Kinetic determination of cobalt and nickel by flow-injection spectrophotometry

Anaiytica Chimica Acta, 283 (1993) 476-480 Elsevier Science Publishers B.V., Amsterdam

476

Kinetic determination of cobalt and nickel by flow-injection spectrophotometry M.A.Z. Arruda, E.A.G. Zagatto and N. Maniasso Centro de Energia Nuclear na Agricultura, Universidade de So Pa&, P.O. Box 96, 13400 Rracicaba SP (Brazil) (Received 12th March 1992; revised manuscript received 13th May 1992)

Abstract

Reactor replacement is proposed as means of realizing different sample residence times in flow-injection analysis. As an application, an improved spectrophotometric determination of cobalt and nickel is proposed. The method exploits differences in reaction rates between cobalt and nickel citrate complexes and 4_(2-pyridylaxo)resorcinol (PAR). The influence of pH, reagent concentrations, temperature, ionic strength, flow-rates and masking agent addition were investigated. Selectivity was enhanced by adding EDTA plus pyrophosphate and exploiting kinetic discrimination. The system is very stable and allows the determination of cobalt and nickel up to 5.00 mg 1-r with a sampling frequency of 40 hh’. For tool steel samples characterized by low nickel/cobalt concentration ratios, the system can be simplified, allowing kinetic masking of nickel and single cobalt determination. The reagent consumption is 65.3 pg of PAR per determination, and precise results (R.S.D. < 1%) in agreement with inductively coupled plasma atomic emission spectrometry are obtained at a sampling frequency of 100 h-r. Irleywords: Flow injection; Kinetic methods; UV-Visible

In flow-injection analysis, chemical kinetics have been efficiently exploited by recording several peaks after a single injection [l]. To this end, the formation of two reaction zones in a dispersed sample, stream splitting, reversed or oscillating flows, detector relocation [21, stopped-flow systems, etc., have been utilized [3]. As a consequence, flow-injection procedures exploiting differential kinetics [4-61 have been proposed mainly for simultaneous determinations in binary or ternary mixtures [7,8]. This paper describes an alternative procedure for multi-peak recording based on reactor re-

Correspondence to: E.A.G. Zagatto, Centro de Energia Nuclear na Agricultura, Universidade de Sfio Paulo, P.O. Box 96, 13400 Piracicaba SP (Brazil).

spectrophotometry;

Cobalt; Nickel; Steels

placement, which can be accomplished by commutation. The method deals with differences in kinetics, because to each commutation status corresponds a given mean available time for reaction development. As an application, an improved spectrophotometric determination of cobalt and nickel in steels is proposed. The method is based on complexation with citrate followed by time-dependent dissociation and reaction with 4-(2-pyridylazohesorcinol (PAR) [9]. Stream splitting and thermostated water-baths used in the original work to emphasize kinetic discrimination are not needed. Moreover, the contribution of nickel to the analytical signal can be modified or almost suppressed, allowing system simplification for tool steel analysis where only the determination of cobalt is required.

ooO3-2670/92/$06.00 Q 1992 - Elsevier Science Publishers B.V. All rights reserved

M.A.Z. Anuda et al./Anal.

Chh.

Acta

283 (1992) 476-480

EXPERIMENTAL

Solutions

All solutions were prepared with analytical-reagent grade chemicals and distilled, deionized water. Working standard solutions (0.00-5.00 mg Co 1-l and/or 0.00-1.00 mg Ni 1-l) also contained 0.01 M HCl in order to match the mean sample acidity. To minimize pH gradients, 0.01 M HCl was used as sample carrier stream C (Figs. 1 and 2). Reagents R,, R, and R, were 0.01 M sodium citrate, 0.1 M sodium tetraborate and 0.1 M Na,EDTA plus 0.01 M sodium pyrophosphate, respectively. Reagent R, (0.001 M PAR) was freshly prepared by dissolving 25.92 mg of PAR in 100 ml of water. Samples were solubilized by accurately weighing ca. 0.5 g of finely ground steel filings (50-100 mesh) and placing them in crucibles with 5.0 ml of aqua regia [HNO,-HCl (1 + 3, v/v)]. The mixture was heated to complete dissolution (16O”C, ca. 15 min) and, after cooling, the volume was made up to 100 ml with water. Before injection, the samples were diluted lOOO-fold manually with 0.01 M HCl. Flow-injection system The system for the simultaneous determination of cobalt and nickel (Fig. 1) was constructed using components described elsewhere [lo] and provided two different reaction times depending on the presence of P, or PZ in the analytical path. The sample plug selected by the loop L, was transported by carrier stream C and merged with reagent R, for quantitative metal-citrate complexation inside coil B,. After additions of R, and R,, metal-PAR complexes were formed inside PZ under buffered alkaline conditions. As the metal-citrate decomplexation rate is slower for nickel than cobalt [9], a long P2 coil was needed to increase the mean available time for Ni-PAR formation. R, was thereafter added, and foreign ions were masked by EDTA inside B,. The coloured zone was monitored at 520 nm, with the recorded peak height reflecting the contributions of cobalt and nickel.

