In-tip solid-phase microextraction: a method for determination of sulphonamide residues in environmental water samples

ABSTRACT This study involved the development of an in-pipette tip solid-phase microextraction (SPME) method using activated charcoal as an adsorbent. Simultaneous determination of three sulphonamide antibiotic residues in environmental water was performed using this method, coupled with high-performance liquid chromatography-photodiode array (HPLC-PDA) detection. Seven extraction parameters were optimised by one-variable-at-a-time (OVAT) and response surface (RSM) methods. Optimum extraction efficiency was achieved when 10 mL of sample solution at a natural pH (pH 5–6) was loaded through a 1 mL pipette tip packed with 10 mg of adsorbent, washed with 1 mL of hexane and eluted with 500 µL of 1% ammonium in methanol solution. Under the optimised experimental condition, this method manifested good linearity (5–500 µg L−1) at the coefficient of determination, R2 of 0.9992 to 0.9993, high sensitivity (limit of detection, LOD: 0.38–1.14 µg L−1; limit of quantification, LOQ: 1.14–3.35 µg L−1), and satisfactory recoveries (82.8–108.7%) with acceptable reproducibility (RSD < 4.6%). Real sample analysis of five environmental water samples sourced from Universiti Malaya, Kuala Lumpur and Malacca showed that the samples contained either none or traces of the target sulphonamides (SAs) with their concentration lower than the LOQs. In contrast to other methods, the comparison outcomes depicted that this in-tip SPME method was beneficial in terms of its simple, reusable, cost-saving setup together with a significant reduction in chemical consumption.


Introduction
Sulphonamides (SAs) serve as a bacteriostatic to suppress the growth and multiplication of bacteria.Due to its relatively low cost as compared to other antibiotics [1], SAs are applied extensively to control and inhibit bacterial infection on animals in poultry industries.On account of their broad spectrum, SAs are also used in human medications to treat cancer [2], conjunctivitis, and burn-related infections [3].However, SAs are poorly absorbed by organisms [4].Excessive use of SAs in human and veterinary antimicrobial therapy leads to their leeching into environmental water, with the majority excreted as urine and faeces [4,5].As SAs cannot be entirely removed by sewage treatment [1], their pollution in surface water occurs due to the drainage of SA residues from hospitals and poultry industries' sewage.SA contamination in environmental water also arises from the direct administration of antibiotics into fish farms [6].Both direct and indirect drainage of a considerable level of SAs into natural water resources possess potential hazards to human health.
Improper application of SA drugs, such as overdose and fast withdrawal, may result in SA residues in eggs, milk, and meat.Thus, the legislative basis for the maximum residue limits (MRL) of SAs and their analogues in food products have been set by local (FoSIM) and international food standards bodies to ensure and secure public health [7].The average maximum SAs' residue level established by these local and international food standard bodies falls around 100 µg kg −1 for animal food products [8,9].Similarly, waterborne SAs can enter the food chain and accumulate in all kinds of organisms.This may give rise to severe allergic reactions [4] and induce resistance to SAs in certain human body pathogens [1].Patients infected with SA drug-resistance bacteria that cause infections such as bronchitis, pneumonia, and urinary tract infections have a higher risk to experience treatment failure.In this respect, there is a need to develop an efficient and reliable system to analyse residual SAs in environmental water.
Over the years, there is a vast improvement of techniques for SA residue determination in water samples.As the polar SA drugs are non-volatile, the most widely adopted approach for SA analysis is HPLC with detectors such as mass spectrometry (MS) or photodiode array (PDA) spectrophotometry [4].However, in the trace-level analysis of residual SAs, the sensitivity and accuracy of the chromatographic technique are highly limited by the significant matrix interferences that existed in the analysis of real environmental water samples [10].Hence, more efforts are required on sample preparation to improve the sensitivity of the SA detection method in water samples.
On account of the world's demand for green chemistry, sample preparation methods such as liquid-liquid extraction (LLE), solid-liquid extraction (SLE), and solid-phase extraction (SPE) should be designed in microscale to minimise the usage and disposal of organic solvents and sorbents.Extraction using traditional SPE cartridges often encounters some practical downsides, such as high solvent and sorbent consumption.The modification of classic SPE to in-tip solid-phase microextraction (SPME) provides the capability to get rid of these problems while keeping the advantages of SPE [11][12][13][14].Its relatively more straightforward extraction setup and low sample intake ease the sampling process as compared to the somewhat bulky conventional SPE cartridges [11].As a result, the overall sample pre-treatment time can significantly decrease with the microscale application using the pipette tip.
This study seeks to develop a relatively simple and cost-effective SA extraction system using the in-tip SPME technique using commercially available activated charcoal as sorbent.The first part of the project is to assess the validity of the proposed extraction method in the analysis of sulfachlorpyridazine (SCP), sulfamethoxazole (SMX), and sulfadimethoxine (SDM).Then, the optimisation of the proposed method coupled with HPLC-PDA detection would be carried out using one-variable-at-a-time (OVAT) and response surface (RSM) methods.The former is a classical experimental design method involving only one-factor testing at a time, and the latter is a statistical approach in which it utilises statistical analysis software to carry out the design of experiment (DoE) and data modelling to optimise the desired outcome for several experiment parameters and responses in one goal.In contrast to OVAT, RSM can simultaneously explore the significance of multiple variables and study the interaction among those variables and one or more response variables.In the second part of this project, the competency of the proposed method will be evaluated for real sample analysis on different types of environmental water samples.

