Analyst, December, 1979, Vol. 104, $9. 1119-1123 A bso r pt iornet r i c Deter m i nation of Low Oxygen Concentrations in Power-station Waters 1119 Part 11.” AutoAnalyzer Continuous Automatic Method using the Technicon G. I. Goodfellow, D. F. Libaert and H. M. Webber Central Electricity Research Laboratories, Kelvin Avenue, Leatherhead, Surrey, K T 2 2 7SE A continuous automatic method for the absorptiometric determination of low oxygen concentrations in power-station waters using the Technicon AutoAnalyzer is described. Keywords Oxygen determination ; absorfitiowtetry ; water analysis ; continuous analysis Within the Central Electricity Research Laboratories, a method was required for the continuous determination of low concentrations (<50 pg 1-l) of oxygen in water from high- pressure/high-temperature corrosion research rigs.Because of the limited capacity of these rigs it was possible to abstract sample streams at flow-rates of only a few millilitres per minute ; this restriction virtually eliminated the use of any commercial dissolved oxygen on-line instrument. Goodfellow and Webberl (Part I of this series) have developed a manual absorptiometric technique for determining oxygen using leuco-methylene blue. This paper describes the modifications to the technique necessary to enable it to be used on a continuous basis with a Technicon AutoAnalyzer. For a detailed discussion of the various aspects of the technique the reader is referred to Part I. Experimental and Results Apparatus AutoAnaZyxer components. The proportioning pump, recorder and colorimeter were standard AutoAnalyzer I system components.The recorder was fitted with a range- expansion unit so that the chart width could be made to correspond to transmission ranges of 0 to 100, 50 to 100, 75 to 100 or 90 to 100%. The colorimeter was fitted with a flow- through cell having an optical path length of 50 mm. All glass coils, tubing and glass connections were standard items; Tygon pump tubing was used throughout. The term “reaction system” denotes the assembly of tubing, glass coils, etc., used to achieve the formation of the desired coloured product. The manual method described in Part I1 was shown to be suitable for determining oxygen up to concentrations of 100 pg 1-1 in water and the method was adapted to the AutoAnalyzer.Because of the permeability of plastic tubing to atmospheric oxygen it is necessary to construct the system in glass or stainless steel with butt-joint connections made with thick-walled PVC tubing. I t is also necessary to provide the sample at a positive pressure and use the pump after the reaction and measuring systems for metering purposes only. To add the reagent, it is necessary for it to pass through pump tubing where some oxygen contamination occurs; however, by subsequently passing the reagent through a delay coil (ca. 20 min) auto-reduction of any oxidised reagent occurs before its addition to the sample stream. It was known from previous work1,2 that both the chemical reaction rates and the electrical components of the AutoAnalyzer are affected by temperature fluctuations. To stabilise the system the components were housed in a Perspex box as is shown in Fig.2. The recorder was housed separately. Additionally, that part of the system used for the chemical reaction was protected from light. Reaction system. A diagrammatic representation of the reaction system is shown in Fig. 1. * For Part I of this series, see p. 1105.1120 Analyst, Vol. 104 A perfusor pump is required to inject known amounts of air-saturated water into the sample stream for calibration purposes. A pump suitable for this purpose, with a number of fixed flow-rates, for generating oxygen concentrations of about 4, 8, 20, 40 and 80 pg 1-l (and capable of much greater flow-rates if required), is manufactured by B. Braun, Melsungen, West Germany; the British agents are F.T. Scientific Instruments Ltd., Tewkesbury, Gloucestershire. GOODFELLOW et al. : ABSORPTIOMETRIC DETERMINATION OF Perfusor pump. Hypodermic needle, 25 mm X 23G (stainless steel) Luer fitting 0.42 ml min-' 2.5 ml min-' Cation-exchange resin column 3.9 mI min-1 Fig. 1. Reaction system for the AutoAnalyzer. Reagents Water. Distilled water passed through a mixed-bed de-ionisation unit is suitable for use in reagent preparations. Leuco-methylene blue. Solution A : dissolve 0.154 g of methylene blue (technical dye grade) and 0.81 g of glucose (analytical-reagent grade) in 44 ml of water and dilute the solution to 5 1 with glycerol (analytical-reagent grade). This solution is stable for several weeks. Solution B : dissolve 100 g of potassium hydroxide (analytical-reagent grade) in water and dilute to 200 ml with water.To 700 ml of solution A add 7 ml of solution €3. Mix by shaking and allow the solution to decolorise (ca. 30 min). This is sufficient reagent for 28 h continuous running of the AutoAnalyzer, and should not be used later than 1 week after preparation. Low-oxygen and air-saturated water. Prepare these solutions as described in Part 1.l Preparation of the Calibration Graph Setting the base line The base line on the recorder is set using low-oxygen water in place of the sample. Determining the response characteristics Having set the base line, the sensitivity and linearity of response to changes in oxygen concentration are determined by injecting, with the