- Email: [email protected]
A critical comparison of constant and pulsed flow systems exploiting gas diffusion
Talanta 148 (2016) 596–601
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
A critical comparison of constant and pulsed flow systems exploiting gas diffusion Claudineia Rodrigues Silva a,b, Camelia Henriquez a,b, Rejane Mara Frizzarin a,b, Elias Ayres Guidetti Zagatto a,n, Victor Cerda b a b
Center for Nuclear Energy in Agriculture, University of Sao Paulo, 13400-970 Piracicaba, SP, Brazil Laboratory of Environmental Analytical Chemistry, University of Balearic Islands, 07122 Palma de Mallorca, Spain
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2015 Received in revised form 24 September 2015 Accepted 25 September 2015 Available online 28 September 2015
Considering the beneficial aspects arising from the implementation of pulsed flows in flow analysis, and the relevance of in-line gas diffusion as an analyte separation/concentration step, influence of flow pattern in flow systems with in-line gas diffusion was critically investigated. To this end, constant or pulsed flows delivered by syringe or solenoid pumps were exploited. For each flow pattern, two variants involving different interaction times of the donor with the acceptor streams were studied. In the first one, both the acceptor and donor streams were continuously flowing, whereas in the second one, the acceptor was stopped during the gas diffusion step. Four different volatile species (ammonia, ethanol, carbon dioxide and hydrogen sulfide) were selected as models. For the flow patterns and variants studied, the efficiencies of mass transport in the gas diffusion process were compared, and sensitivity, repeatability, sampling frequency and recorded peak shape were evaluated. Analysis of the results revealed that sensitivity is strongly dependent on the implemented variant, and that flow pattern is an important feature in flow systems with in-line gas diffusion. & 2015 Elsevier B.V. All rights reserved.
Keywords: Gas diffusion Flow analysis Pulsed flow Constant flow
1. Introduction Constant or pulsed flows have been generally exploited in flow analysis, and the related flow patterns may influence the performance of the analyzer. These flow patterns depend on the fluid propeller device, generally peristaltic or piston pumps for constant flows, or solenoid pumps for pulsed flows. Under constant flow, all fluid elements are displaced following parallel trajectories, and the pronounced radial gradient of the linear flow velocities leads to a parabolic profile of velocities inside the sample zone [1]. Consequently, laminar flow conditions are established. On the other hand, a pulsed flow is established by successive and sudden insertions of small solution aliquots, leading to a chaotic movement of the fluid elements. Turbulent mixing is then noted during the pulse insertions [2], and this aspect favors the radial mass transport and homogenization of solutions, thus improving the mixing conditions and reducing the sample dispersion. As the differences in linear speeds of the fluid lines are reduced, tailing effects are minimized. Influence of flow pattern in flow analysis has been emphasized in different applications [3], and the superior performance of n
Corresponding author. E-mail address: [email protected] (E.A.G. Zagatto).
http://dx.doi.org/10.1016/j.talanta.2015.09.064 0039-9140/& 2015 Elsevier B.V. All rights reserved.
pulsed flow systems has been demonstrated in relation to system portability [4], slow chemical reactions [5], heat transfer [6], solid phase extraction involving fluidized beds [7] and immobilized reagents [8]. In comparison with constant flow systems, pulsed flow systems are generally characterized by improved sensitivity, repeatability, analytical frequency, reduced sample and reagent consumptions and versatility. Gas diffusion (GD) is a worldwide exploited analyte separation/ concentration process and its implementation in flow analysis may overcome some drawbacks inherent to GD under batch wise conditions [9]. Several analytical procedures with in-line GD, relying on either constant or pulsed flows, have been recently reported [10], and the superior performance of pulsed flow systems in relation to constant flow systems was already stressed [11]. However, a systematic investigation of the influence of flow pattern in analytical systems with in-line GD was not yet carried out. Selectivity and sensitivity are dependent on the physical characteristics of the donor and acceptor streams (e.g. flow rate, chemical composition, pressure, and flow direction), manifold design, reagent concentrations and GD-cell geometry (including type of semipermeable membrane) [12,13]. Moreover, variants involving acceptor and donor streams continuously flowing (AF), acceptor stream stopped and donor stream flowing during GD (AS), exploitation of stream segmentation and use of oscillating streams have been exploited [14,15,16]. With the AS variant, in-line
C.R. Silva et al. / Talanta 148 (2016) 596–601
597
concentration of the volatile species, thus sensitivity, is generally favored [14,17]. The aim of this work was then to critically compare the influence of flow pattern in flow systems exploiting in-line GD. To this end, a multi-syringe flow injection system and a multi-pumping flow system, involving constant or pulsed flows, were designed as similar as possible to each other, the only difference being the fluid propelling device, piston or solenoid pumps for delivering constant or pulsed flows, respectively. Ammonia, ethanol, carbon dioxide, hydrogen sulfide were selected as model volatile species, and the AF and AS variants were considered. System performance was evaluated under all the investigated situations, in terms of efficiency of mass transport during GD, repeatability, sampling frequency and recorded peak shape.
2. Experimental 2.1. Solutions All solutions were prepared with analytical-grade chemicals from Scharlau SA (Barcelona, Spain) and deionized water (resistivity 418.2 MΩ cm) provided by a Milli-Q system. The reagents and standard solutions are presented in Table 1. 2.2. Apparatus A model Bu4S burette from Crison Instruments S.A. (Alella, Barcelona, Spain) equipped with four 5-mL glass syringes, model TLL SYR from Hamilton (Bonaduz, Switzerland), was used for establishing the constant flow inherent to the system in Fig. 1a. Each syringe head included a three-way solenoid valve accountable for directing the pumped solution towards either the manifold or the solution reservoir for refilling. An additional model STV-3 1/4UKG three-way solenoid valve from Takasago (Nagoya, Japan) was used for inserting the standard solutions. Its central port was connected to the ON port of the syringe head valve through a 2.2 mL holding coil (HC) and its ON and OFF ports were connected to the standard reservoir and to the manifold, respectively. This valve was controlled through an auxiliary supply port of the burette. In the pulsed flow system (Fig. 1b), four model P/N 120SP12205TV solenoid pumps with 25-mL stroke volume from Bio-Chem Inc. (Boonton NJ, USA) were used as liquid drivers. For delivering 1.0 or 2.0 mL min 1 flow-rates, the pumps were actuated in a synchronized way at 0.67 or 1.33 Hz, respectively. The three-way valve was likewise accountable for introducing either the standard (ON) or the carrier (OFF) solutions. The solenoid pumps and valves were controlled by a multi-pumping module from Sciware System SL (Bunyola, Spain). The RC1 and RC2 coiled reactors (length ¼100 cm, inner volume ca 0.5 mL), the holding coil and the transmission lines were made of PTFE (polytetrafluoroethylene) tubing (i.d. ¼0.8 mm), and the
Fig. 1. Flow diagrams of the constant (a) and pulsed (b) flow systems. M ¼ model species solution; Ri¼ reagents; C ¼carrier stream; MSB ¼ multi-syringe burette with Si syringes; Pi ¼solenoid pumps; V ¼three-way solenoid valve; HC ¼holding coil; RCi ¼ coiled rectors; GD ¼gas diffusion cell with the membrane specified by a traced line; reaction coil; D¼flow-through detector; W ¼flask for waste collection. Two syringes (S1, S3) or pumps (P1, P3), three syringes (S1, S2, S3) or pumps (P1, P2, P3) and four syringes or pumps are need for EtOH, NH3 or CO2 and H2S, respectively. For H2S, the acceptor stream is formed by converging the R2 and R3 streams, and mixing them inside the RC2 reactor. For details, see text.
connectors and GD-cell were made of PMMA (polymethylmethacrylate). PTFE tubes (length ¼30 cm, i.d.¼1.5 mm) were used for aspirating the reagent and carrier solutions towards the pulsed flow system. The GD-cell [17] was built-up by juxtaposing two identical rectangular PMMA blocks, each one with a U-shaped flow channel (width, depth, and length: 2.0, 0.7, and 128 mm). A Teflons hydrophobic GD membrane from Lachat Instruments (Loveland CO, USA), recommended for ammonia GD by the manufacturer [18], was placed between the blocks, thus establishing separated donor and acceptor channels (inner volume ¼ 180 mL). The donor and acceptor streams flew though these channels in a countercurrent way. Regarding conductometric detection, a lab-made flow cell [19] was connected to a model 525 conductometer from Crison Instruments S.A. The cell constant was calculated as 0.06 cm 1 by using a 0.01 mol L 1 KCl (1413 mS cm 1 at 25 °C) conductivity and standard solution. Two scales (0.1–200 mS cm 1 0.01–20 mS cm 1) were selected for NH3 and CO2 and their corresponding baselines were adjusted close to the maximum values in order to increase peak height. Measurements were done at every 0.2 s, and temperature correction was not applied because the laboratory temperature was stable enough [11].
