*3.1.2. Optimization of the SPE operation parameters: Multivariate approach*

In order to achieve quantitative preconcentration of analytes by SPE system, the optimization of the most influential parameters, such as sample pH, sample and eluent flow rates, was carried out using 23 full factorial design and response surface methodology (RSM) based on a central composite design. Two level full factorial design (FFD), involving 11 experiments was used for screening of the significant factors for the extraction and preconcentration of UV filters. **Table 2** presents the factorial design matrix and the analytical responses (expressed as average percentage recovery, %R) obtained in each experiment. Analysis of variance (ANOVA) reproduced in the form of Pareto chart was used to investigate the significance of the effects SPE procedure. The Pareto chart of main effects and their interactions produced are shown in **Figure 3**.

It can be seen from **Figure 3** that for simultaneous preconcentration of UV filters, sample pH and EFR and their interactions were statistically significant. The effect of sample flow rate on the analytical response (%R) was not significant at 95% confidence level. The overall results obtained for the screening step indicated that sample pH and EFR required further optimization. Whereas, sample flow rate was fixed at 3.5 mL min−1.

> quadratic equations and 3D surface response plots, the calculation indicated that pH = 7.5 and EFR = 2 mL min−1 provided maximum retention and recovery of the studied analytes. Therefore, the results obtained from both designs, illustrated that the optimum conditions that led to quantitative extraction and preconcetration of UV filters were 7.5, 3.5, 2 mL min−1

**Sample pH SFR (mL/min) EFR (mL/min) UV-01 %R UV-02 %R** 4 3.5 1 66.1 63.4 10 3.5 1 65.7 44.5 4 3.5 3 49.7 16.6 10 3.5 3 25.9 19.2 7 3.5 2 61.3 51.9 7 3.5 2 54.8 59.2 7 3.5 2 49.7 49.7 2.8 3.5 2 78.8 49.6 11.2 3.5 2 42.0 66.0 7 3.5 0.5 100.1 103.1 7 3.5 3.4 32.2 17.4 7 3.5 2 75.0 81.5 7 3.5 2 74.8 82.0 7 3.5 2 75.1 81.8

**Figure 3.** Pareto charts of standardized effects for variables in the preconcentration of (A) benzophenone and (B)

Multivariate-Assisted Solid Phase Extraction Procedure for Simultaneous Preconcentration…

http://dx.doi.org/10.5772/intechopen.75641

51

for sample pH, sample and eluent flow rates, respectively.

**Table 3.** Central composite design matrix and analytical response.

sulisobenzone.

A central composite design matrix consisting of 14 experiments and analytical response based on each of the experimental runs (**Table 3**) was used for further optimization of the SPE method. The analysis of variance (ANOVA) of the predicted response surface quadratic model for the recoveries of UV filters was obtained. The ANOVA results were analysed using quadratic equations (not included) for the models to illustrate the dependence of the analytical response with respect to the evaluated main effects [31].

The 3D response surface plots (**Figure 4**) were used to access the interactive relationship between individual variables (sample pH and EFR) and analytical response [31]. Based on


**Table 2.** Two level (23 ) full factorial design matrix and analytical response. Multivariate-Assisted Solid Phase Extraction Procedure for Simultaneous Preconcentration… http://dx.doi.org/10.5772/intechopen.75641 51

*3.1.2. Optimization of the SPE operation parameters: Multivariate approach*

50 Emerging Pollutants - Some Strategies for the Quality Preservation of Our Environment

tion. Whereas, sample flow rate was fixed at 3.5 mL min−1.

cal response with respect to the evaluated main effects [31].

carried out using 23

are shown in **Figure 3**.

