Determination of reducing sugars by flow injection gravimetry

Determination of reducing sugars by flow injection gravimetry

Analytica Chimica Acta 366 (1998) 119±125 Determination of reducing sugars by ¯ow injection gravimetry Raquel P. Sartini, ClaÂudio C. Oliveira, Elias...

132KB Sizes 0 Downloads 17 Views

Analytica Chimica Acta 366 (1998) 119±125

Determination of reducing sugars by ¯ow injection gravimetry Raquel P. Sartini, ClaÂudio C. Oliveira, Elias A.G. Zagatto*, H. Bergamin Filho1 Centro de Energia Nuclear na Agricultura, Universidade de SaÄo Paulo, P.O. Box 96, Piracicaba SP 13400-970, Brazil Received 26 August 1997; received in revised form 24 November 1997; accepted 3 December 1997

Abstract A ¯ow injection procedure for gravimetric determination of reducing sugars was developed. The system was designed to permit the monitoring of an in-line formed suspension. In view of the high density of the precipitate, a product of oxidation of reducing sugars by Fehling reagent, the Archimedes principle was exploited, so that the precipitate was weighed inside the ¯owing stream. Sample and reagents were simultaneously injected into two convergent carrier streams ¯owing through a reaction coil immersed in boiling water under re¯ux. The formed precipitate was cooled, accumulated on a sintered glass mini-®lter and weighed. An acidic solution was then injected into both carrier streams to promote in-line solubilization of the precipitate. Characteristics of the ®ltering device, geometry of the debubbler, ¯ow rates, wash solution and possibilities of weighing with the ®ltering unit on the plate of the analytical balance or suspended under it were investigated. Furthermore, effects of reagent concentration, temperature, surfactant addition and available time for reaction were studied. The gravimetric ¯ow injection system was very stable and required only 0.64 g Cu per determination. Precipitate drying was not needed. Precise results (r.s.d. usually <0.9%), in agreement with an of®cial titrimetric procedure, were obtained within 0.20 and 1.00% w/v invert sugar. A new weighing was accomplished every sixth minute. # 1998 Elsevier Science B.V. Keywords: Flow injection analysis; Gravimetry; Reducing sugars; Lane-Eynon method; Syrups

1. Introduction In the sugar industry, determination of the reducing sugars and `total reducing sugars' (reducing sugars plus inverted sucrose) [1] may provide information about the quality of the raw matter, process control and characteristics of the produced sugar [2]. The sucrose content in the juice may indicate the degree of sugarcane maturity and is often used to assess the commercial value. Moreover, the automated determination

*Corresponding author. Fax: 019 429 4610. 1 In memoriam. 0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0003-2670(97)00725-3

of the above mentioned analytes in glucose syrups is required for the quality control of production and for trading purposes. For this determination, the Lane±Eynon method [3] involving reducing sugar titration by Cu (II) under alkaline medium and high temperature (1008C) is often used. A product of the reaction is the slightly soluble Cu2O, thus an analytical balance can be used as a detector. A ¯ow injection system with gravimetric detection was recently described [4] and barium determination in synthetic samples was elected as a model application. However, partial and reproducible drying of the barium oxalate was needed. This cumbersome

120

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

step can be avoided in a ¯ow system which permits the gravimetric monitoring of an in-line formed suspension. In this way, the precipitate is accumulated on a mini-®lter and measured inside the main solution, the Archimedes principle then being exploited. The strategy is particularly attractive when high density precipitates like Cu2O (dˆ6.0 g cmÿ3) [5] are concerned. The main objective of this work then was to develop a ¯ow injection system with gravimetric detection for the determination of reducing sugars in glucose syrups and honey.

2. Experimental 2.1. Reagents, standards and samples All solutions were prepared with deionized water and analytical reagent quality chemicals. A modi®ed Fehling solution was formed in-line by merging two solutions (A and B [3]) at equal ¯ow rates. The A solution (R1- Fig. 1) was prepared by dissolving 20.784 g CuSO45H2O in 100 ml of water and the B solution (R2), by dissolving 20.0 g NaOH plus 70.56 g KNaC4H4O64H2O in 100 ml of water.

