Robotic-assisted dynamic large drop microextraction

Robotic-assisted dynamic large drop microextraction

G Model ARTICLE IN PRESS CHROMA-460416; No. of Pages 8 Journal of Chromatography A, xxx (xxxx) xxx Contents lists available at ScienceDirect Jour...

1MB Sizes 0 Downloads 50 Views

G Model

ARTICLE IN PRESS

CHROMA-460416; No. of Pages 8

Journal of Chromatography A, xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Robotic-assisted dynamic large drop microextraction Luis Felipe Rodríguez Cabal, Deyber Arley Vargas Medina, Adriel Martins Lima, Fernando Mauro Lanc¸as, Álvaro Jose Santos-Neto ∗ University of São Paulo, São Carlos, Institute of Chemistry of São Carlos, SP, Brazil

a r t i c l e

i n f o

Article history: Received 4 June 2019 Received in revised form 26 July 2019 Accepted 30 July 2019 Available online xxx Keywords: Automation Sample preparation Multi-batch single drop microextraction Large surface area drops Extraction device Multipurpose cartesian robot

a b s t r a c t By proper design of an innovative extraction device, a lab-made multipurpose autosampler was exploited in the automated performance of the dynamic large drops based microextraction. The pluses of this new analytical strategy were demonstrated in the determination of sulfonamides and fluoroquinolones in surface water samples, by direct immersion single drop microextraction (SDME) and liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) analysis. Operational autosampler features and critical experimental factors influencing SDME, including the extraction mode (static or dynamic), extraction, stirring rate, salt addition, drop size, number of cycles and drop exposition time, were comprehensively investigated using both univariate and multivariate optimization. The lab-made autosampler allowed to performance challenging dynamic and static large drop based SDMEs in an automated and effortless way and with minimal requirements of hardware and software. Large stable drops provided high surface area, enhancing the phase ratio and in consequence increasing the analytes uptake. The best extraction efficiencies were obtained as a result of the synergic interaction between the use of large drops and the automated dynamic mode of extraction. The developed method proved to be a reliable, sensitive, and robust analytical tool, with intraday RSDs ranging between 4.0 and 7.6% (n = 6), and interday RSDs between 4.8 and 9.3% (n = 6), and, LOD and LOQ in the range of 15–50 and 35–100 ng L−1 , respectively. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Simultaneously introduced by Dasgupta [1] and Jeannot & Cantwell [2], in the middle of 1990s, single-drop microextraction (SDME) was the first reported embodiment of liquid-phase microextraction (LPME). Nowadays, this technique is a wellestablished strategy for environmentally-friendly sample preparation. In the most commonly SDME approach, the analytes are extracted from the aqueous sample by a single drop of a waterimmiscible solvent hanging at the tip of a microsyringe needle [3]. The SDME efficiency is controlled by the physicochemical properties of the extraction solvent, the volume ratio between organic and aqueous phases, and the number of the extraction batches, among many others [4]. The nature of the extraction solvent impacts SDME, not only by its partition coefficient but also by its viscosity and volatility, factors on which depends the drop stability. SDME, as other equilibrium techniques, depends on the phase ratio. The higher the volumetric ratio between extraction drop and the sample (Vdrop /Vsmp ), the bigger the amount of analyte extracted. Hence, the use of large drops

∗ Corresponding author. E-mail address: [email protected] (Á.J. Santos-Neto).

could be an analytical strategy for enhancing chromatographic sensitivity [5,6]. Nevertheless, the possibility to use a large volume of extraction solvent is limited by the drop stability. Several lab-made devices have been developed aiming to increase the drop volume and its stability [7,8]. Usually, these reported devices are attached directly to the tip of a microsyringe needle, hindering the full automation of the process. In this paper, we describe a simple and innovative device [9], which: (i) allows the use of large and stable drops and (ii) makes possible the full automation of the SDME process. Multi-batch SDME procedures can be performed using the dynamic extraction mode. In this approach, the microsyringe plunger moves periodically up and down, allowing the sequential exposition/retrieval of the drop, with the partial renovation of the extraction solvent. Hence, extraction recovery increases and the extraction time can be reduced [10]. On the other hand, successful performance of dynamic extractions strongly dependents on the process automation. The microsyringe plunger must be actuated at a constant speed and the drop volume must be controlled with high precision and accuracy. SDME has been previously automated by using both flow systems and robotic fashion setups (autosamplers) [11,12], being the latter the preferred approach in chromatographic applications [10]. Nowadays, autosamplers are an important prospect in LPME

https://doi.org/10.1016/j.chroma.2019.460416 0021-9673/© 2019 Elsevier B.V. All rights reserved.

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model

ARTICLE IN PRESS

CHROMA-460416; No. of Pages 8

L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

2

Table 1 Concentration of sulfonamides and fluoroquinolones found in water samples. Analytes

Matrix

Conc. Found [ng L−1 ]

Method

Ref.