Chemicals, reagents, and materials
Standard solutions of sulfachlorpyridazine (SCP), sulfamethoxazole (SMX), and sulfadimethoxine (SDM) were purchased from Sigma Aldrich (St. Louis, Missouri, USA).The chemical properties of all three SAs are exhibited in Table 1.Meanwhile, HPLC grade acetonitrile (ACN), methanol (MeOH) and glacial acetic acid were supplied by Merck (Darmstadt, Germany).The ultrapure water (18.2MΩ/cm) used throughout this research project was supplied by the Merck Milli-Q system (Lane End, UK).Commercially available activated charcoal and dried n-hexane were purchased from Merck (Darmstadt, Germany). 1 mL, 3 mL syringes, and nylon syringe filters (0.45 µm) were obtained from HmbG Chemicals (Hamburg, Germany).

Standard and working solutions
The standard stock solutions of SCP, SMX, and SDM (1000 μg L −1 ) were prepared in HPLC grade MeOH solution.The stock solutions were stored inside amber glass vials at 4°C in the refrigerator.Standard working solutions of SAs were prepared fresh by diluting the prior-mentioned stock solution with ultrapure water whenever required.

Instrumentation and chromatographic conditions
The pH for stock standard solutions, samples, and mobile phases was adjusted using a Mettler-Toledo FiveEasy pH metre (Columbus, Ohio, USA).The IR spectrum for activated charcoal was recorded by a Perkin Elmer Spectrum 400 Fourier Transform Infrared (FTIR) device (Perkin Elmer, USA) using attenuated total reflectance (ATR) mode.The SA separation was carried out on a Shimadzu LC-20AT High-Performance Liquid Chromatography (HPLC) system (Japan) equipped with a dual-plunger tandem-flow solvent delivery module, solvent degasser unit, and column oven.A Synchronic C18 column (150 mm x 4.6 mm, particle size: 5.0 µm, USA) was used to separate the SA analytes under reversed-phase isocratic condition.The sample (10 µL) was auto-injected into the column with the oven temperature kept constant at 25°C throughout the 15 min run time.The mobile phase was the 1% acetic acid mixture in ultrapure water and acetonitrile (70:30, v/v) with a flow rate of 1.0 mL min −1 .The SA detection was performed on an SPD-M20A PDA detector (Shimadzu, Japan) at 269 nm.

In-tip SPME procedure
Activated charcoal (10 mg) was filled into a dried 1 mL pipette tip using cotton wool as a filter at both ends, as illustrated in Figure 1.The tip packed with activated charcoal was used as the SPME cartridge to carry out microextraction.A 3 mL syringe was connected to the end of the tip to facilitate sample drawing.
For microextraction, the tip cartridge was first pre-conditioned with MeOH (1.0 mL) and ultrapure H 2 O (1.0 mL) consecutively to ensure consistent interactions between the analytes and sorbent.Subsequently, the pre-treated sample solution (10.0 mL) was loaded into the tip cartridge.The extracted sample was discharged, and the sorbent was washed with dried hexane (1.0 mL) to remove the co-adsorbed interfering species.Compared with other polar solvents, hexane was adopted as the washing solvent as it had poor eluting capability to remove the polar SAs adsorbed on activated charcoal.Then, the analyte was eluted with 1% NH 4 + in MeOH solution (0.5 mL).After filtration, the eluate was injected into the HPLC-PDA system for SA detection.