Table 1 Composition of the involved solutions. Table refers to the flow systems in Fig. 1. Stream Model species NH3
CO2
EtOH
M
5.0–20.0 mg L 1 (as NH4Cl) 2.0–10.0 mmol L 1 (as NaHCO3) 10.0–60.0% (v/v)
C R1 R2
Water 25 mmol L 1 NaOH 25 μmol L 1 HCl
Water 25 mmol L 1 H2SO4 10 mmol L 1 NaOH
R3
–
–
Water – 0.3 mol L 1 K2Cr2O7 in 4.0 mol L 1 H2SO4 –
H2S 5.0–25.0 mg L 1 S2 (as Na2S 9H2O) in 25 mmol L 1 NaOH Water 0.5 mol L 1 HCl 5.0 mmol L 1 DMPD in 1.0 mol L 1 HCl 50 mmol L 1 Fe3 þ (as FeCl3 6H2O) in 1.0 mol L 1 HCl
598
C.R. Silva et al. / Talanta 148 (2016) 596–601
An USB-2000 miniature CCD spectrophotometer from Ocean Optics (Dunedin FL, USA) and a highly stable bright-white LED light source were used. Both were aligned with the optical path of a model 75.1 SOG 10-mm flow-through cuvette from Starna (Hainault, UK). Integration times were set as 30 and 75 ms for ethanol and H2S. The AutoAnalysis 5.0 software from Sciware Systems S.L. was used for instrumental control as well as for data acquisition and treatment. 2.3. Procedure Influence of flow pattern in the GD process was investigated with the flow systems in Fig. 1, which were designed with the same strategy for standard/reagent insertions and the same manifold architecture. For each model species (ammonia, carbon dioxide, ethanol and hydrogen sulfide), four procedures involving different flow patterns (constant/pulsed) and variants (AF/AS) were considered. In general, these procedures comprised: – Sample insertion. Either the model species (M) or the carrier (C) solution was pumped by P1 or S1, towards the analytical path during pre-defined time intervals, via the V three-way valve (Fig. 1). A sample zone was then established inside the M/C stream; – Formation of the volatile species. The M/C stream merged with the R1 reagent, which was pumped by P2 or S2. The donor stream, where the model species in the ionic form was converted to a volatile form, was then established. This stream was directed towards the GD-cell through the RC1 reaction coil, and wasted; – Diffusion of the volatile species. The formed NH3, CO2 or H2S volatile species and ethanol underwent diffusion through the hydrophobic membrane towards the acceptor stream; – Collection of the diffused species. The volatile species were collected into the acceptor stream as its ionic form, whereas ethanol was collected without conversion. A secondary sample zone was then established. The R2 stream pumped by P3 or S3 (Fig. 1), eventually combined with the R3 stream, was either flowing simultaneously with the donor stream (AF variant) or stopped during passage of the sample zone through the GD-cell (AS variant). – Measurement. The acceptor stream was directed towards detection cell and wasted. Passage of the secondary sample zone through the flow cell led to a transient modification in the monitored signal, which was recorded as a peak. Height of this peak constituted the measurement basis. Acid–base equilibria were involved in the formation of the gaseous ammonia, carbon dioxide or hydrogen sulfide in the donor stream as well as for their collections in the acceptor stream. The NH+4 and CO2− 3 ionic species established into the acceptor stream were conductometrically monitored. Regarding sulfide, the acceptor stream was a N,N-dimethyl-p-phenylenediamine (DMPD) plus Fe(III) acidic solution that reacted with S2 yielding methylene blue, which was spectrophotometrically monitored at 668 nm. Details of the involved chemistry are given elsewhere [20]. Ethanol was sufficiently volatile to be separated from the liquid phase without further conversion or temperature increment. Thus, the R1 reagent (Fig. 1) was not required and the M/C stream constituted itself as the donor stream. The collected ethanol reduced Cr(VI) to Cr(III), which was spectrophotometrically monitored at 600 nm [21]. Regarding system dimensioning, the R1 reagent was added by confluence aiming at an efficient homogenization with the M/C
stream. Its concentration was set in excess for improving the gas diffusion. Addition of this reagent via zone merging configuration [1] was not performed to avoid an eventual return of the volatile species back to the donor stream. In flow systems where an analyte separation/concentration step is not implemented, variations in flow rates do not affect sample dispersion in a pronounced manner. In contrast, flow rates are of utmost relevance in relation to in-line GD, as they determine the mean residence times of the fluid elements of the donor and acceptor streams inside the GD-cell, thus the time of contact with the membrane and the available time for the collection of the gaseous species by a given fluid element (tGD). Variations in flow rates may therefore alter the mass transfer of the volatile species, thus the analytical sensitivity. Influence of flow rates was then investigated under two extreme situations, associated with the AF and AS variants. In AF, the same flow rate was selected for the acceptor and donor streams, and tGD was minimal, whereas in AS, the acceptor stream was stopped (flow rate¼0) during passage of the entire sample zone through the GD-cell. In this later situation, tGD was maximal and transference of the gaseous species to the acceptor stream was improved. For both flow patterns and variants, the donor stream flow rate was fixed as 2.0 mL min 1. Other relevant parameters such as e.g. inserted volume, injection mode, reactor dimensions, temperature, reagent concentrations, hydrophobic membrane and GD-cell geometry were maintained, regardless of the flow pattern or variant involved. 2.4. The phase transfer factor In analytical procedures involving GD, the analyte is generally not completely transferred to the acceptor phase due to insufficient available time, low capacity of the collection medium and/or physico-chemical characteristics involved (e.g. distribution coefficients, diffusion coefficients, equilibrium constants, and molecule properties). Consequently, an “enrichment factor”, the ratio of analyte concentrations in the donor and acceptor media, should be estimated. In flow analysis however the concentration gradients along the sample zone, the lack of a steady analyte concentration, and the low tGD value may impair the accurate prediction of the enrichment factor. A phase transfer factor (FP) was then considered for roughly evaluating the mass transfer of the species of interest in flow analysis [12]. FP is the ratio of numbers of moles of the model species in the acceptor solution and in the inserted aliquot. The number of moles of the considered model species in the inserted aliquot was straightforwardly obtained, as standard solutions were used and the inserted volume was known. The number of moles collected by the acceptor stream was evaluated by taking into account the recorded peak area related to the secondary sample zone. To this end, the A vs t function (A¼instant absorbance or conductance; t ¼time) should be transformed into a C vs V function (C ¼concentration; V ¼volume). Abscise values were then multiplied by flow rate (here 2.00 mL min 1) and ordinate values were transformed in concentration. For spectrophotometric monitoring, the analytical response was considered. For this purpose, different standard solutions were successively used instead of the acceptor stream and the A values were modified according to the obtained analytical curve. For conductometric monitoring, this transformation was attained as in earlier work [11].
3. Results and discussion Flow-based methods are essentially kinetics, i.e. steady state situations are often not attained, resulting in partial development
C.R. Silva et al. / Talanta 148 (2016) 596–601
599
Table 2 Analytical characteristics F P ¼phase transfer factor, in %; r.s.d.¼ relative standard deviation of peak heights, in %; SR ¼sampling rate, in h 1; Sens.¼ sensitivity in μ S L cm 1 g 1, mS L cm 1 mmol 1, Abs (%v/v) 1 or¼ Abs L mg 1, relying on the NH3, CO2, EtOH or H2S concentration ranges specified in Table 1. Other data refer to 20 mg L 1 NH3, 10 mmol L 1 CO2, 20% v/v EtOH and 25 mg L 1 H2S. For details, see text.
Fp
Pulsed Constant
Sens.
