Open Access Published by De Gruyter May 19, 2021

Determination of four parabens in cosmetics by high-performance liquid chromatography with magnetic solid-phase and ionic dispersive liquid–liquid extraction

Xun Gao, Kai Xu, Miaomiao Chi, Jiaojiao Li, Lingzhe Suo, Lin Zhu, Kexin Chen and Jingqing Mu


To determine the trace amount of four benzoic acid esters in cosmetics, ionic dispersive liquid–liquid microextraction (DLLME) and magnetic solid-phase extraction were combined and optimized. After solvent optimization, 1-octyl-3-methylimidazolium hexafluorophosphate was selected as the extraction solvent to form hydrophobic droplets in the process of ionic DLLME, followed by removal of ions from the sample solution containing Fe3O4@GO nano-materials. The magnetic nano-materials were characterized using Fourier transform infrared spectroscopy, scanning electron microscopy, transmission electron microscopy, and vibrating sample magnetometer. Some parameters affecting the efficiency of extraction were optimized using Box-Behnken design. Under optimized conditions, the limit of detection for all the preservatives was less than 0.135 mg/L and the accuracy ranged from 88.5% to 101%. This technology could determine the trace amount of preservatives in cosmetics with comparatively higher accuracy and sensitivity.


[C 8MIM][PF 6]

1-octyl-3-methylimidazolium hexafluorophosphate


analysis of variance


Box-Behnken design




coefficient of variation


capillary electrophoresis


cloud point extraction


dispersive liquid–liquid microextraction


ethyl 4-hydroxybenzoate

Fe 3O 4@GO

magnetic graphene oxide


fast liquid chromatography


Fourier transform infrared spectroscopy


graphene oxide


high-performance liquid chromatography


ionic liquids


limit of detection


limit of quantification


microwave-assisted extraction


methyl 4-hydroxybenzoate


magnetic solid-phase extraction


propyl p-hydroxybenzoate


relative standard deviation


scanning electron microscope


solid-phase extraction


transmission electron microscopy


ultrasonic-assisted extraction


ultra-high-performance liquid chromatography


vibrating sample magnetometer

1 Introduction

Currently, most of the active ingredients in cosmetics are nutrients, which may degrade easily because of microbial contamination, failing product specification [1]. Because of their comparatively wider antibacterial spectrum, strong antibacterial activity, and good solubility in water, and because they are colorless, odorless, and cheaper, benzoate preservatives are widely used in the production of cosmetics [2,3,4,5]. According to recent medical research results, excessive paraben may cause irritation to the human skin and respiratory and endocrine systems [6,7,8,9]. In addition, exposure to paraben in some cosmetic products has been proved to be associated with birth size [10]. The safety issue of paraben in cosmetics has attracted increasing attention, and in many countries, the accepted limits of preservatives are strictly controlled. Taking p-hydroxybenzoic acid as an example, the authorized maximum content of single esters and ester mixtures should not exceed 0.4% (v/v) and 0.8% (v/v), respectively [11].

As the formulations of cosmetic products are relatively complicated and diverse, the reported methods showed comparatively low sensitivity, accuracy, and efficiency [12]. Thus, sample preparation, including preconcentration and purification, are the key steps to improve accuracy and sensitivity when using high-performance liquid chromatography (HPLC) and ultra-high-performance liquid chromatography (UHPLC) based on the ultraviolet-visible detector and diode array detector [13,14,15,16].

Different extraction technologies for benzoate preservative have been reported, such as vortex extraction, supercritical fluid extraction, matrix solid-phase dispersion, and quick, easy, cheap, effective, rugged, and safe. To reduce energy cost and organic solvent usage without compromising extraction efficiency, dispersive liquid–liquid microextraction (DLLME) was optimized to concentrate the target analytes in extracting solvent [17]. Furthermore, ionic liquids (ILs), known as green solvents, were used as extracting solvents in ILs-DLLME with their unique properties, such as high chemical and thermal stability, low flammability, and excellent solvation ability for a wide range of compounds [18,19,20]. Moreover, to avoid time-consuming centrifugation in ILs-DLLME, the magnetic nano-materials were applied to absorb the target analytes by hydrophobic interaction which was known as magnetic solid-phase extraction (MSPE). Magnetic graphene oxide (Fe3O4@GO) nano-materials, with high extraction ability and its magnetic properties, could improve efficiency in purification [21,22,23,16].