Sample preparation
Five different environmental water samples were sourced from Universiti Malaya (UM) and Malacca state.The tap water was collected from the tap in a UM laboratory, whereas the lake water was collected from Varsity Lake, University of Malaya.The river water samples were collected from three rivers located in Malacca, namely Jebat River, Rambai River, and a river linked to Indah Water Consortium's wastewater cannel.All water samples were filtered through 0.45 µm nylon syringe filters to remove any solid suspension.Then, the sample pH was measured, and the samples were stored in glass containers wrapped with aluminium foil under refrigeration at 4.0°C prior to analysis.

Characterisation of activated charcoal
The functional groups present in activated charcoal were analysed using ATR-FTIR spectroscopy.The infrared (IR) spectrum of dried activated charcoal was compared before and after the extraction procedure (Figure 2) to observe any changes of the functional groups due to SA adsorption.The peak and broad bands in the IR spectrum were assigned, as shown in Table S1, referring to the IR absorption table available at [17].Both FTIR spectra showed no significant difference in the functional groups of activated charcoal before and after SA extractions.

Optimisation of extraction parameters
Based on the in-tip SPME procedure, seven extraction parameters associated with the sample loading and elution stages were chosen.These parameters included the sample pH, type of desorption solvent, effect of modifier (NH 4 + ) in desorption solvent, volume of desorption solvent, adsorbent mass, sample volume, and the sample loading-elution strokes per extraction.The first five parameters were optimised using the OVAT approach, and the last three parameters mentioned above were optimised by the RSM method.In the optimisation stage, ultrapure water spiked with 1000 µg L −1 of each SCP, SMX, and SDM stock standard solution was used as the sample.The target experiment variables and the corresponding levels for OVAT and RSM experiment designs are listed in Table S2.

OVAT optimisation 3.2.1.1. Sample pH.
Generally, the stability of the neutral SA species is easily altered by pH because of the two ionisable functional groups in SAs [18].SAs comprise a basic aromatic amine and an acidic nitrogen atom, resulting in two different pK a ranges (strongest basic pK a : 1.39-2.44;strongest acidic pK a : 5.9-8.81).Consequently, any working pH lower than the strongest basic pK a or higher than the strongest acidic pK a will reduce the SA extraction efficiency [18] as more charged molecules are formed.Hence, it is crucial to ensure SAs remain dominant in neutral forms to maximise the hydrophobic interactions between SAs and sorbents [18].In this study, the sample pH was optimised over the range of pH 2-10.Following the result shown in Figure 3(a), extraction of sample solution at pH 6 achieved the highest peak areas for all three SAs.The experiment outcome inferred that the slightly acidic pH condition at the range of pH 4 to 6 was more favourable for SA extraction using activated charcoal sorbent.This finding was homologous to the previously reported papers conducted by Ning's and Xu's research groups [19,20].Therefore, no pH adjustment was needed prior to microextraction as the natural pH of the water sample spiked with SAs was found in the range of pH 5-6.

Type of desorption solvent.
In this case, a good desorption solvent must have strong interactions with the SA polar functional groups to displace all three types of SA drugs adsorbed onto the sorbent in minimal volume.On that account, polar solvents such as methanol (MeOH), acetonitrile (ACN), 1% acetic acid-methanol (v/v), 1% acetic acidacetonitrile (v/v), 1% NH 4 + in methanol (v/v), and 1% NH 4 + in acetonitrile (v/v) were used as the desorption solvents, respectively.Desorption efficiencies of these solvents were compared based on the resulting peak area for all three target SAs as shown in Figure 3(b).Besides, the effect of increasing acidity and alkalinity of the desorption solvent was also investigated with 1% acetic acid or ammonium (NH 4 + ) into MeOH and ACN, respectively.1% NH 4 + in methanol (v/v) had the highest desorption capability as it yielded the highest peak areas for all SA drugs.This phenomenon suggested that alkaline condition diminished the acidic SA compounds' affinity towards activated charcoal, thus facilitating the elution mechanism.

