Reversed-phase HPLC separation of complex mixtures of trace metals as dibenzyldithiocarbamate chelates

T&ma, Vol. 35, No. 9, pp. 685-691, 1988 in Great Britain. All rights reserved

0039-9140/88

$3.00 + 0.00

Copyright 0 1988Pergamon Press plc

Printed

REVERSED-PHASE HPLC SEPARATION OF COMPLEX MIXTURES OF TRACE METALS AS DIBENZYLDITHIOCARBAMATE CHELATES C. BAIOCCHI*, A. MARCHETTO, G. SAINI and P. BERTOLO Dipartimento di Chimica Analitica, Universita’ di Torino, Via P. Giuria 5, 10125 Torino, Italy

Laboratorio

G. PETTITI di Anestesiologia e Rianimazione, Universita’ di Torino, C.so Bramante 90, 10100 Torino, Italy

(Received 14 June 1987. Revised 29 April 1988. Accepted 12 May 1988)

Summary-The dibenzyldithiocarbamate chelates of Cd(H), Pb(II), Bi(III), Hg(II), Ni(II), Cu(II), As(III), Fe(III), Co(II1) and In(II1) are separated by reversed-phase HPLC in isocratic conditions. The procedure is simple, rapid, and gives satisfactory separations with high efficiency and sensitivity at mobile phase compositions very rich in organic modifier (85-88% CH,CN). The detection limits range from 1.4 to 14 pg/l. The elution order is correlated with the ability of the central metal atom to affect the electronic distribution of the ligand, which has readily polarizable donor atoms. Infrared spectroscopy data corroborate this assumption.

The presence of the two benzyl groups ensures high absorptivity and their relative bulkiness contributes to the stability of the complexes formed. An excess of ligand was added to the mobile phase in order to minimize decomposition during the separation.

An especially promising approach for quantitative multi-element determinations of trace metals involves pre-column formation of chelates followed by reversed-phase liquid chromatographic separation. The dithiocarbamates form thermodynamically stable complexes readily extractable from aqueous systems into organic solvents.’ With such character-

istics they have long been used for analysis by extraction followed by spectrophotometric or atomicabsorption spectrometric determination. For the same reasons these complexes are particularly suitable for HPLC separation, allowing the determination of trace levels of metal ions even in complex matrices such as sea-water and other marine samples. Separations of various dialkyldithiocarbamate chelates on hydrocarbon stationary phases have been reported.2m’4 Chelates of pyrrolidinedithiocarbamate,” N-methyl-2-naphthylmethyldithiocarbamate,’6 bis(2-hydroxyethyl)dithiocarbamate,’7~’8 N-methylN-2-sulphoethyldithiocarbamate’9 and tetramethylenedithiocarbamate*O have also been studied. The detection technique was mainly ultraviolet spectrophotometry, but some attention was also paid to electrochemical detection, because many metal dithiocarbamate complexes are electroactive at various working electrodes. A limitation of the method arises from the possible decomposition of the complexes in the column, particularly when very low concentrations are involved, as in trace analysis. The ligand used in the present work is sodium dibenzyldithiocarbamate (DBzDTC). *To whom all correspondence

EXPERIMENTAL Instrumentation

The liquid chromatograph was a Varian LC 5060, with a UV 100 spectrophotometric detector and a VISTA 401 Data System. The detector was operated at 254 nm. The analytical columns used were a Varian MCH-10 (lo-pm particles, 300 x 4.6mm, carbon load 8%, no endcapping) and an Alltech Adsorbosphere (lo-pm particles, 250 x 4.6mm, carbon load 19%. fully end-capped) indicated hereafter as columns A and B, respectively. The ultraviolet spectroscopic properties of the samples were monitored in the range 19@6OOnm. The chromatographic analysis had to be performed as soon as possible after preparation of the sample. All glassware was conditioned with l.OM nitric acid and rinsed several times with water from the Milli-Q system to minimize contamination. The infrared spectra were obtained with solid samples of the metal complexes after crystallization from chloroform. The spectral range explored was 4000400cmat a scan speed of 10 cm-‘/see, with a resolution of 4 cm-’ at an absorbance of 0.50. Reagents

HPLC-grade acetonitrile (Merck) and water purified with the Millipore Milli-Q system were used as the mobile phase constituents. Analytical grade sodium dibenzyldithiocarbamate was obtained from Fluka. Individual O.OlM solutions of each metal salt were made by dilution of stock lOOO-mg/l.standard solutions (C. Erba “For atomic absorption”, Merck “Titrisol”). The ligand stock solution (1 .O x 10m3M)was prepared by dissolving the

should be addressed. 685

C. BAIOCCHI et al.

686 1.:

l.!

