- Email: [email protected]
derivatization for the determination of transition metal cations in natural water samples
Talanta 136 (2015) 75–83
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Sequential Injection Chromatography with post-column reaction/derivatization for the determination of transition metal cations in natural water samples Burkhard Horstkotte n, Patrícia Jarošová, Petr Chocholouš, Hana Sklenářová, Petr Solich Department of Analytical Chemistry, Faculty of Pharmacy, Charles University in Prague, Heyrovského 1203, 500 05 Hradec Králové, Czech Republic
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2014 Received in revised form 27 December 2014 Accepted 3 January 2015 Available online 12 January 2015
In this work, the applicability of Sequential Injection Chromatography for the determination of transition metals in water is evaluated for the separation of copper(II), zinc(II), and iron(II) cations. Separations were performed using a Dionex IonPACTM guard column (50 mm 2 mm i.d., 9 mm). Mobile phase composition and post-column reaction were optimized by modified SIMPLEX method with subsequent study of the concentration of each component. The mobile phase consisted of 2,6pyridinedicarboxylic acid as analyte-selective compound, sodium sulfate, and formic acid/sodium formate buffer. Post-column addition of 4-(2-pyridylazo)resorcinol was carried out for spectrophotometric detection of the analytes' complexes at 530 nm. Approaches to achieve higher robustness, baseline stability, and detection sensitivity by on-column stacking of the analytes and initial gradient implementation as well as air-cushion pressure damping for post-column reagent addition were studied. The method allowed the rapid separation of copper(II), zinc(II), and iron(II) within 6.5 min including pump refilling and aspiration of sample and 1 mmol HNO3 for analyte stacking on the separation column. High sensitivity was achieved applying an injection volume of up to 90 mL. A signal repeatability of o 2% RSD of peak height was found. Analyte recovery evaluated by spiking of different natural water samples was well suited for routine analysis with sub-micromolar limits of detection. & 2015 Elsevier B.V. All rights reserved.
Keywords: Sequential Injection Chromatography Transition metal cations Post-column reaction Spectrophotometric detection On-column stacking
1. Introduction Non-separating flow techniques (FT) are powerful tools for the automation of analytical procedures, especially where high sample throughput, high repeatability, and the avoidance of sample contamination during processing are very important such as in ambient monitoring of trace compounds. This is due to the (computer-)controlled handling of all solutions in a closed tubing network or “manifold”, continuous/discontinuous system cleaning by a carrier flow, and precise timing between all procedure steps such as mixing and detection. With respect to solution handling, mixing conditions, flow pattern, and system configuration, different non-separating FT can be distinguished, with Flow Injection Analysis (FIA) [1], Sequential Injection Analysis (SIA) [2], and Multi-Syringe Flow Injection Analysis (MSFIA) [3] being the ones with the highest impact on this work. The combination of FT for sample handling and preparation with liquid chromatography (LC) is advantageous because despite all its n
Corresponding author. Tel.: þ 420 495 067 504; fax: þ 420 495 067 164. E-mail address: [email protected] (B. Horstkotte).
http://dx.doi.org/10.1016/j.talanta.2015.01.001 0039-9140/& 2015 Elsevier B.V. All rights reserved.
potentials, the most definite disadvantage of all FT is the difficulty of differentiation or gradual separation of analytes. Generally, selective reagents and detection techniques, interference masking, or observance of the reaction kinetic allow discrimination of the analyte from sample matrix components or enable multicomponent analysis by changing the reagent or the reaction conditions appropriately. Further separations, such as gas diffusion or extraction, allow enhancing the method's selectivity. However, without coupling FT with chromatographic or electrophoretic separations, mono- or oligo-analyte methods with potential crosssensitivity to other compounds are typical [4,5]. In this context, the development of monolithic columns [6] and their integration into FT analyzer systems allowing low-pressure chromatographic separations [7] was a milestone for the development of more versatile analytical procedures based on FT. Consequently, the combination of the different FT led to the proposal of Sequential Injection Chromatography (SIC) [8] and later Multi-Syringe Chromatography (MSC) [9] and Flow Injection Chromatography (FIC) [10]. The separation performance even using gradient approaches [11] is similar or inferior to HPLC, mainly because the applicable pressure and column length in SIC are limited. The main
76
B. Horstkotte et al. / Talanta 136 (2015) 75–83
advantages of SIC and the related technique are the possibility of automated in-system sample treatment such as sample clean-up [12], the flexibility of operation such as the use of reagents for analyte derivatization or on-column stacking, the ease of method set-up, and the lower cost and transportability of the analyzer system enabling its use for monitoring and field operation. Applications of SIC and related techniques used mostly commercial silica-based reverse-phase monolithic columns and were focused mainly on the separation of organic analytes. Lately, monolithic columns available with new functionalities and core-shell columns [13] are opening the field of application and operation of SIC [14]. Two comprehensive reviews can be found elsewhere [15,16]. While separation of metal cations with post-column addition of either chromogenic or chemiluminescence-generating reagents are state-of-the-art in HPLC [17,18], in FT, multi-parameter methods are mainly based on the use of ion-selective reagents. Due to the lack of commercial monolithic columns with cationexchange functionality, either dynamic coating of reverse phases [19] or permanent modification [20] have been used to enable cation separation. In the present work, the potential and application of a crosslinked divinylbenzene stationary phase modified by a bilayer of anion- and cation-exchange functionalized latex for low-pressure separation of transition cations in SIC are studied. As an alternative to monolithic columns, only the guard column of an HPLC column assembly [21] was used and rapid and low-pressure separation of copper(II), zinc(II), and iron(II) was achieved. Firstly in SIC, post-column addition of a chromogenic reagent was applied to enable selective analyte detection [22]. For this reason, an additional syringe pump was used to provide a confluent flow of an alkaline solution of 4-(2-pyridylazo)resorcinol (PAR) that is typically used as a chromogenic reagent for sensitive detection of transition metal cations by the formation of red-colored complexes [17,23–25].
2. Material and methods 2.1. Reagents All solutions were prepared with demineralized quality water (resistivity418 MΩ cm) and all reagents were of quality “for analysis” and purchased from Sigma Aldrich (Prague, Czech Republic).
For the preparation of the chromogenic reagent, the following aqueous stock solutions were prepared: 50 mmol L 1 of 4-(2-pyridylazo)resorcinol (PAR) in 100 mmol L 1 NaOH, 4.0 mol L 1 ammonium hydroxide, and 1.0 mol L 1 NaHCO3. For the preparation of the mobile phase, the following stock solutions were prepared: 250 mmol L 1 2,6-pyridinedicarboxylic acid (PDCA) in 750 mmol L 1 NaOH, 2 mol L 1 formic acid, and 500 mmol L 1 Na2SO4. In addition, 2.0 mol L 1 NaOH was used for pH adjustment of mobile phase and post-column reagent. Aliquots of the stock solutions of PAR and PDCA were stored in the dark at 18 1C until use. All working solutions were prepared daily. Further stock solutions of the transition metal cations of interest, namely Co(II), Cu(II), Fe(II), and Zn(II) of 1 mmol L 1, were prepared in 10 mmol L 1 nitric acid. Nitrate salts were used for Co(II) and Zn (II) and sulfate salts for Cu(II), and Fe(II). Standard solutions were prepared daily with 4 mmol L 1 of nitric acid. Freshwater samples were used to evaluate the applicability of the method. They were acidified with a surplus of nitric acid (12 mmol L 1) to overcome the buffer capacity of dissolved bicarbonates and carbonates, achieving a final pH of about 2 being similar to the standard solutions. The samples were used unfiltered but after settling of sediments overnight.
2.2. Instrumentation The analyzer configuration is schematically shown in Fig. 1. All experiments were carried out on a Sequential Injection Chromatography system (SIChrom™) from FIAlab company (Bellevue, WA, USA). It integrated a medium pressure, biocompatible piston pump (PP) from Sapphire Engineering (IDEX Corporation, Oak Harbor, WA, USA) of 4 mL total dispense volume (3 inlets) and an eightport selection valve (SV1, type Cheminert, 12U-0484H, 5000 psi) from VICI (Valco Instruments Co. Inc., Houston, TX, USA). A second selection valve of the same type (SV2) and an additional syringe pump (SP, type Cavro XL) from Tecan (San Jose, CA, USA) equipped with a 2.5 mL glass syringe were used to enable piston pump refilling and addition of the post-column reagent, respectively. All instrumentation was controlled via RS232 interface. The second outlet of the PP was connected to a pressure safety valve (PSV) of 500 psi release threshold.
Fig. 1. Scheme of the SIC analyzer system. D: detection flow cell, HC1 and HC2: holding coils (150 cm, 0.8 mm i.d. and 30 cm 0.5 mm i.d., PEEK), PP: piston pump, PSV: pressure safety valve, SP: syringe pump with air bubble, SV1 and SV2: selection valves, V: syringe head valve, X: confluence (T-connector). Dimensions of PEEK tubing with 0.25 mm i.d. A: 25 cm, B: 5 cm, C: 15 cm, E1: 5 cm. Dimensions of FEP tubing (0.8 mm i.d.), E2 & I: 100 cm, G: 30 cm, H: 30 cm. Detail drawing: detection flow cell with flow inlet (L) and outlet (M) and optical fiber connections (N & O).