Sulfonamides Sulfamethazine Sulfamethoxazole Sulfamethoxazole Fluoroquinolones Norfloxacin Norfloxacin

Pig farm pound Water samples Sewage River waters River waters Wastewaters Water streams

447-6662 82.5 76 -1720 510 265-459 11200 239-310

SPE-LC-MS/MS SPE-LC-MS/MS SPE-LC-MS/MS SPE-LC-UV SPE-LC- FLD SPE-LC-FLD SDME-LC-MS/MS

[28] [29] [30] [31] [31] [32] This work

automation and a wide diversity these instruments are commercially available [13], even with the capability to provide multiple sample processing in 96-well plate format [14,15]. Nevertheless, the high prices of the commercial autosamplers is still a limiting factor in his spread use an implementation. Open-source robotics has become a topic of high interest in the development of dedicated, autosamplers [16], and diverse sample treatment systems [17–22]. By proper design of the extraction devices, we explored the capabilities of a dedicated lab-made autosampler [23,24] in the improvement of the solvent-based microextraction and the findings are shown herein. The aim of this study has been demonstrating the analytical gains taken of the use of large and stable drops in the performance of dynamic SDMEs by exploiting the automation of the technique. As a proof-of-concept, the developed automated large drop based method was assessed in the microextraction of sulfonamides (SAs) and fluoroquinolones (FQs) in surface water (Table 1, [25–28]), followed by LC–MS/MS analysis. Factors affecting the feasibility, precision and the extraction efficiency were carefully studied. The developed method proved to be an innovative approach towards the automated use of solvents for sample preparation, in the determination of emergent organic pollutants. 2. Experimental section 2.1. Reagents and materials Sulfadiazine (SDZ), sulfathiazole (STZ), sulfamerazine (SMR), norfloxacin (NOR) and enrofloxacin (ENR) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The stock standard solutions were prepared by dissolving the analytes (1 mg mL−1 ) in acetonitrile containing phosphoric acid (0.1%), while the working solutions were prepared daily in ultrapure water. Sulfamethoxazole (13 C) was used as an internal standard. Standard solutions were stored in amber bottles at −20 ◦ C. Methanol, acetonitrile, dichloromethane (DCM), chloroform (Chl) and ethyl acetate (EtOAc) (chromatographic analysis grade) were purchased from Tedia (Fairfield, OH, USA). Formic acid (MS grade, 98%) was purchased from Sigma–Aldrich (St. Louis, MO, USA), and Na2 SO4 used for ionic strength study was purchased from J.T.Baker (Phillipsburg, NJ, USA). Ultrapure water was produced in a Milli-Q purification system (Millipore, USA). For method development, standard solutions of each antibiotic (50 ng mL−1 ) were prepared in lab-made aqueous media simulating a typical sewage composition [29]. Stream samples were collected, from three different streams from São Carlos (SP, Brazil). Sampling sites and location details are shown in Fig. S1 (Supplementary Information). The samples were collected and stored in plastic flasks (500 mL) at −20 ◦ C. Before the analysis, a filtration was made through a 0.22-␮m cellulose membrane. 2.2. Extraction procedures for automated SDME All the SDME procedures, including filling the syringe with the extraction solvent, withdrawing the solvent to the syringe barrel,

and final extract dispense, were auto-performed with a lab-made autosampler [23] (Fig. 1B). Programs, for automated performance of dynamic and static SDME, were written in the Arduino Integrated Development Environment (IDE) [30], and are attached as Supplementary information. To make compatible the formation of large drops and the process automation, the 10 mL sample reservoirs were equipped with a lab-made polyfluoroetilene (PTFE) adjustable needle endpoint drop-expander [9], as explained in Fig. 1A. For static SDME, the lab-made autosampler was equipped with a 500 ␮L, removable needle, Gastight 1750 microsyringe (Hamilton, USA). The autosampler filled the syringe with the extraction phase (plunger flow rate: 20 ␮L s−1 ), pierced the vial septum, drove the needle through the drop-expander and slowly depressed the plunger to expose a drop into the 8-mL sample solution, under magnetic stirring (1 cm stir bar). After the defined extraction time, the drop was withdrawn into the barrel and the extract dispensed into a 500-␮L conic vial. The extract was dried under a gentle nitrogen stream and reconstituted into 50 ␮L of ultrapure water. All the involved static SDME variables were studied and optimized as described in the next Section (2.3). For dynamic SDME, the lab-made autosampler was equipped with the same 500-␮L Gastight microsyringe. As in the static SDME, the autosampler filled the syringe with the extraction phase, pierced the vial septum, drove the needle through the dropexpander and slowly depressed the plunger to expose a drop into the 8-mL sample solution, under magnetic stirring (1 cm stir bar). After a defined static time, the drop was slowly withdrawn into the barrel of the syringe and, subsequently, re-exposed (plunger flow rate: 20 ␮L s−1 ). This exposure-withdrawal procedure was repeated a defined number of times (cycles) before the final extract was collected, dried under a nitrogen stream and reconstituted into 50 ␮L of ultrapure water for injection in the chromatographic system. All the involved dynamic SDME variables were studied and optimized as described in the next Section (2.3). 2.3. SDME procedure optimization Univariate and multivariate analysis, including a central composite design, was used to comprehensively access the effect of the variables on SDME performance. Data were processed by Statistica 13 (StatSoftInc, Tulsa, USA, 2013). The process optimization was carried out in four stages: (i) Type of extraction solvent (EtOAc, DCM, Chl and its equivolumetric mixtures), the volume of the extractant drop (30–90 ␮L), stirring rate (200–600 rpm) and effect of the salt addition were evaluated by univariate analysis. In this stage, all the extractions were carried out under a generic dynamic extraction mode, using 300 ␮L as the total volume of extraction solvent and 70 exposure-withdrawal cycles (20 s of static drop exposition). (ii) Comparative performance of the two extraction modes was studied using four different total extraction times. A 90-␮L drop of dichloromethane/ethyl acetate mixture (1:1, v/v) was exposed into the 8-mL sample solution, in a sequence of

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model CHROMA-460416; No. of Pages 8

ARTICLE IN PRESS L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

3

Fig. 1. Instrumental setup employed for automated performance of SDME. A) Picture of the lab-made PTFE adjustable needle endpoint drop-expander; B) Picture of the lab-made autosampler and the integrated use with the drop-expander to form the large and stable drops.