Fig. 1. Didactic presentation (upper) and flow diagram (lower) of the proposed system. ICˆinjector commutator, the alternative position specified by the dashed area; Sˆsample aspirated at 3.9 ml minÿ1; R1 and R2ˆA and B components of modified Fehling reagent at 0.8 ml minÿ1; Cˆwater carrier streams at 2.5 ml minÿ1; R3ˆacidic solution aspirated at 3.9 ml minÿ1; R4ˆsurfactant reagent at 0.4 ml minÿ1; L1, L2, L3, and L4ˆ400 cm (about 2 ml) sampling loops; B1ˆ300 cm heated coil; B2ˆ30 cm transmission line; B3ˆ100 cm coil; Dˆdebubbler with an outlet flow rate of 0.3 ml minÿ1; Fˆfiltration unit; x,y and zˆconfluences; Wˆwaste.

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

The wash solution (R3) was a 4.0 mol lÿ1 HCl or a 4.0 mol lÿ1 HCl plus 6.0 mol lÿ1 LiCl solution, prepared by dissolving 25.4 g LiCl in 100 ml of 4.0 mol lÿ1 HCl. The R4 stream was a 2.0% w/v Tween-80 solution and the C carrier streams were water. The standard stock solution, 10.00% w/v of invert sugar was prepared by dissolving 9.50 g of dry re®ned sucrose in 100 ml of water and treating with 5 ml of concentrated HCl. After 3 days, the volume was completed to 1000 ml [2]. Working standard solutions within 0.20 and 1.00% w/v invert sugar were daily prepared by appropriated water dilutions of the standard stock solution. Glucose syrups and honey were provided by a local sugar industry. About 1.0 g of the syrups or 0.5 g of honey were accurately weighed, dissolved in about 50 ml of water and the volume was ®lled up to 100 ml with water. Sugar-cane juices were ®ltered through a 0.45 mm cellulose membrane ®lter and 10-fold diluted with water. 2.2. Apparatus The systems comprised a Mettler Toledo AB 204 analytical balance with a device to hang the ®ltration unit under the plate, an Ismatec IPC-12 peristaltic pump with Tygon and Acid-¯ex pumping tubes, a manually operated injector-commutator [6], a Tshaped glass debubbler with an inner volume of about 400 ml (Fig. 2), a glass ®ltering device (Fig. 2) placed inside a ®ltration unit with a low inner volume (800 ml) and connectors. Mini-®lters with speci®c porosities were build up by the partial melting of the glass ®llings (106, 150 or 210 mm) in order to get porous sintered glass discs. The manifold portion before the debubbler was built up with PTFE tubing (i.d. 0.8 mm, wall thickness 0.2 mm) so that B1 coil could be immersed into water boiling inside a 1000 ml round-bottom ¯ask connected to a 70 cm re¯ux condenser and placed on a heating mantle [7]. The manifold portion after the debubbler was built up with polyethylene tubing (i.d. 0.8 mm) of the non-collapsible wall type; it was placed inside silicone tubing to minimize variations in measured mass due to tube tensioning.

121

Fig. 2. Debubbler (left) and filtration unit. 1ˆinlet stream; 2ˆoutlet air removal; 3ˆoutlet stream; 4ˆglass, wall thickness 0.5 mm; 5ˆsintered glass with 0.106 mm porosity.

2.3. Flow diagram The proposed system is outlined in Fig. 1. Sample and reagent solutions were simultaneously injected into convergent carrier streams and a surfactant solution was added immediately to the merging point. The processed sample zone reached the B1 coil and sugar oxidation proceeded inside it at about 958C. The formed precipitate passed through a debubbler and was then accumulated in the ®ltration unit ®tted under the plate of the analytical balance. In this way, the mass was continuously monitored. After achievement of the steady state signal, the injector-commutator was switched to the alternative position, inserting plugs of the R3 acidic solution into both carrier streams. Passage of the acidic zone through the ®ltration unit promoted the solubilization of the precipitate which was then wasted. When baseline mass returned to zero, next sample was injected. The system was designed with symmetric convergent streams and a high carrier-to-con¯uent ¯ow rate ratio at con¯uence point z (Fig. 1). B1 and B2 coil lengths were ®xed as 300 and 30 cm. Preliminary observations con®rmed that R1 and R2 reagents should not be combined. The sodium hydroxide plus sodium potassium tartrate and the copper sulphate solutions were stable but, after combination, the copper tartrate

