- Email: [email protected]
A flow-batch internal standard procedure for iron determination in hydrated ethanol fuel by flame atomic absorption spectrometry
Talanta 70 (2006) 522–526
A flow-batch internal standard procedure for iron determination in hydrated ethanol fuel by flame atomic absorption spectrometry Jos´e Edson da Silva a , F´abio Andr´e da Silva a , M. Fernanda Pimentel b,∗ , Ricardo Saldanha Honorato c , Valdinete Lins da Silva b , Maria da Conceic¸a˜ o B.S.M. Montenegro d , Alberto N. Ara´ujo d b
a Depto de Qu´ımica Fundamental (UFPE), Recife, Brazil Depto de Engenharia Qu´ımica (UFPE), Rua Prof. Arthur de S´a s/n, Cidade Universit´aria, 50740-521 Recife, Brazil c Departamento de Pol´ıcia Federal-SR/DPF/PE, Recife, Brazil d REQUIMTE/Depto de Qu´ımica-F´ısica, Fac. Farm´ acia (UP), Porto, Portugal
Received 21 September 2005; received in revised form 29 December 2005; accepted 29 December 2005 Available online 13 February 2006
Abstract A flow-batch manifold coupled to a flame atomic absorption spectrometer was evaluated to assess the iron content by the internal standard method in hydrated ethanol used as fuel in automotive industry. For this assessment official methods require calibration procedures with matrix matching, making it difficult to obtain accurate results for samples adulterated by the addition of water. Nickel was selected as the internal standard since it is usually absent in samples and because it requires similar conditions of atomization. After procedure optimization, which requires about 4.25 mL of sample and standard per measurement, it was possible to get linear analytical response for iron concentrations between 0.12 and 1.40 mg L−1 and a detection limit of 0.04 mg L−1 . Eighteen samples were collected randomly from fuel stations in Pernambuco (Brazil) and iron concentration was determined using the proposed procedure. Comparison of results obtained (0.20–1.50 mg L−1 ) showed a mean standard error of 3.9%, with 3.8% and 2.3% calculated for the mean variation coefficients of the proposed method and the reference procedure, respectively. For adulterated samples (0.12–0.64 mg L−1 ), the mean standard error was 4.8% when compared with the standard addition method. These results allowed concluding that the proposed procedure is adequate to accomplish the determination of iron in ethanol fuel in a large scale basis with a sampling rate of about 10 h−1 . © 2006 Elsevier B.V. All rights reserved. Keywords: Hydrated ethanol fuel; Iron; Internal standard calibration; Flame atomic absorption spectrometry
1. Introduction Since the first petrol crisis in the middle of the seventies, a number of countries have adopted policies to search for other more economical energy sources. This was the case of Brazil, which envisioned the use of ethanol obtained from sugar fermentation both for alternative fuel and to increase employment opportunities in the industry related to its production [1]. As a result vehicle engines solely using this type of fuel and more recently flexfuel engines have been developed at the same time as a widespread distribution of hydrated ethanol fuel (HEF)
∗
Corresponding author. Tel.: +55 81 21267291; fax: +55 81 21267235. E-mail address: [email protected] (M.F. Pimentel).
0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2005.12.057
in street gas stations has been effected. An issue of particular concern with HEF usage is its contamination with iron during the distillation, storage and distribution processes. When iron is present above 5 mg L−1 , it induces the formation of deposits during the explosion stage, thus reducing the life of an engine using this fuel [2]. In face of this situation, the Brazilian normalizing committee, Associac¸a˜ o Brasileira de Normas e T´ecnicas (ABNT), has proposed an analytical procedure based on flame atomic absorption spectrometry which is based on the use of calibrating solutions prepared in ethanol 95% (v/v) solvent [3]. This simple procedure, however, does not take into account the usual variations in ethanol content of the fuel, which can be between 69% and 99%, thus causing significant accuracy problems in the results obtained. Technicians often resort to standard additions method to overcome this problem, albeit with the
J.E. da Silva et al. / Talanta 70 (2006) 522–526
drawbacks of increasing the hazards related with samples handling, being labor intensive and difficult to afford in large scale analysis. Procedures based on the use of internal standards are commonly used in chromatographic [4,5] and inductively coupled plasma atomic emission spectrometry to overcome biased results by the instrumental drift and correction for non-spectroscopic matrix effects [6]. This use has lately been extended to inductively coupled plasma-mass spectrometry (ICP-MS) to improve isotope ratio precision [7,8] and inductively coupled plasma optical emission spectrometry with axial detection in order to improve precision and minimize matrix effects, mainly due to other elements present in samples [9]. Internal standard procedures are based on sample spiking with a known amount of absent chemical species which physical–chemical behavior is similar to the analyte, being both measured preferably at the same time. Although it is impossible to fulfil this last condition with traditional instrumentation, use of an internal standard method has been reported since the very beginning of atomic absorption spectrometry, based on the development of instruments with multichannel capabilities [10–14]. Thus, procedures for the determination of Al, Ca, Cr, Cu, Fe, Mg, Mn, Ni, Si and Zn in cement, feeds and alloys by flame atomic absorption spectrometry (FAAS) were described to minimize the different nebulization yields observed in samples and calibration standards [15–17]. For the same reason, references to internal standard procedures in electrothermal atomization are also scarce, except concerning the use of cobalt in iron determinations [13], copper and indium for the determination of lead and selenium in water [17], tallium and bismuth for the assessment of lead in biological materials [10] and arsenium for the determination of selenium content in mineral waters [18]. In the present work, the internal standard method approach is once more revisited to assess iron levels in HEF. The reliability of the results provided is compared with the protocols under current use. Recently, flame atomic absorption spectrometers were placed on the market, enabling the exploitation of the different analytical wavelengths emitted by multi-element hollow cathode lamps or by several single-element lamps, thus providing short time sequential determinations of more than one element per assay [19,20]. In this way, internal standard calibration can be envisaged as a simple means to circumvent the previously cited problems, namely those related to sample matrix variability when the direct calibration protocol is used or with the standard additions method analytical rate. In order to provide safer sample handling and expeditious sample preparation and measurement, the application of a flow-batch manifold is extended to allow a fully automated procedure for iron determination in HEF. The flow-batch concept which has been recently proposed [21,22] is based on a programmed flow preparation of solutions into a small-volume open container. Due to the time required for signal acquisition at both wavelengths, this procedure enables the automated preparation of large volumes with homogeneous content in both the analyte and the internal standard.
