Anal Bioanal Chem (2002) 374 : 354–358 DOI 10.1007/s00216-002-1460-2
A P P L I C AT I O N N O T E
G. Alavoine · B. Nicolardot
Comparison of potentiometric titration, IR spectrophotometry and segmented micro-flow analysis to determine inorganic C in alkaline solutions Received: 04 March 2002 / Revised: 14 June 2002 / Accepted: 24 June 2002 / Published online: 22 August 2002 © Springer-Verlag 2002
Abstract A segmented micro-flow analysis procedure was described for the determination of total carbonates in alkaline solutions. The method is based on CO2 diffusion from an acidified sample through a silicone membrane. The gas is then collected in a coloured acceptor solution. The decolouration of the solution due to the trapped CO2 is then determined by spectrophotometry at a wavelength of 550 nm. The method was evaluated for the range 0– 1000 mg C CO2 L–1 at a sampling frequency of 60 h–1. The method was repeatable (RSD better than 1%), accurate (1.5 and 2.5% for standards) and the limit of detection was close to 6 mg C CO2 L–1. The results were also compared with those obtained by the potentiometric reference method (PT) and an infrared spectrometric method (IRS). The measurements obtained showed good agreement with those obtained using PT and IRS methods, the PT method being the most accurate. The SFA method appeared to be a really efficient and a more suitable technique for routine C CO2 determination, although some results showed that for standard solutions, the measured concentrations were significantly different from those obtained with the PT method for concentrations higher than or equal to 800 mg C CO2 L–1. For 0.25 M NaOH samples, the three methods tested gave similar results in the range 0–600 mg C CO2 L–1. Keywords Carbonate · Carbon dioxide · Gas permeation · Segmented flow analysis · Spectrophotometry · Potentiometric titration
Introduction The determination of CO2 by continuous flow with spectrophotometric detection or other techniques in different
G. Alavoine (✉) · B. Nicolardot INRA – Unité d’agronomie de Châlons-Reims, Centre de Recherche Agronomique, 2 esplanade Roland Garros, BP 224, 51686 Reims cedex 2, France e-mail:
[email protected]
types of matrix has already been the subject of numerous publications (Table 1). Over the last two decades, a lot of work has been done to improve flow injection analysis (FIA). Unlike segmented flow analysis (SFA), in which the segmentation of the reagents and the sample by air bubbles ensures the homogeneity of the mixture and prevents carryover into the analytical conduits, FIA is based on the injection of highly reproducible volumes of samples under a laminar flow of reagents. The often-described advantages of an unsegmented stream are control of sample dispersion, resolution of the baseline between each sample, absence of any stabilisation time and high sampling frequency [1]. However, in the case of more complex chemical determinations requiring long reaction times (more than 20 s), clean-up step, on-line distillation or UV digestion, the flow injection analysis often turned to out be less effective than the air segmentation flow [2]. This explains why SFA is still much used. Skeggs et al. [3] were the first authors to publish an SFA method to determine CO2 in blood plasma. The procedure was based on the use of a debubbler as gas separator, which allowed the removal of CO2 from the solution before absorbing it again in a buffered phenolphthalein solution. By using this automatic method, 40 analyses could be performed in an hour. This method was also used by Chaussod et al. [4], who adapted the procedure for measuring CO2 trapped in NaOH solutions coming from soil incubations in controlled conditions. This procedure was seldom used and replaced by dialysers equipped with silicone rubber [5] or a PTFE membrane, the latter always being used in flow injection analysis (Table 1). Furthermore, these authors also used other coloured indicators instead of phenolphthalein. The debubbler is no longer used for the SFA method due to the tight linearity range and the fortuitous decolouration of the acceptor solution due to small volumes that pass from the acid stream to the buffered solution. The aim of this paper is to describe an optimised procedure to determine CO2 trapped in alkaline solution (0.25 M NaOH). This procedure is based on a segmented micro-flow analysis method by using a CO2 diffusion
355
Table 1 Methods used for the determination of carbonates in solution Ref.