In this paper, IL-DLLME-MSPE, which was a method rarely reported in the literature, was developed for the extraction of four types of benzoic acid esters from cosmetics. IL-DLLME remarkably cut the energy and cost by reducing the usage of organic solvents and improved the ease of operation, and MSPE greatly improved the efficiency of separation and extraction, which solved the problem of long-time centrifugation, freezing, or manual collection of large amounts of organic solvents in DLLME. The optimized method was proved to be easy operating, cost-effective, and eco-friendly with acceptable accuracy and precision, which can be applied in the routine assay for preservatives in cosmetics [24,25,26].

2 Experimental

2.1 Instruments

HPLC analysis was performed using Agilent 1200 high performance liquid chromatograph (California, USA) equipped with a Diamonsil C18 column (4.6 mm × 150 mm, 5 μm) (Beijing, China). The column temperature was kept at 35°C. The injection volume was set at 10 μL. The analytes were separated using isocratic elution with a 70:30 (v/v) mixed solution of methanol and water at a flow rate of 0.8 mL/min. The detection wavelength was set at 254 nm. FT-IR spectra were recorded on a Bruker IFS-66 Fourier transform infrared spectroscopy (Karlsruhe, Germany). Magnetization measurements were performed on a Quantum Design Vibrating Sample Magnetometer (VSM, San Diego, USA). The structure and morphology of the prepared nano-materials were characterized by a Helio Nanolab 600i Scanning Electron Microscope (SEM, Hillsboro, USA). The transmission electron microscopy (TEM) images of the prepared nano-materials were obtained using a JEM-2100F electron microscopy (Tokyo, Japan).

2.2 Materials

Deionized water was obtained from Wahaha Group Co., Ltd. (Hangzhou, China). HCl, NaOH, and absolute alcohol (purity, 95%) were supplied by Hengxing Chemical Reagent Factory (Tianjin, China). Chromatographic pure methanol, acetonitrile, and acetone were purchased from Shandong Yuwang Industrial Co., Ltd. (Yucheng, China). FeCl3·6H2O (purity, 99%) and FeCl2·4H2O (purity, 99%) were provided by Ailan Chemical Technology Co., Ltd. (Shanghai, China). Graphene oxide (GO) was supplied by Nanjing Ji Cang Nano Technology Co., Ltd. (Nanjing, China). Methyl 4-hydroxybenzoate (MP), ethyl 4-hydroxybenzoate (EP), propyl p-hydroxybenzoate (PP), butyl-paraben (BP) standard, and 1-octyl-3-methylimidazolium hexafluorophosphate ([C8MIM][PF6]) were purchased from Shanghai Aladdin Chemistry Co., Ltd. (Shanghai, China). The standard solutions of MP, EP, PP, and BP with a concentration of 100 μg/mL were prepared in a volumetric flask and stored away from light at 4°C. The standard solutions were diluted with methanol and prepared every day.

2.3 Preparation of Fe3O4@GO nano-materials

According to the chemical co-precipitation method, Fe3O4@GO nano-materials were prepared [27]. Briefly, 0.5 g of GO powder and 100 mL of distilled water were taken in a 250 mL three-neck round bottom flask, followed by mechanical stirring until the water bath temperature increased evenly from 20°C to 70°C in 3 min. About 2.16 g of FeCl3·6H2O and 0.80 g of FeCl2·4H2O were weighed and dissolved in 40 mL of distilled water with vigorous stirring at 70°C for 5 min. Then the pH of this suspension was adjusted to 12.0 with an ammonia solution (28%, g/mL) to conjugate the Fe3O4 nano-materials onto GO sheets. After mechanically stirring at 70°C for 60 min, the black magnetic material was separated from the suspension using a magnet and washed thrice with distilled water and ethanol. To avoid complete growth and oxidation of the nano-material crystals, the above reaction needs to be carried out in a nitrogen environment. Fe3O4@GO nano-materials were achieved after drying under vacuum at 60°C for 6 h and grounded using a mortar for further use.