523
2. Experimental 2.1. Reagents and solutions Analytical grade chemicals were used without further purification, and all solutions were prepared using double deionized water. Nickel and iron calibrating solutions were prepared by rigorous dilution of 1000 mg L−1 of standard stock solutions available on the market (Titrisol, Merck). The ethanolic standards were prepared according to the ABNT NBR 11331 procedure. The density of P.A. Ethanol (99.9%) was first adjusted with deionized water to 0.82 g mL−1 to obtain a 95% ethanol mixture, which was then used to prepare iron standards with 0.20, 0.50, 0.80, 1.10 and 1.40 mg L−1 , by dilution of a 100 mg L−1 aqueous iron solution. 2.2. Equipment The flow manifold (Fig. 1), was comprised of a peristaltic pump Ismatec IPC, model 78001-10, (Switzerland), a 10-ports rotary valve VICI Cheminert NCR 0138 (Valco, Houston), a NResearch 161 T031 three-way solenoid valve (Stow, MA), and a Varian Fast-Sequential, model 220 FS, atomic absorption spectrometer. Upon activation of the solenoid valve the detector aspirated the solutions under stirring placed by the manifold in a home-made open cylindrical PTFE mixing chamber possessing a cavity with a 2 cm internal diameter and 4 cm high in which a 1.2 mm flow path was drilled in the bottom. The different devices were connected using PTFE tubing with 1.2 mm i.d., having the length of 10 m for the holding coil, 40 cm for the transmission line between the rotary valve and the mixing chamber, 8 cm between the mixing chamber and the solenoid valve and further 30 cm up to the detector. To control the different devices home-made software was implemented on a microcomputer connected by means of a PCL 711S Advantech interface card. Signal acquisition was carried out by sending a start command to the SpectrAA software of the spectrometer.
Fig. 1. Flow-batch system scheme. FAAS-FS, fast-sequential atomic absorption spectrometer; PC, microcomputer; DW, deionized water; SV, three-way solenoid valve; MC, mixture chamber; AI, air input; S, sample; IS, internal standard; CS, calibration iron standard; RV, rotary valve; W, waste; HC, holding coil; PP, peristaltic pump.
524
J.E. da Silva et al. / Talanta 70 (2006) 522–526
2.3. Procedures
solution towards the mixing chamber, where it is stirred for 15 s before measurements, another air bubble is aspirated.
Atomic absorption determinations were performed sequentially, using a time window of 4 s, at the wavelengths of 284.3 nm for iron, 232.0 nm for nickel (bandwidth of 0.2 nm) and 228.8 nm for cadmium (bandwidth of 0.5 nm). Acetylene and air flows were settled to, respectively, 1.80 and 13.0 L min−1 . In these conditions an aspiration flow-rate of 9 and 10 mL min−1 into the pneumatic nebulizer was registered for water and ethanol (95%, v/v), respectively. In order to preserve the spectrometer usage either for conventional measurements as well for the automated set-up developed, a three-way solenoid valve was connected to the entrance tubing of the nebulizer, thus allowing conventional use when the valve was inactive, or aspiration of the mixing chamber content when activated during flow-batch measurements. To minimize the time spent in the aspiration and propelling of the different solutions through the flow-path of the manifold, previous experiments had been performed using water and ethanol, to verify the maximum allowable flow rates at which the driven volumes were independent from the physical characteristics of the solution. Finally, the flow-rates of 3.5 and 3.2 mL min−1 were settled for water and 95% (v/v) ethanolic containing solutions, respectively. Afterwards, the analytical cycle described in Table 1 was developed. Each analytical cycle started with the aspiration of an air bubble through port 2 of the rotary valve into the coil HC, thus avoiding samples or standards dispersion in the water used as carrier. Calibrating solutions were prepared in-line, by sequential aspiration into the coil HC of between 0.45 and 3.15 mL of a 2.00 mg L−1 iron solution presented at port 5 of the rotary valve, 0.25 mL of a 10 mg L−1 nickel solution available in port 4 and between 3.80 and 1.10 mL of water though port 6. To assess iron concentration in HEF samples the same analytical cycle was adopted (Table 1, steps 13–16). The volume of each aspired sample was 4.25 mL, to which it was added 0.25 mL of the internal standard. Before propelling the aspirated
Table 1 Flow-batch system analytical cycle for iron determination in ethanol fuel Step 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Event
Direction
Volume (mL)
Time (s)
Air Calibration standard Internal standard Deionized water Air Solutions to the chamber Stirring Signal acquisition Chamber cleasing with deionized water Chamber emptying Sample change Discharge Air Sample aspiration Internal standard Proceed to step 5
Aspiration Aspiration Aspiration Aspiration Aspiration Forward
– 0.45–3.15 0.25 3.80–1.10 – 4.50
1.5 7.7–54.0 4.3 20.0–65.0 1.5 80.0
– FAAS aspiration Forward
– – 4.50
15.0 37.5 80.0
FAAS aspiration Aspiration Forward Aspiration Aspiration Aspiration –
– –
51.5 5 5 1.5 79.7 4.3 –
– 4.25 0.25 –
