Continuous and stopped flow injection for catalytic determination of total iodine in urine

Abstract

This paper describes the use of flow injection (FI) techniques for the determination of iodine in urine, based on the catalytic effect of iodide in the redox reaction between Ce(IV) and As(III). The proposed procedures minimize errors in the conventional batch method arising from the reading of absorbance at a fixed time after addition of Ce(IV) reagent. Two FI systems, for the continuous and stopped modes of operation were assembled. In the continuous-FI system, a thermostated bath was used to increase the sensitivity. However this is not necessary for the stopped-FI system. The two systems are comparable in terms of sensitivity, sample throughput and detection limit. The continuous-FI and the stopped-FI exhibited detection limits (3σ) of 2.3 and 3 μg I l⁻¹ respectively. Both systems have equal sample throughputs of 35 samples h⁻¹. Calibration plots for both techniques are linear. The FI procedures provide very short analysis times compared to the batch procedure. Using the linear regression test, there is no significant difference between the results from the four methods, i.e., continuous-FI, stopped-FI, conventional method and ICP-MS. The proposed methods are readily applicable for automation and can be an alternative to the conventional procedure for the survey of the iodine deficiency disorder. A condition for sample digestion is also proposed to reduce the amount of chloric acid required for complete digestion. Kinetic information of the reaction can also be obtained from the stopped flow mode.

---

Introduction

Iodine contents in urine have been widely used as a marker for status assessment of iodine deficiency disorder (IDD). In 1992, the ICCIDD/WHO/UNICEF,¹ gave the following values as a guide for IDD status: <20 μg I l⁻¹ (severe); 20–49 μg I l⁻¹ (moderate); 50–100 μg I l⁻¹ (mild) and >100 μg I l⁻¹ (normal). There have been a number of documents that describe methods for determination of inorganic iodine content and total content in urine.

For the quantitation of free iodide, different modes of high performance liquid chromatography (HPLC) such as ion² and ion-pair reversed-phase^3–5 were proposed with either electrochemical detection^3–5 or post-column reaction.² Yabu et al. have proposed a method to measure iodide content in urine using an iodide-selective electrode.⁶

Total content of iodine can be determined by using a hyphenated technique such as inductively coupled plasma mass spectrometry (ICP-MS). Usually dilution of sample with water⁷ or other types of reagent such as nitric acid^7,8 and ammonia⁹ is adequate for this technique. Variations in the signals of the detection mass, are corrected by addition of some internal standards, e.g., europium,⁸ rhodium or indium.⁷ An ICP-MS method for determination of urinary iodine using isotope dilution with iodine-129 was presented by Haldimann et al.⁹

Surrounded by other methods, the most common method used in laboratories dealing with monitoring of IDD status is a spectrometric method with catalytic effect of iodide on a redox reaction between Ce(IV) and As(III).^1,10 The method is more preferable than other methods because it is more practical as a survey tool. Sandell and Kolthoff first described the use of this reaction for determination of trace iodide catalyst in 1934.¹¹ Rodriguez and Pardue proposed some possible mechanisms based on their experimental data achieved in 1969.¹²

Mushtakova et al.¹³ proposed a Sandell and Kolthoff method for determination of free iodide in urine. Their method requires no urine digestion step prior to spectrometric detection of Ce(IV). However extensive investigation of interfering effects from potential species in urine is yet required for this method.

Description of the classical method was provided by Dunn et al.^1,10 The procedure is usually carried out batchwise. Interferences to the redox reaction in urine must be eliminated prior to the step of spectrometric measurement. Digestion with chloric acid is more often used than the alkaline ashing method.¹⁴ For chloric acid digestion, iodine species are oxidised to iodate ions after digestion. Unpublished results¹⁵ apparently indicated that chloride ion added in the arsenious solution helps in conversion of iodate to iodide within less than a minute. The results have also demonstrated that the disappearance of Ce(IV) is a first-order process under the conditions of the conventional method.