2Fig. 1. Flow diagram of the system for the simultaneous determination of cobalt and nickel. S = Sample at 2.0 ml mm-‘; L, and L, = 20- and 100~cm sampling loops (100 and 500 ~1); C = carrier stream (0.01 M HCI, 2.9 ml min-‘1; R,, R,, R, and R, = 0.1 M Na,B,O,, 0.01 M sodium citrate, 0.001 M PAR and 0.1 M Na,EDTA+O.Ol M Na,P,O, at 0.42 ml min-*; B, and B, = 20 and 50-cm reactors; P, and P2 = l- and 4OO-cmpaths, A = detector (520 nm); x, y, z and k = confluence points; W = waste; I = optional stream at 2.0 ml mitt- ’ for studying interferences; black arrows = sites where pumping is applied. Boxed components are linked to movable portions of the commutator with the next position specified by the dashed area.

By switching the commutator, the content of loop L, was intercalated in stream C and the sample plug similarly processed. As Co-citrate decomplexation was quantitative within a few seconds [9] and Ni-citrate decomplexation was relatively slow, P2 was replaced with the injector path (PI), resulting in a very short time for decomplexation and reaction with PAR. Therefore, the recorded signal reflected mainly the cobalt concentration in the sample. Nickel determination was then based on peak heights associated with P, and P,. The simplified system for cobalt determination in tool steels (Fig. 2) was similarly operated. To minimize the formation of Ni-PAR, a twofold increase in flow-rates was effected and P, was

6

R,

R4

w

Fig. 2. Flow diagram of the system proposed for cobalt determination in tool steels. L = 50-1311sampling loop; P, = lo-cm reactor. Other components and symbols as in Fig. 1.

478

chosen as short as possible (10 cm), yet sufficient for providing suitable conditions for Co-PAR formation. Procedure PAR and sodium citrate concentrations were varied within the ranges 0.0001-0.01 and O.OOl1.0 M. PAR solutions with different ages and storage conditions (presence or absence of light) were used. The influence of ionic strength was studied by using an extra stream (2.0 ml min-’ at confluence x) which added sufficient NaCl to provide ionic strengths of l-3 for the solution inside B,. The effects of temperature in the range 0-50°C were studied by immersing coil B, in a thermostated water-bath. The chemistry involved was studied either by varying the length of Pz (10-500 cm) or by placing standard solutions in a situation of “infinite volume” and stopping the peristaltic pump after achievement of a steady state [3]. Whenever required, the pump speed was varied in order to provide flow-rates of 25-200% relative to those in Fig. 1. A merging zones approach [ll] was used to investigate selectivity. Concentrations of potentially interfering ions [lO.OO mg W(V1) I-‘, 5.00 mg Cu(I1) I-‘, 50.00 mg Fe(II1) l-l, 5.00 mg V(V) l-l, 5.00 mg M&I) l-‘, 5.00 mg Mo(VI) 1-l and 5.00 mg Cr(VI) 1-l; solution I, Fig. 11 were well above those usually found in solubilized tool steels. In these experiments, the pH was varied (5.0-11.0) by adding HCl or NaOH to R, and the ionic strength inside B, was always adjusted to 1.0 by adding NaCl to the buffer solution.

RESULTS AND DISCUSSION

For 0.1 mM PAR, decomplexation rates of Co- and Ni-citrate were too low and kinetic discrimination could not be exploited owing to the small analytical signals (ca. 0.017 and 0.009 absorbance for Co and Ni). With R, = 0.001 M PAR, a stable baseline of about 0.06 absorbance was attained. For R, > 0.01 M PAR, high baselines were observed and kinetic discrimination became less pronounced. PAR reagent prepared

M.A.Z.“Amda

et al. /Ad.