3. Results and discussion Aiming to assess the influence of the ethanol content on the solutions aspirated into the atomic absorption spectrometer, different sets of iron calibrating solutions were prepared in an ethanol/water solvent ranging from 69% to 99% (v/v). As expected, significant (at the 95% confidence level) different linear calibration slopes were obtained, between 0.0239 ± 0.010 L mg−1 for the set with lowest ethanol content and 0.0207 ± 0.015 L mg−1 for the highest. This decrease of the slope values could be explained by the change of the flame atomization temperature resulting of the increase of the fuel:air stoichiometric ratio. The statistical analysis of the calibration data obtained [23] revealed, at the confidence level of 95%, a t-value of 2.15, below the theoretical value of 2.45, when the slopes obtained for 95% and 99% (v/v) ethanol/water were compared, indicating that for small variations of the ethanol content in samples the official standard calibration procedure was able to provide acceptable results. In spite of this, our experience shows that samples with ethanol content below 95% are very often encountered, thus requiring the development of an alternative and more robust procedure. 3.1. Internal standard selection In our previous experiments both nickel and cadmium had not been detected in the HEF samples and hence could be used to implement a new internal standard protocol. To select which of them was the more appropriate specie, the relative standard deviation of absorbencies ratio was considered. Therefore, samples with an iron concentration of 0.50 mg L−1 were prepared in water and 95% (v/v) ethanol/water solvents, spiked both with nickel and cadmium up to the concentration of 0.50 and 0.20 mg L−1 , respectively, using the flow-batch procedure described above. The relative standard deviation values estimated from four replicates for the iron/nickel ratios in water and in 95% (v/v) ethanol/water were 2.0% and 3.8%, respectively. For the iron/cadmium ratio, these values increased to 7.6% and 7.7%, respectively. In order to further assess if nickel was an appropriate choice, on-line prepared calibrating solutions with iron concentrations in the range up to 1.40 mg L−1 , in the presence or absence of 95% (v/v) ethanol were tested. Considering only the signals provided for iron, slopes of 0.0358 and 0.0213 L mg−1 were found for the calibration curves in water and 95% (v/v) ethanol, respectively. A t-test shows that these values were significantly different at the 95% confidence level. When the absorbance ratio was used as the dependent variable, the two slopes (2.08 L mg−1 for water and 2.12 L mg−1 for 95%, v/v, ethanol) were statistically indistinguishable. Within-batch reproducibity was also compared for the standard and the proposed internal standard procedure, considering a 9-h analytical run and one single initial calibration procedure. In this evaluation two real samples (content of ethanol of 95%, v/v) containing, respectively, 0.83 mg L−1 (S1) and 1.10 mg L−1 (S2) of iron
J.E. da Silva et al. / Talanta 70 (2006) 522–526
were processed in 30 min intervals, using the flow-batch set-up, for the addition of the internal standard to a fixed concentration of 0.50 mg L−1 . During the analytical run a continuous drift of instrumental conditions was registered, thus rendering results between 0.57 and 0.99 mg L−1 for sample S1 and from 0.79 to 1.37 mg L−1 for sample S2 (relative standard deviation of 13.1%), with the standard method based only on the analytical signals enabled by the iron content. Using to the internal standard additions method, the variation around the target value was reduced to 4.7% and 2.7%, respectively, thus compensating for the instrumental drifts.
525
Table 3 Iron concentration in ethanol fuel adulterated samples obtained by the proposed procedure and the ABNT official method Sample
N O P Q R S
Iron concentration (mg L−1 ) Ethanol content (%)
ABNT official method
Standard addition method
Proposed procedure
84 82 77 75 72 70
0.39 ± 0.01 0.34 ± 0.01 0.54 ± 0.02 0.02 ± 0.01 0.50 ± 0.01 0.12 ± 0.01
0.39 ± 0.01 0.44 ± 0.02 0.64 ± 0.01 0.12 ± 0.01 0.56 ± 0.02 0.38 ± 0.01
0.40 ± 0.01 0.47 ± 0.01 0.67 ± 0.01 0.12 ± 0.01 0.53 ± 0.01 0.34 ± 0.01
3.2. Performance of the internal standard procedure in real sample analysis The flow-batch set-up developed enabled to adjust the linear interval of calibration by simply reducing the sample volume aspirated through side port 4. When considering the maximum allowable aspirated volume of 3.15 mL (please, see Section 2.3) a linear calibration plot up to 1.40 mg L−1 of iron is yielded, corresponding to the equation AbsFe/Ni = (1.63 ± 0.10)Fe2+ − (0.008 ± 0.005) and R2 = 0.9996. The limit of detection (LD) and quantification (LQ) were estimated by Eqs. (1) and (2): √ 3 × MQr LD = (1) α √ 10 × MQr LQ = (2) α where α is the slope and MQr is the residual mean square of the analytical curve with internal calibration. The LD and LQ calculated were 0.04 and 0.12 mg L−1 , respectively. Eighteen HEF samples were randomly collected from gasoline service stations around the state of Pernambuco, and assayed using both the proposed procedure and the ABNT official method. Adulterated samples were further analyzed using the standard addition method. The results for the unadulterated samples (ethanol content from 92% to 95%) are presented in Table 2.