exposure-withdrawal cycles (10, 30, 50 and 70 cycles) for dynamic mode, and in an equivalent extraction time (5 , 15, 26 and 38 , respectively) for static mode. The complete performed experimental design is attached as supplementary information (Table S1). (iii) The effect of the total volume of extractant solvent into the SDME syringe and its interaction with the extraction mode were investigated by a 22 factorial design. The two levels for each factor were set up as 100 ␮L (−) and 300 ␮L (+) for the total volume of extractant solvent and, static (−) and dynamic (+) for the extraction mode. A 90-␮L drop of dichloromethane/ethyl acetate mixture (1:1, v/v) was exposed into the 8-mL sample solution, in a sequence of 70 exposure-withdrawal cycles (20 s of static drop exposition) in dynamic experiments, and for an equivalent extraction time of 38.5 min in the static mode. The complete performed experimental design is attached as supplementary information (Table S2). (iv) Finally, the dynamic SDME process was optimized as a function of the number of extraction cycles and the static exposure time of each drop. A central composite design was carried out, √ √ over the experimental region between 30 (− 2) to 109 (+ 2) √ √ exposure-withdrawal cycles and 5 (− 2) to 25 s (+ 2) of static exposure time of each drop. The complete performed experimental design is attached as supplementary information (Table S3). 2.4. Chromatographic analysis The chromatographic optimization stage was carried out in a Shimadzu (Kyoto, Japan) LC system equipped with CBM-20A communication bus module, SIL-20AC autosampler, DGU-20AS degasser, two LC 20-AD pumps, CTO-20A column oven and SPD20A UV–vis detector monitoring at 270 and 290 nm. Water and acetonitrile were selected as mobile phases, both contained 0.1% of formic acid. The flow rate was set at 0.200 mL min−1 . Initially, an isocratic elution was programmed with 7% of acetonitrile for 5 min. From 5 to 15 min a gradient elution was performed by linearly increasing the acetonitrile content from 7 to 32%, then changing linearly to 7% of organic modifier in 4 min. For the validation and application stages, a more sensitive online-SPE-LC–MS/MS system was used. The analytes were trapped on an SPE column (2.1 × 20 mm, 25 ␮m) packed with a hydrophilic-

lipophilic balanced polymer (Oasis HLB) from Waters. The LC–MS analysis was performed on an Agilent 1260 series (|Palo Alto, CA), equipped with binary LC pumps, an ALS 1200 autosampler, a thermostatted column compartment 1200 coupled to a quadrupole-linear ion trap 5500 QTRAP (AB SCIEX, Foster, CA) with electrospray ionization. MRM mode was used for quantitative and qualitative identification. Separations were accomplished through a Poroshell 120 EC-C18 column (100 mm 2.1 mm i.d., 2.7-␮m particle size) from Agilent Technologies, and an acetonitrile/water gradient used as mobile phase. The flow rate of mobile phase was set at 0.6 mL min−1 . An Analyst data acquisition software ver. 1.6.1 was used for data acquisition and processing. The online SPE-LC system was configured as described by Gomes et al. [29]. 2.5. Mass spectrometry parameters Selected fragments for each analyte and parameters like collision energy (CE), declustering potential (DP) and collision cell exit potential (CXP) are summarized in Table S4. Ion spray source was operated in positive ion (PI) mode with an entrance potential fixed at 10 V. The QTRAP tandem MS was operated in selected reaction monitoring (SRM) mode using a dwell time of 20 ms per channel. Three SRM transitions were monitored for each compound (Table S4); being the resolution at the first and third quadrupole (Q1 and Q3) set to unitary and inter-scanning delay was 5 ms. 3. Results and discussion Typical solvent-based extractions are equilibrium-based processes and may present variable efficiencies. To enhance analyte recovery, both equilibrium and kinetic extraction considerations should be taken into account and the influencing parameters must be carefully modulated [31]. To obtain the most favorable SDME condition, we used the peak area of the analytes (under UV detection) as the chromatographic response to evaluate the extraction efficiency. 3.1. Univariate optimization 3.1.1. Effect of the extraction solvent A suitable extractant solvent for SDME direct immersion should be water-immiscible, able to provide stable drops and capable to

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model CHROMA-460416; No. of Pages 8

ARTICLE IN PRESS L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

4

provide high enrichment factors of the target analytes with appropriate selectivity. Generally, the principle “like dissolves like” guide the solvent selection. Sulphonamides (SAs) and fluoroquinolones (FQs) are amphoteric compounds, with hydrophobic character at pH 6–8, but with notable differences in polarity [32]. The studied SAs range in polarity from log kow −0.09 to log kow 0.14 and the selected FQs between log kow 1.03 and log kow 1.1. Simultaneous determination of SAs and FQs had been performed in most of the cases by solid-phase extraction [33]. In SDME applications, SAs had been extracted with ionic liquids [34] and FQs with basic aqueous solutions in three-phase systems, by immersion of the acceptor drop in a thin flat layer of Chl or DCM/toluene, in single drop liquid–liquid–liquid microextraction approaches (SD-LLLME) [35,36]. To the best of our knowledge, there is no previous report that describes the simultaneous extraction of SAs and FQs by SDME. For the simultaneous extraction of SAs and FQs, three organic solvent candidates, DCM, Chl, EtOAc, and their equivolumetric mixtures, were evaluated. The best extraction efficiency was obtained when a DCM/EtOAc (1:1, v/v) mixture was used. As shown in Fig. S2, the DCM/EtOAc (1:1, v/v) mixture provided the best uptakes for all selected SAs and, in the case of FQs, similar extraction efficiency when compared to the other tested solvents. Consequently, the mixture DCM/EtOAc (1:1, v/v) was selected as the extractant solvent for further experiments. 3.1.2. Stirring rate effect The rate constant (k) of analyte uptake by an extractant drop mainly depends on the diffusion coefficient of the analyte in the sample and the thickness of the stagnant layer around the drop [3]. Hence, sample-stirring allows the constant renovation of the stagnant layer, being that a simple and important parameter to modulate the analyte mass transference in LPME [37]. Magnetic stirring can increase the mass transfer between the sample and the drop, but at the same time, it can cause drop dislodgement. Stirring rates were evaluated at 0, 200, 400 and 600 rpm. As shown in Fig. S3, peak areas of the target compounds increase with the increase of the stirring rates. Nevertheless, at stirring rates > 600 rpm, drop stability decreases considerably. Hence, a stirring rate of 600 rpm was used in this study. 3.1.3. Effect of salt addition When an electrolyte is added to an aqueous solution the solubility of the organic molecules decrease; this effect is known as salting out effect [38]. The study of salt addition effect was carried out through the addition of Na2 SO4 . In this case, salt addition caused drop instability, probably due to the increase in the sample density. As a consequence, we choose to extract both sulfonamides and fluoroquinolones in surface water samples without salt addition. 3.1.4. Effect of the drop size In liquid–liquid extraction, the fraction of analyte extracted (E) is directly proportional to the phase ratio ( ), according to Eq. (1). E=