After chloric acid digestion, the sample is added with As(III) solution which contains sodium chloride. Finally after an accurate volume of Ce(IV) solution is added to this solution the reading of absorbance (405–420 nm) must be taken at a fixed time for all the samples and the standards. Hence technician error can easily occur in the conventional batch method. Automation of the procedure has been reported through the use of an autoanalyser, which is based on the air-segmented flow (ASF) technique.^4,16–19 The use of flow injection (FI) techniques offers various advantages.²⁰ Nevertheless, there has been no report of a FI system to use in conjunction with this type of sample preparation.

In 1996, Yaping et al.²¹ proposed a FI method for the determination of iodine in urine. The method employed the Sandell and Kolthoff reaction but with the use of a different type of reagent (H₂SO₄ + KMnO₄ + K₂Cr₂O₇) for on-line digestion. With this mixture of oxidant, and with the detection wavelength at 480 nm, spectral interference is at risk especially from KMnO₄. Another disadvantage of the method is that the method unnecessarily used a toxic brucine as a reagent stream.

In this article, two methods developed for the determination of urinary iodine based on flow injection techniques are described. The first method is proposed for use in the normal continuous mode. The system can be modified to operate under a stopped-FI mode.²⁰ Both methods are readily applicable for automation to reduce technician error, which often arises in the convention batch method.

Experimental

Manifolds for continuous and stopped flow injection

The configuration of the FI system used in the continuous mode was modified from the system reported in previous work²² and is shown in Fig. 1. Concentrations of Ce(IV) and As(III) were optimized from the conventional method used by Dunn et al.^1,10 An Ismatec peristaltic pump, model IS7610 was used for propelling reagents and sample plug. A Rheodyne injection valve, model 7l25, fitted with Teflon loop (1.0 mm id) was employed for injections of standards and samples. A glass reaction coil, 100 cm, was made of glass tubing with id 1.0 mm by curling this glass tube into a spiral shape around an opened glass cylinder (id 1.5 cm and 17.5 cm in length). Two pieces of the coil were connected together to give a 200 cm coil length. Cole Parmer Tygon tubings with id of 0.75 mm, 2.25 mm od and 0.75 mm wall, were used for the FI assembly. A PolyScience water bath, model 2L-M, was used for controlling the temperature of the glass coil immersed in the bath. A Shimadzu spectrophotometer, model UV-2-01, Japan, with a tungsten lamp and a Philips flow cell of 0.01 ml volume was used for monitoring the absorbance of Ce(IV) at 420 nm. An Alltech chart recorder, model LR 93025, USA, was used for recording the signal from the spectrophotometer. The recorder was set at 50 mV f.s.d. (full scale deflection).

Fig. 1 The manifold of the continuous-FI system used for determination of iodine in urine.

The system of the continuous-FI shown in Fig. 1 was modified to operate under the stopped-FI mode. In the stopped mode, the glass reaction coil was replaced by a coil made of Tygon tube (100 cm). The water bath was not used in this system. The pumping rate for each channel was 1 ml min⁻¹.

Reagents

All chemicals used in this work were AR grade, supplied by Merck (Germany) except indium nitrate. Deionized distilled water was obtained from a Milli-Q system, Millipore, USA, and was used throughout.

Iodine standards. A stock of iodine standard (1000 mg I l⁻¹) was prepared by dissolving 0.1685 g of potassium iodate in water and making up to the mark of a volumetric flask (100 ml). Further dilutions were made for working solutions of appropriate concentrations.

Chloric acid 28% m/v. Potassium chlorate (250 g) was dissolved in 455 ml of deionized distilled water in an Erlenmeyer flask. The mixture was heated until a clear solution was obtained; 187.5 ml of perchloric acid 70% v/v was slowly added into this solution with constant stirring. This solution mixture was stored in a freezer overnight and was filterred through a filter paper (Whatman no. 1 or equivalent), into a Buchner funnel.^1,10 The final volume of filtrate was approximately 850 ml. This solution contained approximately 28% m/v of chloric acid and was used for digesting all the urine samples and iodine working solutions. The solution of chloric acid was stored at 4 °C in a refrigerator.