**Table 2.** Two level (23

In order to achieve quantitative preconcentration of analytes by SPE system, the optimization of the most influential parameters, such as sample pH, sample and eluent flow rates, was

a central composite design. Two level full factorial design (FFD), involving 11 experiments was used for screening of the significant factors for the extraction and preconcentration of UV filters. **Table 2** presents the factorial design matrix and the analytical responses (expressed as average percentage recovery, %R) obtained in each experiment. Analysis of variance (ANOVA) reproduced in the form of Pareto chart was used to investigate the significance of the effects SPE procedure. The Pareto chart of main effects and their interactions produced

It can be seen from **Figure 3** that for simultaneous preconcentration of UV filters, sample pH and EFR and their interactions were statistically significant. The effect of sample flow rate on the analytical response (%R) was not significant at 95% confidence level. The overall results obtained for the screening step indicated that sample pH and EFR required further optimiza-

A central composite design matrix consisting of 14 experiments and analytical response based on each of the experimental runs (**Table 3**) was used for further optimization of the SPE method. The analysis of variance (ANOVA) of the predicted response surface quadratic model for the recoveries of UV filters was obtained. The ANOVA results were analysed using quadratic equations (not included) for the models to illustrate the dependence of the analyti-

The 3D response surface plots (**Figure 4**) were used to access the interactive relationship between individual variables (sample pH and EFR) and analytical response [31]. Based on

**Expt. pH SFR (mL/min) EVR (mL/min) UV-01 (%R) UV-02 (%R)**

 4 2 1 34.39 33.7 10 2 1 39.63 39.1 4 5 1 51.01 53.5 10 5 1 72.71 75.4 4 2 3 12.22 16.0 10 2 3 8.84 9.1 4 5 3 11.58 10.6 10 5 3 10.24 17.6 7 3.5 2 83.55 72.8 7 3.5 2 78.88 77.5 7 3.5 2 81.08 78.6

) full factorial design matrix and analytical response.

full factorial design and response surface methodology (RSM) based on

**Figure 3.** Pareto charts of standardized effects for variables in the preconcentration of (A) benzophenone and (B) sulisobenzone.

quadratic equations and 3D surface response plots, the calculation indicated that pH = 7.5 and EFR = 2 mL min−1 provided maximum retention and recovery of the studied analytes. Therefore, the results obtained from both designs, illustrated that the optimum conditions that led to quantitative extraction and preconcetration of UV filters were 7.5, 3.5, 2 mL min−1 for sample pH, sample and eluent flow rates, respectively.


**Table 3.** Central composite design matrix and analytical response.

respectively. Whereas for sulisobenzone, the DLR, R2

**3.3. Validation and application**

ranged from 99.3 to 100.7%.

**Analyte(s) Matrix Analytical** 

Benzophenone-3 Water SBSE-LC-MS/

Benzophenone-2 Human serum DLLME-

Benzophenone-3 Sea water DLLME-

Benzophenone Tap water DDA-IL-

Benzophenone and sulisobenzone

LOQ-250 μg L−1, 0.9991, 0.15 μg L−1, 0.50 μg L−1 and 55, respectively. Furthermore, the intraday (repeatability; n = 10) and interday (reproducibility; n = 7 working days) precisions of the SPE method, expressed in terms of relative standard deviation (% RSD), ranged 3.1–3.3 and 4.5–5.2%, respectively. The analytical performance of the proposed method was compared to other methods that reported in the literature **Table 4**. It can be seen that the performance of the current method was comparable or better that those reported in the literature. In addition, the LODs obtained using the current method were lower compared to [25, 32].

Multivariate-Assisted Solid Phase Extraction Procedure for Simultaneous Preconcentration…

The accuracy of the SPE/UV procedure was evaluated using a sunscreen lotion with a sulisobenzone content of 1.75% (w/w). The recovered sulisobenzone was 1.69 ± 0.07% (w/w) meaning that the percentage recovery was 96.6%. Therefore, the determined values by SPE/UV were in the acceptable range. In addition, the accuracy and matrix effects were investigated by analysing spiked real waste water samples and the results are shown in **Table 5**. From the recoveries shown in **Table 4**, it can be seen that the SPE/UV procedure described was not affected by the matrix effects as the recoveries for both benzophenone and sulisobenzone

As seen on **Tables 5** and **6**, there was a significant amount of both UV filters on the influent. This can be explained by the fact that the Daspoort waste water treatment plant treats domestic waste water. Therefore as explained by [37], personal care products are usually applied to

Octicrylene Wastewater MEPS-GC-MS 0.25–20 0.081 - 7 [33] Benzophenone-3 Water CE-ESI-MS 300–20000 150 3400 1.5–6.5 [25]

Benzophenone Sunscreen SPE-GC-MS 10–2000 4.4 25.3 4.6–5.5 [36]

MEPS = microextraction in packed syringe, CE-ESI = capillary electrophoresis-electro spray ionisation, SBSE-LC-MS/ MS = stirbar sorptive extraction-liquid chromatography tandem mass spectroscopy, DLLME-UPLC = dispersive liquidliquid microextraction-ultra pressure liquid chromatography, GC-MS = gas chromatography-mass spectroscopy, SPE = solid phase extraction, DDA-IL =double dispersant assisted-ionic liquid, UV/vis = ultraviolet-visible spectrophotometry.