KD  1 + KD 

(1)

where KD is the distribution constant [31]. In microextraction,  values usually range from 0.001 to 0.01. Hence, even though low analyte recoveries are normally observed, high analytes concentrations in the final organic extract normally lead to enhanced analytical responses [31]. This statement works very well for gas chromatography, where the extractant solvent (nature and volume) is compatible with the injection conditions. For liquid chromatography, most of the organic extractant solvents used are not compatible with the reversed-phase mode of elution. Therefore, the organic extracts have to be dried and reconstituted to be compatible with the initial liquid chromatography mobile phase

composition [39,40]. In this case, higher phase ratios can be used and the extract dried and reconstitute in a small portion of injection solvent, resulting in enhanced analyte recoveries. Additionally, in SDME the interfacial area increases proportionally with the drop volume. The larger the drop, the more efficient mass transfer process is [41]. On the other hand, the use of hanging large drops is limited by its stability [42]. The maximum volume of a drop suspended at the tip of a vertical capillary can be expressed by Eq. (2): Vdmax =

2Rm g

(2)

where Rm is the external capillary radius,  is the interfacial tension,  is the density difference between the interior droplet phase and the exterior matrix phase, and g is the gravitational acceleration [42]. Several attempts to increase the size and the superficial area of extractant drops have been reported [8]. Recently, the use of drops with volumes higher than 30 ␮L in SDME has been accomplished employing Lab-In-Syringe automated systems [43]. In this study, we developed a PTFE adjustable needle endpoint drop-expander, which allows the use of large drop volume, reducing the probability of its detachment. In the case of traditional SDME, the drop is a small hanging ball with high detachment probability. The use of dropexpanders leads to the formation of drops with the shape of half flat ball, strongly adhered to the stem, reducing the probability of drop detachment [8]. The developed drop-expander has an Rm = 0.35, being able to provide drops with theoretical Vmax of 125 ␮L, 339 ␮L and 592 ␮L with Chl, DCM and EtOAc, respectively. Nevertheless, under stirring conditions (600 rpm), the experimental Vmax obtained was 120 ␮L and drops with volume >90 ␮L were susceptible to detachment and not suitable for prolonged extractions. To obtain the best compromise between recovery, drop stability and time, using the lab-made autosampler and the developed dropexpander, we studied the extraction efficiency of extractant drops with volumes ranging from 30 to 90 ␮L. The drops were stable and suitable for static and dynamic extractions in all the cases. As shown in Fig. 2, as the volume of the extractant drop increases, the chromatographic areas increase, too. Analytical responses improved between 40–60% when the drop size changed from 30 ␮L to 90 ␮L. The best extraction performance was obtained at 90 ␮L. Hence, this drop volume was selected for the performance of further experiments. 3.2. Comparison between extraction modes: dynamic vs static SDMEs can be performed in a static and/or dynamic mode of extraction [10]. While static procedures are carried out by the continuous exposition of the extractant drop, dynamic extractions are performed out in a sequence of cycles of drop exposition/retrieval. In most of the cases, SDMEs are carried out manually, with suitable performance in static mode of the extraction. Manual performance of dynamic extractions is an arduous and prone to errors task. To overcome this challenge and investigate the full potential of the dynamic extraction mode, we exploited an automated system able to perform a large number of exposition/retrieval cycles, with rigorous and reliable control of the exposure time and the plunger movement. At low times of extraction, the performance of static and dynamic procedures are similar. However, in prolonged extractions (>15 min) the dynamic mode of extraction provide enhanced analyte uptake (Fig. 3). 3.2.1. Study of the total extraction time To optimize the extraction time is important to obtain a reasonable compromise between extraction performance and sample throughput. The amount of analyte extracted should increase proportionally with the extraction time and maximum extraction

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model CHROMA-460416; No. of Pages 8

ARTICLE IN PRESS L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

5

Fig. 2. Effect of the drop volume on the extraction efficiency.