Arsenious acid 0.05 and 0.1 M. As₂O₃ (10 g) and NaCl (47 g) were dissolved in 500 ml of water with heating on a hot plate. After cooling to room temperature, 27.8 ml of concentrated H₂SO₄ was added to this solution followed by dilution with water to one litre. This 0.1 M arsenious acid solution was used in the FI methods. Arsenious acid 0.05 M was prepared in a similar way except that the weights of As₂O₃ and NaCl were respectively reduced to 5 and 25 g. This more diluted solution of arsenious acid was used throughout the batch method.

Ceric ammonium sulfate 0.008 and 0.038 M. Ceric ammonium sulfate 0.008 M was used as a reagent stream in the FI methods. This solution was prepared by dissolving 5 g of Ce(NH₄)₄(SO₄)₄·2H₂ O in 1 l of 1.75 M H₂SO₄. Ceric ammonium sulfate 0.038 M was prepared in a similar way using 24 g of Ce(NH₄)₄(SO₄)₄·2H₂ O dissolved in 3.5 M H₂SO₄. This latter solution was used for the determination of urinary iodine in the batch method.

Indium nitrate 1 μg In l⁻¹. Indium nitrate (Fluka, Switzerland) 0.655 g was dissolved and diluted to 250.00 ml in 2% v/v nitric acid. This solution was used as the internal standard for the determination of iodine in urine by inductively coupled plasma mass spectrometry (ICP-MS).

Sample. Casual urine samples were collected from students and were selected to cover the range of iodine concentration studied. The samples were frozen and thawed before use.

Determination of urinary iodine by catalytic methods

Digestion of urine samples. The general method for acid digestion of urine,^1,10 was used with slight modification. The volume ratio of chloric acid solution to urine was reduced from conventionally 3 + 1 to be 1.2 + 1.

Urine (4.00 ml) or working iodine standard was pipetted into a test tube, which contained 4.8 ml of 28% m/v chloric acid. After a 1 h digestion in a heating block at 105–110 °C, the solution was transferred into a volumetric flask and was made up to 10.00 ml with water. The reagent blank was prepared in a similar way using deionized distilled water. There is also enough volume of liquid for replicate injections in the FI methods.

To avoid any course of error, calibration of samples was always made against the set of iodine standards digested in the same run. The working range of standards was the same for the batch and the flow methods, which ranged from 20 to 200 μg I ml⁻¹.

Batch method. Aliquots of 1.0 ml of the digested solutions were transferred into separate test tubes. Determination of iodine was carried out using the method described by Dunn et al.^1,10

Continuous-FI. Operation of the system in Fig. 1 for optimization and analysis was carried out in the usual manner by injecting the liquid sample into a continuous flowing stream of water carrier. The sample zone, after being merged and mixed with streams of arsenious acid and ceric ammonium sulfate, was transported to the detector which recorded profiles of the signal. Besides the catalytic effect of iodine, the reaction zone was also thermally catalysed in the glass coil immersed in the 43 °C bath. Signal profiles of the continuous mode are depicted in Fig. 2a. The drop in absorbance reading was due to the color fading of Ce(IV). For this set up, the maximum drop was achieved at 100 s after injection.

Fig. 2 Profiles of iodine standards obtained from the (a) continuous-FI system and (b) stopped-FI system.

Stopped-FI. For this mode, it is convenient to operate the system at room temperature (26 °C), thus, the bath (Fig. 1) was not used to raise the temperature. Injection of a sample was carried out in a similar way to the continuous mode. At the time of 41 s, when the absorption minimum appeared in the usual continuous mode, the flow was stopped for a period (1 min). The signal was continually recorded during the stopped-flow period. The flow was restarted again at time t′ (101 s after injection) to push away the sample zone resulting in the rise of absorbance back to the baseline. Some of the stopped-FI profiles are depicted in Fig. 2b. In the calibration, the stopped interval must be fixed for all working solutions and the samples.

Interferent studies

Interferences of diverse species which potentially exist in urine (Cl⁻, SO₄²⁻, HCO₃⁻, Ca²⁺, K⁺, Mg²⁺, SCN⁻, Fe³⁺, F⁻, Mn²⁺, (NH₄)₂HPO₄, glucose, urea, oxalic acid, uric acid and ascorbic acid) were examined.

