936 · 108 for S-PPL and 2.149 · 108 for R-PPLrespectively.

The experiment of Section 3.4.2 was performed belowthe maximum concentration of AGP, and free PPL and acomplex coexisted in a mixed solution. That was, Cf wasnot 0. It is obvious that Cb and Cf can be easily calculatedin this experiment according to formulas (5) and (6).

3.4.2. Calculation of association constant (KA) and binding

site number (n)

The simple equilibrium relationship, PPL + AGP =PPL Æ AGP, can be used to illustrate such a binding modelas a small ligand (PPL) bound to a macromolecule (AGP).For an equilibrium system with multiple, independencebinding sites (Scatchard model) the relationship betweenbound and free ligand can be represented as [20,21]:

r ¼ n1K1Cf=ð1þ K1CfÞ þ n2K2Cf ð7Þwhere r is the number of moles of bound PPL per mole ofAGP, n1 is the number of primary binding sites, K1 is theprimary Scatchard association constant, Cf is the concen-tration of free PPL, and n2K2 is a measure of nonspecificsecondary binding. For an equilibrium system with onlyone primary binding site and no secondary binding, a Scat-chard plot of r/Cf versus r is a reliable device for measuringn1 and K1 as long as there is significant saturation of thebinding sites [22]:

r=Cf ¼ n1K1 � K1r ð8Þr ¼ Cb=Cpt ð9Þ

where Cpt is the total concentration of AGP.Collect the fluorescence value of fixed concentration of

PPL after interacted with different concentration of AGP.As shown in Fig. 7, a calibration curve was prepared whichrelated the value of r/Cf to the value of r. From the Scata-chard plot, we can get the value of KA and n. The resultswere shown in Table 1. Obviously there is about one bind-ing site (n = 1.17) when S-PPL binds to AGP at 17 �C, andthe association constant of S-PPL with AGP is about 2.3times higher than that of R-PPL. The result has no signif-icant difference from the value reported before by othermethods [11,23].

AGP consists of two kinds of chiral units: one proteindomain and five oligosaccharide chains. It is well knownthat AGP shows chiral recognition ability to many chiraldrugs [6], but the mechanism has not been distinctlyelucidated. The binding of drug to AGP is thought to occurat the single hydrophobic pocket within the proteindomain, and hydrophobicity is considered as a drivingforce of drug Æ AGP binding. According to the report of

Fig. 7. The Scatchard plot of the interaction between PPL with AGP: (a) S-PPL, CS-PPL = 3.33 · 10�6 mol/L, CAGP (·10�6 mol/L) = 0.21, 0.64, 1.07, 1.50,2.14, 2.78, 4.28, 5.35, at 17 �C; (b) R-PPL, CR-PPL = 2.78 · 10�6 mol/L,CAGP (·10�6 mol/L) = 0.21, 0.64, 1.07, 1.50, 2.14, 2.78, 4.28, 5.35, at 17 �Cand (c) S-PPL, CS-PPL = 3.33 · 10�6 mol/L, CAGP (·10�6 mol/L) = 0.21,0.64, 1.07, 1.50, 2.14, 4.28, 5.35, at 37 �C.

Table 1The values of KA and n of PPL with AGP

t (�C) Isomer KA (·105 M�1) n

17 R-PPL 2.62 0.4117 S-PPL 8.57 1.1737 S-PPL 3.02 0.84

CR-PPL = 2.78 · 10�6 mol/L, CS-PPL = 3.33 · 10�6 mol/L.

Table 2Effect of coexisting ions on the values of KA and n

Ions Ca2+ Cu2+ Zn2+

K 0Að�105 M�1) 6.23 1.84 7.64n 0.96 0.36 1.15K 0A=KA 0.73 0.21 0.89

CS-PPL = 3.33 · 10�6 mol/L, the concentrations of Ca2+, Cu2+ and Zn2+

are all 0.02 mol/L. t = 17 �C. K 0A is the value of association constant withaddition of different ions into the solution.

250 F. Zhang et al. / Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251

Yukihirok [5], AGP showed chiral selectivity to PPL, but itwas not because of the five oligosaccharide chains. So wethink that the protein domain may cause this selectivity.As shown in Table 1, the KA value of R-PPL with AGPwas much less than that of S-PPL. It is indicated that thebinding ability of S-PPL with AGP is much stronger thanthat of R-PPL, and transport ability of S-PPL in blood isstronger than that of R-PPL. Namely the capability of S-PPL arriving to target organ is much stronger than thatof R-PPL. Maybe it is an evidence to prove that the cura-tive effect of S-PPL is better than that of R-PPL.

3.5. Determination of interaction type

The types of interaction between small ligand and mac-romolecule include hydrogen bond or van der waals force,electrostatic action, hydrophobic interaction and so on.

Different drug has dissimilar type of interaction withAGP. According to the equation of thermodynamics, wecan get some thermodynamic parameters of the interactionof S-PPL and AGP. The equation of thermodynamics isexpressed as follows [24,25]:

ln K2=K1 ¼ DHð1=T 1 � 1=T 2Þ=R ð10ÞDG ¼ �RT ln K ð11ÞDS ¼ ðDH � DGÞ=T ð12Þ

where K1, K2 is the association constant of S-PPL withAGP at different temperature, G is Gibbs energy, H is en-thalpy, and S is entropy. If DH > 0 and DS > 0, the mainforce is hydrophobic interaction; if DH < 0 and DS > 0, itis mainly electrostatic action, but if DH < 0 and DS < 0,it is mainly hydrogen bond or van der waals force. Accord-ing to the parameter of thermodynamics, it can be simplydecided which type of interaction between S-PPL andAGP exists (because R-PPL can cause side effect, only S-PPL was examined in the following experiment).