perfusor pump, known amounts of air-saturated water into the stream of low-oxygen water prior to the addition of the reagent.December, 1979 LOW OXYGEN CONCENTRATIONS IN POWER-STATION WATERS.PART 11 1121 Fig. 2 . Layout of reaction-system components: A, sample inlet; B, reagent; C, peristaltic pump; D, mixing coils; E, colorimeter; and F, perfusor syringe-pump. The concentration of oxygen at each calibration point is determined from the equation flow-rate of air-saturated water( 1 h-l) total flow-rate of water (1 h-l) x 103s~ 0 2 (pFLg1-l) = where S,mgl-l is the concentration of oxygen in the air-saturated water at a known temperature and pressure. S, is calculated from the equation SP 760 s, = - where P mm is the observed pressure and S is the solubility at 760 mm and the observed temperature; this value may be obtained from the table of solubilities in Part I,l Table I.The recorder trace of a typical set of calibration points is shown in Fig. 3 and the corre- sponding absorbances are given in Table I. The results show that a linear response is obtained up to 88 pg 1-1 of oxygen; above this concentration the response becomes pro- gressively non-linear although meaningful results can be obtained up to at least 220 pg 1-1 (about 10% transmission). TABLE I RESPONSE TO KNOWN OXYGEN CONCENTRATIONS Concentration of oxygen added/pg 1-I 0 8.8 22 44 88 220 Absorbance corrected Absorbance for the blank 0.045* - 0.098 0.053 0.187 0.142 0.333 0.288 0.599 0.554 1.022 0.977 * Base line arbitrarily set a t 90% transmission. An estimate of the precision of the method was determined by analysing two solutions, of 0 and about 20 pg 1-1 of added oxygen, alternately for 30-min periods over 6 h.During this time the base line (blank) drifted by 0.33 pg l-l, and the short-term noise (maximum range over a 10-min period) at both concentrations was less than 0.25 pg 1-1. The precision data, together with that obtained by the manual method,l are given in Table 11.1122 GOODFELLOW ef al. : ABSORPTIOMETRIC DETERMINATION OF Analyst, “01. 104 1 oo L- 0 30 60 90 1 Time/min 10 Fig. 3. Recorder trace for standard solutions. The numbers on the peaks are the concentration of added oxygen in p g 1-l; the base line is att 0 pg 1-I. TABLE I1 PRECISIONS OBTAINED WITH THE MANUAL AND AUTOANALYZER METHODS Manual method* AutoAnalyzer method? Standard deviationlpg 1-1 (i) a t 0 pg 1-1 (ii) a t ca. 20 p g 1-l Criterion of detectionlpg 1-1 0.40 11.69 0.93 0.11 0.37 0.25 * Calculated from 5 determinations.7 Calculated from 6 determinations made by alternating the 0 and 20 p g 1-1 solutions for 30-min periods. Sampling To avoid the interference effects from iron(I1)i and copper(I1) ions the sample must pass through a cation-exchange column prior to anal.ysis (see Part I1). The size of the column required will depend on the degree of contamination but should be made as small as possible to reduce the delay in response to a minimum. I[f the sample is extracted from a pressurised system it is necessary to arrange for the sample to be taken from a stream flowing to waste at atmospheric pressure. In this instance care inust be taken to ensure that the sample is not contaminated by the back-diffusion of air.Sources of Error Errors in the base-line setting This water should ensure that any bias is less than 1 pg 1-1. However, this setting does not take into account any self-colour in the sample. The true base line for the sample can be obtained by replacing the reagent with 98% glycerol solution, but this is an extremely time-consuming procedure owing to the long rinse-out period of the viscous reagent and is not recommended for routine use. The base line is normally set using low oxygen water in place of the sample.h X m b e Y , 1979 LOW OXYGEN CONCENTRATIONS I N POWER-STATION WATERS. PART I1 1123 E f e c t of temperature Changes in temperature affect the chemical reaction rate and the stability of the Auto- Analyzer system, while the stability of the reagent is adversely affected by an increase in temperature. The first two effects are minimised by enclosing the system in a Perspex box; the amount of reagent prepared (700 ml) ensures that it is replenished daily so that little deterioration does, in fact, occur.Reagent composition hence on the observed response of the AutoAnalyzer system. system is re-calibrated with each new batch of reagent. Small changes in the reagent composition have a marked effect on the reaction rate and It is recommended that the E f e c t of other substances Details of the effects of other substances were given in Part 1.l Of the substances normally present in power-station feed waters only iron(I1) and copper(I1) ions interfere with the determination and these are removed by cation-exchange resin. I t may be found necessary to fit a pre-filter into the sample line to protect the ion-exchange column from premature fouling. Any particulate matter in the sample will affect the absorbance of the solution. This work was carried out at the Central Electricity Research Laboratories and is published by permission of the CEGB. References 1. 2. Goodfellow, G. I., and Webber, H. M., Analyst, 1979, 104, 1105. Dolbear, S. A,, Riley, M., and Tetlow, J. A., Central Electricity Research Laboratories Report No. RD/L/R 1623. NOTE-Reference 1 is to Part I of this series. Received April 27th, 1979 Accepted May 30th, 1979