Chim. Acta 283 (1992) 476-480

immediately before use yielded results in general 10% higher than those obtained with solutions aged for 5-10 days. In addition to the decrease of the signal, baseline drift (ca. 0.004 absorbance h-l) was observed for old R, reagents. No differences in peak heights were found for PAR solutions stored in the presence or absence of light. Therefore, the use of R, reagents older than 3 days is not recommended. With 0.001 M citrate, kinetic discrimination was not observable because the citrate/PAR concentration ratio was not sufficient to promote pronounced differences in decomplexation rates. In the range 0.001-0.01 M citrate, an increased R, concentration decreased the nickel signal without a significant effect on the cobalt signal. Further, when R, was varied from 0.01 to 1.0 M citrate, the nickel signal decreased by about 12%. These results can be attributed to the mass effect with the consequent lowering of decomplexation rates and sensitivity losses. The proposed procedure is slightly dependent on ionic strength. For 1.00 mg Co l-l, peak heights of 0.302,0.300 and 0.270 absorbance were observed for ionic strengths of 1, 2 and 3 inside B,, respectively. The signal related to this cobalt concentration underwent an 8% increase when the water-bath temperature was varied from 0 to 50°C. A 54% increase was observed for nickel, emphasizing the influence of temperature on kinetic discrimination. A water-bath thermostated at 25’C was used thereafter. An ice-bath, although beneficial [9], was not used because the enhancement of the kinetic discrimination was too small to justify addition of an extra apparatus in a system intended for steel-making analysis. Increasing the pH from 7.0 to 9.0 resulted in sensitivity and selectivity improvement also because of the more effective action of EDTA. Higher pH values were not selected because under more alkaline conditions kinetic discrimination became worse. At pH 11.0 a twofold increase in peak height was observed for nickel but the colour of the PAR solution led to a baseline of 0.9 absorbance. Also, effects of foreign ions became more pronounced, probably owing to precipitation as hydroxides. As a compromise between kinetic discrimination, selectivity and sensi-

479

M.A.Z. Armda et al. /AnaL Chitn. Acta 283 (1992) 476-480

tivity, a pH of 8.9 was chosen (coil B,), which corresponds to pH 9.2 in the solution R,. Metal interferences in the absence of cobalt or nickel were mostly negligible, manifesting themselves only at increased analyte concentrations. EDTA masked most of the tested ions and iron(II1) interference was noted only for higher flow-rates (pump speed > 150% of the nominal speed). Improvements in selectivity were investigated further by adding to R, other masking agents such as triethanolamine, EGTA, tartrate, fluoride, KCN and Na,P,O, (0.01 M). For most of the masking agents, inhibition of kinetic discrimination between cobalt and nickel was observed. With triethanolamine or EGTA, reaction development for cobalt and nickel was almost quantitative and interferences of foreign ions were still unacceptable, probably owing to synergistic effects. With Na,P,O,, iron interference was reduced to about one tenth and the analytical signals underwent only a 5% reduction. This can be explained by the fact that the stability constant of Fe(III)-Na,P,O, is twice that of Fe(III)-EDTA

WI. The combination

of R, and R, was tried in order to eliminate one pumping tube, but the analytical signals decreased to about 70%. This

TABLE 1 Comparative results for cobalt and nickel contents (mg I-‘) in dissolved alloys as determined by the proposed procedure (Fig. 1) and by inductively coupled plasma atomic emission spectrometry (ICP-AES) Sample

Proposed procedure ’ co

Ni

Co

Ni

1 2 3 4 5 6 7

2.73 (3.2) 1.4OCl.l) 3.30 ( < 1) 3.15 (1.4) 3.88 (1.9) 2.00 (2.1) 4.09 (2.3) 4.83 (1.2) 5.24 (0.8) 3.81 (1.0)

- ’ - c 0.39 (4.5) 0.47 (3.0) 0.55 (2.6) 0.29 (5.3) 0.12 (7.0) 0.91 (1.3) 0.87 (0.9) 0.95 (1.2)

2.67 1.38 3.36 3.20 4.00 2.05 3.90 4.81 5.20 3.86

0.08 0.10 0.41 0.50 0.50 0.30 0.15 0.89 0.86 0.99

al d a2 d a3 d

ICP-AES b

a Numbers in parentheses are relative standard deviations (%I for five injections. b R.S.D. of ICP-AES results was about 1.7%. ’ R.S.D. > 10%. d a, = Synthetic samples.

TABLE 2 Comparative results for cobalt contents (mg I-‘) in solubilized tool steels as determined by the proposed procedure (Fig. 2) and by ICP-AES Sample

Proposed procedure a

ICP-AES a

1 2 3 4 5 6 7

1.09 (1.7) 0.60 (1.8) 0.60 (0.6) 0.39 (2.7) 0.32 (1.4) 1.09 (0.3) 0.53 (0.1)

1.13 (1.6) 0.64 (1.8) 0.65 (1.5) 0.38 (2.1) 0.3lC3.3) 1.06 (1.4) 0.58 (0.8)

a Numbers in parentheses are relative standard deviations (%I for five injections.