Pulsed Constant
r.s.d.
Pulsed Constant
SR
Pulsed Constant
Fig. 2. Recorded tracings related to A: ammonia (20.0 mg L 1 NH+ 4 ); B: ethanol (20.0% (v/v)); C: carbon dioxide (10.0 mmol L 1 HCO− 3 ) and D: hydrogen sulfide (25.0 mmol L 1 S2 ) model species. Left and right portions: AS and AF variants. Full and traced lines: constant and pulsed flows.
of the involved processes. Even so, precise results are obtained, as reproducible handling conditions are inherent to the flow analyzers. With partial GD, the analytical signal is strongly dependent on different parameters, amongst them the flow pattern and the tGD value. Analysis of the recorded peak shape and evaluation of the phase transfer factor play then a relevant role in the context. 3.1. Recorded peak shape In flow analysis without a separation/concentration step, increasing the analytical path increases sample dispersion and tailing effects for both flow patterns, thus altering the recorded peak shape. For pulsed flows, influence of this parameter is less pronounced, due to the chaotic movement of the fluid elements that improves fluid homogenization, radial mass transport and mixing conditions [5]. Nevertheless, if the analytical path includes a GDcell, the recorded peak shape becomes dependent also on other related aspects, such as e.g. management of the acceptor and donor streams (AS or AF variants), cell geometry and membrane characteristics. This explains why the influence of flow pattern was not so pronounced (Fig. 2). Regarding acceptor and donor stream management, the recorder tracings in Fig. 2 were strongly influenced by the involved variant, AS or AF, for both flow patterns. With AS, the recorded peaks were higher and thinner, because the entire sample zone was focused in a low volume of the acceptor stream that was stopped inside the GD-cell acceptor channel (maximal tGD). Consequently, the analyte concentration in this volume was higher,
NH3
CO2
EtOH
H2S
AF AS AF AS
0.13 0.22 0.11 0.18
0.15 0.87 0.16 0.70
0.02 0.04 0.02 0.04
1.17 3.47 0.94 2.28
AF AS AF AS
2.2 3.3 1.5 2.4
0.09 0.45 0.10 0.39
0.014 0.020 0.014 0.019
0.024 0.063 0.014 0.038
AF AS AF AS
0.8 0.6 1.0 2.3
4 10 0.7 1.2 2.0
2.4 1.7 1.5 2.8
0.9 3.4 0.9 3.9
AF AS AF AS
75 39 51 32
47 35 20 15
49 32 39 29
75 39 31 16
leading to a higher peak whereas the low volume of the GD acceptor channel determined the thinner recorded peak. The peak shape reflected also a lower washing time and a less symmetrical peak for the AS variant, and this effect is typical for flow systems exploiting in-line analyte concentration/separation steps [12]. The better peak symmetry associated with the AF variant is explained by recalling that larger dispersion the better peak symmetry [22]. With both donor and acceptor streams flowing, the sample dispersion before GD manifested itself also in the secondary analyte zone. Analysis of Fig. 2A reveals the beneficial influence of pulsed flows for both the AS and AF variants. Regarding the ammonia model species, higher peaks were recorded for pulsed flows, and this aspect is due to the inherent chaotic movement of the fluid elements, which improved the mixing conditions. Since R1 and M solutions were faster and more efficiently combined, a higher conversion rate of the considered species into its volatile form was attained, and the analytical repeatability was better (Table 2). Another aspect was the more efficient radial mass transport inherent to pulsed flows: the ammonia molecules in the bulk of the donor stream were more efficiently transferred towards the vicinity of the semipermeable membrane. Moreover, the pulsed fashion of the donor stream led to a mechanical movement of the flexible GD membrane that contributed to the homogenization of the acceptor solution. This effect improved the transfer of the newly collected molecules towards the acceptor stream bulk, thus avoiding the formation of a passive layer close to the membrane. As the ammonium ion did not accumulate near the acceptor stream/membrane interface, drawbacks associated with passivation effects were minimized, allowing high concentration factors to be attained. Similar tendencies manifested themselves for the other model species (Fig. 2), as a more symmetric peak shape for AF and a pronounced tailing effect for AS were noted. Considering that the selected model species interacted with the hydrophobic membrane in different manners due to their different geometries, polarities and/or dimensions, influence of flow pattern was however not so pronounced. Some peculiarities should be highlighted.
600
C.R. Silva et al. / Talanta 148 (2016) 596–601
Regarding EtOH (Fig. 2B), the beneficial influence of the pulsed flows was not evident, and higher peaks with less pronounced tailing effects were recorded for constant flow. Two aspects are probably involved: (i) as no reagent is needed for EtOH release, mixing conditions in the donor stream becomes less restrictive; (ii) the membrane vibrations caused by the flow pulsations might affect the ethanol transference in a different manner as compared to ammonia. It should be emphasized that, in spite of the membrane vibrations caused by the pulsed flows, membrane lifetime was not a relevant issue, as the same membrane could be used for several weeks. For CO2 (Fig. 2C), the recorded tracings related to pulsed flow and AF variant were characterized by poorly defined peaks and low repeatability, corroborating earlier findings [11]. The noisy recorded signal could probably be a consequence of the susceptibility of the conductometric detector to the sudden modifications in flow rates. The effect could be circumvented by adding dampers into the acceptor stream [11] but the procedure was not applied here for comparative purposes. Amongst the investigated model species, CO2 manifested itself as the less dependent on the flow pattern, and the differences in monitored signals related to AF and AS were more pronounced for CO2. In fact, the peak recorded for AF was broader and characterized by a long washing time, as a larger sample inserted volume was required specifically for this model species. A longer washing time is then evident in Fig. 2. It should be stressed that scale in Fig. 2 was maintained: consequently, it is not easy to observe the very smooth peak recorded by CO2 in relation to AF and constant flow. These characteristics were somewhat impaired when pulsed flows were applied, as stair-like peaks were always recorded. On the other hand, a higher CO2 amount was collected by the portion of the acceptor stream stopped inside the GD-cell, thus influence of the variant on the analytical sensitivity underwent a remarkable improvement. In comparison with the other investigated model species, the influence of AF or AS variant in the analytical sensitivity was most impressive in relation to CO2. In fact, a 5-fold sensitivity was attained. The beneficial influence of the AS variant and the pulsed flows were also evident for H2S, as the analytical signal almost doubled (Fig. 2D), and the general tendency above reported for ammonia was followed. The pronounced baseline drift noted during flow stopping in the AS variant was a consequence of the slow interaction of the R3 and R4 reagents. This effect was reproducible, manifested itself in absence of H2S, and might be considered as the analytical blank. For the AF variant, peak height related to constant flow was slightly higher.
less sensitive AF variant. As a rule, exploitation of pulsed flows improved the gas diffusion process, probably because of the higher mass transport towards the membrane and vice-versa, as well as the lower passivation effect involved. The term “passivation” is used here to specify the effects arising from the establishment of a thinner stagnant layer at the membrane/acceptor stream interface. Nevertheless, the system variant was the most relevant aspect in the context, especially in relation to H2S, NH3 and CO2. This can be explained by recalling that the tGD value was increased with AS, thus favoring the mass transport and increasing FP. The beneficial influence of pulsed flows was more expressive in AS, and is also due to the membrane movements promoted by the pulsations. This effect did not hold for constant flow, as improved phase transfer factor was noted only for EtOH and H2S in AS. With regard to the influence of flow pattern, better sensitivity was attained for pulsed flows (Table 2), which demonstrates that the above-mentioned favorable characteristics of this flow pattern improved the kinetics of gas diffusion. No defined tendency in measurement repeatability was observed by modifying flow pattern and/or system variant (Fig. 2 and Table 2). The relative standard deviations of peak heights were in general lower than 4%. The erratic measurements obtained for CO2 in relation to pulsed flows and AF variant were associated to the deformation of the recorded peak shown in Fig. 2C. The hypothesis of higher analytical frequency of pulsed flow systems relatively to constant flow systems was confirmed by analyzing three consecutive recorded peaks. The improved sampling rate noted for pulsed flows relied on the use of solenoid pumps instead of the syringe pumps, which required a time period for syringe refilling. Anyhow, this difference was due to instrumental constraints rather than flow pattern, and would not exist if a peristaltic pump were used instead the syringe pump. Regardless of this mechanical limitation, a significant difference in washing time was reported for pulsed and constant flows [4], and better analytical frequency was attained when solenoid pumps were used. In the present work, the increment in analytical frequency by exploiting pulsed flows occurred for all the volatile species studied. It is worthwhile to note that the system variant manifested itself as the most important factor on the analytical frequency. With AF, better sampling rate was attained for both flow patterns in relation to AS. Its combination with pulsed flow led to significantly faster flow procedures, especially for NH3 and H2S where the analytical frequency almost doubled.