Method and Detection
Coloured indicator
Matrix
Range of Concentration (mg C L –1)
RSD (%)
Sampling Frequency (h–1)
[3] [4] [8]
Gas separator (debubbler) in SFA. Spectrophotometry
Cresol red Phenolphthalein Cresol red
Plasma NaOH Serum
120–480 @750 200–380
–a
40 40 20
[1] [5]
Silicone membrane separation in FIA. Spectrophotometry
Cresol red
Plasma Serum
120–600 60–480
1.8–3.4 0.9
90 50
[6] [7]
PTFE or PVDF membrane separation in FIA. Spectrophotometry
Cresol red Bromothymol blue Bromocresol purple
Pure water Wines
13.7–72 68–1091
– 2.4
70 30–40
Beer and soft drinks
500–2700
1
–
Cresol red
River water, sea water
0.06–14.4 2.4–12 – 0.26–46.9
0.8 2 0.2 –
15 – – 20
[9] [10] [11] [12] [13]
Tubular PTFE membrane separation in FIA. Spectrophotometry
[14]
Titration. Potentiometry
Industrial sodium aluminate solutions
1100–5700
3
–
[15]
PTFE membrane separation in FIA. Electrical conductivity
Natural water
0.6 – 60
1.8
–
[16] [17]
Reextraction via ligand exchange. Spectrophotometry
Mineral, sea waters
6–20 12–20
0.3–2.2 3.8
–
[18]
Kinetic FIA. Spectrophotometry
Water
0.12–12
1–6
20
[19]
PTFE membrane separation in FIA. Passive acoustic emission
Detergents, waters, paper liquors
0.042–240
–
–
[20]
PTFE membrane separation in FIA. Bulk acoustic wave Impedance sensor
Natural water, waste water
0.12–240
0.95
45
[21]
Fourier transform infrared spectrometry (FTIR)
Natural water, waste water
3–200
1.3
15
[22]
Flow injection micro titration. Spectrophotometry
Industrial sodium aluminate solutions
@10,000
0.28
45
This paper
Silicone membrane separation in segmented micro flow analysis. Spectrophotometry
0.25 M NaOH
5–1000
0.2–1.0
60–70
a
Phenolphthalein
Not available.
through a silicone rubber membrane and is able to analyse 60–70 samples per hour. This analytical method considers a total flow that is two times less than the classical continuous flow, and the manifold is made up with 1 mm input diameter tubing instead of 2 mm. The method will be evaluated and compared with a potentiometric reference method (PT) and an infrared spectrometric method (IRS).
Materials and methods Standard solutions The standard solutions used for the calibration of the SFA method and the IRS method were prepared in deionised water with analytical grade Na2CO3. Due to the absence of matrix effect (results not
shown), it was more practical to prepare and preserve these solutions from the atmospheric CO2 by using a water matrix. Six standard solutions (0–1000 mg C L–1) for the SFA method and only four standards (0–100 mg C L–1) for the IRS method were prepared, considering the working range of the apparatus. Standard solutions of 200, 600 and 1000 mg C L–1 were also used to check the repeatability of the SFA measurements by analysing each of them nine times. For the comparison between the three methods, six standards (0–1000 mg C L–1) were prepared using NaHCO3 dissolved in a 0.25 M NaOH matrix. NaHCO3 was used instead of Na2CO3 to neutralise OH–, thereby making it possible to determine CO32– by back titration by using the PT method. For the three methods, the six alkaline standards were analysed four times and each solution was covered with parafilm™ to avoid trapping the atmospheric CO2 before analysis.