Effect of modifier in desorption solvent.
The previous study in Section 3.2.1.2had shown that the SA desorption efficiency can be improved by adding modifiers such as acetic acid and ammonium into the desorption solvent.Herein, another investigation was carried out to study the effect of modifier concentration in desorption solvent on the SA desorption efficiency.The NH 4 + concentration (modifier effect) was optimised from 1% to 5% v/v in methanol.in methanol in the elution step.Results showed that the resolution of the peaks decreased, whereby splitting in SA peaks started to occur as the ammonium concentration increased to 3%.At higher NH 4 + concentrations, the peaks for all studied SAs were found to be completely split into two different peaks with a close retention time.Thus, this may elucidate that the ionisation of SAs occurred in a very basic environment, leading to the separation of ionised SA species from their neutral form.Both 1% and 2% NH 4 + in methanol (v/v) gave relatively high desorption efficiency.To minimise the ionisation effect, 1% NH 4 + was adopted as the optimum concentration of modifier added to the desorption solvent.

Volume of desorption solvent. The volume of solvent was another critical
parameter that would impact the sample elution steps desorption efficiency.The eluent volume must be sufficient to displace all the analytes from the sorbent simultaneously without causing the reduction in extraction efficacy.Five eluent volumes were studied at 0.5, 1.0, 1.5, 2.0, and 2.5 mL.Note that the effective elution could only be achieved at a minimum eluent volume of 0.5 mL.Any volume lower than 0.5 mL would cause incomplete analyte desorption as some of desorption solvents would remain soaked in cotton wool.Figure 3(d) shows the effect of the five eluent volumes.
According to the results, the peak areas for all studied SAs were significantly higher when 0.5 mL of eluent solvent was used for elution.Consequently, 0.5 mL was opted as the optimum desorption volume for the following experiments as this was the volume adequate for all the SAs to be completely displaced from the sorbent.

RSM optimisation 3.2.2.1. Mass of sorbent, sample volume, and sample loading-elution strokes per extraction.
The mass of sorbent was one of the critical factors that would influence the analyte extraction efficiency.For the extraction of SAs, commercial activated charcoal was used as the adsorbent.The mass of activated charcoal packed into the SPME pipette tip cartridge was optimised from 1 to 10 mg.Concurrently, the volume of sample solution loaded through the cartridge was also an important parameter to obtain good sensitivity for SA detection at trace level.Thus, the effect of sample volume was evaluated from 1 to 10 mL.Nevertheless, the effect of the sample loading-elution strokes per extraction on the extraction efficiency was evaluated from 1 to 5 cycles simultaneously with the two parameters mentioned above using the central composite design (CCD) model.