4. Conclusions 3.2. The phase transfer factor The manifold geometry (including the membrane characteristics) and system dimensioning are relevant in calculating the FP value as the number of moles of the considered species collected by the acceptor solution divided by the number of moles present in the original inserted aliquot. As these characteristics were maintained, data in Table 2 can be taken into account for comparative purposes. The most evident feature in this Table is that the FP values were rather different for the investigated model species, emphasizing the differences in e.g. polarity, solubility, diffusion coefficients and size of the molecule. The mass transfer during the GD step, thus the analytical sensitivity, were dependent on the system variant and, to a lesser extent, on the flow pattern (Table 2 and Fig. 2). In the AS variant, a specific volume of the acceptor stream confined inside the GD-cell interacted with the continuously flowing donor stream, decreasing the acceptor/donor volumetric ratio, thus improving sensitivity. In contrast, a higher volumetric ratio was inherent to the potentially
Flow pattern plays a relevant role in flow systems with in-line gas diffusion, and the magnitude of its influence is dependent on the system variant and flow pattern. The chaotic movement of fluid elements and high radial mass transfer inherent to pulsed flows improved the mixing conditions with consequent enhancement of the gas diffusion. Nevertheless, the most important parameter in flow systems with in-line gas diffusion was the system variant, AF or AS, thus the tGD value. The influence of flow pattern was more evident for AS exploiting pulsed flows, which led to a pronounced sensitivity enhancement. However, modifications in flow pattern caused a slight effect in the sensitivity for the AF variant. Peak shapes and dispersion of the secondary zone were more affected by the acceptor and donor solution management rather than the involved flow pattern. In addition, use of pulsed flows improved the sampling rate, thus reducing the analytical time. In short: exploitation of pulsed flow is recommended for the automation of analytical procedures involving gas diffusion. The
C.R. Silva et al. / Talanta 148 (2016) 596–601
system variant should be judiciously selected as a compromise between sensitivity and sampling rate.
Acknowledgments This work was supported by the Spanish “Ministerio de Economía y Competitividad, Spain” through the project CTQ201347461-R, and by the National Research Council (CNPq, proc. 305378/2011-2), by and the S. Paulo Research Foundation (FAPESP, proc. 2011/23498-9) Brazilian funding agencies and CNPq, Brazil postdoctoral fellowships to C. Henríquez, R.M. Frizzarin and C.R. Silva (proc. 313660/2014-0, 201514/2014-1 and 160484/2013-8, respectively) are greatly appreciated.
References [1] Chapter 3: fundamentals, in: E.A.G. Zagatto, C.C. Oliveira, A. Townshend, P. J. Worsfold (Eds.), Flow Analysis with Spectrophotometric and Luminometric Detection, Elsevier, Amsterdam, 2012, pp. 45–93. [2] P.R. Fortes, M.A. Feres, M.K. Sasaki, E.R. Alves, E.A.G. Zagatto, J.A.V. Prior, J.L. M. Santos, J.L.F.C. Lima, Evidences of turbulent mixing in multi-pumping flow systems, Talanta 79 (2009) 979–983. [3] P. González, M. Knochen, M.K. Sasaki, E.A.G. Zagatto, Pulsed flows in flow analysis: potentialities, limitations and applications, Talanta 143 (2015) 419–430. [4] D.A. Weeks, K.S. Johnson, Solenoid pumps for flow injection analysis, Anal. Chem. 88 (1996) 2717–2719. [5] A.C.B. Dias, J.L.M. Santos, J.L.F.C. Lima, C.M. Quintella, A.M.V. Lima, E.A. G. Zagatto, A critical comparison of analytical flow systems exploiting streamlined and pulsed flows, Anal. Bioanal. Chem. 388 (2007) 1303–1310. [6] E.R. Alves, M.A. Feres, J.L.F.C. Lima, E.A.G. Zagatto, Exploiting pulsed flows for heating improvement: application to determination of total reducing sugars in molasses and sugar cane juices, Curr. Anal. Chem. 5 (2009) 65–69. [7] M.F.T. Ribeiro, A.C.B. Dias, J.L.M. Santos, J.L.F.C. Lima, E.A.G. Zagatto, Fluidized beds in flow analysis: use with ion-exchange separation for spectrophotometric determination of zinc in plant digests, Anal. Bioanal. Chem. 384 (2006) 1019–1024.
601
[8] M.K. Sasaki, M.A. Feres, E.A.G. Zagatto, Flow systems with MnO2-coated open tubular reactors for spectrophotometric determination of ascorbic acid in pharmaceutical products, Anal. Lett. (2015), in press, 10.1080/00032719.2014. 979353. [9] B. Karlberg, G.E. Pacey, Flow Injection Analysis. A Practical Guide, Elsevier, Amsterdam (1989), p. 111–118. [10] Chapter 8: sample handling, in: E.A.G. Zagatto, C.C. Oliveira, A. Townshend, P. J. Worsfold (Eds.), Flow Analysis with Spectrophotometric and Luminometric Detection, Elsevier, Amsterdam, 2012, pp. 295–448. [11] C. Henríquez, B. Horstkotte, V. Cerdà, A highly reproducible solenoid micropump system for the analysis of total inorganic carbon and ammonium using gas-diffusion with conductimetric detection, Talanta 118 (2014) 186–194. [12] Z.-L. Fang, Flow Injection Separation and Preconcentration, VCH Verlag Chemie, Weinheim (1993), p. 129–159. [13] M.D. Luque de Castro, Membrane-based separation techniques: dialysis, gas diffusion and pervaporation (Chapter 8), in: S.D. Kolev, I.D. McKelvie (Eds.), Advances in Flow Injection Analysis and Related Techniques, Elsevier, Amsterdam, 2008, pp. 204–234. [14] J. Klimundova, R. Forteza, V. Cerdà, A multisyringe flow injection system coupled with a gas diffusion cell for ammonium determination, Int. J. Environ. Anal. Chem. 83 (2003) 233–246. [15] R.M. Frizzarin, F.R.P. Rocha, A multi-pumping flow-based procedure with improved sensitivity for the spectrophotometric determination of acid-dissociable cyanide in natural waters, Anal. Chim. Acta 758 (2013) 108–113. [16] S.D. Kolev, P.R.L.V. Fernandes, D. Satinsky, Petr Solich, Highly sensitive gasdiffusion sequential injection analysis based on flow manipulation, Talanta 79 (2009) 1021–1025. [17] C. Henríquez, B. Horstkotte, V. Cerdà, Conductometric determination of ammonium by a multisyringe flow injection system applying gas diffusion, Int. J. Environ. Anal. Chem. 93 (2013) 1236–1252. [18] 〈http://www.lachatinstruments.com/products/products.asp〉 (accessed 08.15.). [19] R.T. Elsholz, T.C. Rodrigues, V. Cerdà, M. Tubino, Niedrigkostenmessverfahren für Acetat: Konduktometrische Untersuchungen der Gasdiffusion von Acetat mittels MSFIA, GIT Labor-Fachz. 1 (2008) 110–113. [20] Y.-J. Yuan, H. Kuriyama, Determination of hydrogen sulfide in a yeast culture solution by flow analysis utilising methylene blue spectrophotometric detection, Biotechnol. Lett. 22 (2000) 795–799. [21] C.R. Silva, T.F. Gomes, V.A.F. Barros, E.A.G. Zagatto, A multi-purpose flow manifold for the spectrophotometric determination of sulphide, sulphite and ethanol involving gas diffusion: application to wine and molasses analysis, Talanta 113 (2013) 118–122. [22] J. Ruzicka, E.H. Hansen, Flow injection analysis. Part X. Theory, techniques and trends, Anal. Chim. Acta 99 (1978) 37–76.