Ho Ni Pb CO AS CU

Abs

Abs

C

cm

250

300

X (nm)

350 X

400

450

(nm)

Fig. 1. Spectra of DBzDTC and its metal chelates appropriate amount of reagent in 100.0 ml of chromatography grade acetonitrile that had been deoxygenated by passage of gentle stream of nitrogen through it. DBzDTC chelates in concentrations varying in the range 1.O x 10-5-1.0 x IO-‘M were formed by adding appropriate amounts of metal ion stock solutions to 50.0 ml of a 1: 1 v/v water-acetonitrile mixture that was 4.0 x 10m5M in DBzDTC.

In Fig. 2, chromatograms corresponding to injections of (a) freshly prepared and (b) 24 hr old solutions of DBzDTC and Pb(DBzDTC), are shown. The

(a)

RESULTS AND DISCUSSlON

The ultraviolet spectroscopic properties of DBzDTC and its complexes are shown in Fig. 1. The

spectra differ from metal to metal. In the absence of a common wavelength of maximum absorbance for all the chelates, it was decided to use 254 nm as the detection wavelength, to make the method easily applicable to chromatographic systems equipped with a fixed wavelength detector. The chromatographic studies were performed with either 1: 1 v/v acetonitrile-water solutions of the DBzDTC complexes or acidic solutions (pH 2.5) of salts of the following species: Cd(H), Pb(II), Bi(III), Zn(II), Hg(II), Ni(II), Cu(II), As(III), Fe(III), Co(I1). The Co(I1) was rapidly aerially oxidized to Co(II1) species stabilized by complexation with the dithiocarbamate. Few of the complexes were inert enough to give adequate resolution of peaks in a simple water-acetonitrile mixture. The addition of a tertiary amine (triethylamine, TEA) to give a 0.094 concentration in the aqueous component in order to shield the free silanol groups of the stationary phase,” and of an excess of ligand (2.0 x 10m4M) to the organic component of the mobile phase, resulted in very satisfactory chromatographic results. The pH of the final solution containing TEA was adjusted to 4.5 with 0.05M acetate buffer. The mobile phase composition providing the best resolution was different for the two columns used and the shapes and resolution of the peaks showed significant differences, depending on the stationary phase.

Sample

:

DBz DTC

Sample : DBzDTC+

Pb

(b)

LFig. 2. Chromatograms of the ligand and its Pb(lI) complex: (a) freshly prepared, (b) after 24 hr.

HPLC separation of trace metals

retention time of the chelate is very long and a new peak with a retention time of 5.0 min appears; by analogy with similar cases,15 this peak may be attributed to the formation of a dimer of the free ligand by oxidation according to the scheme

RN-,//”

4

2

+

Ox +

2H+

types of column are reported. As expected, the efficiency and selectivity were different in the two cases. Columns A and B differed in carbon load (8 and 19% respectively) and column B was end-capped with trimethylsilane. Both factors affect the number