**Table 4.** Comparison of the analytical figures of merit of the current method and those reported in the literature.

Wastewater SPE-UV/vis 0.50–250 0.15-0.28 50 and 55 3.1–5.2 Current

**DLDLR (μg L−1)**

**LOD (μg L−1)**

0.005–0.5 0.0009 - 3–7 [34]

0.6–40 0.2 - 1.9–13.1 [35]

0.1–0.5 0.03 262 <15 [25]

0.002–1.5 0.0013 - 3.5–5.3 [32]

**EF % RSD Refs.**

work

**method**

MS

UPLC-MS/ MS

GC-MS

DLLME

, LOD, LOQ, and EF were found to be

http://dx.doi.org/10.5772/intechopen.75641

53

**Figure 4.** Response surfaces obtained for (A) benzophenone and (B) sulisobenzone after extraction and preconcentration by SPE.

The effect of pH on the extraction and preconcentration of UV filters can be seen from **Tables 2** and **3**. Acidic pH resulted in lower recoveries. This is because benzophenone and sulisobenzone have pKa values of 7.5 and 7.6, respectively, meaning that in acidic pH they are more likely to accept H<sup>+</sup> ions resulting in lower recoveries as they end up with an overall positive charge. In more alkaline conditions (pH 10), the analytes are easily displaced on the adsorbent, resulting in little or no adsorption.

The central pH (7) showed the highest recoveries for both UV-filters. This is a result of the interaction of the analytes' negative charge with the positive charges of the adsorbent prior to elution with methanol.

The optimum conditions obtained by the multivariate approach were confirmed experimentally. Under these conditions (7.5, 3.5, 2 mL min−1 for sample pH, sample and eluent flow rates, respectively), quantitative recoveries ranging from 96 to 98.6% were obtained. These recoveries were compared with the predicted recoveries values (95.6 and 98.1% for benzophenone and sulisobenzone) obtained using the RSM model. It was then concluded that the results obtained by RSM model were valid since there was no significant difference at a 95% confidence level between the experimental and predicted values.

### **3.2. Analytical performance**

Under the determined optimum experimental conditions, the analytical performances of the developed method for preconcentration and determination of UV-filters were investigated. The calibration curves were obtained after a set of standard solutions (0 to 350 μg L−1) was processed using the described SPE procedure. The concentrations of the analytes in the eluent solutions were quantified with the aid of a UV spectrophotometer. The limits of detection and quantification were calculated using the expressions: LOD = 3*Sd* ⁄ *<sup>b</sup>* and LOQ= 10*Sd* ⁄ *<sup>b</sup>*, where *Sd* is the standard deviation of 10 replicate measurements at lower concentrations of calibration curves and *b* is the slope of each calibration curves. Dynamic linear ranges (DLR), correlation coefficient (R2 ), enrichment factor (EF), LOD and LOQ for benzophenone were determined to be LOQ-250 μg L−1, 0.9990, 50, 0.28 μg L−1 and 0.93 μg L−1, respectively. Whereas for sulisobenzone, the DLR, R2 , LOD, LOQ, and EF were found to be LOQ-250 μg L−1, 0.9991, 0.15 μg L−1, 0.50 μg L−1 and 55, respectively. Furthermore, the intraday (repeatability; n = 10) and interday (reproducibility; n = 7 working days) precisions of the SPE method, expressed in terms of relative standard deviation (% RSD), ranged 3.1–3.3 and 4.5–5.2%, respectively. The analytical performance of the proposed method was compared to other methods that reported in the literature **Table 4**. It can be seen that the performance of the current method was comparable or better that those reported in the literature. In addition, the LODs obtained using the current method were lower compared to [25, 32].