Central composite design.
The low, middle, and high levels of the sorbent mass, sample volume, and extraction cycle are listed in Table S3 together with the corresponding coded units.
To study the effect of the three aforementioned extraction variables, a 3-factor, 3-level central composite design (CCD) model was used in the RSM optimisation method.This RSM optimisation was accomplished in 16 experiment runs (without replication) that had two centre points.Note that the experiment run for each condition was randomised to minimise the probability of getting biased data.The experiment design (DoE) displayed in Table S4 was conducted using the JMP Pro 15 software.
The variance analysis (ANOVA) was used to examine the significance level of the studied parameters on the extraction efficiency.The prob > F value, which was smaller than α = 0.05, suggesting at least one factor was significant in this model (Table S5).The actual value predicted shown in Figure S1 was the plot of the actual response (peak area) against the predicted response by the quadratic model built from the obtained data.The corresponding R 2 value (goodness-of-fit measure) of this plot close to 1 depicted that the data was well fitted to the quadratic model.
Moreover, the quality of the model could be further assessed by the coefficient of determination (R Squared) and adjusted coefficient of determination (R Squared Adj.) in the summary of fit table (Table S6).For all three SAs, these two values were close to 1, thus revealing a good relationship between the experiment data and the fitted model.Furthermore, the satisfied R Squared and adjusted R Squared values gave confidence that the obtained predicted equations of the regression models (Equations S1, S2, and S3) were valid to correlate the experimental data.
The effect summary (Table S7) generated from the obtained data illustrated the overall significance of each parameter effect on the extraction efficiency of all studied SAs in the order of sample volume ≫ sorbent mass > extraction cycle.F ratio is a statistical signal-tonoise (S/N) ratio that can be calculated from the ratio of the mean square model (variance derived from the regressed model) and mean square error (unexplained variance).When there was no relationship between the response (peak area of SAs) and any of the studied parameters, the F ratio would be approximately 1 as both variances were more or less the same.On the contrary, in the case of F ratio greater than 1, at least one of the regressed models' coefficients was significant to create some variation in the response.In this case, the degree of freedom and the number of parameters associated with the effects were only 1 in each case, as no replication was done for each experiment run.Since all F ratios calculated for all the sources of variation in Table S7 were greater than 1, this implied that the model was significant.The obtained responses in the regressed model were not caused by noise.
Using the calculated p-values, sample volume was the only significant extraction variable for the SA extraction based on the proposed in-tip SPME method.This was supported by the prob > F value of sample volume (<0.0001), lower than α = 0.05, indicating that the model was significant due to at least one significant parameter in the experiment.Meanwhile, the effect of sorbent mass, extraction cycle, and interaction between these three factors were less pronounced, with their prob > F values greater than α.

Interaction effect.
Nevertheless, the interaction effects among the sample volume, sorbent mass and extraction cycles were evaluated using the regressed model.As the interaction plots for each SA gave similar trends, only the interaction plot for SCP is displayed in Figure 4 for illustration purpose.The interaction plot revealed that as the sorbent mass increased from 1 to 10 mg, a desirable high level of responses would only be achieved at a high sample volume (10 mL).Along with the increment of sorbent mass to a high level (10 mg), the effective number for sample loading-elution strokes per extraction was reduced from 5 to 1 time.This proposed that less time (less extraction cycles) is required for an extraction using high sorbent mass and large sample volume.
Meanwhile, both sorbent mass and extraction cycles showed similar outcomes when interacted with the sample volume.The linear graphs obtained from these two interactions predicted that the response signal would increase along with the increment in the sample volume.Since the linear lines did not reach the plateau, this suggested that the critical value (maximum SCP peak area) was out of the data range.Hence, the range for the sample volume had to be expanded to a higher value to achieve a higher peak area.From a practical point of view, the optimum sample volume was selected to consider the required sensitivity and sufficient extraction time.Therefore, the experiment proceeded using 10 mL as the chosen sample volume to ensure adequate sensitivity and shorten the experiment's duration.
Acceptable responses would only be possible for extraction cycles at a high sample volume (10 mL).There was no great difference between the response signal obtained at such high sample volume at the low or high number of sample loading-elution strokes per extraction.Hence, one extraction cycle would be favoured to extract a 10 mL sample solution to establish a satisfactory extraction efficiency within a shorter period.S2 illustrate the SCP peak area changes against individual parameters.Similar to the interaction plot, the response surface for all three SAs looked alike.Thus, only the SCP response surface plots were discussed as the representative in this RSM study.Figure S2 (a) shows a twisted ridge surface created from the interaction between the extraction cycle's linear effect and the sorbent mass' quadratic effect on the SA peak areas.This finding was parallel to the fact that the SCP peak area would decrease as the extraction cycles increased at a high level of sorbent mass, as discussed before, based upon their interaction plots.