4 RN-CHS 1

__f

's-

2

RN-/ 1

-

R,N-8

and detection

"\-NR \S-Ss/

It is difficult to identify what change in the structure or composition of the complex gives rise to the change in retention time. The chelates of all the other metal ions examined showed similar changes in the peaks after aging of their solutions. Moreover, chromatographic runs of samples that had undergone extraction, evaporation to dryness and reconstitution in an acetonitrile-water mixture gave the same chromatograms as the aged solutions. The solvent extraction obviously facilitates oxidation by dissolved and/or aerial oxygen. The instability of the solutions necessitated frequent preparation of new standards. To avoid this, in situ formation of complexes by direct injection of slightly acidic solutions of the metal ions was tried. The efficiency and selectivity of the separation of DBzDTC complexes formed in situ was exactly the same as for those formed prior to injection. A possible difficulty with in situ complex formation is that during the residence time in the column the less reactive ions in an injected mixture might not be able to react completely with the ligand in the mobile phase. The example in Fig. 3 shows that this was the case for the Sn(IV) and Cd(H) complexes. In the same figure it can be seen that the detection sensitivity obtained in the case of the pre-column chelation was better, mainly for the peaks in the middle of the chromatogram (chromatogram A was recorded at a sensitivity eight times greater than that for chromatogram B). This effect may also be due to incomplete reaction. The solutions for the pre-column system could be made more stable by careful degassing (for about 15 min), but the resulting increase in sensitivity did not seem large enough to justify the work and careful handling involved. The simpler option of in situ formation of the complexes was adopted as our standard procedure and Fig. 4 shows typical chromatograms obtained with the two columns used. In Table 1 the chrodata

+ 20H'S

'S'

matographic

687

limits for the two

=

of free silanol groups, which may complicate the retention mechanism and influence the separation efficiency and selectivity. In the absence of tertiary amine in the mobile phase, column A (which was richer in free silanol groups) was not able to provide peaks of acceptable shape for most of the complexes. In contrast, for column B the presence of a buffer to control the pH was enough to give very good peaks for all the complexes examined. However, aqueous solutions of TEA were used in both cases for uniformity. Calibration plots for all the metal ions examined exhibit satisfactory linearity in the concentration range examined. Table 2 lists the numerical results.

(A) Fig. 3. Effect of reactivity competition on the in situ chelation of a mixture of metal ions. Chromatogram A: in situ chelation (detector sensitivity 0.0156 absorbance for full scale deflection). B: pre-column chelation (detector sensitivity 0.125). (1) Cd(H), (2) Sn(IV), (3) Pb(II), (4) Zn(II), (5) I-WI), (6) WIV), (7) WII), (8) Cu(II), (9) As(III), (10) Co(III), (I 1) Fe (III) and (12) In(U) in B only. Flow-rate 1.5 ml/min. All metal concentrations 1.O x 10-SM.

688

C. BAICKCHI et al.

Fig. 4. Comparison of separations performed on different columns. A: column A, CH,CN (+2.0 x 10m4M DBzDTC)/H,O (+O.OlMTEA) (85: 15); flow-rate 1.2 ml/min, detector sensitivity 0.0156. B: column B, CH,CN (+2.0 x 10m4M DBzDTC)/H,O (+O.OlM TEA) (88: 12); flow-rate 1.5 ml/min, detector sensitivity 0.031. The metals common to both experiments were (2) Pb(II), (3) Zn(II), (4) Hg(II), (5) Se(IV), (6) Ni(II), (7) Cu(II), (11) In(II1). Fe(U) (9) and Co(III) (10) were used only in B, and Bi(III) (1) and As(III) (8) only in A. All concentrations were 5.0 x 10m6M. Table 1. Chromatographic

retention and efficiency parameters and detection limits of the metal complexes

Column

Parameter

Bi

Pb

Se

Zn

Hg

Ni

Cu

As

In

1.23

A+

k’ Plate height, mm DL pgcgll.

1.59 0.11 1.8

2.05 0.12 1.4

2.65 0.32 1.8

3.47 0.16 5.8

4.10 0.16 2.8

5.51 0.10 3.1

6.58 0.15 2.2

7.49 0.18 5.4

1.38 0.051 1.4

5.08 0.040 1.4

2.93 0.048 1.5

3.33 0.049 4.4

4.43 0.039 2.0

7.02 0.039 3.1

Bt *Column iColumn

14.0

k’ Plate height, mm DL pg/l. A: CH,CN B, CH,CN

85%; flow-rate 88%; flow-rate

Fe

10.71 9.28 0.036 0.042 3.4 -

Co

9.28 0.042 -

1.2 ml/min. 2.0 ml/min.