The studies were made using the continuous-FI system shown in Fig. 1. The sixteen chemical species were separately added into 1 ml aliquots of 1000 μg I l⁻¹ potassium iodate solution in a volumetric flask. These solution mixtures were made up to the mark with water to 10.00 ml. After the digestion, the signal obtained from these solutions were compared to the signal achieved from a pure standard of potassium iodate (100 μg I l⁻¹). A species was considered to interfere if its presence resulted in a signal alteration of greater than ±10%.

Determination of urinary iodine by ICP-MS

An Elan 6000 inductively coupled plasma mass spectrometer, Canada, with the recommended operation conditions, was used in the determination of iodine as a comparative method. Sample introduction to the nebulizer was carried out using a Gilson peristaltic pump.

In the analysis, 10-fold dilution with 0.16 M nitric acid was made on the samples. In, added as indium nitrate, was used as the internal standard to correct for non-spectral interference and for signal instability.

Results and discussion

Modification of the digestion method

For this work, the volume ratio of chloric acid solution to urine was reduced from conventionally 3 + 1^1,10 to be 1.2 + 1. The results preliminarily studied on five samples using the manifold in Fig. 1 have shown that there was no significant variation in the signals over the ratios of 4 + 1, 3 + 1, 2 + 1 and 1 + 1 (all made up with water to 5.00 ml). This indicated complete digestion for all conditions tested.

With the current volume ratio, concentrations of iodine after digestion lay within the appropriate range of standards for the FI methods as well as the batch method. There was also enough volume of liquid for replicate injections in the FI methods.

Optimization of FI parameters

For a catalytic reaction, concentrations of reagents play an important role on the kinetic patterns. In the Sandell and Kolthoff method, it is desirable to keep the concentration of Ce(IV) well below the As(III) concentration.

Optimization of the flow injection method was carried out using the set up shown in Fig. 1 in continuous mode. The concentration of Ce(IV) was varied over the range of 6.0 × 10⁻³ to 2.0 × 10⁻² M when the concentration of As(III) was fixed at 0.1 M. Replicate injections of 100 μg I l⁻¹ solution were made. The results have shown that the size of signal increased to approximately 0.2 absorbance value as the concentration of Ce(IV) was increased up to 1.5 × 10⁻² M. Above this concentration, the signal stayed fairly constant. The concentration of Ce(IV) at 8.0 × 10⁻³ M was chosen, because at this concentration acceptable sensitivity was obtained. Under this condition the molar ratio between Ce(IV) and As(III) was 1∶12.5 (the pumping rates of these two reagent streams were equal).

Chantore et al.²² have found that raising the temperature of the reaction zone can increase the sensitivity. For the continuous-FI system, the effect of temperature on sensitivity was studied at two bath temperatures. Linear calibrations were obtained from both temperatures, e.g., ΔA = 0.0010[I] + 0.049, r² = 0.998 (26 °C) and ΔA = 0.0016[I] + 0.133, r² = 0.997 (43 °C). The intercepts represent signals of the reagent blank. The results have shown that the sensitivity was better achieved at 43 °C than at room temperature (26 °C), in terms of slope and intercept of calibration.²³ Temperatures greater than 43 °C although giving higher sensitivity can decrease the solubility of air and sometimes cause air bubbles inside the glass coil. Thus for the continuous-FI, the bath temperature was fixed at 43 °C.

The set up condition of the stopped mode was similar to the condition illustrated in Fig. 1 except that the flow rate could be made faster to 3 ml min⁻¹. The results in Table 1 show that prolonging of the stopped time increased the sensitivity. For this work, a stopped time of 1 min was sufficient to provide reasonable absorbance readings over the working range of 50 to 200 μg I l⁻¹.