Contents lists available at ScienceDirect
Talanta journal homepage: www.elsevier.com/locate/talanta
A critical comparison of constant and pulsed flow systems exploiting gas diffusion Claudineia Rodrigues Silva a,b, Camelia Henriquez a,b, Rejane Mara Frizzarin a,b, Elias Ayres Guidetti Zagatto a,n, Victor Cerda b a b
Center for Nuclear Energy in Agriculture, University of Sao Paulo, 13400-970 Piracicaba, SP, Brazil Laboratory of Environmental Analytical Chemistry, University of Balearic Islands, 07122 Palma de Mallorca, Spain
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2015 Received in revised form 24 September 2015 Accepted 25 September 2015 Available online 28 September 2015
Considering the beneficial aspects arising from the implementation of pulsed flows in flow analysis, and the relevance of in-line gas diffusion as an analyte separation/concentration step, influence of flow pattern in flow systems with in-line gas diffusion was critically investigated. To this end, constant or pulsed flows delivered by syringe or solenoid pumps were exploited. For each flow pattern, two variants involving different interaction times of the donor with the acceptor streams were studied. In the first one, both the acceptor and donor streams were continuously flowing, whereas in the second one, the acceptor was stopped during the gas diffusion step. Four different volatile species (ammonia, ethanol, carbon dioxide and hydrogen sulfide) were selected as models. For the flow patterns and variants studied, the efficiencies of mass transport in the gas diffusion process were compared, and sensitivity, repeatability, sampling frequency and recorded peak shape were evaluated. Analysis of the results revealed that sensitivity is strongly dependent on the implemented variant, and that flow pattern is an important feature in flow systems with in-line gas diffusion. & 2015 Elsevier B.V. All rights reserved.
Keywords: Gas diffusion Flow analysis Pulsed flow Constant flow
1. Introduction Constant or pulsed flows have been generally exploited in flow analysis, and the related flow patterns may influence the performance of the analyzer. These flow patterns depend on the fluid propeller device, generally peristaltic or piston pumps for constant flows, or solenoid pumps for pulsed flows. Under constant flow, all fluid elements are displaced following parallel trajectories, and the pronounced radial gradient of the linear flow velocities leads to a parabolic profile of velocities inside the sample zone [1]. Consequently, laminar flow conditions are established. On the other hand, a pulsed flow is established by successive and sudden insertions of small solution aliquots, leading to a chaotic movement of the fluid elements. Turbulent mixing is then noted during the pulse insertions [2], and this aspect favors the radial mass transport and homogenization of solutions, thus improving the mixing conditions and reducing the sample dispersion. As the differences in linear speeds of the fluid lines are reduced, tailing effects are minimized. Influence of flow pattern in flow analysis has been emphasized in different applications [3], and the superior performance of n
Corresponding author. E-mail address: [email protected] (E.A.G. Zagatto).
http://dx.doi.org/10.1016/j.talanta.2015.09.064 0039-9140/& 2015 Elsevier B.V. All rights reserved.
pulsed flow systems has been demonstrated in relation to system portability [4], slow chemical reactions [5], heat transfer [6], solid phase extraction involving fluidized beds [7] and immobilized reagents [8]. In comparison with constant flow systems, pulsed flow systems are generally characterized by improved sensitivity, repeatability, analytical frequency, reduced sample and reagent consumptions and versatility. Gas diffusion (GD) is a worldwide exploited analyte separation/ concentration process and its implementation in flow analysis may overcome some drawbacks inherent to GD under batch wise conditions [9]. Several analytical procedures with in-line GD, relying on either constant or pulsed flows, have been recently reported [10], and the superior performance of pulsed flow systems in relation to constant flow systems was already stressed [11]. However, a systematic investigation of the influence of flow pattern in analytical systems with in-line GD was not yet carried out. Selectivity and sensitivity are dependent on the physical characteristics of the donor and acceptor streams (e.g. flow rate, chemical composition, pressure, and flow direction), manifold design, reagent concentrations and GD-cell geometry (including type of semipermeable membrane) [12,13]. Moreover, variants involving acceptor and donor streams continuously flowing (AF), acceptor stream stopped and donor stream flowing during GD (AS), exploitation of stream segmentation and use of oscillating streams have been exploited [14,15,16]. With the AS variant, in-line
C.R. Silva et al. / Talanta 148 (2016) 596–601
597
concentration of the volatile species, thus sensitivity, is generally favored [14,17]. The aim of this work was then to critically compare the influence of flow pattern in flow systems exploiting in-line GD. To this end, a multi-syringe flow injection system and a multi-pumping flow system, involving constant or pulsed flows, were designed as similar as possible to each other, the only difference being the fluid propelling device, piston or solenoid pumps for delivering constant or pulsed flows, respectively. Ammonia, ethanol, carbon dioxide, hydrogen sulfide were selected as model volatile species, and the AF and AS variants were considered. System performance was evaluated under all the investigated situations, in terms of efficiency of mass transport during GD, repeatability, sampling frequency and recorded peak shape.
2. Experimental 2.1. Solutions All solutions were prepared with analytical-grade chemicals from Scharlau SA (Barcelona, Spain) and deionized water (resistivity 418.2 MΩ cm) provided by a Milli-Q system. The reagents and standard solutions are presented in Table 1. 2.2. Apparatus A model Bu4S burette from Crison Instruments S.A. (Alella, Barcelona, Spain) equipped with four 5-mL glass syringes, model TLL SYR from Hamilton (Bonaduz, Switzerland), was used for establishing the constant flow inherent to the system in Fig. 1a. Each syringe head included a three-way solenoid valve accountable for directing the pumped solution towards either the manifold or the solution reservoir for refilling. An additional model STV-3 1/4UKG three-way solenoid valve from Takasago (Nagoya, Japan) was used for inserting the standard solutions. Its central port was connected to the ON port of the syringe head valve through a 2.2 mL holding coil (HC) and its ON and OFF ports were connected to the standard reservoir and to the manifold, respectively. This valve was controlled through an auxiliary supply port of the burette. In the pulsed flow system (Fig. 1b), four model P/N 120SP12205TV solenoid pumps with 25-mL stroke volume from Bio-Chem Inc. (Boonton NJ, USA) were used as liquid drivers. For delivering 1.0 or 2.0 mL min 1 flow-rates, the pumps were actuated in a synchronized way at 0.67 or 1.33 Hz, respectively. The three-way valve was likewise accountable for introducing either the standard (ON) or the carrier (OFF) solutions. The solenoid pumps and valves were controlled by a multi-pumping module from Sciware System SL (Bunyola, Spain). The RC1 and RC2 coiled reactors (length ¼100 cm, inner volume ca 0.5 mL), the holding coil and the transmission lines were made of PTFE (polytetrafluoroethylene) tubing (i.d. ¼0.8 mm), and the
Fig. 1. Flow diagrams of the constant (a) and pulsed (b) flow systems. M ¼ model species solution; Ri¼ reagents; C ¼carrier stream; MSB ¼ multi-syringe burette with Si syringes; Pi ¼solenoid pumps; V ¼three-way solenoid valve; HC ¼holding coil; RCi ¼ coiled rectors; GD ¼gas diffusion cell with the membrane specified by a traced line; reaction coil; D¼flow-through detector; W ¼flask for waste collection. Two syringes (S1, S3) or pumps (P1, P3), three syringes (S1, S2, S3) or pumps (P1, P2, P3) and four syringes or pumps are need for EtOH, NH3 or CO2 and H2S, respectively. For H2S, the acceptor stream is formed by converging the R2 and R3 streams, and mixing them inside the RC2 reactor. For details, see text.
connectors and GD-cell were made of PMMA (polymethylmethacrylate). PTFE tubes (length ¼30 cm, i.d.¼1.5 mm) were used for aspirating the reagent and carrier solutions towards the pulsed flow system. The GD-cell [17] was built-up by juxtaposing two identical rectangular PMMA blocks, each one with a U-shaped flow channel (width, depth, and length: 2.0, 0.7, and 128 mm). A Teflons hydrophobic GD membrane from Lachat Instruments (Loveland CO, USA), recommended for ammonia GD by the manufacturer [18], was placed between the blocks, thus establishing separated donor and acceptor channels (inner volume ¼ 180 mL). The donor and acceptor streams flew though these channels in a countercurrent way. Regarding conductometric detection, a lab-made flow cell [19] was connected to a model 525 conductometer from Crison Instruments S.A. The cell constant was calculated as 0.06 cm 1 by using a 0.01 mol L 1 KCl (1413 mS cm 1 at 25 °C) conductivity and standard solution. Two scales (0.1–200 mS cm 1 0.01–20 mS cm 1) were selected for NH3 and CO2 and their corresponding baselines were adjusted close to the maximum values in order to increase peak height. Measurements were done at every 0.2 s, and temperature correction was not applied because the laboratory temperature was stable enough [11].