122

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

complex was not stable enough to prevent a slow copper precipitation and turbidity could be observed within a few days. 2.4. Procedure Initially, a simple ¯ow system was designed to carry out the preliminary tests involving triplicate injections of either glucose or fructose standard solutions. Different debubblers to withdraw the air bubbles evolved under high temperature and different ®ltering devices were tested with this system. Possibility of weighing with the ®ltering unit placed on the balance plate or hanging under it, in¯uence of temperature and conditions for precipitate solubilization were also investigated. In this system, the sample volume selected by a 100±800 cm sampling loop was intercalated into a 3.2 ml minÿ1 water carrier stream and the established sample zone merged with the in-line formed Fehling reagent (2.0 ml minÿ1). Sugar was oxidized inside a 300 cm coil immersed in a water bath and the resulting suspension passed through the debubbler. The precipitate was further retained on the mini-®lter, evaluated and further solubilized by an acidic solution which was injected instead of the sample. The glucose or fructose standard solutions were always injected in triplicate. In order to improve the washing of the system, the above system was modi®ed permitting the solubilization step to be accomplished by the intermittent addition of the acidic solution [6]. This solution was added when the injector±commutator was moved to the sampling position and recycled in the other position. Furthermore, introduction of the sample and reagent in convergent streams, in merging zones con®guration [6], was performed to investigate the in¯uence of surfactant addition, washing step, reagent concentrations and reaction available time. Thereafter, the system in Fig. 1 was built-up and the main analytical characteristics evaluated by using invert sugar standards. The system was then applied to analyses of real samples. 3. Results and discussion 3.1. Conditions for system set-up The analytical balance did not stabilize when the ®ltration unit was placed on its plate, because the

lateral windows remained partially open and the inand outlet polyethylene tubes of the ®ltering chamber remained tensioned. Stability was improved by hanging the ®ltration unit under the balance, however oscillations were observed yet. Better stabilization was attained by changing the polyethylene by silicone tubes, which are more ¯exible. In this situation, the balance indicated 0.00.1 mg even with the peristaltic pump working. However, for higher sample amounts (m>20 mg), precipitate adherence on the silicone tube walls was veri®ed and reproducibility was not satisfactory. The drawback was minimized by using polyethylene tubing involved by silicone tubes, situation elected for this work. The need for the external silicone tube is due to its ability in insulating the main stream from external in¯uences such as vibrations, air currents, etc. The geometry and material of the debubbler were important parameters, since the air removal, amount of precipitate accumulated on the ®lter and washing time depended on the characteristics of this unit. With a T-shaped glass debubbler [7] and an aspiration rate of 0.3 ml minÿ1, bubbles were only partially removed by deteriorating the measurement precision and/or baseline stability. On the other hand, higher ¯ow rates were more ef®cient but part of the formed precipitate was wasted together with the air phase. Use of a perspex debubbler with larger (1300 ml) inner volume improved bubble elimination, but precipitate settlement and adherence to the inner walls were veri®ed. Again, small particles were aspirated too and washing time was unacceptably long. A small (400 ml) glass debubbler (Fig. 2) was then designed. It should be stressed that a debubbler in the analytical path was a positive factor in detector stabilization, because it actuated also as a pulse dampener [8]. The inner volume of the ®ltration unit could not be selected at will. When it was too low (<300 ml) system clogging was observed for the more concentrated standards, as there was no suitable space for proper precipitate accumulation. Compactation of the precipitate was then veri®ed with a consequent increase in hydrodynamic pressure as well as dif®culties in precipitate removal. Since large dead volumes in the analytical path should be avoided, the inner volume of the ®ltration unit was selected as 800 ml which was enough for the proper measurement and precipitate