Table 1 Sensitivity and sample throughput of the stopped-FI system as determined by the stop interval

Stop interval/s	Slope (× 10^−4)	Intercept	r^2	Sample throughput/ sample h^−1
0	3.0	0.019	0.999	88
20	6.0	0.039	0.999	59
40	9.0	0.067	0.999	44
60	12	0.095	0.999	35
120	17	0.188	0.996	22
180	21	0.265	0.996	16

---

Sensitivity and sample throughput

In the studies, criteria for consideration of sensitivity are based on the slope of calibration and the calibration intercept.²³

Due to some differences such as configuration, the operating procedure, and under the conditions studied, the flow rates, the sensitivities given by the two modes, are incomparable. Nevertheless, it was observed that the sensitivity of the continuous system is greater than that obtained from the condition of the stopped mode. For example, the regression equations are ΔA = 0.0016[I] + 0.133, r² = 0.989 (for the continuous-FI, 43 °C) and ΔA = 0.0011[I] + 0.091, r² = 0.996 (for the stopped-FI, 26 °C).

In the stopped-FI mode, it is expected that the size of signal will be enlarged if the time spent by the reaction zone is prolonged. The results in Table 1 demonstrate that increasing the interval length resulted in larger size of signals including the signal of the reagent blank (intercept). Also the results have indicated that the slope became greater with time. Thus, prolonging of the stopped time increases the sensitivity of the stopped mode.

Besides the sensitivity, sample throughput is often a parameter used for consideration. This parameter is an indication of speed of analysis. The throughputs as determined by stopped interval are summarized in Table 1. Although the stopped time of 180 s delivered the highest sensitivity, the analysis was time consuming. For this work, the stopped interval of 1 min was chosen as it resulted in a reasonable sensitivity and acceptable throughput of sample (35 samples h⁻¹). This sample throughput is approximately equal to that given by the continuous-FI system.

Limits of detection

Limits of detection were determined by ten replicate injections of a reagent blank into both systems. The limits (3σ)²⁴ were found to be 2.3 and 3 μg I l⁻¹ for the continuous-FI and the stopped-FI respectively. Therefore these two systems are capable of detection of iodine in severe IDD samples (<20 μg I l⁻¹).

Interferences

The results in Table 2 show that signal alteration of all the species was less than the margins (±10%), even when the first seven species were studied at remarkably high levels. The rest of the foreign species that were formerly tested in another FI system were not considered to interfere in the present system. Although less chloric acid was used with this modified condition, serious interferents reported for this reaction such as SCN⁻ and ascorbic acid¹⁷ were eliminated.

Table 2 Effect of foreign species studied based on alteration of signal obtained from triplicate injections of iodine standard, 100 μg I l⁻¹

Foreign species	Added as	Concentration level tested/M	Signal alteration (%)
a Three-fold of the typical levels in urine.   b These levels of foreign ions were previously tested on a continuous-FI system proposed by Yaping et al.^21
Cl^−	NaCl	1.5^a	+8.1
SO4^2−	Na2SO4	0.2^a	+5.2
HCO3^−	NaHCO3	0.06^a	+0.9
(NH4)2HPO4	(NH4)2HPO4	0.144^a	−4.8
Ca^2+	CaCl2	0.084^a	0.0
Glucose	C6H12O6	2.34 m^a	+0.9
Urea	CO(NH2)2	9.9 m^a	+2.2
K^+	KCl	9.9 m^b	+0.9
Mg^2+	MgCl2	20.6 m^b	+1.8
SCN^−	KSCN	0.3 m^b	+4.0
Fe^3+	FeCl3	0.36 m^b	+7.3
F^−	NaF	0.53 m^b	−0.8
Mn^2+	MnSO4	36.4 μ^b	−0.9
Oxalic acid	C2H6O6	4.0 m^b	0.0
Uric acid	C5H4O3N4	11.9 m^b	−4.0
Ascorbic acid	C6H8O6	0.57 m^b	+0.9

---

Recovery

A recovery study was made on 14 samples using both techniques of flow injection as summarized in Table 3. From the results, recovery ranged from 84 to 111% or 99% on average. This satisfactorily good recovery reflects that interfering species were completely eliminated during digestion. This also indicated that the modified procedure of acid digestion is applicable for this method.