Response surface plot. The response surface plots in Figure
The formation of a ridge response surface in Figure S2 (b) could be rationalised by combining the linear effect of sample volume and the quadratic effect of the sorbent mass.According to Figure S2 (b), the SCP peak area plot against the sample volume resulted in a positive straight line, indicating the linear relationship between sample volume and SCP peak area.On the contrary, a positive quadratic graph was shown on the two-dimensional surface of SCP peak area versus sorbent mass.This depicted that after a certain saturation point, using the same sample volume, the peak area of SCP would not significantly differ even if the sorbent mass increased.
Next, the flat response surface presented in Figure S2 (c) was generated solely due to the main effects of both sample volume and extraction cycle, hinting that the interaction effect of these two parameters was negligible.This could be confirmed by the effect test's prob > F value (0.2283) for the interaction between sample volume and extraction cycles, which was larger than α = 0.05 (Table S7).The SCP peak area plot against the sample volume yielded a much steeper gradient as compared to the graph of SCP peak area against extraction cycles.This was agreed with the prob > F values for both the parameters in Table S7, revealing that sample volume was the more critical parameter for SCP extraction, in contrast to the number of sample loading-elution strokes per extraction cycle.

Deduction of the optimised condition for SA extraction. A prediction profiler
was utilised to deduce the optimum experiment condition to predict the changes in the peak area that arise from the variation in the parameters.The sample loading-elution strokes per extraction were reduced from 5 to 1 in the prediction profiler (Figure 5).As a result, the experiment parameters including 10 mg sorbent mass, 10 mL sample volume, and 1 extraction cycle were selected as the optimum condition for SA extraction in order to establish reliable analytical results with good method sensitivity.

Reusability of the SPME tip cartridge
The reusability of activated charcoal was evaluated by examining the loss in its extraction efficiency towards SCP, SMX, and SDM in five successive extraction cycles (Figure 6).After the third extraction, there was a drastic reduction (>25%) in the peak area for all three SAs and the decreasing trends continued for the subsequent extractions.This was presumably due to the degradation of activated charcoal's adsorption surface after repeated washing with diluted 1% acetic acid solution.Therefore, this suggested that the SPME tip cartridge could only be used a maximum of three times without any substantial loss in the SAs' extraction efficiency.Despite the limited usage lifetime, the cheap raw materials required to build the SPME tip cartridge made it economical to use multiple SPME tips at once.

Analytical performance of the developed method
In essence, the extraction parameters optimised in the earlier stage listed in Table S8 were applied in method validation and real sample analysis.For method validation, a series of experiments were carried out on water samples spiked with SA standard solutions under optimised conditions to obtain linear ranges, LOD, LOQ, intra/inter-day precision, and preconcentration factors.
All three matrix-matched calibration curves (n = 3) for SCP, SMX, and SDM yielded good linearity by achieving an excellent coefficient of determination (R 2 ) ranging from 0.9992 to 0.9993.The calibration curves of SCP and SDM gave linear ranges from 30 to 500 µg L −1 , whereas the calibration curve of SMX covered a more comprehensive range of concentrations from 5 to 500 µg L −1 .
Furthermore, LOD and LOQ were determined using the blank residual method developed based on the S/N ratio of 3.3 and 10, respectively.Method detection limits of SAs in water fell within the range of 0.38-1.14µg L −1 for LODs, and 1.14-3.35µg L −1 for LOQs.The method detection limits were found to be lower than the instrument detection limits (SCP ≥ 70 µg L −1 , SMX ≥ 50 µg L −1 , and SDM ≥ 70 µg L −1 ) established from the direct injection of SAs' standards into the HPLC coupled to PDA detection.This showed that the developed in-tip SPME method possessed a higher sensitivity to detect SAs at much lower concentrations, which cannot be achieved by direct sample analysis.
Simultaneously, the repeatability of the optimised method was investigated from the intra-and inter-day precisions at 300 µg L −1 concentration of each SA in water.The intraand inter-day relative standard deviations (RSD) were obtained by performing five replicates of extractions on a water sample fortified with 300 µg L −1 of SAs on the same day and after three consecutive days.The intra-day RSD was ranged between 3.47% and 4.32%, and the inter-day RSD was 7.31-7.52%.As summarised in Table 2, the relatively low LODs (0.38-1.14 µg L −1 ) and RSD (≤7.52%) manifested that the developed method was reliable to provide reproducible results at trace-level analysis of SAs.
Nevertheless, the pre-concentration factor (PF) was calculated using a water sample spiked with 300 µg L −1 of SA standard mixture to assess the extraction efficiency of the activated charcoal to the SAs.As exhibited in Table 2, the developed method generated PF values ranging from 41 to 261.The extraction of SMX using the developed method resulted in a relatively higher PF at 261.This could be due to the larger mass transfer of SMX molecules from the sorbent to the eluting solvent (1% NH 4 + in MeOH) by virtue of its smaller molecular weight than the SCP and SDM molecules [21].