B. Horstkotte et al. / Talanta 136 (2015) 75–83
Lateral ports of SV1 were connected to the separation column, sample reservoir, 1 mmol L 1 nitric acid, and to SV2. SV2 allowed the selection of different mobile phases from lateral ports 2 to 7. On port 8, 50 mmol L 1 sodium hydroxide solution was placed and used for column cleaning and storage. The central ports of both SV were connected to the PP. On both SV, one lateral port was permanently closed (port 7 on SV1, port 1 on SV2). This configuration enabled refilling the PP from SV2 with SV1 in position 7, so that study of the mobile phase composition was possible by placing different mobile phases on each port of SV2. For all other flow operations performed on SV1, SV2 was in the closed position 1. Tubing connections for solution supply were made of 0.8 mm i.d. PTFE tubing. For pressure application, PEEK tubing of different diameters was used. The holding coil consisted of a 150 cm PEEK tube of 0.8 mm i.d. (HC1, 750 mL). A second 30 cm PEEK tube of 0.5 mm i.d. (HC2, 60 mL) was used to reduce sample dispersion during aspiration and injection. Connections between SV1, column, post-column reagent confluence, and detector were made as short as possible and of 0.25 mm i.d. PEEK tubing. All ferrules and fittings used for the chromatographic part were made of PEEK, while TEFZEL and Acetal pieces were used for low-pressure connections. A miniature fiber-optic spectrophotometer from OceanOptics Inc. (Dunedin, FL, USA), type USB4000, was used for detection. A bright white LED, mounted onto an optical fiber SMA connector (FIAlab), was used as a light source. Both the light source and the spectrophotometer were connected via 0.6 mm i.d. optical fibers from FIAlabs to a micro-volume detection flow cell (shown in Fig. 1 in detail) with 1 cm optical path length made of UltemTM polymer from the same company. 2.3. Data acquisition and evaluation Initially, detection wavelengths of 510 nm, 520 nm, and 530 nm were simultaneously recorded and used as analytical wavelengths [24–26]. Here, the red-colored complexes of the analytes and PAR were measured against a reference wavelength of 560 nm where neither the reagent nor the analyte–PAR complex showed any significant absorbance. The high light intensity of the LED allowed the detector an integration time as short as 6 ms, averaging over 30 single measurements and an effective measuring frequency of 3 Hz. Spectral smoothing over 9 pixels of the CCD array sensor was carried out. Peak height values were used for data evaluation, throughout. Instrumentation control as well as data evaluation was carried out using the FIAlab software, version 5.9.321, provided with the SIC instrument. The software allowed simultaneous measurement at four different wavelengths, control of procedures, variables, loops, and if-then instructions. This gave high versatility for automated optimization and maintenance. Posterior data smoothing (SavizkyGoley, 2-order polynomial, 11 points) and data evaluation were managed using MS-Excel. 2.4. Separation column and post-column reagent addition All separations were performed using a Dionex IonPACTM CG5A guard column 50 mmx2 mm i.d.) from Thermo Fischer Scientific company (www.thermoscientific.com) [27]. The stationary phase consisted of 9 mm cross-linked DVB (55%) particles with a bilayer of anion-exchange and cation-exchange functionalized latex. The column was stored in 50 mmol L 1 NaOH if not used for more than 1 day. The column inlet was connected to the lateral port 3 of SV1 and outlet downstream to a confluence (X, T-connector, 0.5 mm i.d.) of minimal dead volume for post-column addition of the chromogenic reagent provided by SP. A PEEK tube for mixing the mobile phase and the reagent was placed between the confluence and the
77
detection flow cell (15 cm, 0.25 mm i.d.). In order to avoid bubble formation by outgassing of the solution in the detection cell, a short PEEK tube of 5 cm, 0.25 mm i.d., was placed downstream as a restrictor. 2.5. Operation protocol All dispense steps performed through the column with the PP, i.e. applying significant pressure, were followed by a waiting time of 4 s to release the pressure before changing the position of the selection valves. The operation protocol is given as Supplementary material 1. The analytical method started with simultaneous filling of both the PP and the SP with mobile phase (SV2, port 2) and postcolumn reagent, respectively. The PP was filled with 2.67 mL of mobile phase and at high flow rate in order to provoke partly degassing of the mobile phase in the piston pump void. After pressure release, possibly formed bubbles inside the pump or the aspiration tube were expulsed via the mobile phase reservoir. Then, 300 mL of 1 mmol L 1 nitric acid was aspirated (SV1, port 2) by the PP and 150 mL was pushed into the column (SV1, port 3). Then, another 150 mL of 1 mmol L 1 nitric acid followed by the sample volume to be injected was aspirated. By dispensing in total 2500 mL through the column at a flow rate of 480 mL min 1, the sample was injected and chromatographic separation was performed. The first volume of diluted nitric acid was used to improve initial analyte stacking on the column while the second volume was intended to create an initial gradient and flush out the sample matrix, i.e. inorganic salts. After an operation time of the PP of 60 s, post-column reagent addition was started. After a stabilization time of 10 s, blank measurement was performed and data acquisition was performed over 245 s. All separations were carried out in triplicate. Additional procedures were used for system maintenance such as cleaning the tubes to the different solution reservoirs at both selection valves, cleaning the PP and SP, and flushing the column for cleaning, equilibration, or storage.
3. Results and discussion 3.1. Optimization of the analyzer configuration Because continuous degassing of the mobile phase and other solutions was not feasible, the piston pump was placed slightly inclined (ca. 101) as shown in Fig. 1 in order to avoid any air bubble aspirated from SV2 from entering the holding coil and later the column. Ultrasonic degassing of the mobile phase was tested prior to its usage but was found not to improve the separation performance significantly. During the optimization of the mobile phase composition, SV2 was used as autosampler, which allowed to test up to 7 different mobile phases in stand-alone operation. Since the pressure to yield the recommended flow rate of 1200 mL min 1 was not applicable with the used instrumentation, the highest possible flow rate without overcoming 500 psi, i.e. the break-through pressure of the safety restrictor, was 480 mL min 1. At lower flow rates of the SP than 480 mL min 1, significant baseline noise was obtained. This is because the post-column reagent itself adsorbs, although slightly, on the analytical wavelength. At very low flow rates of the SP, the step-wise rotation of the driving stepmotor becomes notable, i.e. the flow shows a slight pulsation. This fluctuation in the flow rate and thus of the mixing ratio of the mobile phase and the post-column reagent becomes notable. To improve the baseline noise, a flow rate of 480 mL min 1 was applied for the SP as soon it was found that not more than 2500 mL
78
B. Horstkotte et al. / Talanta 136 (2015) 75–83
of the mobile phase was needed for the separation. From the tested approaches to further reduce the SP pulsation, the best results (baseline noise reduction by 25%) were obtained when the SP was placed upside down, allowing a permanent air cushion to be present inside the syringe and act as pressure damper. An additional advantage was that any air bubble arising from reagent degassing would remain in the syringe pump and not be expulsed towards the detection cell.
3.2. Optimization of mobile phase composition Mobile phase optimization was carried out in two steps. First, optimization of the mobile phase composition proposed in the technical note from the column producer [28] was done by the modified SIMPLEX method [29–31], which included the following parameters: concentrations of PDCA, formic acid, and Na2SO4 and the pH value. Second, univariant studies of the individual parameters were done to verify and optionally to improve the achieved results. PDCA was chosen as the selective component in the mobile phase over the alternative use of oxalate [23,24], which is known to lead to an accumulation of Fe(II) on the column [28]. The optimization was done using the recommended post-column chromogenic reagent. Since only the guard column was used (i.e. reduction of column length by factor of 6 compared to the technical note), optimization of peak separation was the main objective during mobile phase optimization. Thus, the Simplex optimization of peak resolution as desirability function was carried out. For this, Cu(II) and Co(II) were used as analytes since even with the recommended mobile phase, partial peak resolution was achieved. The results and conditions are given in Table 1. After the initial 5 vertices, the Simplex tended immediately towards lower pH values losing this dimension due to repeated reduction and the prior fixed pH threshold. Furthermore, it was found that reduced Na2SO4 and PDCA content yielded higher peak resolution. Since vertex 12 was identical to vertex 9, the optimization was stopped after 12 cycles. The vertex yielding the best result and the finally applied conditions for univariant study of the formic acid concentration are bolded. For univariant studies, Fe(II) was used in addition to Cu(II) and Co(II), since it showed the longest retention time of the analytes of Table 1 Simplex optimization of the mobile phase composition*. Conditions: chromogenic reagent: 75 mmol L 1 NaHCO3, 126 mmol L 1 NH4OH, 50 mmol L 1 NaOH, 0.5 mmol L 1 PAR. Injection: 10 ml of Cu and Co, each 10 mmol L 1 in 4 mmol L 1 HNO3. Detection at 510 nm against 560 nm. No.
PDCA [mmol L 1]
Formic acid [mmol L 1]
pH
Sodium sulfate [mmol L 1]
RS
1 2 3 4 5 6 7 8 9 10 11 12
4.00 5.85 4.44 4.44 4.44 4.88 5.12 5.24 4.28 4.76 4.84 4.28
50.0 55.5 73.1 55.5 55.5 61.3 64.5 66.2 66.2 60.2 63.4 66.2
3.8 4.2 4.2 5.6 5.6 3.3 3.4 3.3 3.2 3.3 3.3 3.2
4.00 4.44 4.44 4.44 5.85 4.74 3.04 1.38 3.47 3.47 3.50 3.47
1.38 1.19 1.24 0.60 0.95 2.11 2.52 2.52 1.59 2.29 1.59 1.59
Finally 5.00 applied
60.0
3.3
3.00
*Recommended composition for 30 cm column assembly [28]: 7.0 mmol L 1 PDCA, 74 mmol L 1 formic acid, 5.6 mmol L 1 potassium sulfate, 66 mmol L 1 potassium hydroxide.
interest. At first, the influence of the formic acid content on peak resolution was studied in the range from 30 to 70 mmol L 1. The results and conditions are given in Table 2. The effect on the resolution of the Cu(II) and Co(II) peaks was low while the resolution of the Co(II) and the Fe(II) decreased considerably with higher formic acid concentration. This was because the retention of cations with weak complex formation with PDCA such as Fe(II) depends strongly on the salt content in the mobile phase. On the other hand, the peak height of Cu(II) decreased with higher concentration of formic acid in the mobile phase, which is most likely related to the altered mobile phase buffer capacity and thus suboptimal pH for the addition of post-column reagent under these conditions. Finally, a formic acid content of 30 mmol L 1 was selected for all following work. Second, the influence of the content of sodium sulfate on the elution was studied in the range from 10 to 1 mmol L 1. Other salts as additive to the mobile phase were not tested. The optimization was started with the highest Na2SO4 concentration. The results and conditions are given in Table 3. With lower Na2SO4 content, the retention times and peak resolution increased considerably while peaks became lower and wider. This was due to a very low release rate of the retained analytes and longer time of separation, which favored longitudinal diffusion inside the column (data not shown). For 1 mmol of Na2SO4, the eluent strength was insufficient for the elution of both Co(II) and Fe(II) in one run. Since peak resolution was the main criterion of the optimization, 3 mmol L 1 of sodium sulfate was chosen. Third, the influence of the content of PDCA on the elution was studied in the range of 7–1 mmol L 1. The conditions and results are given in Table 4. The peak resolution of copper and iron and the retention times increased considerably and in approximation linearly with lower PDCA content. On the other hand, the peak heights were Table 2 Univariant study of the effect of formic acid concentration on peak resolution. Conditions: chromogenic reagent: 75 mmol L 1 NaHCO3, 126 mmol L 1 NH4OH, 50 mmol L 1 NaOH, 0.25 mmol L 1 PAR. Mobile phase: 5 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Injection: 10 ml of Cu, Co, and Fe(II) of 10, 10, and 20 mmol L 1, respectively, in 4 mmol L 1 HNO3. Detection at 510 nm against 560 nm. Formic acid [mmol L 1]
30 40 50 60 70 80
Peak height [mAU] Mean (n ¼3) 7 SD
Resolution Retention time [s]
Cu(II)
Co(II)
Fe(II)
Cu/ Co
Co/ Fe
Cu (II)
Co (II)
Fe(II)
727 10 617 4 587 4 617 1 557 8 587 7
22 70.2 18 73 24 72 21 71 18 71 23 74
337 2 357 4 357 4 327 4 317 5 367 3
2.26 2.43 2.43 2.42 2.45 2.10
2.31 2.19 1.88 1.63 1.63 1.41
80.8 88.8 83.2 84.5 70.0 73.3
122.0 134.1 126.4 127.7 108.2 109.9
172.4 173.2 160.8 158.0 135.6 136.5
Table 3 Univariant study of the effect of sodium sulfate concentration on peak resolution. Conditions as in Table 2 with 40 mmol L 1 formic acid and 4 mmol L 1 PDCA, pH 3.4. Na2SO4 [mmol L 1]
10 7 4 2 1
Peak height [mAU] Mean (n¼ 3) 7 SD
Resolution
Retention time [s]
Cu(II)
Co(II)
Fe(II)
Cu/ Co
Co/ Fe
Cu (II)
Co (II)
Fe(II)
1077 7 797 1 677 1 617 1 397 1
407 9 307 1 247 2 227 2
697 10 547 1 407 2 287 1
1.16 1.63 2.31 2.86
1.30 1.32 1.80 2.46
45.2 58.9 82.0 122.3 185.1
63.9 84.7 120.5 183.1
84.3 110.1 156.9 234.4
B. Horstkotte et al. / Talanta 136 (2015) 75–83
found maximal between 4 and 5 mmol L 1 and decreased with higher PDCA concentration. This later observation was found due to the competition between PDCA and the post-column reagent PAR, both forming complexes with the metal cations. For 1 mmol L 1, elution of Fe(II) could not be observed, taking into account that the maximum elution volume applicable with the piston pump in one stroke was 3.8 mL. As a compromise of time of analysis and peak resolution, a PDCA concentration of 4 mmol L 1 was chosen. Finally, lower pH values of the mobile phase than the original Simplex threshold were tested. To adjust pH values lower than the pKa of formic acid down to pH 2.5, nitric acid was added to compensate the hydroxide ions originating from the PDCA stock solution. It was observed that the retention times increased but no significant improvement was achieved for either peak heights or peak resolution (data not given).