Table 3 Percentage of the recovery measured by using the flow injection systems^a

Urine sample	Iodine content/ μg I l^−1	Added/ μg I l^−1	Found/ μg I l^−1	Recovery (%)
a Determination of samples A to F and G to N were respectively carried out using the continuous mode (n = 2) and the stopped mode (n = 3).
A	31 ± 2	100.00	131 ± 0	100
B	78 ± 2	100.00	178 ± 0	100
C	83 ± 2	100.00	167 ± 1.5	84
D	106 ± 5	100.00	206 ± 3	100
E	147 ± 0	100.00	256 ± 3	109
F	330 ± 2	100.00	440 ± 6	111
G	15 ± 1	50.00	67 ± 0.6	110
H	33 ± 1.5	50.00	79 ± 0.6	85
I	51 ± 0.6	50.00	100 ± 1	98
J	52 ± 0.6	50.00	105 ± 0.6	107
K	53 ± 0.6	50.00	103 ± 0.6	99
L	81 ± 2.1	50.00	126 ± 2.5	93
M	86 ± 0.6	50.00	133 ± 2.7	96
N	94 ± 3.5	50.00	146 ± 0.6	101

---

Comparison between the batch and the flow methods

For the batch method, calibration is a plot between the absorbance reading against iodine concentration. The reading must be taken at a constant interval after the addition of Ce(IV) solution to the mixture of digested sample and As(III) solution.^1,10 Pino et al.¹⁹ used the plot between the change in percentage transmission and the concentration. None of these calibration plots is linear.

The calibration plot as suggested for use in the continuous-FI system was the plot between the size of signal (ΔA) versus iodine concentration. Under the conditions used, the calibration was linear which means that within 100 s the size of signal was directly proportional to the rate of reaction. Similar to the continuous-FI method, linear calibration was also obtained from the stopped-FI method. It is thus an advantage of the proposed methods.

Twenty samples of urine were analysed for iodine contents using the conventional method and the continuous flow injection method. The two procedural approaches gave good agreement in the results. The plot between the mean values of both techniques showed the absence of any analytical bias. The Pearson coefficient of correlation (n = 24) was found to be 0.952 and a reasonable linear regression line was obtained (y = 0.938x + 4.63).

Precision of the measurement between the batch and the flow method was compared. Ten replicates of injection (100 μg I l⁻¹) were made on the continuous-FI set up. Both methods contributed satisfactorily low values of RSD (1.6% and 0.4% for the batch and the flow methods respectively).

Although the batch method did not give a prominently high value of RSD, the results could be worse in the hands of different operators. The RSD results have shown that the continuous flow mode has allowed for a better control of timing than the batch method.

Repeatability of the stopped FI system was also measured and the RSD was found to be 1.2%. This RSD value was obtained when the pump was turned on and off manually. The precision index could have been improved if the system had been fully automated.

Excluding the step of digestion, the two systems of FI have given the analysis time for a sample of approximately 1 min 40 s. The analysis time is therefore much less than the time required in the batch method (25 to 35 min per sample).

Method validation

Iodine contents in eleven samples were determined by using four different procedures: three based on the Sandell and Kolthoff reaction, i.e., the conventional batch method, continuous-FI and stopped-FI methods and another by using ICP-MS. The results in Table 4 demonstrate that the results from the ICP-MS were greater than those from the catalytic methods. This perhaps suggested that there is some loss of iodine during the digestion. However, the results of all four methods are mostly consistent. It is observed for some samples, having contents below 20 μg I l⁻¹ (e.g. samples 1 and 3), that the precision of the catalytic methods were low. These samples were below the calibration ranges of the catalytic methods. This was not found for sample 2.

Table 4 Contents of iodine in urine determined on eleven samples using four methods of analysis, i.e., continuous-FI, stopped-FI, batch and ICP-MS (n = 3)