All the chromatographic runs were done in triplicate and the relative standard deviations of the peak heights were at worst [Se(W) and As(II1) chelates] about 3%. Table 1 also lists the detection limits, which were evaluated as the concentrations corresponding to a signal to noise ratio of two. As can be seen, they were satisfactorily low, taking into account that they were obtained under the separation conditions for the mixture. Since real samples containing all the metals examined are unlikely to be common, it is possible to improve the detection sensitivity of single ions by choosing more suitable operating conditions (e.g., different organic modifier concentration, lower flow velocity, more suitable wavelength). In this connection, the chromatographic behaviour of the Cd(DBzDTC)I chelate did not allow its determination in conditions compatible with those for the other chelates. However, its determination was possible with a suitable mobile phase (70:30 v/v CH,CN/NH,OAc buffer). It is worth noting that very good selectivity was

achieved at high organic modifier concentrations in the mobile phase, though such compositions normally give poor selectivity.

Table 2. Calibration plot parameters Column

Complex

Slope, mm. 1.mg-’

Intercept, mm

Correlation coefficient

A

Bi(III) Pb(II) Se(W) Zn(II) H8(II) Ni(II) Cu(II) As(W) In(III) Pb(II) Zn(II) Hg(II) Ni(I1) Se(IV) Cu(II) In(M)

3.3 18.0 2.6 10.6 10.7 4.8 4.5 3.0 6.3 27.7 12.6 15.4 3.3 7.7 4.9 11.9

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2

0.989 0.971 0.981 0.984 0.988 0.987 0.985 0.979 0.983 0.978 0.986 0.965 0.966 0.981 0.975 0.985

B

HPLC separation of trace metals

689

I

-

7

d

1

I= s

690

C. BAIOC-CHIet al.

A rationalization of the results may be attempted in terms of the electron-density distribution in the chelating ring, which is the part of a metal complex molecule that is mainly responsible for differences in the chromatographic properties of this type of compound.22 The central metal ion can affect the protonacceptor ability of the donor atoms of the ligands (i.e., their ability to give dipolar interaction with the solvent molecules) as a consequence of the more or less ionic character of the metal-ligand bond. In addition, it can alter the electronic distribution of the ligand molecules (which are generally highly conjugated) and so influence the lipophilicity of the nonpolar part of the compound. This influence is rather small for nitrogen and oxygen as donor atoms, since these are strongly electronegative and not easily polarized, so in such cases only small differences in chromatographic behaviour are caused by the different metal ions. To separate the metal complexes it is then necessary to employ a mobile phase with a low organic content, which often results in very long retention times and poor peak shapes. In contrast, the donor sulphur atoms of DBzDTC are only weakly electronegative and the central metal ions can markedly affect the electronic distribution and character of the metal-ligand bond and hence govern the chromatographic selectivity. The degree of ionic character of the metal-ligand bonds depends on the effective positive charge localized on the metal ion. However, the values obtained for this charge by means of parameters such as the covalency index values Xfr (where Xf, and r are the electronegativity and radius of the metal ion) do not correlate with the elution order of the chelates examined. Information obtained from the infrared spectra is more useful. It is well known that in dithiocarbamates the C-N bond is partly double-bond in character owing to the shift of a nitrogen electron-pair toward the chelate ring. Such a shift affects the ability of the nitrogen atom to interact with polar solvent molecules and obviously depends on the nature of the central metal ion. The position of the absorption band assigned to the stretching vibrations of the C-N bond in the infrared spectra of dibenzyldithiocarbamate complexes, such as those shown in Fig. 5, confirms an intensification of double-bond character. The larger the vcN value, the smaller the ability of the nitrogen atom to interact with mobile phase molecules, and the larger the retention times. Table 3 shows that the proposed correlation is strictly observed for tervalent metal ions, but the elution order of bivalent ions is not as expected in the case of Ni and Zn. The Ni and Zn chelates differ from the others in being coordination unsaturated. Two co-ordination positions are ligand-free and can be occupied by mobile phase molecules that are then in the inner co-ordination sphere of the metal ion. The resulting stronger interactions with the solvent cause shorter retention times than expected. The Co(II1) and Fe(II1) complexes,