3.3. Optimization of post-column reagent Once the mobile phase was optimized, the composition of the post-column reagent was studied. This task was reasonable because a more diluted and about two pH units more acidic mobile phase Table 4 Univariant study of the effect of PCDA concentration on peak resolution. Conditions as in Table 2 but with 40 mmol L 1 formic acid and 3 mmol L 1 sodium sulfate, pH 3.4. PDCA [mmol L 1]
7 6 5 4 3 2 1
Peak height [mAU] Mean (n¼ 3) 7 SD
Resolution
Cu(II)
Cu(II)/Co Co(II)/Fe Cu (II) (II) (II)
Co (II)
Fe(II)
2.46 2.93 2.90 2.79 3.06 3.15 3.30
104.1 114.1 120.1 120.3 135.1 145.9 166.7
140.0 150.3 155.1 148.5 167.9 197.5
467 4 567 2 647 6 647 5 607 1 537 3 467 4
Co(II)
12 7 3 29 7 1 32 7 5 28 7 5 23 7 1 22 7 2 23 7 5
Fe(II)
22 73 25 72 27 72 33 72 31 71 32 78
Retention time [s]
79
compared to the recommended mobile phase [28] was found to be optimal due to the shorter separation column used in this work. Maintaining the concentration of ammonium hydroxide constant at 150 mmol L 1, an experiment was designed for testing
Table 5 Study of the effect of sodium bicarbonate and sodium hydroxide concentration in the chromogenic reagent on peak height. Conditions: mobile phase: with 40 mmol L 1 formic acid, 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Other components in reagent: 150 mmol L 1 NH4OH and 0.15 mmol L 1 PAR. Injection: 30 mL of Cu and Co 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. NaOH [mmol L 1]
0 15 30 45 60 75 90
NaHCO3 [mmol L 1]
150 130 110 90 70 50 30
pH
Peak height [mAU] Mean (n¼3)7SD
9.29 9.39 9.50 9.63 9.77 9.94 10.2
Cu
Co
46 72 54 71 56 74 59 71 61 71 65 75 72 79
117 0.3 137 1 147 1 167 0.2 187 4 177 3 187 3
Baseline noise [mAU]
1.6 1.4 1.1 2.5 9.5 12.8 35.1
Table 6 Study of the effect of PAR concentration in the post-column reagent. Conditions: mobile phase: 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Other components in reagent: 150 mmol L 1 NH4OH, 90 mmol L 1 NaHCO3, 45 mmol L 1 NaOH. Injection: 30 ml of Cu and Co 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. PAR [mmol L 1] Peak height [mAU] Mean (n¼ 3) 7SD Baseline noise [mAU]
2.18 2.55 2.32 2.30 2.07 2.55
71.3 78.8 82.0 86.8 90.8 98.3 104.9
0.05 0.10 0.15 0.20 0.25 0.30 0.40
Cu
Co
407 1 457 2 447 0.4 477 1 547 1 677 1 68 7 1
87 0.5 117 0.8 107 0.3 117 0.6 147 0.9 217 0.4 277 0.6
0.7 0.9 1.5 2.3 2.5 2.3 3.4
Fig. 2. Study of the effect of the injection volume on peak height and peak resolution (RS) using a standard of each 10 mmol L 1 Cu(II) and Co(II). Conditions: mobile phase: 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Reagent: 0.3 mmol L 1 PAR, 150 mmol L 1 NH4OH, 90 mmol L 1 NaHCO3, and 45 mmol L 1 NaOH. Injection: 30 ml of Cu(II) and Co(II) 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. Below: scheme of injection tested modes.
80
B. Horstkotte et al. / Talanta 136 (2015) 75–83
different combinations of sodium hydroxide and sodium bicarbonate concentrations. The conditions and results are given in Table 5. The signal height for Cu(II) and Co(II) peaks increased with higher pH value; however, also the background signal of the reagent increased significantly, as indicated by a higher baseline noise. As a compromise, 45 mmol L 1 NaOH and 90 mmol L 1 NaHCO3 were then chosen for further work. Finally, the effect of the concentration of PAR in the postcolumn reagent was studied in the range of 0.05–0.4 mmol L 1 with the conditions and results given in Table 6. Both signal height and baseline noise increased with higher PAR concentration. A concentration of 0.3 mmol L 1 was chosen as a compromise between baseline noise and the signal heights, which increased only slightly for a higher PAR concentration. 3.4. Injection volume and sample stacking The influence of injection volume on peak heights and peak resolution was studied and two different injection modes were tested. In the first, only a defined volume of sample was aspirated and injected into the column, directly followed by the mobile phase. In the second approach, finally applied in this work and described in Section 2.5, at first, 150 mL of 1 mmol L 1 nitric acid was injected into the column and a second volume of nitric acid was aspirated before the sample (separating the mobile phase from the sample). The second volume was used to flush the sample matrix out of the column and by this to improve the robustness of the method with respect to the ionic strength of the sample. Also, stacking of the analytes on the column was intended to decrease peak broadening and thus to improve the peak resolution. The volumes of nitric acid exceeded the dead volume of the column twice so that a complete filling of the column with nitric acid during sample loading and a complete flushout of the sample matrix before mobile phase entering can be assumed. The injection scheme and the results are depicted in Fig. 2 with the experimental conditions indicated. For both injection modes, the peak resolution decreased with increasing volume of sample; however, the resolution was improved, proving that analyte stacking was achieved by the gradient application. Therefore, this injection mode was adopted further on. A linear correlation between the injection volume and peak heights was fulfilled in approximation up to 90 mL.
ranges with square correlation coefficients r2 between 0.9974 and 0.9992 without significant loss of peak resolution. Despite the fact that all solutions had to be prior metered by aspiration and sample dispersion in the HC, peak height repeatability values were high. For Cu(II), Zn(II), and Fe(II), RSD values for calibration standards were 1.8%, 2.0%, and 3.2% for an injection volume of 30 mL and 0.5%, 1.0%, and 2.0% for an injection volume of 90 mL, respectively. Examples of peak separations for both injection volumes are given in Fig. 3. Lower LOD values were obtained than in a previous work using the same guard column with an analytical column (five-fold longer) and 520 nm as analytical wavelength [24]. Compared to 510 nm as an analytical wavelength used elsewhere [25,26], sensitivity values were about 90% and 65% lower for 520 nm and 530 nm, respectively. However, the LOD values were the lowest for 530 nm. This was due to a significant lower contribution of the absorbance of the reagent, which caused part of the baseline noise. Therefore, 530 nm was adopted further on and a baseline noise of o1 mAU was achieved. The number of theoretical plates was up to 625 and 910 for injection volumes of 30 and 90 mL, respectively, which were, considering the length of the column, acceptable and in a similar range as reported in previous works for transition metal separation [24,25]. The interference of the sample matrix was evaluated by measuring different water samples and calculating the recovery values. The samples were immediately acidified by the addition of 12 mmol L 1 nitric acid reaching a similar pH value of 2.5–3.0 as for the MilliQ standard solutions. After degassing and overnight storage to allow sedimentation of particles, the samples were measured without any further treatment. Water samples including mineral, tap, river, and fountain waters were measured to estimate the applicability of the method. Spiking was done to determine the analyte recovery for the different
3.5. Analytical performance and sample analysis Calibration linearity was studied for injection volumes of 30 mL and 90 mL comparing 510 nm, 520 nm, and 530 nm as analytical wavelengths. The values for calibration curve slope and limit of detection (LOD), calculated as threefold the noise level of the baseline, are given in Table 7. Linearity was proven for the indicated
Fig. 3. Repeated injections of A: 20 mmol L 1 Cu(II) and Zn(II) and 38.4 mmol L 1 Fe (II), injection volume 30 mL, B: 30 mmol L 1 Cu(II) and Zn(II) and 57.6 mmol L 1 Fe (II), injection volume 30 mL, C: 5 mmol L 1 Cu(II) and Zn(II) and 9.6 mmol L 1 Fe(II), injection volume 90 mL, D: 10 mmol L 1 Cu(II) and Zn(II) and 19.2 mmol L 1 Fe(II), injection volume 90 mL. Conditions as given in Fig. 2, injection volume 30 mL. Detection at 530 nm against 560 nm.