356
Fig. 1 SFA manifold used for the CO2 determination. (Gas diffusion unit = 4.3’’ dialyser equipped with a silicone rubber membrane)
Alkaline samples To measure the soil respiration, experimental samples were obtained following the procedure published by Chaussod et al. [4] and these came from various soil incubation experiments. Soil samples, equivalent to 50 g dry soil, were incubated under controlled conditions (soil moisture and temperature) in 600 mL plasma flask (one sample per flask). CO2 produced by soil was trapped in each flask by 10 mL 0.25 M NaOH; traps were periodically replaced. Eighty-three NaOH samples were chosen to compare the SFA and the IRS method and these were analysed twice. Segmented flow analysis method Analytical procedure Total inorganic carbon was determined by an SFA method using a continuous flow TRAACS 2000 analyser (Bran+Luebbe, Norderstedt, Germany). An improved and simplified SFA manifold was used (Fig. 1). The sample was aspirated at a sampling rate of 60 h–1 (sample to wash ratio=1.7) and mixed with a stream of acid diluent (donor), at the point where the released carbon dioxide enters the gas phase. Thereafter, the stream was dialysed against a bufferphenolphthalein solution (acceptor) using a 4.3’’ dialyser containing a silicone rubber membrane permeable to gasses but not to H3O+ ions (for CO2, at 25 °C a silicone membrane is 300 times more permeable than a PTFE membrane). The carbon dioxide gas absorbed by the acceptor solution causes a decrease in pH, which induces a decrease in indicator colour proportional to the concentration of CO2. The colour was then measured at 550 nm, corresponding to the maximum absorbance for this indicator (maximum absorbance from 549.7 nm to 552.6 nm). Elsewhere, the procedure did not require the use of soda lime as CO2 absorber to eliminate CO2 from the segmented air, since no alteration of the baseline was observed (absorbance and stability). The working range of the acceptor solution and the linearity range of the method were determined by titration of 100 mL of acceptor solution with the addition of 0.05 M NaOH or 0.05 M HCl. pH and absorbance values at 550 nm were recorded simultaneously. All the measurements could be automatically corrected for baseline and sensitivity drifts, by respectively analysing a 0 mg C L–1 and 600 mg C L–1 standard solution every 12 samples in the analytical sequences.
Reagents To prepare the reagents, analytical grade chemical products and CO2free deionised water were used. The donor solution was an acid solution (pH=1.2) allowing analysis of the 0.25 M alkaline sample. The solution was prepared by dissolving 6.8 g NaH2PO4, H2O, 27 mL 85% H3PO4, and 0.5 mL Triton 50% diluted in isopropanol, with water then added to reach a final volume of 1000 mL. The acceptor solution (1000 mL) was prepared by taking 67 mL of a stock buffer solution and mixing it with 1.4 mL 1% phenolphthalein stock solution (1 g phenolphthalein diluted in 100 mL ethanol) and 0.4 mL Triton 50%. The stock buffer solution was a mixture of 5.3 g L–1 Na2CO3 and 8.4 g L–1 NaHCO3. Maintaining a CO2-free air-segmentation procedure was not necessary.
Potentiometric titration method CO32– trapped in 0.25 M sodium hydroxide samples was determined by back titration using a 719 digital titrator (Metrohm, Herisau, Switzerland) with combined glass electrode, automatic burette and mechanical stirrer. To titrate 20 mL of these alkaline solutions, excess barium chloride was added (5 mL 5% BaCl2) to precipitate the Na2CO3 to BaCO3. The OH– left in solution was titrated by hydrochloric acid (0.2 M) up to the end-point of 8.62 corresponding to the maximum precipitation of BaCO3. For this pH value, almost all free CO32– was precipitated and all OH– was neutralised by H3O+. For an endpoint pH value less than 8.62, dissolution of BaCO3 occurs. Finally, the initial concentration of hydroxide was determined by using the same procedure. Infrared spectrometric method Total inorganic carbon (TIC) was determined using a 1010 total organic carbon analyser (O.I. Analytical, College Station, TX) equipped with a 1015 autosampler. The NaOH sample (1 mL) was introduced in the digestion vessel. The exact volume of the sample was then acidified with 2 mL of 5% H3PO4 acid solution prepared from freshly deionised water to convert inorganic carbon to carbon dioxide. The CO2 was purged from the sample with nitrogen carrier gas (99.998% purity and 300 mL min–1 flow rate) and then detected by a non-dispersive infrared detector (NDIR) calibrated to directly display the mass of CO2 measured; the detector was linearised over a range of 0–125 µg C. Eight analyses were performed in an hour.