Table 1 Composition of the involved solutions. Table refers to the flow systems in Fig. 1. Stream Model species NH3
CO2
EtOH
M
5.0–20.0 mg L 1 (as NH4Cl) 2.0–10.0 mmol L 1 (as NaHCO3) 10.0–60.0% (v/v)
C R1 R2
Water 25 mmol L 1 NaOH 25 μmol L 1 HCl
Water 25 mmol L 1 H2SO4 10 mmol L 1 NaOH
R3
–
–
Water – 0.3 mol L 1 K2Cr2O7 in 4.0 mol L 1 H2SO4 –
H2S 5.0–25.0 mg L 1 S2 (as Na2S 9H2O) in 25 mmol L 1 NaOH Water 0.5 mol L 1 HCl 5.0 mmol L 1 DMPD in 1.0 mol L 1 HCl 50 mmol L 1 Fe3 þ (as FeCl3 6H2O) in 1.0 mol L 1 HCl
598
C.R. Silva et al. / Talanta 148 (2016) 596–601
An USB-2000 miniature CCD spectrophotometer from Ocean Optics (Dunedin FL, USA) and a highly stable bright-white LED light source were used. Both were aligned with the optical path of a model 75.1 SOG 10-mm flow-through cuvette from Starna (Hainault, UK). Integration times were set as 30 and 75 ms for ethanol and H2S. The AutoAnalysis 5.0 software from Sciware Systems S.L. was used for instrumental control as well as for data acquisition and treatment. 2.3. Procedure Influence of flow pattern in the GD process was investigated with the flow systems in Fig. 1, which were designed with the same strategy for standard/reagent insertions and the same manifold architecture. For each model species (ammonia, carbon dioxide, ethanol and hydrogen sulfide), four procedures involving different flow patterns (constant/pulsed) and variants (AF/AS) were considered. In general, these procedures comprised: – Sample insertion. Either the model species (M) or the carrier (C) solution was pumped by P1 or S1, towards the analytical path during pre-defined time intervals, via the V three-way valve (Fig. 1). A sample zone was then established inside the M/C stream; – Formation of the volatile species. The M/C stream merged with the R1 reagent, which was pumped by P2 or S2. The donor stream, where the model species in the ionic form was converted to a volatile form, was then established. This stream was directed towards the GD-cell through the RC1 reaction coil, and wasted; – Diffusion of the volatile species. The formed NH3, CO2 or H2S volatile species and ethanol underwent diffusion through the hydrophobic membrane towards the acceptor stream; – Collection of the diffused species. The volatile species were collected into the acceptor stream as its ionic form, whereas ethanol was collected without conversion. A secondary sample zone was then established. The R2 stream pumped by P3 or S3 (Fig. 1), eventually combined with the R3 stream, was either flowing simultaneously with the donor stream (AF variant) or stopped during passage of the sample zone through the GD-cell (AS variant). – Measurement. The acceptor stream was directed towards detection cell and wasted. Passage of the secondary sample zone through the flow cell led to a transient modification in the monitored signal, which was recorded as a peak. Height of this peak constituted the measurement basis. Acid–base equilibria were involved in the formation of the gaseous ammonia, carbon dioxide or hydrogen sulfide in the donor stream as well as for their collections in the acceptor stream. The NH+4 and CO2− 3 ionic species established into the acceptor stream were conductometrically monitored. Regarding sulfide, the acceptor stream was a N,N-dimethyl-p-phenylenediamine (DMPD) plus Fe(III) acidic solution that reacted with S2 yielding methylene blue, which was spectrophotometrically monitored at 668 nm. Details of the involved chemistry are given elsewhere [20]. Ethanol was sufficiently volatile to be separated from the liquid phase without further conversion or temperature increment. Thus, the R1 reagent (Fig. 1) was not required and the M/C stream constituted itself as the donor stream. The collected ethanol reduced Cr(VI) to Cr(III), which was spectrophotometrically monitored at 600 nm [21]. Regarding system dimensioning, the R1 reagent was added by confluence aiming at an efficient homogenization with the M/C
stream. Its concentration was set in excess for improving the gas diffusion. Addition of this reagent via zone merging configuration [1] was not performed to avoid an eventual return of the volatile species back to the donor stream. In flow systems where an analyte separation/concentration step is not implemented, variations in flow rates do not affect sample dispersion in a pronounced manner. In contrast, flow rates are of utmost relevance in relation to in-line GD, as they determine the mean residence times of the fluid elements of the donor and acceptor streams inside the GD-cell, thus the time of contact with the membrane and the available time for the collection of the gaseous species by a given fluid element (tGD). Variations in flow rates may therefore alter the mass transfer of the volatile species, thus the analytical sensitivity. Influence of flow rates was then investigated under two extreme situations, associated with the AF and AS variants. In AF, the same flow rate was selected for the acceptor and donor streams, and tGD was minimal, whereas in AS, the acceptor stream was stopped (flow rate¼0) during passage of the entire sample zone through the GD-cell. In this later situation, tGD was maximal and transference of the gaseous species to the acceptor stream was improved. For both flow patterns and variants, the donor stream flow rate was fixed as 2.0 mL min 1. Other relevant parameters such as e.g. inserted volume, injection mode, reactor dimensions, temperature, reagent concentrations, hydrophobic membrane and GD-cell geometry were maintained, regardless of the flow pattern or variant involved. 2.4. The phase transfer factor In analytical procedures involving GD, the analyte is generally not completely transferred to the acceptor phase due to insufficient available time, low capacity of the collection medium and/or physico-chemical characteristics involved (e.g. distribution coefficients, diffusion coefficients, equilibrium constants, and molecule properties). Consequently, an “enrichment factor”, the ratio of analyte concentrations in the donor and acceptor media, should be estimated. In flow analysis however the concentration gradients along the sample zone, the lack of a steady analyte concentration, and the low tGD value may impair the accurate prediction of the enrichment factor. A phase transfer factor (FP) was then considered for roughly evaluating the mass transfer of the species of interest in flow analysis [12]. FP is the ratio of numbers of moles of the model species in the acceptor solution and in the inserted aliquot. The number of moles of the considered model species in the inserted aliquot was straightforwardly obtained, as standard solutions were used and the inserted volume was known. The number of moles collected by the acceptor stream was evaluated by taking into account the recorded peak area related to the secondary sample zone. To this end, the A vs t function (A¼instant absorbance or conductance; t ¼time) should be transformed into a C vs V function (C ¼concentration; V ¼volume). Abscise values were then multiplied by flow rate (here 2.00 mL min 1) and ordinate values were transformed in concentration. For spectrophotometric monitoring, the analytical response was considered. For this purpose, different standard solutions were successively used instead of the acceptor stream and the A values were modified according to the obtained analytical curve. For conductometric monitoring, this transformation was attained as in earlier work [11].
3. Results and discussion Flow-based methods are essentially kinetics, i.e. steady state situations are often not attained, resulting in partial development
C.R. Silva et al. / Talanta 148 (2016) 596–601
599
Table 2 Analytical characteristics F P ¼phase transfer factor, in %; r.s.d.¼ relative standard deviation of peak heights, in %; SR ¼sampling rate, in h 1; Sens.¼ sensitivity in μ S L cm 1 g 1, mS L cm 1 mmol 1, Abs (%v/v) 1 or¼ Abs L mg 1, relying on the NH3, CO2, EtOH or H2S concentration ranges specified in Table 1. Other data refer to 20 mg L 1 NH3, 10 mmol L 1 CO2, 20% v/v EtOH and 25 mg L 1 H2S. For details, see text.
Fp
Pulsed Constant
Sens.