2.4 Sample preparation

We have purchased five different brands of cosmetics from a local shopping mall (Shenyang, Liaoning), including lotion and facial cleanser. Weighed 5 g of lotion and 5 g of facial cleanser, and measured 1 mL of toning lotion and 1 mL of cleansing water to 100 mL centrifugal tube separately. This was mixed with 50 mL of distilled water and the matrix dispersed by vortexing for 2 min. After sonication for 10 min, the suspension was filtered through a 0.45 μm filter membrane and stored at 4°C for further use.

2.5 Pretreatment of IL-DLLME and MSPE

About 4.0 g of NaCl was weighed and dissolved completely in the above samples, and the pH was adjusted to 3.2 with 0.1 mol/L HCl. Then, 30 μL of [C8MIM][PF6] and acetonitrile were added to the suspension before shaking for 2 min. After that, 32.7 mg of Fe3O4@GO was added to the dispersion solution, and sonicated for 6 min. After separation by a magnet, magnetic nano-materials were resuspended in 2 mL of absolute ethanol and sonicated for 5 min. Finally, Fe3O4@GO were magnetically separated from the suspension again, and the supernatant was collected and filtered with a 0.22 μm filter membrane before injecting into HPLC for further analysis.

3 Results and discussion

3.1 Characterization of Fe3O4@GO nanoparticles

The dimension and surface morphologies of the nano-materials were characterized by SEM, TEM, FT-IR, and VSM.

The surface morphologies of GO and Fe3O4@GO from SEM are shown in Figure 1a and b. Figure 1a show a typically smooth and disorderly folded sheet-like structure of GO. In comparison, the SEM image of Fe3O4@GO in Figure 1b show a rougher surface because a large number of Fe3O4 nano-materials have been attached to the surface of GO. The structural characteristics of GO and Fe3O4@GO observed by TEM are shown in Figure 1c, and the GO was sheet-like in shape with high transparency and wrinkled edges. Fe3O4 nano-materials were located at the cross-section of Fe3O4@GO nano-materials (Figure 1d) and observed as dark spots that distributed homogeneously on the smooth surface of the GO sheet.

Figure 1 (a) SEM image of GO; (b) SEM image of Fe3O4@GO; (c) TEM image of GO; and (d) TEM image of Fe3O4@GO.

Figure 1

(a) SEM image of GO; (b) SEM image of Fe3O4@GO; (c) TEM image of GO; and (d) TEM image of Fe3O4@GO.

Figure 2 shows the FT-IR spectra of GO and Fe3O4@GO. The most distinct peak at 3436.6 cm−1 was attributed to the O–H stretching vibration. The characteristic peaks of GO that appeared at 1720.8, 1629.6, 1401.1, and 1052.1 cm−1 were corresponded to the carbonyl (C═O) stretching, sp2 carbon skeletal network, C–OH group stretching, and C–O–C in the epoxide group stretching vibration, respectively. According to the spectra of MGO, the new presence of absorption peak at 1396.4 cm−1 indicated an additional vibration band in samples, confirming the formation of the chemical bond between the carboxyl group and Fe. In addition, the peak at 584.5 cm−1 was attributed to the Fe–O–Fe bond vibration. Considering the above evidence, we could conclude that the covalent bonds between magnetic nanoparticles and GO were successfully synthesized.

Figure 2 FT-IR spectrum of GO and Fe3O4@GO.

Figure 2

FT-IR spectrum of GO and Fe3O4@GO.

The magnetic property of Fe3O4@GO at 20°C was investigated using VSM, and the hysteresis loop diagram is shown in Figure 3. The magnetization hysteresis loops of Fe3O4@GO were S-like curves, indicating a superparamagnetic capability of the synthesized nano-materials. The saturate magnetization value of Fe3O4@GO was 47.15 emu/g and enough to be separated with conventional magnets that were sufficient for magnetic separation with conventional magnets. Therefore, Fe3O4@GO was able to be separated rapidly from the aqueous matrix.