Table 7 Calibration curve characteristics for different injection volumes and detection wavelengths. Conditions as given in Fig. 2. Analyte Tested linear range [mmol L 1]
Mean RSD (n ¼3) (%)
510 nm
520 nm
530 nm
Slope [AU L mol 1]
LOD [mmol L 1]
Slope [AU L mol 1]
LOD [mmol L 1]
Slope [AU L mol 1]
LOD [mmol L 1]
30 mL Cu(II) Zn(II) Fe(II)
5.0–50 5.0–50 9.6–90
1.8 2.0 3.2
4.5 10 3 5.8 10 3 1.3 10 3
4.0 3.5 2.6
5.7 10 3 1.7 10 3 8.6 10 4
3.3 1.8 2.2
6.4 10 3 1.9 10 3 9.8 10 4
0.6 0.5 2.1
90 mL Cu(II) Zn(II) Fe(II)
2.5–30 2.5–30 4.8–60
0.5 1.0 2.0
1.7 10 2 2.4 10 2 5.1 10 3
1.1 0.8 3.6
1.6 10 2 4.7 10 2 2.1 10 2
0.9 0.6 2.9
1.2 10 2 3.3 10 2 1.4 10 2
0.2 0.2 0.8
B. Horstkotte et al. / Talanta 136 (2015) 75–83
81
Table 8 Concentration and recovery data from water sample analysis. Sample
Ion
30 mL Added [mmol L 1]
Tap water 1
Tap water 2
Mineral water 1
Mineral water 2
Fountain water
River water 1
River water 2
90 mL Found [mmol L 1]
Recovery
Added [mmol L 1]
Found [mmol L 1]
Recovery
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.78 14.87 10.50
95.6% 89.8% 109.4%
5.0 5.0 19.2
5.08 13.97 21.90
101.6% 107.4% 114.1%
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6 5.0 5.0 9.6
4.71 5.05 12.48
94.1% 100.9% 130.0%
5.0 5.0 19.2
3.82 4.92 18.14
76.3% 98.6% 94.5%
10.38
8.60
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.94 5.39 10.77
98.7% 107.9% 112.2%
5.0 5.0 19.2
4.91 5.28 20.18
95.2% 105.6% 105.1%
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.55 4.69 9.71
90.9% 93.8% 101.2%
5.0 5.0 19.2
5.05 4.32 17.32
100.9% 86.3% 90.2%
Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe
0.15
2.14 0.33 5.0 5.0 9.6
7.26 5.42 9.78
5.0 5.0 9.6
0.56 4.90 5.56 11.46
5.0 5.0 9.6
0.85 4.72 5.34 11.54
1.92 0.41 102.3% 101.8% 101.9%
97.9% 111.2% 113.6%
94.4% 106.8% 111%
5.0 5.0 19.2
6.73 5.56 20.53
96.4% 103.1% 106.9%
5.0 5.0 9.6
0.14 1.67 4.67 5.04 11.87
93.4% 97.9% 106.3%
5.0 5.0 9.6
0.12 0.67 4.74 5.04 9.55
94.7% 98.4% 92.4%
Empty cells correspond to no addition or below LOD, respectively
matrices. The results are given in Table 8 and examples of standard and sample analysis are given in Fig. 4. Applying an injection volume of 30 mL, standard spiking was done with concentration levels of 5 mmol L 1 for Cu(II) and Zn(II) and 9.6 mmol L 1 for Fe(II). Recovery values of 96.373.7%, 101.576.4%, and 111.379.6%, respectively, were obtained using an injection volume of 30 mL. Applying an injection volume of 90 mL, recovery values for the same spiking levels were 94.178.4%, 99.877.6%, and 101.479.0%, respectively. In conclusion, the method was applicable to the tested water samples with acceptable recovery of the analytes.
3.6. Final discussion on the method’s performance While the separation on the guard column plus analytical column assembly (30 cm in total) took 14 min [28], the here-presented method with the guard column alone took only 5 min. The entire procedure including SP and PP refilling, stacking with nitric acid, and sample aspiration took less than 6.5 min. Only 2.5 mL of mobile phase and 2.0 mL post-column reagent were required per analysis.
Despite the fact that the ratio of mobile phase to post-column reagent was lower (i.e. higher dilution of the eluting analytes) and a 40% lower injection volume was applied, an about 20% better sensitivity than the original method was achieved using 30 mL as injection volume. Sensitivity was even doubled for an injection volume of 90 mL. These improvements were explained by the achieved sample stacking and a lower peak broadening by diffusion as the separation column was shorter. An important advantage of application of a lower pH of the mobile phase as described in this work is the fact that oxygen removal from the column using sodium sulfite solution as recommended by the producer was not necessary, since Fe(II) oxidation is slower in acidic conditions. In fact, we never observed any significant signal, which could be traced as Fe(III) caused by oxidation of Fe(II) used as standard in the system. Real sample analysis was carried out using different drinking and tap waters as samples, both with 90 mL and with 30 mL as injection volumes. On the other hand, the method is limited by the possible peak resolution to the analysis of waters, where only a few transition
82
B. Horstkotte et al. / Talanta 136 (2015) 75–83
times have to be expected. This would also allow the use of a smaller injection volume and, consequently, a higher peak resolution without an unacceptable loss of sensitivity.
4. Conclusion The usefulness of Sequential Injection Chromatography for the determination of different transition cations on a commercial ionexchanger guard column was demonstrated as a quick alternative to longer HPLC separation. A simple, rapid, and versatile analyzer system was developed. It was successfully applied to the analysis of real samples being drinking (tap and bottled) and mineral waters. The high repeatability, speed, and acceptable analyte recovery are adequate for screening and monitoring purposes. Approaches to improve the method repeatability, to lower the baseline noise level, and to improve the limit of detection as well as to increase peak resolution and sensitivity were studied and discussed. The potential of sequential injection chromatography as on-line preparation tool for transition metal stacking and separation even using a short column was further demonstrated.
Acknowledgments B. Horstkotte was supported by a postdoctoral fellowship of the project CZ.1.07/2.3.00/30.0022 supported by the Education for Competitiveness Operational Program (ECOP) and co-financed by the European Social Fund and the state budget of the Czech Republic. The authors gratefully acknowledge the financial support of the Czech Ministry of Education, Youth and Sports project VES13 Kontakt II LH13023.
Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2015.01.001. References
Fig. 4. Repeated injections of calibration standards (1: 5 mmol L 1 Cu(II), 5 mmol L 1 Zn(II), 9.6 mmol L 1 Fe(II), 2: 10 mmol L 1 Cu(II), 10 mmol L 1 Zn(II), 19.2 mmol L 1 Fe(II)) and samples. Conditions as given in Fig. 3.
metals are present in significant concentration levels such as in the case of tap water, where copper and iron are the ions of major interest. The method might further be applied to the analysis of industrial process waters, where iron and zinc monitoring could be a useful tool to detect corrosion. Study of minor component in natural waters such as Co(II) is not possible since its peak cannot be resolved from the other analytes and a longer separation column would be needed. This work demonstrated for the first time that simple cation separation using sequential injection chromatography (SIC) is possible and by choosing an appropriate selective post-column reagent, the system could be used for stand-alone analysis and monitoring purposes of other metals. This work also reported for the first the use of a chromogenic post-column derivatization in SIC. The method, and especially the LOD, could further be improved applying in-line pre-concentration and matrix elimination, since especially for samples of elevated salinity, changes in retention
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
J. Růžička, E.H. Hansen, Anal. Chim. Acta 78 (1975) 145–157. J. Růžička, G.D. Marshall, Anal. Chim. Acta 237 (1990) 329–343. F. Albertús, B. Horstkotte, A. Cladera, V. Cerdá, Analyst 124 (1999) 1373–1381. Flow injection analysis, in: S.D. Kolev, I. McKelvie (Eds.), Comprehensive Analytical Chemistry, Elsevier, Netherlands, 2008. Advances in Flow Methods of Analysis, in: M. Trojanowicz (Ed.), Wiley-VCH, Weinheim, 2008. G. Guiochon, J. Chromatogr. A 1168 (2007) 101–168. J. Huclová, D. Šatínsky, R. Karlíček, Anal. Chim. Acta 494 (2003) 133–140. D. Šatínsky, P. Solich, P. Chocholouš, R. Karlíček, Anal. Chim. Acta 499 (2003) 205–214. H.M. Gonzalez-San Miguel, J.M. Alpizar-Lorenzo, V. Cerda-Martin, Talanta 72 (2007) 296–300. J.R. Santos, A.O.S.S. Rangel, Anal. Chim. Acta 715 (2012) 57–63. M. Fernández, M. Miró, H.M. González, V. Cerdà, Anal. Bioanal. Chem. 391 (2008) 817–825. S.K. Hartwell, A. Kehling, S. Lapanantnoppakhun, K. Grudpan, Anal. Lett. 46 (2013) 1640–1671. G. Guiochon, F.J. Gritti, J. Chromatogr. A 1218 (2011) 1915–1938. P. Chocholouš, L. Kosařová, D. Šatínský, H. Sklenářová, P. Solich, Talanta 85 (2011) 1129–1134. P. Chocholouš, P. Solich, D. Šatínský, Anal. Chim. Acta 600 (2007) 129–135. M. Fernandez, H.M. Gonzalez-San Miguel, J.M. Estela, V. Cerda, Trends Anal. Chem. 28 (2009) 336–346. C. Sarzanini, J. Chromatogr. A 850 (1999) 213–228. R.E. Shepherd, Coord. Chem. Rev. 247 (2003) 159–196. F. Maya, J.M. Estela, V. Cerdà, Microchim. Acta 173 (2011) 33–41. E. Sugrue, P. Nesterenko, B. Paull, Analyst 128 (2003) 417–420. E. Santoyo, S. Santoyo-Gutierrez, S.P. Verma, J. Chromatogr. A 884 (2000) 229–241. C.K. Zacharis, P.D. Tzanavaras, Anal. Chim. Acta 798 (2013) 1–24. R. Cassidy, L. Sun, J. Chromatogr. A 654 (1993) 105–111.
B. Horstkotte et al. / Talanta 136 (2015) 75–83
[24] Y.F. Lasheen, A.F. Seliman, A.A. Abdel-Rassoul, J. Chromatogr. A 1136 (2006) 202–209. [25] L. Barron, M. O’Toole, D. Diamond, P.N. Nesterenko, B. Paull, J. Chromatogr. A 1213 (2008) 31–36. [26] E. Pobozy, R. Halko, M. Krasowski, T. Wierzbicki, M. Trojanowicz, Water Res. 37 (2003) 2019–2026. [27] IonPac CS5A Analytical Column Manual, 〈http://www.dionex.com/en-us/products/columns/ic-rfic/transition-metal-packed/ionpac-cs5a/lp-73276.html〉 01.11.13, 2013.
83
[28] TN 10: Determination of Transition Metals by Ion Chromatography, 〈http:// www.dionex.com/en-us/products/columns/ic-rfic/transition-metal-packed/ ionpac-cs5a/lp-73276.html〉 01.11.13, 2013. [29] J. Nelder, R Mead, Comput. J. 7 (1965) 308–313. [30] D. Betteridge, A.P. Wade, A.G. Howard, Talanta 32 (1985) 709–722. [31] D. Betteridge, A.P. Wade, A.G. Howard, Talanta 32 (1985) 723–734.