Fig. 3. Relative extraction efficiency of the dynamic and static SDME modes.

efficiency could be obtained under equilibrium conditions. However, in SDME the time needed to reach equilibrium may be prolonged, according to the Eq. (3): Co (t) = Coequ (1 − e−kt )

(3)

where k is the rate constant, and depend on the interfacial area and the mass-transfer coefficients for the drop and the sample solution [3,37,44]. At the static mode of extraction, a linear relationship between the amount of extracted analyte and the extraction time has been previously described for small extractant drops (<20 ␮L) [10]. In this study, using a DCM/EtOAc (1:1) 90 ␮L drop and 300 ␮L of total extraction solvent, just ENR showed that previously described behavior, when the extractions were carried out in the static mode (Fig. 3). For NOR and all tested SAs the expected relation between the mass uptake and the extraction time was not observed. Additionally, recoveries of ENR were superior to all other analytes at all the tested extraction times. ENR is the most hydrophobic compound among the investigated analytes (log kow 1.1) and the most

prone to partitioning to the organic solvent. Due to the partial miscibility of the EtOAc in water, the volume of the extractant drop didn’t remain constant during the static extractions. Shrinkage of the extractant drop could take place with partial resolubilization of the more polar analytes, impeding their growing enrichment across the extraction times. Therefore, only the more hydrophobic analytes remain preferentially in the organic phase, accumulating across the time. At the dynamic extraction mode, a linear relationship between analyte recovery and the number of extraction cycles has also been reported [10]. We observed similar behavior for all tested analytes (Fig. 3). Drop shrinkage effect was compensated by the solvent recycling process. By appropriate programming of the labmade autosampler, it was possible to redefine the end position of the syringe plunger after each cycle. The plunger of the syringe descends slowly, cycle after cycle, allowing to use the remaining solvent in the syringe barrel to compensate the loss of volume of the drop (0.5 ␮L by cycle), maintaining constant the interfacial area across the whole extraction process. The overall result depicted by

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model CHROMA-460416; No. of Pages 8 6

ARTICLE IN PRESS L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

Fig. 4. Pareto chart for the evaluation of (i) dynamic vs static mode, (ii) the total volume of extraction solvent loaded into the syringe and (iii) their interaction as shown by the STZ extraction efficiency.

Fig. 3 shows a far superior extraction efficiency for the dynamic extraction mode when compared to the static mode, under the same extraction time frame. 3.2.2. Study of the total volume of extractant solvent According to Eq. (1), enhanced analyte uptakes can be achieved by increasing the volume of the total extraction phase. Therefore, it is possible to take advantage of the total volume of extractant solvent loaded into the microsyringe, when the dynamic mode of extraction is used. In this case, the interfacial area is periodically renewed, offering a dilute fresh surface at each cycle and resembling a multi-batch liquid–liquid extraction in miniaturized scale. To evaluate the effectiveness of the total volume loaded into the microsyringe, in association with the dynamic extraction, two different volumes (100 and 300 ␮L) were studied and the uptakes obtained using the static and dynamic modes were compared using a 22 factorial experimental design. Fig. 4 shows the Pareto charts obtained for sulfathiazole (STZ), indicating the statistical significance of both effects and its positive interaction on the chromatographic responses at 95% confidence level; similar results were obtained for all evaluated analytes and their Pareto charts have been attached as supplementary information (Fig. S4). With a lower but still significant effect, the increase of the total volume of the extraction solvent yield improved analyte uptakes, at both extraction modes. The dynamic extraction mode leads to superior analyte extractions at both low (100 ␮L) and high (300 ␮L) levels of total extractant solvent. The significant and positive interaction of both factors pointed out that the most favorable extractions were obtained utilizing the association between the dynamic extraction mode and the total extraction volume of 300 ␮L. 3.2.3. Multivariate optimization of the dynamic SDME process In dynamic SDME, a short exposure time, a reduced number of extraction cycles, or both, can negatively affect the extraction efficiency. However, prolonged extraction times, excessive extraction cycles, or both, can cause the re-dilution of the analyte into the sample, drop dislodgment, and low analytical throughput. Thus, the exposure time and the number of cycles in which the equilibrium of extraction is reached must be optimized. For determination of the optimum number of cycles and exposure time of the drop, a central composite design was carried out. All the experiments were randomly done and are summarized in Table S3. The number of cycles and the exposure time were both statistically significant (p < 0.05). To obtain an overall condition for all analytes, the param-

Fig. 5. Desirability response surface for the factors (i) time of exposition of each drop formed into the sample solution and (ii) number of exposure-withdrawal cycles of the drop into the sample solution.

eters were observed using the desirability function (DF). The DF allows predicting the best combination of several cycles and exposure time of the drop, to obtain the maximum response values for the analytes (Fig. 5). According to the desirability response surface, the analytical response depends upon the linearly on the exposure time and exponentially on the number of cycles. Additionally, the torsion of the response surface and Pareto charts showed a positive interaction between these variables, meaning that the best combination between the factors should take into account their synergy (number of cycles and exposure time). The generated mathematical models showed that for each exposure time, the maximum extraction performance is obtained after 100 expositions/retrieval cycles. When a few extraction cycles are carried out, the increase in the exposure time does not lead to enhanced analyte uptakes. The exposure time became significant above approximately 40 extraction cycles and its effect increase with the number of extraction cycles. Under the employed experimental conditions, the maximum extraction capability was obtained after 109 expositions/retrieval cycles and 25 s of exposure time. To avoid an excessively long sample preparation time, the compromised condition selected for further experiments were 70 extraction cycles and 20 s of exposition time. 3.3. Analytical performance and sample analyses The analytical curves were obtained at six different concentration levels by spiking blank sewage water samples and analyzing by the optimized SDME method. Linear dynamic ranges (LDRs), limits of detection (LODs) limits of quantification (LOQs), enrichment factors (EF) and the extraction recovery (ER) were determined and are summarized in Table 2. LODs and LOQs were established as the lowest concentrations that produced a signal three and ten times higher than the signal noise, respectively. Accuracy, inter-day and intraday precision were established by spiking samples at three different concentration levels (100, 1500, 3000 ng L−1 ) on triplicates and three different days. Enrichment factors were estimated as the ratio between the concentration of the analytes in the drop and the sample at the end of the extraction process (EF = Cdrop /Csample ). The concentration of the analytes in the drop was calculated from a calibration curve obtained by direct injection of the analytes in a