Table 2 Iron concentration in ethanol fuel unadulterated samples obtained by the proposed procedure and the ABNT official method Sample
A B C D E F G H I J L M
Iron concentration (mg L−1 ) Proposed procedure
ABNT official method
0.31 ± 0.01 0.61 ± 0.01 0.70 ± 0.01 0.65 ± 0.01 0.49 ± 0.03 1.58 ± 0.04 0.24 ± 0.02 0.19 ± 0.01 0.29 ± 0.01 0.51 ± 0.04 0.84 ± 0.01 1.20 ± 0.12
0.29 ± 0.01 0.56 ± 0.01 0.71 ± 0.01 0.57 ± 0.01 0.50 ± 0.01 1.53 ± 0.05 0.21 ± 0.01 0.20 ± 0.01 0.36 ± 0.02 0.50 ± 0.01 0.83 ± 0.01 1.08 ± 0.07
An assessment of the correlation of the results obtained using the proposed method with the values obtained using the official procedure yields the equation Yproposed = (1.06 ± 0.08)XABNT − (0.01 ± 0.06), R2 = 0.9891, which confirms the good agreement between the procedures, since the ideal values of zero intercept and unit slope fall within the respective confidence intervals. The mean coefficients of variation (calculated from 12 replicate samples) of the proposed method and the reference procedure were 3.8% and 2.3%, respectively. The mean relative error of the proposed procedure is 3.9%. For the adulterated samples (ethanol content from 70% to 84%), the results obtained are presented in Table 3. In this case an assessment of the correlation of the results obtained using the proposed method with the values obtained with the standard addition method (SAM) yields the equation Yproposed = (1.02 ± 0.21)XSAM − (0.01 ± 0.10), R2 = 0.9769, which also confirms the good agreement between the procedures. As expected, the ABNT official method failed to predict the iron content for samples adulterated by water addition. 4. Conclusions The proposed automated procedure, enabled operator free assays of iron content in hydrated ethanol fuel samples, thus minimizing the handling of potential hazardous materials. Coupling fast-sequential atomic absorption spectrometry, with internal standard calibration, overcomes usually observed instrumental drift and ethanol content variation in samples. Moreover, this procedure enables a simpler and more economical preparation of aqueous calibrating solutions instead of analytical grade ethanol calibrators. Finally, the use of a flow-batch set-up enables further implementation of other determination procedures by simple a computer change of the steps comprising the analytical cycle and the use of the detection equipment without having to disconnect the developed set-up. Acknowledgments The authors thank the FINEP-CTPETRO and CAPESGRICES for financial support. Research fellowships granted by the Brazilian agency CNPq is also gratefully acknowledged.
526
J.E. da Silva et al. / Talanta 70 (2006) 522–526
References [1] A.E. Wheals, L.C. Basso, D.M.G. Alves, H. Amorim, Focus 17 (1999) 482. [2] Agˆencia Nacional do Petr´oleo, Brasil. Portaria No. 126, 2002. ´ [3] Associac¸a˜ o Brasileira de Normas T´ecnicas (ABNT), NBR 11331. Alcool et´ılico–Determinac¸a˜ o do teor de ferro por espectrofotometria de absorc¸a˜ o atˆomica, 1990. [4] M.E. Cullum, M.H. Zile, Anal. Biochem. 153 (1986) 23. [5] B. Charles, S. Chulavatnatol, Biomed. Chromatogr. 7 (1993) 204. [6] A.S. Al-Ammar, R.K. Gupta, R.M. Barnes, Spectrochim. Acta Part B 54 (1999) 1849. [7] J.J. Thompson, R.S. Houk, Appl. Spectrosc. 41 (1987) 801. [8] A.S. Al-Ammar, R.M. Barnes, J. Anal. At. Spectrom. 16 (2001) 327. [9] E.J. dos Santos, A.B. Herrmann, V.L.A. Frescura, J. Anal. At. Spectrom. 20 (2005) 538. [10] B. Radziuk, N. Romanova, Y. Thomassen, Anal. Commun. 36 (1999) 13. [11] L.R.P. Butler, A. Strasheim, Spectrochim. Acta 21 (1965) 1207.
[12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
S.B. Smith, J.A. Blasi, F.J. Feldman, Anal. Chem. 40 (1968) 1525. T. Takada, K. Nakano, Anal. Chim. Acta 107 (1979) 129–136. T. Takada, K. Nakano, Spectrochim. Acta Part B 36 (1981) 735. F.J. Feldman, J.A. Blasi, S.B. Smith, Anal. Chem. 41 (1969) 1095. F.J. Feldman, Anal. Chem. 42 (1970) 719. I. Atsuya, K. Itoh, K. Ariu, Pure Appl. Chem. 63 (1991) 1121. A.P. Oliveira, J.A. Gomes Neto, J.A. N´obrega, P.V. Oliveira, Talanta 64 (2004) 334. H. Projahn, U. Steeg, J. Sanders, E. Vanclay, Anal. Bioanal. Chem. 378 (2004) 1083. F.A.C. Amorim, S.C. Ferreira, Talanta 65 (2005) 960. R.S. Honorato, M.C.U. Ara´ujo, R.A.C. Lima, E.A.G. Zagatto, R.A.S. Lapa, J.L.F.C. Lima, Anal. Chim. Acta 396 (1999) 91. R.S. Honorato, J.M.T. Carneiro, E.A.G. Zagatto, Anal. Chim. Acta 441 (2001) 309. W. Penninckx, D.L. Massart, P. Vankeerberghen, J. Smeyers-Verbeke, J. Anal. At. Spectrom. 10 (1995) 207.