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
Sequential Injection Chromatography with post-column reaction/derivatization for the determination of transition metal cations in natural water samples Burkhard Horstkotte n, Patrícia Jarošová, Petr Chocholouš, Hana Sklenářová, Petr Solich Department of Analytical Chemistry, Faculty of Pharmacy, Charles University in Prague, Heyrovského 1203, 500 05 Hradec Králové, Czech Republic
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2014 Received in revised form 27 December 2014 Accepted 3 January 2015 Available online 12 January 2015
In this work, the applicability of Sequential Injection Chromatography for the determination of transition metals in water is evaluated for the separation of copper(II), zinc(II), and iron(II) cations. Separations were performed using a Dionex IonPACTM guard column (50 mm 2 mm i.d., 9 mm). Mobile phase composition and post-column reaction were optimized by modified SIMPLEX method with subsequent study of the concentration of each component. The mobile phase consisted of 2,6pyridinedicarboxylic acid as analyte-selective compound, sodium sulfate, and formic acid/sodium formate buffer. Post-column addition of 4-(2-pyridylazo)resorcinol was carried out for spectrophotometric detection of the analytes' complexes at 530 nm. Approaches to achieve higher robustness, baseline stability, and detection sensitivity by on-column stacking of the analytes and initial gradient implementation as well as air-cushion pressure damping for post-column reagent addition were studied. The method allowed the rapid separation of copper(II), zinc(II), and iron(II) within 6.5 min including pump refilling and aspiration of sample and 1 mmol HNO3 for analyte stacking on the separation column. High sensitivity was achieved applying an injection volume of up to 90 mL. A signal repeatability of o 2% RSD of peak height was found. Analyte recovery evaluated by spiking of different natural water samples was well suited for routine analysis with sub-micromolar limits of detection. & 2015 Elsevier B.V. All rights reserved.
Keywords: Sequential Injection Chromatography Transition metal cations Post-column reaction Spectrophotometric detection On-column stacking
1. Introduction Non-separating flow techniques (FT) are powerful tools for the automation of analytical procedures, especially where high sample throughput, high repeatability, and the avoidance of sample contamination during processing are very important such as in ambient monitoring of trace compounds. This is due to the (computer-)controlled handling of all solutions in a closed tubing network or “manifold”, continuous/discontinuous system cleaning by a carrier flow, and precise timing between all procedure steps such as mixing and detection. With respect to solution handling, mixing conditions, flow pattern, and system configuration, different non-separating FT can be distinguished, with Flow Injection Analysis (FIA) [1], Sequential Injection Analysis (SIA) [2], and Multi-Syringe Flow Injection Analysis (MSFIA) [3] being the ones with the highest impact on this work. The combination of FT for sample handling and preparation with liquid chromatography (LC) is advantageous because despite all its n
Corresponding author. Tel.: þ 420 495 067 504; fax: þ 420 495 067 164. E-mail address: [email protected] (B. Horstkotte).
http://dx.doi.org/10.1016/j.talanta.2015.01.001 0039-9140/& 2015 Elsevier B.V. All rights reserved.
potentials, the most definite disadvantage of all FT is the difficulty of differentiation or gradual separation of analytes. Generally, selective reagents and detection techniques, interference masking, or observance of the reaction kinetic allow discrimination of the analyte from sample matrix components or enable multicomponent analysis by changing the reagent or the reaction conditions appropriately. Further separations, such as gas diffusion or extraction, allow enhancing the method's selectivity. However, without coupling FT with chromatographic or electrophoretic separations, mono- or oligo-analyte methods with potential crosssensitivity to other compounds are typical [4,5]. In this context, the development of monolithic columns [6] and their integration into FT analyzer systems allowing low-pressure chromatographic separations [7] was a milestone for the development of more versatile analytical procedures based on FT. Consequently, the combination of the different FT led to the proposal of Sequential Injection Chromatography (SIC) [8] and later Multi-Syringe Chromatography (MSC) [9] and Flow Injection Chromatography (FIC) [10]. The separation performance even using gradient approaches [11] is similar or inferior to HPLC, mainly because the applicable pressure and column length in SIC are limited. The main
76
B. Horstkotte et al. / Talanta 136 (2015) 75–83
advantages of SIC and the related technique are the possibility of automated in-system sample treatment such as sample clean-up [12], the flexibility of operation such as the use of reagents for analyte derivatization or on-column stacking, the ease of method set-up, and the lower cost and transportability of the analyzer system enabling its use for monitoring and field operation. Applications of SIC and related techniques used mostly commercial silica-based reverse-phase monolithic columns and were focused mainly on the separation of organic analytes. Lately, monolithic columns available with new functionalities and core-shell columns [13] are opening the field of application and operation of SIC [14]. Two comprehensive reviews can be found elsewhere [15,16]. While separation of metal cations with post-column addition of either chromogenic or chemiluminescence-generating reagents are state-of-the-art in HPLC [17,18], in FT, multi-parameter methods are mainly based on the use of ion-selective reagents. Due to the lack of commercial monolithic columns with cationexchange functionality, either dynamic coating of reverse phases [19] or permanent modification [20] have been used to enable cation separation. In the present work, the potential and application of a crosslinked divinylbenzene stationary phase modified by a bilayer of anion- and cation-exchange functionalized latex for low-pressure separation of transition cations in SIC are studied. As an alternative to monolithic columns, only the guard column of an HPLC column assembly [21] was used and rapid and low-pressure separation of copper(II), zinc(II), and iron(II) was achieved. Firstly in SIC, post-column addition of a chromogenic reagent was applied to enable selective analyte detection [22]. For this reason, an additional syringe pump was used to provide a confluent flow of an alkaline solution of 4-(2-pyridylazo)resorcinol (PAR) that is typically used as a chromogenic reagent for sensitive detection of transition metal cations by the formation of red-colored complexes [17,23–25].
2. Material and methods 2.1. Reagents All solutions were prepared with demineralized quality water (resistivity418 MΩ cm) and all reagents were of quality “for analysis” and purchased from Sigma Aldrich (Prague, Czech Republic).
For the preparation of the chromogenic reagent, the following aqueous stock solutions were prepared: 50 mmol L 1 of 4-(2-pyridylazo)resorcinol (PAR) in 100 mmol L 1 NaOH, 4.0 mol L 1 ammonium hydroxide, and 1.0 mol L 1 NaHCO3. For the preparation of the mobile phase, the following stock solutions were prepared: 250 mmol L 1 2,6-pyridinedicarboxylic acid (PDCA) in 750 mmol L 1 NaOH, 2 mol L 1 formic acid, and 500 mmol L 1 Na2SO4. In addition, 2.0 mol L 1 NaOH was used for pH adjustment of mobile phase and post-column reagent. Aliquots of the stock solutions of PAR and PDCA were stored in the dark at 18 1C until use. All working solutions were prepared daily. Further stock solutions of the transition metal cations of interest, namely Co(II), Cu(II), Fe(II), and Zn(II) of 1 mmol L 1, were prepared in 10 mmol L 1 nitric acid. Nitrate salts were used for Co(II) and Zn (II) and sulfate salts for Cu(II), and Fe(II). Standard solutions were prepared daily with 4 mmol L 1 of nitric acid. Freshwater samples were used to evaluate the applicability of the method. They were acidified with a surplus of nitric acid (12 mmol L 1) to overcome the buffer capacity of dissolved bicarbonates and carbonates, achieving a final pH of about 2 being similar to the standard solutions. The samples were used unfiltered but after settling of sediments overnight.
2.2. Instrumentation The analyzer configuration is schematically shown in Fig. 1. All experiments were carried out on a Sequential Injection Chromatography system (SIChrom™) from FIAlab company (Bellevue, WA, USA). It integrated a medium pressure, biocompatible piston pump (PP) from Sapphire Engineering (IDEX Corporation, Oak Harbor, WA, USA) of 4 mL total dispense volume (3 inlets) and an eightport selection valve (SV1, type Cheminert, 12U-0484H, 5000 psi) from VICI (Valco Instruments Co. Inc., Houston, TX, USA). A second selection valve of the same type (SV2) and an additional syringe pump (SP, type Cavro XL) from Tecan (San Jose, CA, USA) equipped with a 2.5 mL glass syringe were used to enable piston pump refilling and addition of the post-column reagent, respectively. All instrumentation was controlled via RS232 interface. The second outlet of the PP was connected to a pressure safety valve (PSV) of 500 psi release threshold.
Fig. 1. Scheme of the SIC analyzer system. D: detection flow cell, HC1 and HC2: holding coils (150 cm, 0.8 mm i.d. and 30 cm 0.5 mm i.d., PEEK), PP: piston pump, PSV: pressure safety valve, SP: syringe pump with air bubble, SV1 and SV2: selection valves, V: syringe head valve, X: confluence (T-connector). Dimensions of PEEK tubing with 0.25 mm i.d. A: 25 cm, B: 5 cm, C: 15 cm, E1: 5 cm. Dimensions of FEP tubing (0.8 mm i.d.), E2 & I: 100 cm, G: 30 cm, H: 30 cm. Detail drawing: detection flow cell with flow inlet (L) and outlet (M) and optical fiber connections (N & O).