Real sample analysis
The developed in-tip SPME method was applied to real environmental water samples to determine the SA content.Five environmental water samples, including tap water (UM laboratory), UM Varsity Lake water, Jebat River water (Malacca), Rambai River water (Malacca), and Indah Water Consortium's river water (Malacca), were collected and analysed under optimised conditions, respectively.Results showed all water samples were safe from SA contamination as the developed method could not detect any SA from any of these samples.Preliminary analysis of these blank samples proved they were analyte-free.A recovery study was conducted by fortifying the blank tap water samples at three different SA concentrations (50, 100 and 300 µg L −1 ) from low to high levels.The recovery study results are tabulated in Table S9.
The recoveries obtained from fortified tap water samples were within the range of 82.8-108.7%,with acceptable RSD at 0.2% to 4.6%.However, the detection of SAs in all environmental water samples yielded a negative result.On that account, these water samples either contained no SAs or had SAs at concentrations below the detection limits.Figure 7 demonstrates the chromatograms of tap water with different fortification levels.

Comparative study of SA detection methods
To evaluate the analytical performance of the developed method, its validation parameters were compared with previously reported works for SA determination in environmental water samples in Table 3.The comparative study revealed that carbon-based adsorbents such as activated charcoal, graphene, and magnetic carbon nanotubes possessed superior adsorption capacity.These carbon-based adsorbents worked better on SAs in contrast to Hydrophilic-lipophilic balance (HLB) [1] in SPE and ionic liquid in twophase aqueous extraction (IL-ATPE) [22].This could be explained by the large surface area exposed to adsorption exhibited by the carbon-based adsorbents.The comparative results indicated that the proposed method was beneficial as compared to other methods as its simplicity made the overall extraction process lower in cost, more environmentally friendly, and faster with good reproducibility.

Conclusion
In summary, a simple miniaturised in-tip SPME coupled with HPLC-PDA was successfully developed and applied for simultaneous detection of three SAs (SCP, SMX, and SDM) in environmental water samples.This method manifested good linearity (5-500 µg L −1 ) at the coefficient of determination, R 2 of 0.9992 to 0.9993, high sensitivity (limit of detection, LOD: 0.38-1.14µg L −1 ; limit of quantification, LOQ: 1.14-3.35µg L −1 ), and satisfactory recoveries (82.8-108.7%)with acceptable reproducibility (RSD < 4.6%).Thus, the method validation had manifested that the constructed method could produce a reliable result with acceptable sensitivity and good reproducibility under optimised conditions.Real sample analysis of five environmental water samples showed that all water samples were freed from SA pollution.
Compared with other previously published works, commercially activated charcoal was an effective adsorbent for solid-phase microextraction due to the large surface area of activated charcoal available for SA adsorption, resulting in high extraction efficiency.This in-tip SPME method can shorten the whole analysis duration by eliminating the sorbent synthesis and sample centrifugation step.Its simplicity and low chemical consumption also made this method more cost-effective and environmentally friendly.

Figure 2 .
Figure 2. FTIR spectra of activated charcoal before (a) and after extraction (b).

Figure 3 (
c) conveys the extraction results using different concentrations of NH 4 +

Figure 3 .
Figure 3. Optimisation parameters of the effect of sample pH (a), the effect of different types of desorption solvents (b), the effect of modifier (c), the effect of desorption volume (d).

Figure 4 .
Figure 4. Interaction plot and the effects of these interactions on the peak area of SCP.

Figure 5 .
Figure 5. Prediction profile after modification with optimised conditions.

Figure 6 .
Figure 6.Effect of the number of extraction cycles on the extraction efficiency.

Table 1 .
Characteristic of target SAs.
a Abbrev.: abbreviation.b pK a (Strongest acidic): acidic dissociation constant of the sulphanilamide group in SAs.c pK a (Strongest basic): basic dissociation constant of the amino group in SAs.d Ref.: references.

Table 2 .
Analytical performance parameter of the represented in-tip SPME method.