A flow-batch internal standard procedure for iron determination in hydrated ethanol fuel by flame atomic absorption spectrometry Jos´e Edson da Silva a , F´abio Andr´e da Silva a , M. Fernanda Pimentel b,∗ , Ricardo Saldanha Honorato c , Valdinete Lins da Silva b , Maria da Conceic¸a˜ o B.S.M. Montenegro d , Alberto N. Ara´ujo d b
a Depto de Qu´ımica Fundamental (UFPE), Recife, Brazil Depto de Engenharia Qu´ımica (UFPE), Rua Prof. Arthur de S´a s/n, Cidade Universit´aria, 50740-521 Recife, Brazil c Departamento de Pol´ıcia Federal-SR/DPF/PE, Recife, Brazil d REQUIMTE/Depto de Qu´ımica-F´ısica, Fac. Farm´ acia (UP), Porto, Portugal
Received 21 September 2005; received in revised form 29 December 2005; accepted 29 December 2005 Available online 13 February 2006
Abstract A flow-batch manifold coupled to a flame atomic absorption spectrometer was evaluated to assess the iron content by the internal standard method in hydrated ethanol used as fuel in automotive industry. For this assessment official methods require calibration procedures with matrix matching, making it difficult to obtain accurate results for samples adulterated by the addition of water. Nickel was selected as the internal standard since it is usually absent in samples and because it requires similar conditions of atomization. After procedure optimization, which requires about 4.25 mL of sample and standard per measurement, it was possible to get linear analytical response for iron concentrations between 0.12 and 1.40 mg L−1 and a detection limit of 0.04 mg L−1 . Eighteen samples were collected randomly from fuel stations in Pernambuco (Brazil) and iron concentration was determined using the proposed procedure. Comparison of results obtained (0.20–1.50 mg L−1 ) showed a mean standard error of 3.9%, with 3.8% and 2.3% calculated for the mean variation coefficients of the proposed method and the reference procedure, respectively. For adulterated samples (0.12–0.64 mg L−1 ), the mean standard error was 4.8% when compared with the standard addition method. These results allowed concluding that the proposed procedure is adequate to accomplish the determination of iron in ethanol fuel in a large scale basis with a sampling rate of about 10 h−1 . © 2006 Elsevier B.V. All rights reserved. Keywords: Hydrated ethanol fuel; Iron; Internal standard calibration; Flame atomic absorption spectrometry
1. Introduction Since the first petrol crisis in the middle of the seventies, a number of countries have adopted policies to search for other more economical energy sources. This was the case of Brazil, which envisioned the use of ethanol obtained from sugar fermentation both for alternative fuel and to increase employment opportunities in the industry related to its production [1]. As a result vehicle engines solely using this type of fuel and more recently flexfuel engines have been developed at the same time as a widespread distribution of hydrated ethanol fuel (HEF)
∗
Corresponding author. Tel.: +55 81 21267291; fax: +55 81 21267235. E-mail address: [email protected] (M.F. Pimentel).
0039-9140/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2005.12.057
in street gas stations has been effected. An issue of particular concern with HEF usage is its contamination with iron during the distillation, storage and distribution processes. When iron is present above 5 mg L−1 , it induces the formation of deposits during the explosion stage, thus reducing the life of an engine using this fuel [2]. In face of this situation, the Brazilian normalizing committee, Associac¸a˜ o Brasileira de Normas e T´ecnicas (ABNT), has proposed an analytical procedure based on flame atomic absorption spectrometry which is based on the use of calibrating solutions prepared in ethanol 95% (v/v) solvent [3]. This simple procedure, however, does not take into account the usual variations in ethanol content of the fuel, which can be between 69% and 99%, thus causing significant accuracy problems in the results obtained. Technicians often resort to standard additions method to overcome this problem, albeit with the
J.E. da Silva et al. / Talanta 70 (2006) 522–526
drawbacks of increasing the hazards related with samples handling, being labor intensive and difficult to afford in large scale analysis. Procedures based on the use of internal standards are commonly used in chromatographic [4,5] and inductively coupled plasma atomic emission spectrometry to overcome biased results by the instrumental drift and correction for non-spectroscopic matrix effects [6]. This use has lately been extended to inductively coupled plasma-mass spectrometry (ICP-MS) to improve isotope ratio precision [7,8] and inductively coupled plasma optical emission spectrometry with axial detection in order to improve precision and minimize matrix effects, mainly due to other elements present in samples [9]. Internal standard procedures are based on sample spiking with a known amount of absent chemical species which physical–chemical behavior is similar to the analyte, being both measured preferably at the same time. Although it is impossible to fulfil this last condition with traditional instrumentation, use of an internal standard method has been reported since the very beginning of atomic absorption spectrometry, based on the development of instruments with multichannel capabilities [10–14]. Thus, procedures for the determination of Al, Ca, Cr, Cu, Fe, Mg, Mn, Ni, Si and Zn in cement, feeds and alloys by flame atomic absorption spectrometry (FAAS) were described to minimize the different nebulization yields observed in samples and calibration standards [15–17]. For the same reason, references to internal standard procedures in electrothermal atomization are also scarce, except concerning the use of cobalt in iron determinations [13], copper and indium for the determination of lead and selenium in water [17], tallium and bismuth for the assessment of lead in biological materials [10] and arsenium for the determination of selenium content in mineral waters [18]. In the present work, the internal standard method approach is once more revisited to assess iron levels in HEF. The reliability of the results provided is compared with the protocols under current use. Recently, flame atomic absorption spectrometers were placed on the market, enabling the exploitation of the different analytical wavelengths emitted by multi-element hollow cathode lamps or by several single-element lamps, thus providing short time sequential determinations of more than one element per assay [19,20]. In this way, internal standard calibration can be envisaged as a simple means to circumvent the previously cited problems, namely those related to sample matrix variability when the direct calibration protocol is used or with the standard additions method analytical rate. In order to provide safer sample handling and expeditious sample preparation and measurement, the application of a flow-batch manifold is extended to allow a fully automated procedure for iron determination in HEF. The flow-batch concept which has been recently proposed [21,22] is based on a programmed flow preparation of solutions into a small-volume open container. Due to the time required for signal acquisition at both wavelengths, this procedure enables the automated preparation of large volumes with homogeneous content in both the analyte and the internal standard.