leads to the conclusion that metal-PAR complexes are formed only inside Pi (or PJ and EDTA addition blocks the further formation of Co- and Ni-PAR. The mean available time for metal-PAR formation, proportional to the lengths of P, and P2, is an important parameter. P, was short enough to impair Ni-PAR formation and to guarantee reproducible results. Regarding P2, for lower values (external loop < 5 cm), additivity was not observed owing to the low nickel contribution to the analytical signal, and the measurement reproducibility was unacceptable (R.S.D. 7%). When Pz was 10 cm, the reproducibility was improved (R.S.D. < 3%) and the additivity was determined to be 98-102%, but the nickel signal was still low. Better reproducibility (R.S.D. 1.1%) and almost quantitative Ni-PAR formation were achieved with Pz = 400 cm. This length could not be reduced to improve the sampling rate, because at 200 cm nickel complexation was only about 70% and the reproducibility was poorer (R.S.D. 2%). In spite of P, and P2 replacement, baseline drift or a double baseline was not observed. As the length of B, was 50 cm and the commutator was switched after the peak maximum, carryover effects due to sample entrapment inside P2 were not detected. With a sample frequency of 40 h-l, only 32.5 of fig PAR were consumed per determination and accurate results were obtained (Table 1).

480

M.A.Z. Ada

TABLE 3 Selectivity for single determination in Fig. 2 Ion

wm CrWIl, Fe(III), Mn(II), Cll(IIJ VW) Ni(I1)

of cobalt using the system Tolerance ratio a 200

MoWI)

20 3 2 1

a Mazimum tolerance ratios of interfering metal concentrations causing a 10% error in measurement related to 0.50 mg Co 1-l.

For single cobalt determination, the flow system was simplified (Fig. 2) and the pumping speed was doubled as sensitivity was not critical and the mean available time for Ni-PAR formation should be reduced. In this context, the length of Pi was decreased to 10 cm and Na,P,O, concentration was increased to 0.1 M. This system does not require thermostating, as only a 3.8% increase in analytical signal per 10°C was calculated. After an 8-h working period, the slope of calibration equation underwent only slight variations ( < 5%). Linearity (r < 0.999; n = 7) was observed in the range 0.0-4.0 mg Co 1-l and the detection limit associated with a 250-~1 sampling loop was calculated according to IUPAC recommendations [13] to be 180 pg Co 1-i. Precise results (R.S.D. < 2%) were obtained at a sampling rate of about 100 h-‘. PAR consumption was calculated to be 65.3 pg per determination. The accuracy and selectivity can be assessed from Tables 2 and 3.

et al. /Anal.

Chim. Acta 283 (1992) 476-480

Partial support of this work by FAPESP (FundaqHo de Amparo a Pesquisa do Estado de Sb Paula) and CNPq (Conselho National de Desenvolvimento Cientifico e TecnoMgico) is greatly appreciated. B.F. Reis, J.A. Nobrega and O.M. Matsumoto are thanked for critical comments.

REFERENCES 1 M.D. Luque de Castro and M. Valcarcel, Trends Anal. Chem., 5 (1986) 71. 2 E.A.G. Zagatto, H. Bergamin F”, S.M.B. Brienza, M.A.Z. Arruda, A.R.A. Nogueira and J.L.F.C. Lima, Anal. Chim. Acta, 261 (1992) 59. 3 M. Valcarcel and M.D. Luque de Castro, Flow-Injection Analysis. Principles and Applications, Horwood, Chichester, 1st edn., 1987. 4 H.A. Mottola and H.B. Mark, Jr., Anal. Chem., 56 (19841 96R. 5 H.A. Mottola, M.D. Perez-Bendito and H.B. Mark, Jr., Anal. Chem., 60 (19881181R. 6 M.D. Luque de Castro and M. Valcarcel, Analyst, 109 (19841413. 7 A. Rios, M.D. Luque de Castro and M. Valcarcel, Anal. Chim. Acta, 179 (19861463. 8 A. Femandez, M.D. Luque de Castro and M. Valcarcel, Anal. Chem., 56 (1984) 1146. 9 D. Betteridge and B. Fields, Fresenius’ Z. Anal. Chem., 314 (1983) 386. 10 M.A.Z. Arruda and E.A.G. Zagatto, Anal. Chim. Acta, 199 (1987) 137. 11 M.A.Z. Arruda, E.A.G. Zagatto, A.O. Jacintho and S.M.B. Brienza, J. Braz. Chem. Sot., 2 (1991) 47. 12 A. Ringbom, Complezation in Analytical Chemistry (Chemical Analysis, Vol. XVI), Interscience, New York, London, 1963. 13 Analytical Methods Committee, Analyst, 112 (1987) 199.