According to the association constant of S-PPL withAGP at different temperature in Table 1, DH, DS and DG

can be calculated. They are �38.96 kJ/mol(DH), �20.82J/mol(DS), and �32.92 kJ/mol(DG), respectively. Basedon the above principle (DH < 0, DS < 0), the interactionof S-PPL with AGP is mainly hydrogen bond or van derwaals force, consistent with the report of Yukihirok [5].Hydrogen bond between AGP and the groups of –OH,–NH and –O– of PPL plays an important role in chiral rec-ognition and fluorescence enhancement. Maybe there areother types of interaction between PPL and AGP, buthydrogen bond or van der waals force is dominative.

3.6. Effect of coexisting ions

There are many metal ions existing in vivo. These ionscan influence the transport of PPL and finally affect thecurative effect of PPL [9]. In this experiment, the effect ofthree ions on KA and n of S-PPL with AGP, includingCa2+, Cu2+, Zn2+, which exist extensively in vivo, wasexamined.

As seen from Table 2, all three ions could make thevalue of KA decrease, in which Cu2+ had the most promi-nent action. The reason is maybe that Cu2+ can combinewith many organ macromolecules so as to compete toreduce the value of KA of PPL with AGP and at last tomake the curative effect of PPL worse. So during

F. Zhang et al. / Journal of Photochemistry and Photobiology B: Biology 86 (2007) 246–251 251

administration of PPL, it is suggested to forbid havingsome food or drink containing these ions.

4. Conclusion

The fluorescence method is highly sensitive and conve-nient in the study of intermolecular interaction. Both oftwo enantiomers of PPL can bind to a1-acid glycoproteinto lead to fluorescence enhancement. But the associationconstant of S-PPL with AGP is about 2.3 times higher thanthat of R-PPL, and the binding site number of S-PPL withAGP is also more than that of R-PPL. The type of interac-tion between PPL and AGP is mainly hydrogen bond orvan der waals force. The reaction between PPL and AGPis exothermic reaction, because of DH < 0 [26]. Many ionscan affect the transport of PPL and at last influence itscurative effect.

References

[1] M.X. Xie, X.Y. Xu, Y.D. Wang, Biochim. Biophys. Acta 1724 (2005)215–224.

[2] Y.X. Du, A. Taga, S. Suzuki, W.Y. Liu, S. Honda, J. Chromatogr. A947 (2002) 287–299.

[3] Y.X. Du, A. Taga, S. Suzuki, W.Y. Liu, S. Honda, J. Chromatogr. A962 (2002) 221–231.

[4] Y.X. Du, S. Honda, A. Taga, W.Y. Liu, S. Suzuki, Chin. J. Chem. 20(2002) 1557–1565.

[5] Y. Kuroda, A. Shibukawa, T. Nakagawa, Anal. Bioch. 268 (1999)9–14.

[6] E. Maria, S. Akimasa, Y. Yuki, Anal. Biochem. 274 (1999)27–33.

[7] J.R. Albani, Carbohydrate Res. 338 (2003) 1097–1101.[8] L. Jason, C. Edward, J. Microchem. 55 (1997) 162–168.[9] Y.Y. Jin, Pharmacology, Fifth ed., People’s Medical Publishing

House, Beijing, 2004.[10] T. Gum, M. Valiente, C. Palet, J. Membr. Sci. 256 (2005)

150–157.[11] H. Xuan, D.S. Hage, Anal. Biochem. 346 (2005) 300–310.[12] J.R. Albani, Spectrochim. Acta Part A 54 (1998) 175–183.[13] D. Sliva, C.M. Cortez, S.R.W. Louro, Spectrochim. Acta Part A 60

(2004) 1215–1223.[14] S. Glasson, R. Zini, P. D’ Athis, J.P. Tiliement, J.R. Boissier, Mol.

Pharmacol. 17 (1980) 187–191.[15] A. Taga, Y.X. Du, S. Suzuki, S. Honda, J. Pharm. Biomed. Anal. 30

(2003) 1587–1593.[16] H.R. Xiao, L.Q. Sheng, Q.L. Liu, Spectrosc. Spect. Anal. 24 (2004)

78–81.[17] G.Z. Ma, F. Tan, Y.J. Jiang, Acta Phys. Chim. Sin. 21 (2005)

123–127.[18] R.C. Lu, A.N. Cao, L.H. Lai, J.X. Xiao, Colloid. Surface. A 278

(2006) 67–73.[19] Y. Cao, X.W. He, Spectrochim. Acta Part A 54 (1998) 883–892.[20] L. Soltes, F. Bree, B. Sebille, Biochem. Pharmacol. 34 (1985)

4331–4335.[21] D. Glasson, R. Zini, P. D’ this, J. Tillement, J. Boissier, Mol.

Phramacol. 17 (1980) 187–191.[22] I. Klatz, Science 217 (1982) 1247–1249.[23] K. Hanada, T. Ohta, M. Hirai, J. Pharm. Sci 89 (2002) 751–757.[24] D.P. Ross, S. Subramanian, Biochem. 20 (1981) 3096.[25] Y.H. Pang, L.L. Yang, S.M. Shuang, C. Dong, M. Thompson, J.

Photochem. Photobiol. B 80 (2005) 139–144.[26] W.Y. Zhong, J.S. Yu, Y.Q. Liang, Spectrochim. Acta Part A 59

(2003) 1281–1288.