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

washing even when the most concentrated standard solution was processed. Regarding the mini-®lter porosity, the lowest one was elected, because unreproducible precipitate losses were observed for the higher porosities. It should be stressed that this ®lter could not be used in connection with the 200 ml ®ltration unit as precipitate compactation was noted. Temperature for the sugar oxidation is also a relevant parameter. Precipitate formation was not observed for glucose or fructose standard solutions when the temperature of the waterbath into which B1 coil was immersed was lower than 508C. A slow precipitate formation was observed for fructose standards at 658C. At >908C, precipitate formation was veri®ed both for glucose and fructose standards; as equal amounts were found, quantitative reaction was strongly suggested. Therefore, it was decided to keep B1 coil immersed in boiling water under re¯ux. In this situation, a 100 cm coil (B3) immersed in current water (25±308C) was placed immediately after the B1 coil for cooling the sample zone. Without it, stabilization of the detector was not suitable. Proper washing was not attained with the simple initial system, as most of the injected acid was neutralized by the Fehling reagent. The drawback persisted by intermittently adding the acidic solution: moreover, a ®ne yellowish precipitate was formed because of the lessening of alkalinity due to improper acid removal before the next sample injection. By exploiting the merging zones approach, this effect was circumvented and the injected acid did not interact with the remaining Fehling reagent and vice-versa. With this geometry, however, higher acid amounts should be injected in order to compensate dilution by the con¯uent water streams. The drawback was overcome by using the system in Fig. 1. 3.2. System dimensioning Sample and reagent volumes should be increased to improve sensitivity. However, precipitate adherence on the B1 inner walls was observed for the most concentrated samples. Presence of air bubbles inside this coil (air segmentation) did not produce better results: after the passage of the sample zone through B1 coil, small particles adherent on the tube walls were yet observed, especially where the cuprous oxide

123

formation was more pronounced. Use of glass, PTFE and silicone tubes yielded similar effects. With the addition of the surfactants within 0.1 and 1.0% w/v, no alterations in the washing step were observed. For 2.0% w/v Brij or PVA [poly (vinyl alcohol)] solutions, part of the formed precipitate still remained in the analytical path, whereas use of Triton X-100, glycerol or Tween-80 solutions improved precipitation and washing steps. In the presence of 5.0% w/v Tween80 solution, no precipitate adherence was observed and a higher precipitate amount was retained on the mini-®lter, denoting the formation of larger particles. Therefore, it was decided to add Tween-80 and its concentration could not be increased inde®nitely. With a 5.0% w/v concentration a ®lm of surfactant was established inside the main reactor. The surfactant concentration was then set as 2.0% w/v. It should be reported that PVA addition leaded to formation of small particles which easily passed through the ®ltration unit. Injection of mixed R1 and R2 solutions with similar concentrations as the Fehling reagent (curve b- Fig. 3) permitted to obtain linear analytical curves up to about 0.40% w/v invert sugar. For 0.80% w/v, formation of small yellowish particles was observed as a consequence of the low reagent concentration. In this situation, besides the lower amount of the formed precipitate, part of it was lost through the ®ltration unit. Use of the reagent solutions with lower concentrations yielded similar results only for the lower standards (Fig. 3). A linear analytical curve up to 0.80% w/v invert sugar was obtained by increasing the reagent concentrations to the above speci®ed values (curve c-Fig. 3): y ˆ 0:0004 ‡ 0:05508x…r ˆ 0:9994; n ˆ 5† where yˆanalytical signal, in mg and xˆinvert sugar concentration, in % w/v. No modi®cations were observed by doubling the reagent concentrations. For higher sugar concentrations, bending of the analytical curve was veri®ed, the effect being more pronounced for lower reagent amounts (curves a,bFig. 3). The effect was probably due to limitations in stoichiometry. A critical step of the procedure is the washing of the system. Total removal of the precipitate could not be attained if R3 concentration and/or injected volumes were not enough. Removal improved by increasing

124

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

since the stream through B1 coil actuated as a refreshing agent. 3.3. Application

Fig. 3. Influence of reagent concentrations. Curves a, b, c, d (&, *, ~, !) correspond to 0.5, 1.0, 1.5, 2.0-fold Fehling reagent concentration [3]. Figure refers to system in Fig. 1 with an injected volume of 4 ml.

both the acidity and the injected volume, the later parameter de®ning the effective available time for washing. With 2.5 ml injected volumes ef®cient washing was accomplished during 1 min. It was veri®ed that ef®cient precipitate removal was attained with 4.0 mol lÿ1 HCl relatively to the lower standards (up to 0.40% w/v invert sugar); in order to avoid the use of excessive acidity, LiCl was tested as it could improve copper chlorocomplexation, thus assisting solubilization. After trial and error experiments, the concentration of R3 was de®ned as 4.0 mol lÿ1 HCl plus 6.0 mol lÿ1 LiCl. The proposed system is little affected by the variations in ¯ow rates. When the speed of rotation of the peristaltic pump was varied in 40%, slight variations in analytical signals (usually <15%) were observed especially for the lowest standard. Lower ¯ow rates deteriorated sampling frequency and higher ¯ow rates could not be used to avoid the excessive hydrodynamic pressure due to the precipitate settlement on the mini-®lter; moreover, with a high ¯ow rate, temperature of the reaction medium would decrease