523
2. Experimental 2.1. Reagents and solutions Analytical grade chemicals were used without further purification, and all solutions were prepared using double deionized water. Nickel and iron calibrating solutions were prepared by rigorous dilution of 1000 mg L−1 of standard stock solutions available on the market (Titrisol, Merck). The ethanolic standards were prepared according to the ABNT NBR 11331 procedure. The density of P.A. Ethanol (99.9%) was first adjusted with deionized water to 0.82 g mL−1 to obtain a 95% ethanol mixture, which was then used to prepare iron standards with 0.20, 0.50, 0.80, 1.10 and 1.40 mg L−1 , by dilution of a 100 mg L−1 aqueous iron solution. 2.2. Equipment The flow manifold (Fig. 1), was comprised of a peristaltic pump Ismatec IPC, model 78001-10, (Switzerland), a 10-ports rotary valve VICI Cheminert NCR 0138 (Valco, Houston), a NResearch 161 T031 three-way solenoid valve (Stow, MA), and a Varian Fast-Sequential, model 220 FS, atomic absorption spectrometer. Upon activation of the solenoid valve the detector aspirated the solutions under stirring placed by the manifold in a home-made open cylindrical PTFE mixing chamber possessing a cavity with a 2 cm internal diameter and 4 cm high in which a 1.2 mm flow path was drilled in the bottom. The different devices were connected using PTFE tubing with 1.2 mm i.d., having the length of 10 m for the holding coil, 40 cm for the transmission line between the rotary valve and the mixing chamber, 8 cm between the mixing chamber and the solenoid valve and further 30 cm up to the detector. To control the different devices home-made software was implemented on a microcomputer connected by means of a PCL 711S Advantech interface card. Signal acquisition was carried out by sending a start command to the SpectrAA software of the spectrometer.
Fig. 1. Flow-batch system scheme. FAAS-FS, fast-sequential atomic absorption spectrometer; PC, microcomputer; DW, deionized water; SV, three-way solenoid valve; MC, mixture chamber; AI, air input; S, sample; IS, internal standard; CS, calibration iron standard; RV, rotary valve; W, waste; HC, holding coil; PP, peristaltic pump.
524
J.E. da Silva et al. / Talanta 70 (2006) 522–526
2.3. Procedures
solution towards the mixing chamber, where it is stirred for 15 s before measurements, another air bubble is aspirated.
Atomic absorption determinations were performed sequentially, using a time window of 4 s, at the wavelengths of 284.3 nm for iron, 232.0 nm for nickel (bandwidth of 0.2 nm) and 228.8 nm for cadmium (bandwidth of 0.5 nm). Acetylene and air flows were settled to, respectively, 1.80 and 13.0 L min−1 . In these conditions an aspiration flow-rate of 9 and 10 mL min−1 into the pneumatic nebulizer was registered for water and ethanol (95%, v/v), respectively. In order to preserve the spectrometer usage either for conventional measurements as well for the automated set-up developed, a three-way solenoid valve was connected to the entrance tubing of the nebulizer, thus allowing conventional use when the valve was inactive, or aspiration of the mixing chamber content when activated during flow-batch measurements. To minimize the time spent in the aspiration and propelling of the different solutions through the flow-path of the manifold, previous experiments had been performed using water and ethanol, to verify the maximum allowable flow rates at which the driven volumes were independent from the physical characteristics of the solution. Finally, the flow-rates of 3.5 and 3.2 mL min−1 were settled for water and 95% (v/v) ethanolic containing solutions, respectively. Afterwards, the analytical cycle described in Table 1 was developed. Each analytical cycle started with the aspiration of an air bubble through port 2 of the rotary valve into the coil HC, thus avoiding samples or standards dispersion in the water used as carrier. Calibrating solutions were prepared in-line, by sequential aspiration into the coil HC of between 0.45 and 3.15 mL of a 2.00 mg L−1 iron solution presented at port 5 of the rotary valve, 0.25 mL of a 10 mg L−1 nickel solution available in port 4 and between 3.80 and 1.10 mL of water though port 6. To assess iron concentration in HEF samples the same analytical cycle was adopted (Table 1, steps 13–16). The volume of each aspired sample was 4.25 mL, to which it was added 0.25 mL of the internal standard. Before propelling the aspirated
Table 1 Flow-batch system analytical cycle for iron determination in ethanol fuel Step 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Event
Direction
Volume (mL)
Time (s)
Air Calibration standard Internal standard Deionized water Air Solutions to the chamber Stirring Signal acquisition Chamber cleasing with deionized water Chamber emptying Sample change Discharge Air Sample aspiration Internal standard Proceed to step 5
Aspiration Aspiration Aspiration Aspiration Aspiration Forward
– 0.45–3.15 0.25 3.80–1.10 – 4.50
1.5 7.7–54.0 4.3 20.0–65.0 1.5 80.0
– FAAS aspiration Forward
– – 4.50
15.0 37.5 80.0
FAAS aspiration Aspiration Forward Aspiration Aspiration Aspiration –
– –
51.5 5 5 1.5 79.7 4.3 –
– 4.25 0.25 –