B. Horstkotte et al. / Talanta 136 (2015) 75–83
Lateral ports of SV1 were connected to the separation column, sample reservoir, 1 mmol L 1 nitric acid, and to SV2. SV2 allowed the selection of different mobile phases from lateral ports 2 to 7. On port 8, 50 mmol L 1 sodium hydroxide solution was placed and used for column cleaning and storage. The central ports of both SV were connected to the PP. On both SV, one lateral port was permanently closed (port 7 on SV1, port 1 on SV2). This configuration enabled refilling the PP from SV2 with SV1 in position 7, so that study of the mobile phase composition was possible by placing different mobile phases on each port of SV2. For all other flow operations performed on SV1, SV2 was in the closed position 1. Tubing connections for solution supply were made of 0.8 mm i.d. PTFE tubing. For pressure application, PEEK tubing of different diameters was used. The holding coil consisted of a 150 cm PEEK tube of 0.8 mm i.d. (HC1, 750 mL). A second 30 cm PEEK tube of 0.5 mm i.d. (HC2, 60 mL) was used to reduce sample dispersion during aspiration and injection. Connections between SV1, column, post-column reagent confluence, and detector were made as short as possible and of 0.25 mm i.d. PEEK tubing. All ferrules and fittings used for the chromatographic part were made of PEEK, while TEFZEL and Acetal pieces were used for low-pressure connections. A miniature fiber-optic spectrophotometer from OceanOptics Inc. (Dunedin, FL, USA), type USB4000, was used for detection. A bright white LED, mounted onto an optical fiber SMA connector (FIAlab), was used as a light source. Both the light source and the spectrophotometer were connected via 0.6 mm i.d. optical fibers from FIAlabs to a micro-volume detection flow cell (shown in Fig. 1 in detail) with 1 cm optical path length made of UltemTM polymer from the same company. 2.3. Data acquisition and evaluation Initially, detection wavelengths of 510 nm, 520 nm, and 530 nm were simultaneously recorded and used as analytical wavelengths [24–26]. Here, the red-colored complexes of the analytes and PAR were measured against a reference wavelength of 560 nm where neither the reagent nor the analyte–PAR complex showed any significant absorbance. The high light intensity of the LED allowed the detector an integration time as short as 6 ms, averaging over 30 single measurements and an effective measuring frequency of 3 Hz. Spectral smoothing over 9 pixels of the CCD array sensor was carried out. Peak height values were used for data evaluation, throughout. Instrumentation control as well as data evaluation was carried out using the FIAlab software, version 5.9.321, provided with the SIC instrument. The software allowed simultaneous measurement at four different wavelengths, control of procedures, variables, loops, and if-then instructions. This gave high versatility for automated optimization and maintenance. Posterior data smoothing (SavizkyGoley, 2-order polynomial, 11 points) and data evaluation were managed using MS-Excel. 2.4. Separation column and post-column reagent addition All separations were performed using a Dionex IonPACTM CG5A guard column 50 mmx2 mm i.d.) from Thermo Fischer Scientific company (www.thermoscientific.com) [27]. The stationary phase consisted of 9 mm cross-linked DVB (55%) particles with a bilayer of anion-exchange and cation-exchange functionalized latex. The column was stored in 50 mmol L 1 NaOH if not used for more than 1 day. The column inlet was connected to the lateral port 3 of SV1 and outlet downstream to a confluence (X, T-connector, 0.5 mm i.d.) of minimal dead volume for post-column addition of the chromogenic reagent provided by SP. A PEEK tube for mixing the mobile phase and the reagent was placed between the confluence and the
77
detection flow cell (15 cm, 0.25 mm i.d.). In order to avoid bubble formation by outgassing of the solution in the detection cell, a short PEEK tube of 5 cm, 0.25 mm i.d., was placed downstream as a restrictor. 2.5. Operation protocol All dispense steps performed through the column with the PP, i.e. applying significant pressure, were followed by a waiting time of 4 s to release the pressure before changing the position of the selection valves. The operation protocol is given as Supplementary material 1. The analytical method started with simultaneous filling of both the PP and the SP with mobile phase (SV2, port 2) and postcolumn reagent, respectively. The PP was filled with 2.67 mL of mobile phase and at high flow rate in order to provoke partly degassing of the mobile phase in the piston pump void. After pressure release, possibly formed bubbles inside the pump or the aspiration tube were expulsed via the mobile phase reservoir. Then, 300 mL of 1 mmol L 1 nitric acid was aspirated (SV1, port 2) by the PP and 150 mL was pushed into the column (SV1, port 3). Then, another 150 mL of 1 mmol L 1 nitric acid followed by the sample volume to be injected was aspirated. By dispensing in total 2500 mL through the column at a flow rate of 480 mL min 1, the sample was injected and chromatographic separation was performed. The first volume of diluted nitric acid was used to improve initial analyte stacking on the column while the second volume was intended to create an initial gradient and flush out the sample matrix, i.e. inorganic salts. After an operation time of the PP of 60 s, post-column reagent addition was started. After a stabilization time of 10 s, blank measurement was performed and data acquisition was performed over 245 s. All separations were carried out in triplicate. Additional procedures were used for system maintenance such as cleaning the tubes to the different solution reservoirs at both selection valves, cleaning the PP and SP, and flushing the column for cleaning, equilibration, or storage.
3. Results and discussion 3.1. Optimization of the analyzer configuration Because continuous degassing of the mobile phase and other solutions was not feasible, the piston pump was placed slightly inclined (ca. 101) as shown in Fig. 1 in order to avoid any air bubble aspirated from SV2 from entering the holding coil and later the column. Ultrasonic degassing of the mobile phase was tested prior to its usage but was found not to improve the separation performance significantly. During the optimization of the mobile phase composition, SV2 was used as autosampler, which allowed to test up to 7 different mobile phases in stand-alone operation. Since the pressure to yield the recommended flow rate of 1200 mL min 1 was not applicable with the used instrumentation, the highest possible flow rate without overcoming 500 psi, i.e. the break-through pressure of the safety restrictor, was 480 mL min 1. At lower flow rates of the SP than 480 mL min 1, significant baseline noise was obtained. This is because the post-column reagent itself adsorbs, although slightly, on the analytical wavelength. At very low flow rates of the SP, the step-wise rotation of the driving stepmotor becomes notable, i.e. the flow shows a slight pulsation. This fluctuation in the flow rate and thus of the mixing ratio of the mobile phase and the post-column reagent becomes notable. To improve the baseline noise, a flow rate of 480 mL min 1 was applied for the SP as soon it was found that not more than 2500 mL
78
B. Horstkotte et al. / Talanta 136 (2015) 75–83
of the mobile phase was needed for the separation. From the tested approaches to further reduce the SP pulsation, the best results (baseline noise reduction by 25%) were obtained when the SP was placed upside down, allowing a permanent air cushion to be present inside the syringe and act as pressure damper. An additional advantage was that any air bubble arising from reagent degassing would remain in the syringe pump and not be expulsed towards the detection cell.
3.2. Optimization of mobile phase composition Mobile phase optimization was carried out in two steps. First, optimization of the mobile phase composition proposed in the technical note from the column producer [28] was done by the modified SIMPLEX method [29–31], which included the following parameters: concentrations of PDCA, formic acid, and Na2SO4 and the pH value. Second, univariant studies of the individual parameters were done to verify and optionally to improve the achieved results. PDCA was chosen as the selective component in the mobile phase over the alternative use of oxalate [23,24], which is known to lead to an accumulation of Fe(II) on the column [28]. The optimization was done using the recommended post-column chromogenic reagent. Since only the guard column was used (i.e. reduction of column length by factor of 6 compared to the technical note), optimization of peak separation was the main objective during mobile phase optimization. Thus, the Simplex optimization of peak resolution as desirability function was carried out. For this, Cu(II) and Co(II) were used as analytes since even with the recommended mobile phase, partial peak resolution was achieved. The results and conditions are given in Table 1. After the initial 5 vertices, the Simplex tended immediately towards lower pH values losing this dimension due to repeated reduction and the prior fixed pH threshold. Furthermore, it was found that reduced Na2SO4 and PDCA content yielded higher peak resolution. Since vertex 12 was identical to vertex 9, the optimization was stopped after 12 cycles. The vertex yielding the best result and the finally applied conditions for univariant study of the formic acid concentration are bolded. For univariant studies, Fe(II) was used in addition to Cu(II) and Co(II), since it showed the longest retention time of the analytes of Table 1 Simplex optimization of the mobile phase composition*. Conditions: chromogenic reagent: 75 mmol L 1 NaHCO3, 126 mmol L 1 NH4OH, 50 mmol L 1 NaOH, 0.5 mmol L 1 PAR. Injection: 10 ml of Cu and Co, each 10 mmol L 1 in 4 mmol L 1 HNO3. Detection at 510 nm against 560 nm. No.
PDCA [mmol L 1]
Formic acid [mmol L 1]
pH
Sodium sulfate [mmol L 1]
RS
1 2 3 4 5 6 7 8 9 10 11 12
4.00 5.85 4.44 4.44 4.44 4.88 5.12 5.24 4.28 4.76 4.84 4.28
50.0 55.5 73.1 55.5 55.5 61.3 64.5 66.2 66.2 60.2 63.4 66.2
3.8 4.2 4.2 5.6 5.6 3.3 3.4 3.3 3.2 3.3 3.3 3.2
4.00 4.44 4.44 4.44 5.85 4.74 3.04 1.38 3.47 3.47 3.50 3.47
1.38 1.19 1.24 0.60 0.95 2.11 2.52 2.52 1.59 2.29 1.59 1.59
Finally 5.00 applied
60.0
3.3
3.00
*Recommended composition for 30 cm column assembly [28]: 7.0 mmol L 1 PDCA, 74 mmol L 1 formic acid, 5.6 mmol L 1 potassium sulfate, 66 mmol L 1 potassium hydroxide.
interest. At first, the influence of the formic acid content on peak resolution was studied in the range from 30 to 70 mmol L 1. The results and conditions are given in Table 2. The effect on the resolution of the Cu(II) and Co(II) peaks was low while the resolution of the Co(II) and the Fe(II) decreased considerably with higher formic acid concentration. This was because the retention of cations with weak complex formation with PDCA such as Fe(II) depends strongly on the salt content in the mobile phase. On the other hand, the peak height of Cu(II) decreased with higher concentration of formic acid in the mobile phase, which is most likely related to the altered mobile phase buffer capacity and thus suboptimal pH for the addition of post-column reagent under these conditions. Finally, a formic acid content of 30 mmol L 1 was selected for all following work. Second, the influence of the content of sodium sulfate on the elution was studied in the range from 10 to 1 mmol L 1. Other salts as additive to the mobile phase were not tested. The optimization was started with the highest Na2SO4 concentration. The results and conditions are given in Table 3. With lower Na2SO4 content, the retention times and peak resolution increased considerably while peaks became lower and wider. This was due to a very low release rate of the retained analytes and longer time of separation, which favored longitudinal diffusion inside the column (data not shown). For 1 mmol of Na2SO4, the eluent strength was insufficient for the elution of both Co(II) and Fe(II) in one run. Since peak resolution was the main criterion of the optimization, 3 mmol L 1 of sodium sulfate was chosen. Third, the influence of the content of PDCA on the elution was studied in the range of 7–1 mmol L 1. The conditions and results are given in Table 4. The peak resolution of copper and iron and the retention times increased considerably and in approximation linearly with lower PDCA content. On the other hand, the peak heights were Table 2 Univariant study of the effect of formic acid concentration on peak resolution. Conditions: chromogenic reagent: 75 mmol L 1 NaHCO3, 126 mmol L 1 NH4OH, 50 mmol L 1 NaOH, 0.25 mmol L 1 PAR. Mobile phase: 5 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Injection: 10 ml of Cu, Co, and Fe(II) of 10, 10, and 20 mmol L 1, respectively, in 4 mmol L 1 HNO3. Detection at 510 nm against 560 nm. Formic acid [mmol L 1]
30 40 50 60 70 80
Peak height [mAU] Mean (n ¼3) 7 SD
Resolution Retention time [s]
Cu(II)
Co(II)
Fe(II)
Cu/ Co
Co/ Fe
Cu (II)
Co (II)
Fe(II)
727 10 617 4 587 4 617 1 557 8 587 7
22 70.2 18 73 24 72 21 71 18 71 23 74
337 2 357 4 357 4 327 4 317 5 367 3
2.26 2.43 2.43 2.42 2.45 2.10
2.31 2.19 1.88 1.63 1.63 1.41
80.8 88.8 83.2 84.5 70.0 73.3
122.0 134.1 126.4 127.7 108.2 109.9
172.4 173.2 160.8 158.0 135.6 136.5
Table 3 Univariant study of the effect of sodium sulfate concentration on peak resolution. Conditions as in Table 2 with 40 mmol L 1 formic acid and 4 mmol L 1 PDCA, pH 3.4. Na2SO4 [mmol L 1]
10 7 4 2 1
Peak height [mAU] Mean (n¼ 3) 7 SD
Resolution
Retention time [s]
Cu(II)
Co(II)
Fe(II)
Cu/ Co
Co/ Fe
Cu (II)
Co (II)
Fe(II)
1077 7 797 1 677 1 617 1 397 1
407 9 307 1 247 2 227 2
697 10 547 1 407 2 287 1
1.16 1.63 2.31 2.86
1.30 1.32 1.80 2.46
45.2 58.9 82.0 122.3 185.1
63.9 84.7 120.5 183.1
84.3 110.1 156.9 234.4
B. Horstkotte et al. / Talanta 136 (2015) 75–83
found maximal between 4 and 5 mmol L 1 and decreased with higher PDCA concentration. This later observation was found due to the competition between PDCA and the post-column reagent PAR, both forming complexes with the metal cations. For 1 mmol L 1, elution of Fe(II) could not be observed, taking into account that the maximum elution volume applicable with the piston pump in one stroke was 3.8 mL. As a compromise of time of analysis and peak resolution, a PDCA concentration of 4 mmol L 1 was chosen. Finally, lower pH values of the mobile phase than the original Simplex threshold were tested. To adjust pH values lower than the pKa of formic acid down to pH 2.5, nitric acid was added to compensate the hydroxide ions originating from the PDCA stock solution. It was observed that the retention times increased but no significant improvement was achieved for either peak heights or peak resolution (data not given).