Figure 3 Magnetization curves of GO and Fe3O4@GO.

Figure 3

Magnetization curves of GO and Fe3O4@GO.

3.2 Optimization of IL-DLLME

To obtain optimum extraction efficiency of preservatives in cosmetics, a series of experiments were designed. Some parameters that may affect the extraction efficiency, such as vortexing time, ionic strength, the volume of extraction solvent, and disperser solvent, were optimized. The types of extraction solvent and disperser solvent have been selected according to the previous study with modifications [18].

3.2.1 Volume optimization of extraction solvent

The volume of ILs also played a critical role in improving the extraction efficiency in IL-DLLME. The volume of the extraction solvent varied from 10 to 50 μL. As shown in Figure 4a, the recoveries decreased slightly when the volume was more than 30 μL. This may be because the excessive volume of [C8MIM][PF6] was easier to adhere to the side-walls of tubes and could not effectively extract the target analytes in the form of tiny droplets. Therefore, the volume of 30 μL of [C8MIM][PF6] was selected.

Figure 4 Optimization of extraction parameters for IL-DLLME including (a) volume of ionic liquid, (b) volume of disperser solvent, (c) time of vortex, and (d) amount of NaCl. In each optimization, the optimal amount of the previous step was used. Other conditions were considered unchanged and the same as the previous step.

Figure 4

Optimization of extraction parameters for IL-DLLME including (a) volume of ionic liquid, (b) volume of disperser solvent, (c) time of vortex, and (d) amount of NaCl. In each optimization, the optimal amount of the previous step was used. Other conditions were considered unchanged and the same as the previous step.

3.2.2 Volume optimization of disperser solvent

The influence of the acetonitrile volume on the extraction efficiency of the four common preservatives in cosmetics was also investigated in the range of 10–50 μL. When the extraction solvent is less than 30 μL, the extraction solvent unable to be evenly dispersed in the water phase in the form of droplets, however, when the volume of disperser solvent is higher than 30 μL, the solubility of the analyte in the aqueous phase and decrease its concentration in the extraction solvent, both of which resulted in poor recovery. Therefore, given the considerations of cost and recovery in Figure 4b, we selected the volume of 30 μL for acetonitrile in further study.

3.2.3 Optimization of vortexing time

The vortexing time is an important factor in the efficiency of dispersing cosmetic samples, and the time ranged from 0 to 5 min was investigated. The results are shown in Figure 4c; when the vortex time was 2 min, the recoveries of the four common preservatives were more than 90%. However, with the increase in time, the recoveries decreased slightly or basically unchanged. Therefore, vortexing for 2 min was used to the next study.

3.2.4 Optimization of ionic strength

In IL-DLLME, the ionic strength can reduce the solubility of target analytes in the aqueous phase and increase the partition coefficient of the [C8MIM][PF6] phase and further affect the recovery of extraction. The influence of ionic strength in this experiment was assessed by the addition of NaCl in the range of 0–6.0 g in initial aqueous sample. Figure 4d shows that the recoveries reached the maximum value when 4.0 g of NaCl was used. However, excessive salt could also increase the solubility of ILs in the aqueous phase, resulting in a decrease in the recoveries.

3.3 Optimization of MSPE conditions

To understand the influence of experimental variables on the MSPE process, the interaction between the main parameters that affect the extraction efficiency of the analyte must be fully considered, and each parameter must be optimized. On the basis of preliminary experiments, the influence of six factors was studied from two levels of extraction process and elution process. The main parameters, such as pH of the sample solution, extraction time, and amount of sorbent, were evaluated using the Box-Behnken design (BBD) with a small number of trials, which is a widely used method to fit a second-order response surface. The other three parameters, such as type and volume of desorption solvent, and desorption time, were optimized by one-factor-at-a-time procedure.