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model

ARTICLE IN PRESS

CHROMA-460416; No. of Pages 8

L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

7

Table 2 Analytical parameters for the proposed automated D-SDME-LC–MS/MS method. Analytes

Linearity (ng L−1 )

Slope

Intercept

r2

LOD (ng L−1 )

LOQ (ng L−1 )

Intra-day RSD (%)

Inter-day RSD (%)

SDZ STZ SMZ NOR ENR

50–4000 50–4000 50–4000 100–3000 100–3000

0.0174 0.0269 0.0311 0.0009 0.0015

0.0220 −0.0009 −0.0027 0.0146 0.0102

0.989 0.998 0.996 0.9981 0.9984

15 15 15 35 35

50 50 50 100 100

7.6 4.4 4.5 4.3 4.0

9.3 8.1 6.7 5.7 4.8

Table 3 Results for the spike and recovery determination of target antibiotics in surface water samples. Analytes

Surface water

SDZ STZ SMZ NOR ENR

Spiked (␮g L−1 )

Found (␮g L−1 )

Recovery (%)

25 25 25 25 25

18 19.2 16.7 15.5 14.3

72 76.8 66.8 62 57.2

mixture of dichloromethane/ethyl acetate (50:50). The extraction recovery was obtained by the product of the EF and the phase ratio (Vdrop /Vsample ) by 100, according to the Eq. (4):



ER =

ndrop nsample





× 100 =

Cdrop

Vdrop

Csample VSample

EF

RF (%)

15.8 17.6 12 10.5 9

23.7 26.4 18 15.75 13.5

of proper devices made possible the improvement of SDME analytical performance. An innovative approach associates large and renewable drops in exposition/retrieval. After proofing this concept, the method was applied with precision, accuracy, selectivity, and sensitivity to analyze sulfonamides and fluoroquinolones in surface water samples. Finally, the developed setup demonstrated potential capabilities for the development of future applications of SDME and in the automation of other miniaturized modes of LPMEs, including online coupling with liquid chromatography and or mass spectrometry instruments [24]. Declaration of Competing Interest The authors have declared no conflict of interest.



× 100

(4)

3.4. Proof of concept The developed method was successfully applied to the determination of the target antibiotics in water streams subjected to sewage contamination in the city of São Carlos (SP, Brazil). Samples from three different spots were analyzed and the recoveries values varied between 76.8 and 57.2% (Table 3). Recoveries in this range have been considered acceptable in microextraction techniques, including SPME, SBSE, MEPS, and others. Furthermore, values up to 80% were considered unusual in LPME procedures according to Rasmussen and Pedersen-Bjergaard [45]. Among the antibiotics evaluated in this study, only norfloxacin was detected in two of three examined points: point 1 (239 ng L−1 ) and point 3 (310 ng L−1 ).

Acknowledgments The authors acknowledge the financial support from grants #2010/19910-9, #2014/03795-0, #2016/21950-5 and #2017/02147-0, from São Paulo Research Foundation (FAPESP). We also acknowledge grants #459326/2014-7, #307293/2014-9 and #311300/2015-4 from the Brazilian National Council for the Development of Science and Technology, CNPq, for additional funding and scholarships, as well as the Coordenac¸ão de Aperfeic¸oamento de Pessoal de Nível Superior-Brasil (CAPES) – Finance Code 001. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.chroma.2019. 460416.

4. Conclusion References The effect of large volume drops and extraction mode (dynamic and static) were comprehensively evaluated on SDME performance. For this purpose, we developed a robot-assisted automated SDME process associated with a dedicated microextraction device able to stabilize large solvent drops. Increased drop volume and total solvent volume (into the syringe) acted synergistically with the dynamic mode of extraction enhancing the extraction efficiency of all tested analytes. The dynamic mode interacted positively with the extraction time and also led to higher extraction efficiencies. On the other hand, static mode did not show this interaction, especially for compounds with a lower partition coefficient. Regarding extraction efficiency, we have demonstrated the advantages of using a total volume larger than the drop volume itself, once the dynamic procedure resulted in the gradual renovation of the solvent-exposed to the sample. The robot assistance allowed to precisely modulate (i) the number of cycles and (ii) the duration of drop exposition to the sample, allowing to overcome the drawbacks of the drop shrinkage effect. Finally, a surface response revealed the interaction of the latter two parameters (number and duration of cycles) allowing to select an optimal analysis condition under a feasible time frame. The use of robotic open source tools and the design

[1] H. Liu, P.K. Dasgupta, Analytical chemistry in a drop, TrAC Trends Anal. Chem. 15 (1996) 468–475, http://dx.doi.org/10.1016/S0165-9936(96)00065-9. [2] M.A. Jeannot, F.F. Cantwell, Solvent microextraction into a single drop, Anal. Chem. 68 (1996) 2236–2240, http://dx.doi.org/10.1021/ac960042z. [3] M.A. Jeannot, F.F. Cantwell, Mass transfer characteristics of solvent extraction into a single drop at the tip of a syringe needle, Anal. Chem. 69 (1997) 235–239, http://dx.doi.org/10.1021/ac960814r. [4] S. Tang, T. Qi, P.D. Ansah, J.C. Nalouzebi Fouemina, W. Shen, C. Basheer, H.K. Lee, Single-drop microextraction, TrAC Trends Anal. Chem. 108 (2018) 306–313, http://dx.doi.org/10.1016/j.trac.2018.09.016. [5] Y. He, H.K. Lee, Liquid-phase microextraction in a single drop of organic solvent by using a conventional microsyringe, Anal. Chem. 69 (1997) 4634–4640, http://dx.doi.org/10.1021/ac970242q. [6] J.M. Kokosa, Recent trends in using single-drop microextraction and related techniques in green analytical methods, TrAC Trends Anal. Chem. 71 (2015) 194–204. ˇ [7] R. Cabala, M. Bursová, Bell-shaped extraction device assisted liquid-liquid microextraction technique and its optimization using response-surface methodology, J. Chromatogr. A 1230 (2012) 24–29, http://dx.doi.org/10.1016/ j.chroma.2012.01.069. [8] C. Ye, Q. Zhou, X. Wang, Improved single-drop microextraction for high sensitive analysis, J. Chromatogr. A 1139 (2007) 7–13, http://dx.doi.org/10. 1016/j.chroma.2006.10.089. [9] D.A.V. Medina, A. Martins, L.F. Rodriguez-Cabal, Pattent application BR 10 2018 009330 4. - Dispositivo acessório para técnicas de microextrac¸ão e uso do dispositivo, 2018.