3.3. Optimization of post-column reagent Once the mobile phase was optimized, the composition of the post-column reagent was studied. This task was reasonable because a more diluted and about two pH units more acidic mobile phase Table 4 Univariant study of the effect of PCDA concentration on peak resolution. Conditions as in Table 2 but with 40 mmol L 1 formic acid and 3 mmol L 1 sodium sulfate, pH 3.4. PDCA [mmol L 1]
7 6 5 4 3 2 1
Peak height [mAU] Mean (n¼ 3) 7 SD
Resolution
Cu(II)
Cu(II)/Co Co(II)/Fe Cu (II) (II) (II)
Co (II)
Fe(II)
2.46 2.93 2.90 2.79 3.06 3.15 3.30
104.1 114.1 120.1 120.3 135.1 145.9 166.7
140.0 150.3 155.1 148.5 167.9 197.5
467 4 567 2 647 6 647 5 607 1 537 3 467 4
Co(II)
12 7 3 29 7 1 32 7 5 28 7 5 23 7 1 22 7 2 23 7 5
Fe(II)
22 73 25 72 27 72 33 72 31 71 32 78
Retention time [s]
79
compared to the recommended mobile phase [28] was found to be optimal due to the shorter separation column used in this work. Maintaining the concentration of ammonium hydroxide constant at 150 mmol L 1, an experiment was designed for testing
Table 5 Study of the effect of sodium bicarbonate and sodium hydroxide concentration in the chromogenic reagent on peak height. Conditions: mobile phase: with 40 mmol L 1 formic acid, 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Other components in reagent: 150 mmol L 1 NH4OH and 0.15 mmol L 1 PAR. Injection: 30 mL of Cu and Co 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. NaOH [mmol L 1]
0 15 30 45 60 75 90
NaHCO3 [mmol L 1]
150 130 110 90 70 50 30
pH
Peak height [mAU] Mean (n¼3)7SD
9.29 9.39 9.50 9.63 9.77 9.94 10.2
Cu
Co
46 72 54 71 56 74 59 71 61 71 65 75 72 79
117 0.3 137 1 147 1 167 0.2 187 4 177 3 187 3
Baseline noise [mAU]
1.6 1.4 1.1 2.5 9.5 12.8 35.1
Table 6 Study of the effect of PAR concentration in the post-column reagent. Conditions: mobile phase: 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Other components in reagent: 150 mmol L 1 NH4OH, 90 mmol L 1 NaHCO3, 45 mmol L 1 NaOH. Injection: 30 ml of Cu and Co 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. PAR [mmol L 1] Peak height [mAU] Mean (n¼ 3) 7SD Baseline noise [mAU]
2.18 2.55 2.32 2.30 2.07 2.55
71.3 78.8 82.0 86.8 90.8 98.3 104.9
0.05 0.10 0.15 0.20 0.25 0.30 0.40
Cu
Co
407 1 457 2 447 0.4 477 1 547 1 677 1 68 7 1
87 0.5 117 0.8 107 0.3 117 0.6 147 0.9 217 0.4 277 0.6
0.7 0.9 1.5 2.3 2.5 2.3 3.4
Fig. 2. Study of the effect of the injection volume on peak height and peak resolution (RS) using a standard of each 10 mmol L 1 Cu(II) and Co(II). Conditions: mobile phase: 4 mmol L 1 PDCA, pH 3.4, 3 mmol L 1 Na2SO4. Reagent: 0.3 mmol L 1 PAR, 150 mmol L 1 NH4OH, 90 mmol L 1 NaHCO3, and 45 mmol L 1 NaOH. Injection: 30 ml of Cu(II) and Co(II) 10 and 10 mmol L 1. Detection at 530 nm against 560 nm. Below: scheme of injection tested modes.
80
B. Horstkotte et al. / Talanta 136 (2015) 75–83
different combinations of sodium hydroxide and sodium bicarbonate concentrations. The conditions and results are given in Table 5. The signal height for Cu(II) and Co(II) peaks increased with higher pH value; however, also the background signal of the reagent increased significantly, as indicated by a higher baseline noise. As a compromise, 45 mmol L 1 NaOH and 90 mmol L 1 NaHCO3 were then chosen for further work. Finally, the effect of the concentration of PAR in the postcolumn reagent was studied in the range of 0.05–0.4 mmol L 1 with the conditions and results given in Table 6. Both signal height and baseline noise increased with higher PAR concentration. A concentration of 0.3 mmol L 1 was chosen as a compromise between baseline noise and the signal heights, which increased only slightly for a higher PAR concentration. 3.4. Injection volume and sample stacking The influence of injection volume on peak heights and peak resolution was studied and two different injection modes were tested. In the first, only a defined volume of sample was aspirated and injected into the column, directly followed by the mobile phase. In the second approach, finally applied in this work and described in Section 2.5, at first, 150 mL of 1 mmol L 1 nitric acid was injected into the column and a second volume of nitric acid was aspirated before the sample (separating the mobile phase from the sample). The second volume was used to flush the sample matrix out of the column and by this to improve the robustness of the method with respect to the ionic strength of the sample. Also, stacking of the analytes on the column was intended to decrease peak broadening and thus to improve the peak resolution. The volumes of nitric acid exceeded the dead volume of the column twice so that a complete filling of the column with nitric acid during sample loading and a complete flushout of the sample matrix before mobile phase entering can be assumed. The injection scheme and the results are depicted in Fig. 2 with the experimental conditions indicated. For both injection modes, the peak resolution decreased with increasing volume of sample; however, the resolution was improved, proving that analyte stacking was achieved by the gradient application. Therefore, this injection mode was adopted further on. A linear correlation between the injection volume and peak heights was fulfilled in approximation up to 90 mL.
ranges with square correlation coefficients r2 between 0.9974 and 0.9992 without significant loss of peak resolution. Despite the fact that all solutions had to be prior metered by aspiration and sample dispersion in the HC, peak height repeatability values were high. For Cu(II), Zn(II), and Fe(II), RSD values for calibration standards were 1.8%, 2.0%, and 3.2% for an injection volume of 30 mL and 0.5%, 1.0%, and 2.0% for an injection volume of 90 mL, respectively. Examples of peak separations for both injection volumes are given in Fig. 3. Lower LOD values were obtained than in a previous work using the same guard column with an analytical column (five-fold longer) and 520 nm as analytical wavelength [24]. Compared to 510 nm as an analytical wavelength used elsewhere [25,26], sensitivity values were about 90% and 65% lower for 520 nm and 530 nm, respectively. However, the LOD values were the lowest for 530 nm. This was due to a significant lower contribution of the absorbance of the reagent, which caused part of the baseline noise. Therefore, 530 nm was adopted further on and a baseline noise of o1 mAU was achieved. The number of theoretical plates was up to 625 and 910 for injection volumes of 30 and 90 mL, respectively, which were, considering the length of the column, acceptable and in a similar range as reported in previous works for transition metal separation [24,25]. The interference of the sample matrix was evaluated by measuring different water samples and calculating the recovery values. The samples were immediately acidified by the addition of 12 mmol L 1 nitric acid reaching a similar pH value of 2.5–3.0 as for the MilliQ standard solutions. After degassing and overnight storage to allow sedimentation of particles, the samples were measured without any further treatment. Water samples including mineral, tap, river, and fountain waters were measured to estimate the applicability of the method. Spiking was done to determine the analyte recovery for the different
3.5. Analytical performance and sample analysis Calibration linearity was studied for injection volumes of 30 mL and 90 mL comparing 510 nm, 520 nm, and 530 nm as analytical wavelengths. The values for calibration curve slope and limit of detection (LOD), calculated as threefold the noise level of the baseline, are given in Table 7. Linearity was proven for the indicated
Fig. 3. Repeated injections of A: 20 mmol L 1 Cu(II) and Zn(II) and 38.4 mmol L 1 Fe (II), injection volume 30 mL, B: 30 mmol L 1 Cu(II) and Zn(II) and 57.6 mmol L 1 Fe (II), injection volume 30 mL, C: 5 mmol L 1 Cu(II) and Zn(II) and 9.6 mmol L 1 Fe(II), injection volume 90 mL, D: 10 mmol L 1 Cu(II) and Zn(II) and 19.2 mmol L 1 Fe(II), injection volume 90 mL. Conditions as given in Fig. 2, injection volume 30 mL. Detection at 530 nm against 560 nm.