The Design-Expert 11.0 statistical software was used for the experimental design and data analysis. Based on the results acquired from the analysis of variance (ANOVA) test (presented in Table 1), the significance of model is mainly calculated by R2 value. As presented in Table 1, the coefficients were very high about 0.9569, which indicated that 95.69% of total variance were explained by quadratic model and only 4.31% of total variables were not explained by the model. The adjusted R2 value was 0.9015. The R2 and adjusted R2 values were closed to 1.0000, which displayed a good correlation between experimental and theoretical results. The model had F values of 17.26 with lower p-values (0.0005), which also indicated the impacts of this model. If the lack of fit value in a model was nonsignificant, it suggested that the quadratic model was valid. Here, the lack of fit values were 0.1613, which specified that our model was significant. Another one was the coefficient of variation (C.V.%); if this value was less than 10%, then the model was highly reliable and reproducible for the experimental studies. The C.V.% value for clenbuterol was 8.89%, less than 10%. Therefore, the model was reliable for experimental studies for the determination of four preservatives in cosmetics.

Table 1

Analysis of variance (ANOVA) test for the second-order regression model

Source Sum of squares Degree of freedom Mean square F-value p-value (Prob > F)
Model 6408.17 9 712.02 17.26 0.0005* significant
A-pH 3969.40 1 3969.40 96.25 <0.0001*
B-Sorbent 335.41 1 335.41 8.13 0.0246
C-time 420.50 1 420.50 10.20 0.0152
AB 61.62 1 61.62 1.49 0.2611
AC 97.02 1 97.02 2.35 0.1689
BC 97.02 1 97.02 2.35 0.1689
A2 489.98 1 489.98 11.88 0.0107
B2 677.78 1 677.78 16.43 0.0048
C2 126.79 1 126.79 3.07 0.1230
Residual 288.68 7 41.42
Lack of fit 198.90 3 66.30 2.95 0.1613 not significant
Pure error 89.78 4 22.45
Cor total 6696.86 16
R2 = 95.69% Adj R2 = 90.15% C.V.% = 8.89

*Significant at 0.01 level.

The pH of the sample solution has a negative effect on the recovery in MSPE process because it can determine the ionization of the target analytes and further affect their recoveries. It is seen from Figure 5 that as the pH increases, the recoveries of four benzoates become more unfavorable. When the pH value was higher than the pKa of benzoate, the target analytes existed in the ionized form could hardly be extracted using organic solvents. The extraction efficiency decreased with the hydrolysis of benzoate at a pH of 3.2, and the results in Figure 5 show that the amounts of Fe3O4@GO was positive to efficiencies of MSPE in the range of 10–32.7 mg. To ensure sufficient contact between the adsorbent and the target analytes, the relative recoveries that are shown in Figure 5 increased with the increase in the extraction time from 1 to 5 min and then remained plateau after 5 min.

Figure 5 Response surface plots for effect of the MSPE extraction for (a) extraction time, (b) amount of sorbent, and (c) pH.

Figure 5

Response surface plots for effect of the MSPE extraction for (a) extraction time, (b) amount of sorbent, and (c) pH.

To elute the adsorbed analytes off the adsorbent completely, eluting conditions, such as the type and volume of solvent, as well as the time for desorption, were evaluated to improve the recovery. Three organic solvents, methanol, acetonitrile, and ethanol, were investigated. It can be seen from Figure 6 that ethanol generated the highest recovery of analytes. The volume of the desorption solvent was also investigated in the range of 3.0–9.0 mL. As shown in Figure 6, the results indicated that 2.0 mL of ethanol was sufficient to elute all the target analytes with an acceptable recovery. In addition, the desorption time in the range of 1–11 min was compared. As shown in Figure 6, benzoate could be efficiently eluted in 5 min.

Figure 6 Optimization of elution parameters for MSPE including (a) type of desorption solvent, (b) volume of desorption, and (c) time of desorption.

Figure 6

Optimization of elution parameters for MSPE including (a) type of desorption solvent, (b) volume of desorption, and (c) time of desorption.

3.4 Validation of the method

Under the optimal experimental conditions, the linearity, precision, accuracy, sensitivity, and specificity of the method were validated.