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416

G Model CHROMA-460416; No. of Pages 8 8

ARTICLE IN PRESS L.F. Rodríguez Cabal et al. / J. Chromatogr. A xxx (xxxx) xxx

[10] G. Ouyang, W. Zhao, J. Pawliszyn, Automation and optimization of liquid-phase microextraction by gas chromatography, J. Chromatogr. A 1138 (2007) 47–54, http://dx.doi.org/10.1016/j.chroma.2006.10.093. ˇ B. Horstkotte, P. Solich, H. Sklenáˇrová, Automated in-syringe [11] I. Srámková, single-drop head-space micro-extraction applied to the determination of ethanol in wine samples, Anal. Chim. Acta 828 (2014) 53–60, http://dx.doi. org/10.1016/j.aca.2014.04.031. ´ S. Babic, ´ A.J.M. Horvat, M. Kaˇstelan-Macan, Sample preparation [12] D.M. Pavlovic, in analysis of pharmaceuticals, TrAC Trends Anal. Chem. 26 (2007) 1062–1075, http://dx.doi.org/10.1016/j.trac.2007.09.010. [13] M. Alexoviˇc, Y. Dotsikas, P. Bober, J. Sabo, Achievements in robotic automation of solvent extraction and related approaches for bioanalysis of pharmaceuticals, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 1092 (2018) 402–421, http://dx.doi.org/10.1016/j.jchromb.2018.06.037. [14] G. Mafra, A.A. Vieira, J. Merib, J.L. Anderson, E. Carasek, Single drop microextraction in a 96-well plate format: A step toward automated and high-throughput analysis, Anal. Chim. Acta 1063 (2019) 159–166, http://dx. doi.org/10.1016/j.aca.2019.02.013. [15] D. Lopes, A.N. Dias, J. Merib, E. Carasek, Hollow-fiber renewal liquid membrane extraction coupled with 96-well plate system as innovative high-throughput configuration for the determination of endocrine disrupting compounds by high-performance liquid chromatography-fluorescence and diode array de, Anal. Chim. Acta 1040 (2018) 33–40, http://dx.doi.org/10. 1016/j.aca.2018.07.032. [16] M.C. Carvalho, R.H. Murray, Osmar, the open-source microsyringe autosampler, HardwareX 3 (2018) 10–38, http://dx.doi.org/10.1016/j.ohx. 2018.01.001. [17] I. Ali, V.K. Gupta, H.Y. Aboul-Enein, A. Hussain, Hyphenation in sample preparation: advancement from the micro to the nano world, J. Sep. Sci. 31 (2008) 2040–2053, http://dx.doi.org/10.1002/jssc.200800123. ˇ P. Solich, J. Sabo, Automation of [18] M. Alexoviˇc, B. Horstkotte, I. Srámková, dispersive liquid–liquid microextraction and related techniques. Approaches based on flow, batch, flow-batch and in-syringe modes, TrAC Trends Anal. Chem. (2017), http://dx.doi.org/10.1016/j.trac.2016.10.003. [19] V.M.U.A. Priye, S. Wong, Y. Bi, M. Carpio, J. Chang, M. Coen, D. Cope, J. Harris, J. Johnson, A. Keller, R. Lim, S. Lu, A. Millard, A. Pangelinan, N. Patel, L. Smith, K. Chan, Lab-on-a-Drone: toward pinpoint deployment of smartphone-enabled nucleic acid-based diagnostics for mobile health care, Anal. Chem. 88 (2016) 4651–4660, http://dx.doi.org/10.1021/acs.analchem.5b04153. [20] 2016 Review Current approaches in sample preparation for trace analysis of, (n.d.). [21] 2014 Microextraction in urine samples for gas, (n.d.). [22] D.A.V. Medina, Á.J. Santos-Neto, V. Cerdà, F. Maya, Automated dispersive liquid-liquid microextraction based on the solidification of the organic phase, Talanta 189 (2018), http://dx.doi.org/10.1016/j.talanta.2018.06.081. [23] D.A.V. Medina, L.F. Rodriguez Cabal, F.M. Lanc¸as, Á.J. Santos-Neto, Sample treatment platform for automated integration of microextraction techniques and liquid chromatography analysis, HardwareX 5 (2019), e00056, http://dx. doi.org/10.1016/j.ohx.2019.e00056. [24] D.A.V. Medina, L.F.R. Cabal, G.M. Titato, F.M. Lanc¸as, Á.J. Santos-Neto, Automated online coupling of robot-assisted single drop microextraction and liquid chromatography, J. Chromatogr. A (2019), http://dx.doi.org/10.1016/j. chroma.2019.02.036. [25] X. Xu, L. Feng, J. Li, P. Yuan, J. Feng, L. Wei, X. Cheng, Rapid screening detection of fluoroquinolone residues in milk based on turn-on fluorescence of terbium coordination polymer nanosheets, Chin. Chem. Lett. 30 (2019) 549–552, http://dx.doi.org/10.1016/j.cclet.2018.11.026. [26] L. Song, H. Zhang, T. Cai, J. Chen, Z. Li, M. Guan, Porous graphene decorated silica as a new stationary phase for separation of sulfanilamide compounds in hydrophilic interaction chromatography, Chin. Chem. Lett. (2018), http://dx. doi.org/10.1016/j.cclet.2018.10.040. [27] S. Yuan, Z. Liu, H. Yin, Z. Dang, P. Wu, N. Zhu, Z. Lin, Trace determination of sulfonamide antibiotics and their acetylated metabolites via SPE-LC-MS/MS in wastewater and insights from their occurrence in a municipal wastewater