Table 7 Calibration curve characteristics for different injection volumes and detection wavelengths. Conditions as given in Fig. 2. Analyte Tested linear range [mmol L 1]
Mean RSD (n ¼3) (%)
510 nm
520 nm
530 nm
Slope [AU L mol 1]
LOD [mmol L 1]
Slope [AU L mol 1]
LOD [mmol L 1]
Slope [AU L mol 1]
LOD [mmol L 1]
30 mL Cu(II) Zn(II) Fe(II)
5.0–50 5.0–50 9.6–90
1.8 2.0 3.2
4.5 10 3 5.8 10 3 1.3 10 3
4.0 3.5 2.6
5.7 10 3 1.7 10 3 8.6 10 4
3.3 1.8 2.2
6.4 10 3 1.9 10 3 9.8 10 4
0.6 0.5 2.1
90 mL Cu(II) Zn(II) Fe(II)
2.5–30 2.5–30 4.8–60
0.5 1.0 2.0
1.7 10 2 2.4 10 2 5.1 10 3
1.1 0.8 3.6
1.6 10 2 4.7 10 2 2.1 10 2
0.9 0.6 2.9
1.2 10 2 3.3 10 2 1.4 10 2
0.2 0.2 0.8
B. Horstkotte et al. / Talanta 136 (2015) 75–83
81
Table 8 Concentration and recovery data from water sample analysis. Sample
Ion
30 mL Added [mmol L 1]
Tap water 1
Tap water 2
Mineral water 1
Mineral water 2
Fountain water
River water 1
River water 2
90 mL Found [mmol L 1]
Recovery
Added [mmol L 1]
Found [mmol L 1]
Recovery
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.78 14.87 10.50
95.6% 89.8% 109.4%
5.0 5.0 19.2
5.08 13.97 21.90
101.6% 107.4% 114.1%
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6 5.0 5.0 9.6
4.71 5.05 12.48
94.1% 100.9% 130.0%
5.0 5.0 19.2
3.82 4.92 18.14
76.3% 98.6% 94.5%
10.38
8.60
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.94 5.39 10.77
98.7% 107.9% 112.2%
5.0 5.0 19.2
4.91 5.28 20.18
95.2% 105.6% 105.1%
Cu Zn Fe Cu Zn Fe
5.0 5.0 9.6
4.55 4.69 9.71
90.9% 93.8% 101.2%
5.0 5.0 19.2
5.05 4.32 17.32
100.9% 86.3% 90.2%
Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe Cu Zn Fe
0.15
2.14 0.33 5.0 5.0 9.6
7.26 5.42 9.78
5.0 5.0 9.6
0.56 4.90 5.56 11.46
5.0 5.0 9.6
0.85 4.72 5.34 11.54
1.92 0.41 102.3% 101.8% 101.9%
97.9% 111.2% 113.6%
94.4% 106.8% 111%
5.0 5.0 19.2
6.73 5.56 20.53
96.4% 103.1% 106.9%
5.0 5.0 9.6
0.14 1.67 4.67 5.04 11.87
93.4% 97.9% 106.3%
5.0 5.0 9.6
0.12 0.67 4.74 5.04 9.55
94.7% 98.4% 92.4%
Empty cells correspond to no addition or below LOD, respectively
matrices. The results are given in Table 8 and examples of standard and sample analysis are given in Fig. 4. Applying an injection volume of 30 mL, standard spiking was done with concentration levels of 5 mmol L 1 for Cu(II) and Zn(II) and 9.6 mmol L 1 for Fe(II). Recovery values of 96.373.7%, 101.576.4%, and 111.379.6%, respectively, were obtained using an injection volume of 30 mL. Applying an injection volume of 90 mL, recovery values for the same spiking levels were 94.178.4%, 99.877.6%, and 101.479.0%, respectively. In conclusion, the method was applicable to the tested water samples with acceptable recovery of the analytes.
3.6. Final discussion on the method’s performance While the separation on the guard column plus analytical column assembly (30 cm in total) took 14 min [28], the here-presented method with the guard column alone took only 5 min. The entire procedure including SP and PP refilling, stacking with nitric acid, and sample aspiration took less than 6.5 min. Only 2.5 mL of mobile phase and 2.0 mL post-column reagent were required per analysis.
Despite the fact that the ratio of mobile phase to post-column reagent was lower (i.e. higher dilution of the eluting analytes) and a 40% lower injection volume was applied, an about 20% better sensitivity than the original method was achieved using 30 mL as injection volume. Sensitivity was even doubled for an injection volume of 90 mL. These improvements were explained by the achieved sample stacking and a lower peak broadening by diffusion as the separation column was shorter. An important advantage of application of a lower pH of the mobile phase as described in this work is the fact that oxygen removal from the column using sodium sulfite solution as recommended by the producer was not necessary, since Fe(II) oxidation is slower in acidic conditions. In fact, we never observed any significant signal, which could be traced as Fe(III) caused by oxidation of Fe(II) used as standard in the system. Real sample analysis was carried out using different drinking and tap waters as samples, both with 90 mL and with 30 mL as injection volumes. On the other hand, the method is limited by the possible peak resolution to the analysis of waters, where only a few transition
82
B. Horstkotte et al. / Talanta 136 (2015) 75–83
times have to be expected. This would also allow the use of a smaller injection volume and, consequently, a higher peak resolution without an unacceptable loss of sensitivity.
4. Conclusion The usefulness of Sequential Injection Chromatography for the determination of different transition cations on a commercial ionexchanger guard column was demonstrated as a quick alternative to longer HPLC separation. A simple, rapid, and versatile analyzer system was developed. It was successfully applied to the analysis of real samples being drinking (tap and bottled) and mineral waters. The high repeatability, speed, and acceptable analyte recovery are adequate for screening and monitoring purposes. Approaches to improve the method repeatability, to lower the baseline noise level, and to improve the limit of detection as well as to increase peak resolution and sensitivity were studied and discussed. The potential of sequential injection chromatography as on-line preparation tool for transition metal stacking and separation even using a short column was further demonstrated.
Acknowledgments B. Horstkotte was supported by a postdoctoral fellowship of the project CZ.1.07/2.3.00/30.0022 supported by the Education for Competitiveness Operational Program (ECOP) and co-financed by the European Social Fund and the state budget of the Czech Republic. The authors gratefully acknowledge the financial support of the Czech Ministry of Education, Youth and Sports project VES13 Kontakt II LH13023.
Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2015.01.001. References
Fig. 4. Repeated injections of calibration standards (1: 5 mmol L 1 Cu(II), 5 mmol L 1 Zn(II), 9.6 mmol L 1 Fe(II), 2: 10 mmol L 1 Cu(II), 10 mmol L 1 Zn(II), 19.2 mmol L 1 Fe(II)) and samples. Conditions as given in Fig. 3.
metals are present in significant concentration levels such as in the case of tap water, where copper and iron are the ions of major interest. The method might further be applied to the analysis of industrial process waters, where iron and zinc monitoring could be a useful tool to detect corrosion. Study of minor component in natural waters such as Co(II) is not possible since its peak cannot be resolved from the other analytes and a longer separation column would be needed. This work demonstrated for the first time that simple cation separation using sequential injection chromatography (SIC) is possible and by choosing an appropriate selective post-column reagent, the system could be used for stand-alone analysis and monitoring purposes of other metals. This work also reported for the first the use of a chromogenic post-column derivatization in SIC. The method, and especially the LOD, could further be improved applying in-line pre-concentration and matrix elimination, since especially for samples of elevated salinity, changes in retention
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
J. Růžička, E.H. Hansen, Anal. Chim. Acta 78 (1975) 145–157. J. Růžička, G.D. Marshall, Anal. Chim. Acta 237 (1990) 329–343. F. Albertús, B. Horstkotte, A. Cladera, V. Cerdá, Analyst 124 (1999) 1373–1381. Flow injection analysis, in: S.D. Kolev, I. McKelvie (Eds.), Comprehensive Analytical Chemistry, Elsevier, Netherlands, 2008. Advances in Flow Methods of Analysis, in: M. Trojanowicz (Ed.), Wiley-VCH, Weinheim, 2008. G. Guiochon, J. Chromatogr. A 1168 (2007) 101–168. J. Huclová, D. Šatínsky, R. Karlíček, Anal. Chim. Acta 494 (2003) 133–140. D. Šatínsky, P. Solich, P. Chocholouš, R. Karlíček, Anal. Chim. Acta 499 (2003) 205–214. H.M. Gonzalez-San Miguel, J.M. Alpizar-Lorenzo, V. Cerda-Martin, Talanta 72 (2007) 296–300. J.R. Santos, A.O.S.S. Rangel, Anal. Chim. Acta 715 (2012) 57–63. M. Fernández, M. Miró, H.M. González, V. Cerdà, Anal. Bioanal. Chem. 391 (2008) 817–825. S.K. Hartwell, A. Kehling, S. Lapanantnoppakhun, K. Grudpan, Anal. Lett. 46 (2013) 1640–1671. G. Guiochon, F.J. Gritti, J. Chromatogr. A 1218 (2011) 1915–1938. P. Chocholouš, L. Kosařová, D. Šatínský, H. Sklenářová, P. Solich, Talanta 85 (2011) 1129–1134. P. Chocholouš, P. Solich, D. Šatínský, Anal. Chim. Acta 600 (2007) 129–135. M. Fernandez, H.M. Gonzalez-San Miguel, J.M. Estela, V. Cerda, Trends Anal. Chem. 28 (2009) 336–346. C. Sarzanini, J. Chromatogr. A 850 (1999) 213–228. R.E. Shepherd, Coord. Chem. Rev. 247 (2003) 159–196. F. Maya, J.M. Estela, V. Cerdà, Microchim. Acta 173 (2011) 33–41. E. Sugrue, P. Nesterenko, B. Paull, Analyst 128 (2003) 417–420. E. Santoyo, S. Santoyo-Gutierrez, S.P. Verma, J. Chromatogr. A 884 (2000) 229–241. C.K. Zacharis, P.D. Tzanavaras, Anal. Chim. Acta 798 (2013) 1–24. R. Cassidy, L. Sun, J. Chromatogr. A 654 (1993) 105–111.
B. Horstkotte et al. / Talanta 136 (2015) 75–83
[24] Y.F. Lasheen, A.F. Seliman, A.A. Abdel-Rassoul, J. Chromatogr. A 1136 (2006) 202–209. [25] L. Barron, M. O’Toole, D. Diamond, P.N. Nesterenko, B. Paull, J. Chromatogr. A 1213 (2008) 31–36. [26] E. Pobozy, R. Halko, M. Krasowski, T. Wierzbicki, M. Trojanowicz, Water Res. 37 (2003) 2019–2026. [27] IonPac CS5A Analytical Column Manual, 〈http://www.dionex.com/en-us/products/columns/ic-rfic/transition-metal-packed/ionpac-cs5a/lp-73276.html〉 01.11.13, 2013.
83
[28] TN 10: Determination of Transition Metals by Ion Chromatography, 〈http:// www.dionex.com/en-us/products/columns/ic-rfic/transition-metal-packed/ ionpac-cs5a/lp-73276.html〉 01.11.13, 2013. [29] J. Nelder, R Mead, Comput. J. 7 (1965) 308–313. [30] D. Betteridge, A.P. Wade, A.G. Howard, Talanta 32 (1985) 709–722. [31] D. Betteridge, A.P. Wade, A.G. Howard, Talanta 32 (1985) 723–734.