The proposed system was very stable and required 0.64 g Cu per determination. About 20 samples could be run per hour, but since for the most concentrated samples the washing time was longer, sampling rate was adjusted to 15 hÿ1. Precipitate drying was not needed and solubilization towards waste was performed in-line. Similar results were obtained for glucose and fructose within 0.20±1.00% w/v and additivity was always found. Typical recorded `masses' are within 10 and 50 mg. When applied to syrup and honey analyses, the system yielded precise results. After eight consecutive processing of a syrup sample with 58.0% w/v invert sugar, the relative standard deviation was estimated as 0.9%. Analysis of Table 1 also reveals the measurement repeatability. Agreement between the results obtained by the proposed system and by an of®cial titrimetric procedure [3] can be assessed from Table 1. The titrimetric procedure however yielded higher values of concentrations for some samples when compared with gravimetric procedure in view of the dif®culties in the visual de®nition of the end point.

Table 1 Results obtained with the proposed system (Flow gravimetry) and with a classical method (Titrimetry). Data in % w/w invert sugar obtained by processing typical samples. Honey and syrup samples 200 and 100-fold (w/w) manually diluted with water prior to injection into the system in Fig. 1. Uncertainties are estimates of relative standard deviations based on three replications. For details, see text Sample

Flow gravimetry

Titrimetrya

Honey 1 Honey 2 Honey 3 Glucose syrup 1 Glucose syrup 2 Glucose syrup 3 Glucose syrup 4 Glucose syrup 5 Glucose syrup 6 Sugar-cane juice

79.81.2 83.36.8 78.57.5 65.40.1 62.20.2 61.20.4 59.90.1 56.80.7 58.70.1 0.940.02

79.84.5 74.63.6 78.50.4 65.40.7 64.10.8 65.70.9 57.01.8 61.60.9 61.20.9 0.920.01

a

Lane-Eynon method.

R.P. Sartini et al. / Analytica Chimica Acta 366 (1998) 119±125

Moreover, titrations were performed always by the same technician. 4. Conclusions The feasibility of the ¯ow gravimetry for the determination of the major constituents in syrups, molasses and similar substances was demonstrated. Cumbersome steps related to classical gravimetry such as manual precipitate drying, ®ltering, washing, etc., are avoided and pertinent glassware such as muf¯es, desiccators and others are not needed. Although gravimetry is concerned, the proposed method can be classi®ed as instrumental, so that calibration curves are needed. Standardization of the Fehling reagent is then not required. Reproducibility is good, the recorded mass for a given standard solution being maintained within 3% during several days. The system is very robust, but the detector should be isolated from factors affecting gravimetry, such as air currents, vibrations, etc. It cannot be miniaturized at will because a minimum analyte amount should be present to provide a suitable mass value. To overcome this limitation, use of piezoelectric sensors [9] is promising.

125

Acknowledgements Support from FAPESP, CNPq, CAPES and FINEP/ PRONEX is greatly appreciated. References [1] M. Hangos-Mahr, E. Pungor, Hung. Ac. Sci. 21 (1982) 307. [2] G.P. Meade, J.C.P. Chen, Cane Sugar Handbook, 10th ed., Wiley, New York, 1977, p. 947. [3] F. Schneider (Ed.), Sugar Analysis, Official and tentative methods recommended by the International Commission for Uniform Methods of Sugar Analysis (ICUMSA). ICUMSA, Peterborough, 1979, pp. 41±73. [4] A.O. Jacintho, M.A.Z. Arruda, E.A.G. Zagatto, B.F. Reis, Anal. Chim. Acta 258 (1992) 129. [5] N.A. Lange (Ed.), Handbook of Chemistry, McGraw-Hill, New York, 1961, pp. 248±249. [6] F.J. Krug, H. Bergamin Filho , E.A.G. Zagatto, Anal. Chim. Acta 179 (1986) 103. [7] I.L. Mattos, E.A.G. Zagatto, A.O. Jacintho, Anal. Chim. Acta 214 (1988) 247. [8] H. Bergamin Filho, B.F. Reis, E.A.G. Zagatto, Anal. Chim. Acta 97 (1978) 427. [9] G.C. Dunham, N.H. Benson, D. Petelenz, J. Janata, Anal. Chem. 67 (1995) 267.