[28]

[29]

[30] [31] [32]

[33]

[34]

[35]

[36]

[37]

[38] [39]

[40]

[41]

[42]

[43]

[44] [45]

treatment plant, Sci. Total Environ. 653 (2019) 815–821, http://dx.doi.org/10. 1016/j.scitotenv.2018.10.417. H. Jin, M. Hee, H. Jin, W. Chan, J. Eok, Development of an immunoaffinity chromatography and HPLC-UV method for determination of 16 sulfonamides in feed, Food Chem. 196 (2016) 1144–1149, http://dx.doi.org/10.1016/j. foodchem.2015.10.014. P.C.F. Lima Gomes, I.N. Tomita, Á.J. Santos-Neto, M. Zaiat, Rapid determination of 12 antibiotics and caffeine in sewage and bioreactor effluent by online column-switching liquid chromatography/tandem mass spectrometry, Anal. Bioanal. Chem. 407 (2015) 8787–8801, http://dx.doi.org/10.1007/s00216015-9038-y. https://www.arduino.cc/, Arduino home. Accessed 15 July 2019. R.E. Majors, Practical aspects of solvent extraction, LCGC 22 (2009) 143–147. P.S. Peixoto, I.V. Tóth, M.A. Segundo, J.L.F.C. Lima, Fluoroquinolones and sulfonamides: features of their determination in water. A review, Int. J. Environ. Anal. Chem. 96 (2016) 185–202, http://dx.doi.org/10.1080/ 03067319.2015.1128539. K. Kipper, M. Lillenberg, K. Herodes, L. Nei, E. Haiba, Simultaneous determination of fluoroquinolones and sulfonamides originating from sewage sludge compost, Transfus. Apher. Sci. 2017 (2017) 20–26, http://dx.doi.org/10. 1155/2017/9254072. X. Guo, D. Yin, J. Peng, X. Hu, Ionic liquid-based single-drop liquid-phase microextraction combined with high-performance liquid chromatography for the determination of sulfonamides in environmental water, J. Sep. Sci. 35 (2012) 452–458, http://dx.doi.org/10.1002/jssc.201100777. W. Gao, G. Chen, Y. Chen, X. Zhang, Y. Yin, Z. Hu, Application of single drop liquid-liquid-liquid microextraction for the determination of fluoroquinolones in human urine by capillary electrophoresis, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 879 (2011) 291–295, http://dx.doi.org/10. 1016/j.jchromb.2010.11.040. V.H. Springer, A.G. Lista, In-line coupled single drop liquid–liquid–liquid microextraction with capillary electrophoresis for determining fluoroquinolones in water samples, Electrophoresis 36 (2015) 1572–1579, http://dx.doi.org/10.1002/elps.201400602. Ł. Marcinkowski, F. Pena-Pereira, A. Kloskowski, J. Namie´snik, Opportunities and shortcomings of ionic liquids in single-drop microextraction, TrAC Trends Anal. Chem. 72 (2018) 153–168, http://dx.doi.org/10.1016/j.trac.2015.03.024. R.E. Majors, Salting-out liquid-liquid extraction (SALLE), LCGC 27 (2009) 526–533. H. Prosen, Applications of liquid-phase microextraction in the sample preparation of environmental solid samples, Molecules 19 (2014) 6776–6808, http://dx.doi.org/10.3390/molecules19056776. S.C. Cobzac, S. Gocan, Sample preparation for high performance liquid chromatography: recent progress, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 1157–1267, http://dx.doi.org/10.1080/10826076.2011.588064. A. Song, J. Yang, Efficient determination of amphetamine and methylamphetamine in human urine using electro-enhanced single-drop microextraction with in-drop derivatization and gas chromatography, Anal. Chim. Acta (2018), http://dx.doi.org/10.1016/j.aca.2018.09.024. D. Carvajal, E.J. Laprade, K.J. Henderson, K.R. Shull, Mechanics of pendant drops and axisymmetric membranes, Soft Matter 7 (2011) 10508–10519, http://dx.doi.org/10.1039/c1sm05703k. ˇ I.H. Srámková, B. Horstkotte, K. Fikarová, H. Sklenáˇrová, P. Solich, Direct-immersion single-drop microextraction and in-drop stirring microextraction for the determination of nanomolar concentrations of lead using automated Lab-In-Syringe technique, Talanta 184 (2018) 162–172, http://dx.doi.org/10.1016/j.talanta.2018.02.101. M.A. Jeannot, F.F. Cantwell, Solvent microextraction into a single drop, Anal. Chem. 68 (1996) 2236–2240, http://dx.doi.org/10.1021/ac960042z. S. Pedersen-Bjergaard, K.E. Rasmussen, Liquid-phase microextraction with porous hollow fibers, a miniaturized and highly flexible format for liquid-liquid extraction, J. Chromatogr. A 1184 (2008) 132–142, http://dx.doi. org/10.1016/j.chroma.2007.08.088.

Please cite this article in press as: L.F. Rodríguez Cabal, et al., Robotic-assisted dynamic large drop microextraction, J. Chromatogr. A (xxxx), https://doi.org/10.1016/j.chroma.2019.460416