Iodine content/μg I l^−1 ± s
Sample	Continuous-FI	Stopped-FI	Batch	ICP-MS
1	5.44 ± 22	11.2 ± 15	18.9 ± 8.1	12.1 ± 15
2	13.3 ± 6.7	14.5 ± 1.5	17.4 ± 3.6	16.9 ± 4.2
3	15.5 ± 11	13.3 ± 13	14.6 ± 19	15.5 ± 1.7
4	31.2 ± 1.1	27.1 ± 4.9	28.8 ± 8.4	38.4 ± 17
5	34.0 ± 6.1	37.1 ± 4.0	32.5 ± 3.9	42.3 ± 2.2
6	37.7 ± 2.7	35.6 ± 2.9	39.5 ± 5.3	74.5 ± 0.61
7	43.4 ± 0.7	49.9 ± 3.5	49.8 ± 4.7	55.3 ± 2.5
8	57.1 ± 4.4	52.5 ± 0.8	59.6 ± 3.7	73.4 ± 0.19
9	78.8 ± 1.3	74.1 ± 2.6	84.3 ± 7.3	87.2 ± 11
10	79.0 ± 2.6	83.9 ± 2.3	76.6 ± 3.4	97.8 ± 1.7
11	82.5 ± 1.3	84.9 ± 1.4	81.8 ± 7.5	105 ± 1.2

---

Applying linear regression,²⁴ comparison was made for each method pair via the plot between the results achieved from the two methods. Parameters of the regression line such as intercept (a), slope (b) and correlation coefficient (r²) were then obtained.

The correlation coefficients (r²) shown in Table 5 indicate reasonably good correlation (r² >0.850). The confidence limits suggest that the slope and the intercept do not differ significantly from the ideal values. Thus, there is no evidence for systematic differences between the results obtained from the four techniques, i.e., continuous-FI, stopped-FI, batch and ICP-MS. This agreement of the results shows that the two modes of flow injection can be an alternative to the conventional batch method.

Table 5 Parameters of the regression lines obtained from the plot between iodine contents measured by each of both techniques. The slope (b) and the intercept (a) are shown with the 95% confidence limits (ts)

Method pair (x,y)	b ± tsb	a ± tsa	r^2
Batch, continuous-FI	1.04 ± 0.27	−4.45 ± 11.96	0.968
Batch, stopped-FI	1.03 ± 0.17	−4.15 ± 8.97	0.959
Stopped-FI, continuous-FI	0.981 ± 0.11	0.860 ± 5.74	0.982
ICP-MS, batch	0.732 ± 0.23	4.87 ± 15.18	0.894
ICP-MS, continuous-FI	0.784 ± 0.27	0.797 ± 17.6	0.923
ICP-MS, stopped-FI	0.789 ± 0.27	1.09 ± 17.7	0.915

---

Stopped-FI mode as a tool for kinetic study

With the use of stopped-FI mode, kinetic information can be fulfilled. In the stopped mode, the fading in color of Ce(IV) is recorded against time. Fig. 3 shows profiles of the kinetic process, which were recorded, from the proposed stopped-FI system. It is observed that the data points agreed well with the exponential fitting. This indicated that under the FI conditions, the disappearance of Ce(IV) is a first-order process. An investigation has been carried out in progress for using the rate constant in determination of iodine.

Fig. 3 Kinetics obtained from the stopped-FI system, showing the disappearance of Ce(IV) at different concentrations of iodine standard in sample zones. Full lines represent the exponential fittings.

Conclusions

Two FI procedures are described for determination of iodine in urine. The principle which is based on the catalytic effect of iodide in the reaction between Ce(IV) and As(III) was adopted from the common batch method used in monitoring IDD status. Continuous-FI and stopped-FI modes can be applied to a routine method and are applicable for automation. The error, which often arises in the batch method from the reading of absorbance at a fixed time, can be quite a problem when there are hundreds of samples collected from a subject area. This error is eliminated through the use of the flow injection technique because addition and mixing of reagents is always reproducible by the control of pumping rate. Analysis time per sample is much shorter for the FI methods when compared to the conventional method.

It was found that the calibration plot is linear for both the FI systems. The signal measured at a fixed time is therefore directly proportional to the rate of reaction.

The sensitivity and sample throughput given by the two systems was not much different. The systems give a satisfactorily low value of detection limit (≤3 μg I l⁻¹), although results from the determination of samples having contents below 20 μg I l⁻¹(severe IDD) can be erratic. However the systems are still suitable for screening these types of samples from the samples above the margin of severe IDD status. If necessary the samples containing this very low level of iodine can be re-determined in the stopped mode by increasing the stopped time.