3. Results and discussion Aiming to assess the influence of the ethanol content on the solutions aspirated into the atomic absorption spectrometer, different sets of iron calibrating solutions were prepared in an ethanol/water solvent ranging from 69% to 99% (v/v). As expected, significant (at the 95% confidence level) different linear calibration slopes were obtained, between 0.0239 ± 0.010 L mg−1 for the set with lowest ethanol content and 0.0207 ± 0.015 L mg−1 for the highest. This decrease of the slope values could be explained by the change of the flame atomization temperature resulting of the increase of the fuel:air stoichiometric ratio. The statistical analysis of the calibration data obtained [23] revealed, at the confidence level of 95%, a t-value of 2.15, below the theoretical value of 2.45, when the slopes obtained for 95% and 99% (v/v) ethanol/water were compared, indicating that for small variations of the ethanol content in samples the official standard calibration procedure was able to provide acceptable results. In spite of this, our experience shows that samples with ethanol content below 95% are very often encountered, thus requiring the development of an alternative and more robust procedure. 3.1. Internal standard selection In our previous experiments both nickel and cadmium had not been detected in the HEF samples and hence could be used to implement a new internal standard protocol. To select which of them was the more appropriate specie, the relative standard deviation of absorbencies ratio was considered. Therefore, samples with an iron concentration of 0.50 mg L−1 were prepared in water and 95% (v/v) ethanol/water solvents, spiked both with nickel and cadmium up to the concentration of 0.50 and 0.20 mg L−1 , respectively, using the flow-batch procedure described above. The relative standard deviation values estimated from four replicates for the iron/nickel ratios in water and in 95% (v/v) ethanol/water were 2.0% and 3.8%, respectively. For the iron/cadmium ratio, these values increased to 7.6% and 7.7%, respectively. In order to further assess if nickel was an appropriate choice, on-line prepared calibrating solutions with iron concentrations in the range up to 1.40 mg L−1 , in the presence or absence of 95% (v/v) ethanol were tested. Considering only the signals provided for iron, slopes of 0.0358 and 0.0213 L mg−1 were found for the calibration curves in water and 95% (v/v) ethanol, respectively. A t-test shows that these values were significantly different at the 95% confidence level. When the absorbance ratio was used as the dependent variable, the two slopes (2.08 L mg−1 for water and 2.12 L mg−1 for 95%, v/v, ethanol) were statistically indistinguishable. Within-batch reproducibity was also compared for the standard and the proposed internal standard procedure, considering a 9-h analytical run and one single initial calibration procedure. In this evaluation two real samples (content of ethanol of 95%, v/v) containing, respectively, 0.83 mg L−1 (S1) and 1.10 mg L−1 (S2) of iron
J.E. da Silva et al. / Talanta 70 (2006) 522–526
were processed in 30 min intervals, using the flow-batch set-up, for the addition of the internal standard to a fixed concentration of 0.50 mg L−1 . During the analytical run a continuous drift of instrumental conditions was registered, thus rendering results between 0.57 and 0.99 mg L−1 for sample S1 and from 0.79 to 1.37 mg L−1 for sample S2 (relative standard deviation of 13.1%), with the standard method based only on the analytical signals enabled by the iron content. Using to the internal standard additions method, the variation around the target value was reduced to 4.7% and 2.7%, respectively, thus compensating for the instrumental drifts.
525
Table 3 Iron concentration in ethanol fuel adulterated samples obtained by the proposed procedure and the ABNT official method Sample
N O P Q R S
Iron concentration (mg L−1 ) Ethanol content (%)
ABNT official method
Standard addition method
Proposed procedure
84 82 77 75 72 70
0.39 ± 0.01 0.34 ± 0.01 0.54 ± 0.02 0.02 ± 0.01 0.50 ± 0.01 0.12 ± 0.01
0.39 ± 0.01 0.44 ± 0.02 0.64 ± 0.01 0.12 ± 0.01 0.56 ± 0.02 0.38 ± 0.01
0.40 ± 0.01 0.47 ± 0.01 0.67 ± 0.01 0.12 ± 0.01 0.53 ± 0.01 0.34 ± 0.01
3.2. Performance of the internal standard procedure in real sample analysis The flow-batch set-up developed enabled to adjust the linear interval of calibration by simply reducing the sample volume aspirated through side port 4. When considering the maximum allowable aspirated volume of 3.15 mL (please, see Section 2.3) a linear calibration plot up to 1.40 mg L−1 of iron is yielded, corresponding to the equation AbsFe/Ni = (1.63 ± 0.10)Fe2+ − (0.008 ± 0.005) and R2 = 0.9996. The limit of detection (LD) and quantification (LQ) were estimated by Eqs. (1) and (2): √ 3 × MQr LD = (1) α √ 10 × MQr LQ = (2) α where α is the slope and MQr is the residual mean square of the analytical curve with internal calibration. The LD and LQ calculated were 0.04 and 0.12 mg L−1 , respectively. Eighteen HEF samples were randomly collected from gasoline service stations around the state of Pernambuco, and assayed using both the proposed procedure and the ABNT official method. Adulterated samples were further analyzed using the standard addition method. The results for the unadulterated samples (ethanol content from 92% to 95%) are presented in Table 2.