357
lowing calibration plot established for a concentration range up to 600 mg C CO2 L–1: A=4.547C+3.532, (r=0.999, n=8) where A, is an arbitrary unit given by the apparatus and which corresponds to absorbance, and C the concentration (mg C CO2 L–1). The three standard deviation limit of detection was equal to 6.2 mg C L–1 for the independent analyses of 400 mg C CO2 L–1 (RSD 0.9%, n=4). The QL, which is the lowest concentration defined by convention to ten standard deviations from the analytical blank was close to 18 mg C CO2 L–1. However, by working with a lower range (0–100 mg C CO2 L–1), it is possible to reach a DL close to 600 µg C CO2 L–1 by just modifying the buffer composition of the acceptor solution. Finally, under our analytical conditions, the maximum baseline or sensitivity drift was always less than 5% during a two-hour sequence analysis. Comparison of SFA, IRS and PT methods Fig. 2 Relationships obtained between pH and absorbance for the acceptor solution (buffered phenolphthalein solution) and between CO2 concentration and absorbance
Results and discussion Evaluation of the capacity and linearity range of the acceptor solution used for SFA The relationship between the absorbance and pH of the coloured buffer is shown in Fig. 2. The acceptor solution was completely decoloured for pH values less than 9, and the absorbance regularly increased with the pH to reach a maximum value for pH 11. However, the coloured buffer was in fact usable when the absorbance decreased proportionally to the pH. Thus, for a very limited pH interval of 9.30–9.95, the acceptor absorbance varied greatly: approximate absorbance variation of 0.8–0.2 corresponding to a concentration 0–1000 mg C CO2 L–1. A better fit was obtained for the curve using a quadratic rather than a linear adjustment (r2=0.99996 for the six standards).
The CO2 concentration values of standards obtained for the three methods using a 0.25 M NaOH matrix were well correlated. The PT reference method provided a linear regression Y(PT)=0.982X+8.603 (r2=0.9998), whereas the SFA and IRS methods provided quadratic regressions which were respectively Y(SFA)=5.21×10–5X2+9.695X+2.418 (r2=0.9998) and Y(IRS)=8.44×10–5X2+9.075X+1.738 (r2= 0.9992), where X is the theoretical concentration and Y the measured concentration. The comparison of the results showed that globally, the PT method was the most accurate. The average differences between the measured values and the theoretical concentrations (for 200–1000 mg C CO2 L–1 standard solutions) were 1.2% for PT, 1.7% for SFA and 2.2% for IRS. The SFA method was the most repeatable method for C CO2 concentrations less than
Analytical features of the SFA procedure developed Multiple determinations of the three standards (200, 600 and 1000 mg C CO2 L–1) showed that the proposed method gives some repeatable results (not shown). Indeed, the measured average concentrations were respectively 196 mg C CO2 L–1 with a relative standard deviation (RSD) equal to 1.02% (n=9), 603 mg C CO2 L–1 (RSD 0.56%, n=9) and 987 mg C CO2 L–1 (RSD 0.18%, n=9). The measurements permitted an estimation of an absolute accuracy (100×[theoretical value–experimental value]/[theoretical value]) of the method close to 2.5 and 1.5% for 200 and 1000 mg C CO2 L–1, respectively. The limit of detection (DL) and the limit of quantification (QL) for this method were calculated from the fol-
Fig. 3 Comparison of CO2 concentration in 0.25 M NaOH samples measured by SFA and IRS methods
358
600 mg L–1 (average RSD 0.7, 0.9 and 1.6% for SFA, IRS and PT, respectively). For concentrations greater than or equal to 600 mg C CO2 L–1, the PT method was most repeatable (average RSD 0.1, 1.0 and 1.1% for PT, SFA and IRS, respectively). Finally, the results obtained with the IRS and PT methods were not statically different (t=0.71, p=0.01, n=24). The SFA and PT methods gave comparable results when the C CO2 concentrations were ≤600 mg C CO2 L–1 (t=1.19, p=0.01, n=16), but they were significantly different when the concentrations were ≥800 mg C CO2 L–1 (t=6.49, p=0.01, n=8). CO2 concentrations determined by both SFA and IRS procedures for the set of 83 0.25 M NaOH samples (70–1000 mg L–1 C CO2) were also well correlated (r=0.998, p