Table 3. Frequencies of the stretching vibrations of the C-N bond in the infrared spectra, and the retention volumes of the DBzDTC chelates v,, ml Chelate

vcN, cm-’

Bi(III) Fe(II1) Co(II1) In(II1) Pb(I1) Hg(II) Zn(I1) Cu(I1) Ni(I1)

Column A

Column B

7.6 23.8 23.8 27.4 5.6 10.0 9.2 19.2 14.4

5.2 17.7 17.7 20.0 5.4 10.0 9.0 15.6 12.0

1476 1480 1480 1485 1465 1473 1479 1482 1486

which have identical vibration frequencies for the C-N bond, have identical retention times at all mobile phase compositions examined. Figure 5 shows examples of some of the spectra. The frequency range typical of C-N bond vibrations in the dithiocarbamates is known23 to be 1490-1470cm-‘. In the present case the peak at 1493 cm-‘, occurring in all the spectra, is typical of the benzyl group, and the adjacent peak is typical of the C-N bond vibration. The pair of peaks in the range 1455-1410cm-’ are outside the range usually attributed to vcN vibration, and in any case are not affected by the presence of different metal ions. A similar

situation

was found

in the case of metal

complexes with diethyldithiocarbamate.22 CONCLUSIONS

The structural factors examined above show the fundamental role played by the central metal ions in characterizing the chromatographic behaviour of DBzDTC complexes. The resulting high selectivity of separation (together with the favourable spectroscopic properties of the chelates) makes this method suitable for trace-metal determination. Furthermore, in situ formation of the complexes is simple and rapid, allowing a large number analysed in a reasonable time.

of samples

to be

REFERENCES

1. B. J. Mueller and R. J. Lovett, Anal. Chem., 1987, 59, 1405. 2. A. M. Bond and G. G. Wallace, ibid., 1981, 53, 1209. 3. S. R. Hutchins, P. R. Haddad and S. Dilli, J. Chromatog., 1982, 252, 185. 4. G. Schwedt, Chromarographia, 1979, 12, 613. 5. E. B. Edward-Inatimi, J. Chromatog., 1983, 256, 253. 6. A. R. Timerbaev, 0. M. Petrukhin and Yu. A. Zolotov, Zh. Analit. Khim., 1986, 41, 2135. 7. H. Irth, G. J. de Jong, U. A. Th. Brinkman and R. W. Frei, Anal. Chem., 1987, 59, 98. 8. G. Drasch, Z. Anal. Chem., 1986, 325, 285.

9. T. Tetsumi, M. Sumi, M. Tanaka and T. Shono, J. Chromatog.,

1985, 348, 389.

10. B. J. Mueller and R. J. Lovett, Anal. Left., 1985, 2399. 11. Idem, Anal. Chem., 1985, 57, 2693.

18,

HPLC separation of trace metals 12. R. M. Smith, A. M. Butt and A. Thakur, Analyst, 1985, 110, 35. 13. S. Inoue, S. Hoshi and M. Mathubara,

691

18. A. Munder and K. Ballsehmiter, Z. Anal. Chem., 1986, 323, 869.

Tulonra, 1985,

32, 44. 14. M. Kaniwa, J. Chromafog., 1987, 405, 263. 15. A. M. Bond and G. G. Wallace, Anal. Chem., 1982.54, 1706. 16. Young-Tzung Shih and P. W. Carr, Anal. Chim. Acta, 1982, 142, 55. 17. J. King and J. S. Fritz, Anal. Chem., 1987, 59, 703.

19. G. Sehwedt and P. Schneider, ibid., 1986, 325, 116. 20. S. Leharne, J. Chem. Educ., 1986, 63, 727. 21. J. S. Kiel, S. L. Morgan and R. K. Abramson, J. Chromalog., 1985, 324&313. 22. A. R. Timerbaev and 0. M. Petrukhin, Anal. Chim. Acta, 1984, 159, 229. 23. L. J. Bellamy, The Infrared Spectra of Complex Molecules, p. 401. Methuen, London, 1975