Table 2 Iron concentration in ethanol fuel unadulterated samples obtained by the proposed procedure and the ABNT official method Sample
A B C D E F G H I J L M
Iron concentration (mg L−1 ) Proposed procedure
ABNT official method
0.31 ± 0.01 0.61 ± 0.01 0.70 ± 0.01 0.65 ± 0.01 0.49 ± 0.03 1.58 ± 0.04 0.24 ± 0.02 0.19 ± 0.01 0.29 ± 0.01 0.51 ± 0.04 0.84 ± 0.01 1.20 ± 0.12
0.29 ± 0.01 0.56 ± 0.01 0.71 ± 0.01 0.57 ± 0.01 0.50 ± 0.01 1.53 ± 0.05 0.21 ± 0.01 0.20 ± 0.01 0.36 ± 0.02 0.50 ± 0.01 0.83 ± 0.01 1.08 ± 0.07
An assessment of the correlation of the results obtained using the proposed method with the values obtained using the official procedure yields the equation Yproposed = (1.06 ± 0.08)XABNT − (0.01 ± 0.06), R2 = 0.9891, which confirms the good agreement between the procedures, since the ideal values of zero intercept and unit slope fall within the respective confidence intervals. The mean coefficients of variation (calculated from 12 replicate samples) of the proposed method and the reference procedure were 3.8% and 2.3%, respectively. The mean relative error of the proposed procedure is 3.9%. For the adulterated samples (ethanol content from 70% to 84%), the results obtained are presented in Table 3. In this case an assessment of the correlation of the results obtained using the proposed method with the values obtained with the standard addition method (SAM) yields the equation Yproposed = (1.02 ± 0.21)XSAM − (0.01 ± 0.10), R2 = 0.9769, which also confirms the good agreement between the procedures. As expected, the ABNT official method failed to predict the iron content for samples adulterated by water addition. 4. Conclusions The proposed automated procedure, enabled operator free assays of iron content in hydrated ethanol fuel samples, thus minimizing the handling of potential hazardous materials. Coupling fast-sequential atomic absorption spectrometry, with internal standard calibration, overcomes usually observed instrumental drift and ethanol content variation in samples. Moreover, this procedure enables a simpler and more economical preparation of aqueous calibrating solutions instead of analytical grade ethanol calibrators. Finally, the use of a flow-batch set-up enables further implementation of other determination procedures by simple a computer change of the steps comprising the analytical cycle and the use of the detection equipment without having to disconnect the developed set-up. Acknowledgments The authors thank the FINEP-CTPETRO and CAPESGRICES for financial support. Research fellowships granted by the Brazilian agency CNPq is also gratefully acknowledged.
526
J.E. da Silva et al. / Talanta 70 (2006) 522–526
References [1] A.E. Wheals, L.C. Basso, D.M.G. Alves, H. Amorim, Focus 17 (1999) 482. [2] Agˆencia Nacional do Petr´oleo, Brasil. Portaria No. 126, 2002. ´ [3] Associac¸a˜ o Brasileira de Normas T´ecnicas (ABNT), NBR 11331. Alcool et´ılico–Determinac¸a˜ o do teor de ferro por espectrofotometria de absorc¸a˜ o atˆomica, 1990. [4] M.E. Cullum, M.H. Zile, Anal. Biochem. 153 (1986) 23. [5] B. Charles, S. Chulavatnatol, Biomed. Chromatogr. 7 (1993) 204. [6] A.S. Al-Ammar, R.K. Gupta, R.M. Barnes, Spectrochim. Acta Part B 54 (1999) 1849. [7] J.J. Thompson, R.S. Houk, Appl. Spectrosc. 41 (1987) 801. [8] A.S. Al-Ammar, R.M. Barnes, J. Anal. At. Spectrom. 16 (2001) 327. [9] E.J. dos Santos, A.B. Herrmann, V.L.A. Frescura, J. Anal. At. Spectrom. 20 (2005) 538. [10] B. Radziuk, N. Romanova, Y. Thomassen, Anal. Commun. 36 (1999) 13. [11] L.R.P. Butler, A. Strasheim, Spectrochim. Acta 21 (1965) 1207.
[12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
S.B. Smith, J.A. Blasi, F.J. Feldman, Anal. Chem. 40 (1968) 1525. T. Takada, K. Nakano, Anal. Chim. Acta 107 (1979) 129–136. T. Takada, K. Nakano, Spectrochim. Acta Part B 36 (1981) 735. F.J. Feldman, J.A. Blasi, S.B. Smith, Anal. Chem. 41 (1969) 1095. F.J. Feldman, Anal. Chem. 42 (1970) 719. I. Atsuya, K. Itoh, K. Ariu, Pure Appl. Chem. 63 (1991) 1121. A.P. Oliveira, J.A. Gomes Neto, J.A. N´obrega, P.V. Oliveira, Talanta 64 (2004) 334. H. Projahn, U. Steeg, J. Sanders, E. Vanclay, Anal. Bioanal. Chem. 378 (2004) 1083. F.A.C. Amorim, S.C. Ferreira, Talanta 65 (2005) 960. R.S. Honorato, M.C.U. Ara´ujo, R.A.C. Lima, E.A.G. Zagatto, R.A.S. Lapa, J.L.F.C. Lima, Anal. Chim. Acta 396 (1999) 91. R.S. Honorato, J.M.T. Carneiro, E.A.G. Zagatto, Anal. Chim. Acta 441 (2001) 309. W. Penninckx, D.L. Massart, P. Vankeerberghen, J. Smeyers-Verbeke, J. Anal. At. Spectrom. 10 (1995) 207.