Pulsed Constant
r.s.d.
Pulsed Constant
SR
Pulsed Constant
Fig. 2. Recorded tracings related to A: ammonia (20.0 mg L 1 NH+ 4 ); B: ethanol (20.0% (v/v)); C: carbon dioxide (10.0 mmol L 1 HCO− 3 ) and D: hydrogen sulfide (25.0 mmol L 1 S2 ) model species. Left and right portions: AS and AF variants. Full and traced lines: constant and pulsed flows.
of the involved processes. Even so, precise results are obtained, as reproducible handling conditions are inherent to the flow analyzers. With partial GD, the analytical signal is strongly dependent on different parameters, amongst them the flow pattern and the tGD value. Analysis of the recorded peak shape and evaluation of the phase transfer factor play then a relevant role in the context. 3.1. Recorded peak shape In flow analysis without a separation/concentration step, increasing the analytical path increases sample dispersion and tailing effects for both flow patterns, thus altering the recorded peak shape. For pulsed flows, influence of this parameter is less pronounced, due to the chaotic movement of the fluid elements that improves fluid homogenization, radial mass transport and mixing conditions [5]. Nevertheless, if the analytical path includes a GDcell, the recorded peak shape becomes dependent also on other related aspects, such as e.g. management of the acceptor and donor streams (AS or AF variants), cell geometry and membrane characteristics. This explains why the influence of flow pattern was not so pronounced (Fig. 2). Regarding acceptor and donor stream management, the recorder tracings in Fig. 2 were strongly influenced by the involved variant, AS or AF, for both flow patterns. With AS, the recorded peaks were higher and thinner, because the entire sample zone was focused in a low volume of the acceptor stream that was stopped inside the GD-cell acceptor channel (maximal tGD). Consequently, the analyte concentration in this volume was higher,
NH3
CO2
EtOH
H2S
AF AS AF AS
0.13 0.22 0.11 0.18
0.15 0.87 0.16 0.70
0.02 0.04 0.02 0.04
1.17 3.47 0.94 2.28
AF AS AF AS
2.2 3.3 1.5 2.4
0.09 0.45 0.10 0.39
0.014 0.020 0.014 0.019
0.024 0.063 0.014 0.038
AF AS AF AS
0.8 0.6 1.0 2.3
4 10 0.7 1.2 2.0
2.4 1.7 1.5 2.8
0.9 3.4 0.9 3.9
AF AS AF AS
75 39 51 32
47 35 20 15
49 32 39 29
75 39 31 16
leading to a higher peak whereas the low volume of the GD acceptor channel determined the thinner recorded peak. The peak shape reflected also a lower washing time and a less symmetrical peak for the AS variant, and this effect is typical for flow systems exploiting in-line analyte concentration/separation steps [12]. The better peak symmetry associated with the AF variant is explained by recalling that larger dispersion the better peak symmetry [22]. With both donor and acceptor streams flowing, the sample dispersion before GD manifested itself also in the secondary analyte zone. Analysis of Fig. 2A reveals the beneficial influence of pulsed flows for both the AS and AF variants. Regarding the ammonia model species, higher peaks were recorded for pulsed flows, and this aspect is due to the inherent chaotic movement of the fluid elements, which improved the mixing conditions. Since R1 and M solutions were faster and more efficiently combined, a higher conversion rate of the considered species into its volatile form was attained, and the analytical repeatability was better (Table 2). Another aspect was the more efficient radial mass transport inherent to pulsed flows: the ammonia molecules in the bulk of the donor stream were more efficiently transferred towards the vicinity of the semipermeable membrane. Moreover, the pulsed fashion of the donor stream led to a mechanical movement of the flexible GD membrane that contributed to the homogenization of the acceptor solution. This effect improved the transfer of the newly collected molecules towards the acceptor stream bulk, thus avoiding the formation of a passive layer close to the membrane. As the ammonium ion did not accumulate near the acceptor stream/membrane interface, drawbacks associated with passivation effects were minimized, allowing high concentration factors to be attained. Similar tendencies manifested themselves for the other model species (Fig. 2), as a more symmetric peak shape for AF and a pronounced tailing effect for AS were noted. Considering that the selected model species interacted with the hydrophobic membrane in different manners due to their different geometries, polarities and/or dimensions, influence of flow pattern was however not so pronounced. Some peculiarities should be highlighted.
600
C.R. Silva et al. / Talanta 148 (2016) 596–601
Regarding EtOH (Fig. 2B), the beneficial influence of the pulsed flows was not evident, and higher peaks with less pronounced tailing effects were recorded for constant flow. Two aspects are probably involved: (i) as no reagent is needed for EtOH release, mixing conditions in the donor stream becomes less restrictive; (ii) the membrane vibrations caused by the flow pulsations might affect the ethanol transference in a different manner as compared to ammonia. It should be emphasized that, in spite of the membrane vibrations caused by the pulsed flows, membrane lifetime was not a relevant issue, as the same membrane could be used for several weeks. For CO2 (Fig. 2C), the recorded tracings related to pulsed flow and AF variant were characterized by poorly defined peaks and low repeatability, corroborating earlier findings [11]. The noisy recorded signal could probably be a consequence of the susceptibility of the conductometric detector to the sudden modifications in flow rates. The effect could be circumvented by adding dampers into the acceptor stream [11] but the procedure was not applied here for comparative purposes. Amongst the investigated model species, CO2 manifested itself as the less dependent on the flow pattern, and the differences in monitored signals related to AF and AS were more pronounced for CO2. In fact, the peak recorded for AF was broader and characterized by a long washing time, as a larger sample inserted volume was required specifically for this model species. A longer washing time is then evident in Fig. 2. It should be stressed that scale in Fig. 2 was maintained: consequently, it is not easy to observe the very smooth peak recorded by CO2 in relation to AF and constant flow. These characteristics were somewhat impaired when pulsed flows were applied, as stair-like peaks were always recorded. On the other hand, a higher CO2 amount was collected by the portion of the acceptor stream stopped inside the GD-cell, thus influence of the variant on the analytical sensitivity underwent a remarkable improvement. In comparison with the other investigated model species, the influence of AF or AS variant in the analytical sensitivity was most impressive in relation to CO2. In fact, a 5-fold sensitivity was attained. The beneficial influence of the AS variant and the pulsed flows were also evident for H2S, as the analytical signal almost doubled (Fig. 2D), and the general tendency above reported for ammonia was followed. The pronounced baseline drift noted during flow stopping in the AS variant was a consequence of the slow interaction of the R3 and R4 reagents. This effect was reproducible, manifested itself in absence of H2S, and might be considered as the analytical blank. For the AF variant, peak height related to constant flow was slightly higher.
less sensitive AF variant. As a rule, exploitation of pulsed flows improved the gas diffusion process, probably because of the higher mass transport towards the membrane and vice-versa, as well as the lower passivation effect involved. The term “passivation” is used here to specify the effects arising from the establishment of a thinner stagnant layer at the membrane/acceptor stream interface. Nevertheless, the system variant was the most relevant aspect in the context, especially in relation to H2S, NH3 and CO2. This can be explained by recalling that the tGD value was increased with AS, thus favoring the mass transport and increasing FP. The beneficial influence of pulsed flows was more expressive in AS, and is also due to the membrane movements promoted by the pulsations. This effect did not hold for constant flow, as improved phase transfer factor was noted only for EtOH and H2S in AS. With regard to the influence of flow pattern, better sensitivity was attained for pulsed flows (Table 2), which demonstrates that the above-mentioned favorable characteristics of this flow pattern improved the kinetics of gas diffusion. No defined tendency in measurement repeatability was observed by modifying flow pattern and/or system variant (Fig. 2 and Table 2). The relative standard deviations of peak heights were in general lower than 4%. The erratic measurements obtained for CO2 in relation to pulsed flows and AF variant were associated to the deformation of the recorded peak shown in Fig. 2C. The hypothesis of higher analytical frequency of pulsed flow systems relatively to constant flow systems was confirmed by analyzing three consecutive recorded peaks. The improved sampling rate noted for pulsed flows relied on the use of solenoid pumps instead of the syringe pumps, which required a time period for syringe refilling. Anyhow, this difference was due to instrumental constraints rather than flow pattern, and would not exist if a peristaltic pump were used instead the syringe pump. Regardless of this mechanical limitation, a significant difference in washing time was reported for pulsed and constant flows [4], and better analytical frequency was attained when solenoid pumps were used. In the present work, the increment in analytical frequency by exploiting pulsed flows occurred for all the volatile species studied. It is worthwhile to note that the system variant manifested itself as the most important factor on the analytical frequency. With AF, better sampling rate was attained for both flow patterns in relation to AS. Its combination with pulsed flow led to significantly faster flow procedures, especially for NH3 and H2S where the analytical frequency almost doubled.