For this work, the condition for urine digestion was developed by modifying the former condition^1,10 to reduce the amount of chloric acid. The proposed condition improves the effect of interferents and provides complete digestion.

Besides, the stopped-FI system can be used to study kinetic profiles of the reaction. Some results have shown that the kinetics is first-order in Ce(IV) concentration under the conditions used.

Acknowledgements

This work was supported by grants from the Thailand Research Fund and the Shell Centenary Education Fund Postgraduate Education and Research Program in Chemistry. We would like to thank Dr Prapin Wilairat for his useful comments.

References

1. J. T. Dunn, IDD newsletter, 1993, 9, 5. Accessed 26th January 1999 from http://avery.ined.virgima.edu/∼jtd/iccidd/idddocs/idd196htm.
2. W. Buchberger and K. Winsauer, Anal. Chim. Acta., 1985, 3, 347.
3. J. Odink, J. J. P. Bogaards and H. Sandman, J. Chromatogr., 1988, 431, 309.
4. J. Rendl, S. Seybold and W. Borner, Clin. Chem., 1994, 40, 908.
5. F. Moussa, M. C. R. Demy, F. Veinberg, F. Depasse, R. Gharbi, J. Y. Hautem and P. Aymard, J. Chromatogr. B, 1995, 667, 69.
6. Y. Yabu, K. Miyai, S. Hayashizaki, Y. Endo, N. Hata, Y. Iijima and R. Fushimi, Endocrinol. Jpn., 1986, 33, 905.
7. V. Poluzzi, B. Cavalchi, A. Mazzoli, G. Alberini, A. Lutman, P. Coan, I. Ciani, P. Trentini, M. Ascanelli and V. Davoli, J. Anal. At. Spectrom., 1996, 11, 731.
8. P. Allain, Y. Mauras, C. Douge, L. Jaunault, T. Delaporte and C. Beaugrand, Analyst, 1990, 115, 813.
9. M. Haldimann, B. Zimmerli, C. Als and H. Gerber, Clin. Chem., 1998, 44, 817.
10. J. T. Dunn, H. E. Crutchfield, R. Gutekunst and A. D. Dunn, Thyroid, 1993, 3, 119.
11. E. B. Sandell and I. M. Kolthoff, J. Am. Chem. Soc., 1934, 56, 1426.
12. P. A. Rodriguez and H. L. Pardue, Anal. Chem., 1969, 41, 1369.
13. S. P. Mushtakova, L. F. Kozhina, L. M. Ivanova, A. K. Myshikina and O. A. Rytova, J. Anal. Chem., 1998, 53, 214.
14. G. Aumont and J. C. Tressol, Analyst, 1986, 111, 841.
15. S. Suwannachoat, P. Wilairat and P. Tongkaew, unpublished work..
16. P. J. Garry, D. W. Lashley and G. M. Owen, Clin. Chem., 1973, 19, 950.
17. W. May, D. Wu, C. Eastman, P. Bourdoux and G. Maberly, Clin. Chem., 1990, 36, 865.
18. K. Tsuda, H. Namda, T. Nomura, N. Yokoyama, S. Yamashita, M. Izumi and S. Nagataki, Clin. Chem., 1995, 41, 581.
19. S. Pino, S. L. Fang and L. E. Braverman, Clin. Chem., 1996, 42, 239.
20. J. Ruzicka and E. M. Hansen, Flow Injection Analysis, Wiley, New York, 2nd edn., 1988..
21. Y.-P. Zhang, D.-X. Yuan, J.-X. Chen, T.-S. Lan and H.-Q. Chen, Clin. Chem., 1996, 42, 2021.
22. W. Chantore, S. Muangkaew, J. Shiowatana and D. Nacapricha, Lab. Rob. Autom., 1999, 11, 37.
23. F. W. Fifield and D. Kealey, Principles and Practice of Analytical Chemistry, Blackie Academic & Professional, London, 4th edn., 1995, p. 11..
24. J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry, Ellis Horwood, New York, 3rd edn., 1993..

---