4. Conclusions 3.2. The phase transfer factor The manifold geometry (including the membrane characteristics) and system dimensioning are relevant in calculating the FP value as the number of moles of the considered species collected by the acceptor solution divided by the number of moles present in the original inserted aliquot. As these characteristics were maintained, data in Table 2 can be taken into account for comparative purposes. The most evident feature in this Table is that the FP values were rather different for the investigated model species, emphasizing the differences in e.g. polarity, solubility, diffusion coefficients and size of the molecule. The mass transfer during the GD step, thus the analytical sensitivity, were dependent on the system variant and, to a lesser extent, on the flow pattern (Table 2 and Fig. 2). In the AS variant, a specific volume of the acceptor stream confined inside the GD-cell interacted with the continuously flowing donor stream, decreasing the acceptor/donor volumetric ratio, thus improving sensitivity. In contrast, a higher volumetric ratio was inherent to the potentially
Flow pattern plays a relevant role in flow systems with in-line gas diffusion, and the magnitude of its influence is dependent on the system variant and flow pattern. The chaotic movement of fluid elements and high radial mass transfer inherent to pulsed flows improved the mixing conditions with consequent enhancement of the gas diffusion. Nevertheless, the most important parameter in flow systems with in-line gas diffusion was the system variant, AF or AS, thus the tGD value. The influence of flow pattern was more evident for AS exploiting pulsed flows, which led to a pronounced sensitivity enhancement. However, modifications in flow pattern caused a slight effect in the sensitivity for the AF variant. Peak shapes and dispersion of the secondary zone were more affected by the acceptor and donor solution management rather than the involved flow pattern. In addition, use of pulsed flows improved the sampling rate, thus reducing the analytical time. In short: exploitation of pulsed flow is recommended for the automation of analytical procedures involving gas diffusion. The
C.R. Silva et al. / Talanta 148 (2016) 596–601
system variant should be judiciously selected as a compromise between sensitivity and sampling rate.
Acknowledgments This work was supported by the Spanish “Ministerio de Economía y Competitividad, Spain” through the project CTQ201347461-R, and by the National Research Council (CNPq, proc. 305378/2011-2), by and the S. Paulo Research Foundation (FAPESP, proc. 2011/23498-9) Brazilian funding agencies and CNPq, Brazil postdoctoral fellowships to C. Henríquez, R.M. Frizzarin and C.R. Silva (proc. 313660/2014-0, 201514/2014-1 and 160484/2013-8, respectively) are greatly appreciated.
References [1] Chapter 3: fundamentals, in: E.A.G. Zagatto, C.C. Oliveira, A. Townshend, P. J. Worsfold (Eds.), Flow Analysis with Spectrophotometric and Luminometric Detection, Elsevier, Amsterdam, 2012, pp. 45–93. [2] P.R. Fortes, M.A. Feres, M.K. Sasaki, E.R. Alves, E.A.G. Zagatto, J.A.V. Prior, J.L. M. Santos, J.L.F.C. Lima, Evidences of turbulent mixing in multi-pumping flow systems, Talanta 79 (2009) 979–983. [3] P. González, M. Knochen, M.K. Sasaki, E.A.G. Zagatto, Pulsed flows in flow analysis: potentialities, limitations and applications, Talanta 143 (2015) 419–430. [4] D.A. Weeks, K.S. Johnson, Solenoid pumps for flow injection analysis, Anal. Chem. 88 (1996) 2717–2719. [5] A.C.B. Dias, J.L.M. Santos, J.L.F.C. Lima, C.M. Quintella, A.M.V. Lima, E.A. G. Zagatto, A critical comparison of analytical flow systems exploiting streamlined and pulsed flows, Anal. Bioanal. Chem. 388 (2007) 1303–1310. [6] E.R. Alves, M.A. Feres, J.L.F.C. Lima, E.A.G. Zagatto, Exploiting pulsed flows for heating improvement: application to determination of total reducing sugars in molasses and sugar cane juices, Curr. Anal. Chem. 5 (2009) 65–69. [7] M.F.T. Ribeiro, A.C.B. Dias, J.L.M. Santos, J.L.F.C. Lima, E.A.G. Zagatto, Fluidized beds in flow analysis: use with ion-exchange separation for spectrophotometric determination of zinc in plant digests, Anal. Bioanal. Chem. 384 (2006) 1019–1024.
601
[8] M.K. Sasaki, M.A. Feres, E.A.G. Zagatto, Flow systems with MnO2-coated open tubular reactors for spectrophotometric determination of ascorbic acid in pharmaceutical products, Anal. Lett. (2015), in press, 10.1080/00032719.2014. 979353. [9] B. Karlberg, G.E. Pacey, Flow Injection Analysis. A Practical Guide, Elsevier, Amsterdam (1989), p. 111–118. [10] Chapter 8: sample handling, in: E.A.G. Zagatto, C.C. Oliveira, A. Townshend, P. J. Worsfold (Eds.), Flow Analysis with Spectrophotometric and Luminometric Detection, Elsevier, Amsterdam, 2012, pp. 295–448. [11] C. Henríquez, B. Horstkotte, V. Cerdà, A highly reproducible solenoid micropump system for the analysis of total inorganic carbon and ammonium using gas-diffusion with conductimetric detection, Talanta 118 (2014) 186–194. [12] Z.-L. Fang, Flow Injection Separation and Preconcentration, VCH Verlag Chemie, Weinheim (1993), p. 129–159. [13] M.D. Luque de Castro, Membrane-based separation techniques: dialysis, gas diffusion and pervaporation (Chapter 8), in: S.D. Kolev, I.D. McKelvie (Eds.), Advances in Flow Injection Analysis and Related Techniques, Elsevier, Amsterdam, 2008, pp. 204–234. [14] J. Klimundova, R. Forteza, V. Cerdà, A multisyringe flow injection system coupled with a gas diffusion cell for ammonium determination, Int. J. Environ. Anal. Chem. 83 (2003) 233–246. [15] R.M. Frizzarin, F.R.P. Rocha, A multi-pumping flow-based procedure with improved sensitivity for the spectrophotometric determination of acid-dissociable cyanide in natural waters, Anal. Chim. Acta 758 (2013) 108–113. [16] S.D. Kolev, P.R.L.V. Fernandes, D. Satinsky, Petr Solich, Highly sensitive gasdiffusion sequential injection analysis based on flow manipulation, Talanta 79 (2009) 1021–1025. [17] C. Henríquez, B. Horstkotte, V. Cerdà, Conductometric determination of ammonium by a multisyringe flow injection system applying gas diffusion, Int. J. Environ. Anal. Chem. 93 (2013) 1236–1252. [18] 〈http://www.lachatinstruments.com/products/products.asp〉 (accessed 08.15.). [19] R.T. Elsholz, T.C. Rodrigues, V. Cerdà, M. Tubino, Niedrigkostenmessverfahren für Acetat: Konduktometrische Untersuchungen der Gasdiffusion von Acetat mittels MSFIA, GIT Labor-Fachz. 1 (2008) 110–113. [20] Y.-J. Yuan, H. Kuriyama, Determination of hydrogen sulfide in a yeast culture solution by flow analysis utilising methylene blue spectrophotometric detection, Biotechnol. Lett. 22 (2000) 795–799. [21] C.R. Silva, T.F. Gomes, V.A.F. Barros, E.A.G. Zagatto, A multi-purpose flow manifold for the spectrophotometric determination of sulphide, sulphite and ethanol involving gas diffusion: application to wine and molasses analysis, Talanta 113 (2013) 118–122. [22] J. Ruzicka, E.H. Hansen, Flow injection analysis. Part X. Theory, techniques and trends, Anal. Chim. Acta 99 (1978) 37–76.