Mixed reference solution, blank solution, and test solution were injected into HPLC for analysis. The results (Figure 7) show that no interference was observed.

Figure 7 Chromatograms of the mixed reference solution.

Figure 7

Chromatograms of the mixed reference solution.

Within the range of 0.5–100 μg/mL for the standard solution, 0.5, 1, 5, 10, 20, 50, and 100 μg/mL of the standard solution in the sequence was added to 0.5 g placebo (cosmetic milk) and introduced into HPLC for analysis after sample preparation. The criteria for the limit of detection (LOD) were set at the signal-to-noise ratio of 3, and the limit of quantification (LOQ) was set at 10.

The linearity of all analytes showed a good correlation in the specified range of 0.5–100 mg/L and the R2 ranged from 0.9980 to 0.9997 in Table 2. Under the optimized condition, the LOD of the four target analytes ranged from 0.064 to 0.135 μg/mL. The sensitivity of this method meets the analysis and detection requirements of four preservatives in cosmetics [11].

Table 2

The analytical performance data for HPLC system

Analytes Retention time (min) Calibration equation Linear range (mg/L) R2 LOQ (mg/L) LOD (mg/L)
MP 3.290 Y = 72.29X + 63.97 0.5–100 0.9997 0.216 0.064
EP 3.995 Y = 67.73X + 52.63 0.5–100 0.9995 0.228 0.068
PP 5.233 Y = 64.05X + 47.70 0.5–100 0.9989 0.231 0.069
BP 7.296 Y = 58.55X + 41.88 0.5–100 0.9980 0.451 0.135

Accuracy and precision were conducted by spiking the samples with mixed standards at three different levels (n = 3 at each level), and six replicates of each concentration were made in parallel for 3 days. The mixed reference solution was added to a 0.5 g blank matrix to make final concentrations of 1, 10, and 50 μg/mL. The recovery and relative standard deviations (RSDs) are presented in Table 3, the intra-day RSDs for all the analytes were no more than 4.3%, and the inter-day RSDs were no more than 5.6%. The above results showed that the accuracy and precision of this method are good.

Table 3

Intra-day and inter-day recoveries and relative repeatability standard deviation of analytes in cosmetic samples

Analytes Spiked (mg/L) Inter-day (n = 3) Intra-day (n = 6)
Relative recovery (%) RSD (%) Relative recovery (%) RSD (%)
MP 1 88.3 2.1 85.6 4.5
10 87.9 2.8 86.6 3.9
50 84.1 1.7 84.2 3.2
EP 1 93.2 4.3 90.2 4.2
10 92.3 2.4 91.8 3.8
50 89.7 2.9 89.3 1.7
PP 1 99.2 3.7 98.4 5.6
10 98.6 2.9 97.7 3.1
50 97.3 1.5 97.8 1.2
BP 1 96.7 2.1 96.8 4.7
10 95.3 2.6 96.1 2.2
50 93.7 1.8 94.6 0.9

3.5 Assay of products

To verify the applicability of the optimized and validated method, six different types of cosmetic products were analyzed, including paste, emulsion, and liquid formula. The results are presented in Table 4, and the chromatogram of the real sample is shown in Figure 8. The “Safety Technical Specifications for Cosmetics” (2015 edition) stipulates that the total amount of parabens in cosmetics shall not exceed 0.5%, and the amount of monomers added shall not exceed 0.4% [28]. After analysis, most of the four benzoate compounds in the cosmetic samples did not exceed the specified limit, and only one sample of the makeup remover was detected to be out of the specification MP.

Table 4

Concentrations of four parabens in real samples

Sample Content of the analytes (mg/kg)
Astringent 1 138.2 74.53 48.13 103.3
Astringent 2 ND ND ND ND
Cleansing water 1 4,290 ND ND ND
Cleansing water 2 6.422 ND ND ND
Lotion 272.7 ND 0.9510 ND
Facial cleanser ND ND ND 31.04

ND: not detected.

Figure 8 Chromatogram of the real sample.

Figure 8

Chromatogram of the real sample.

3.6 Comparison of pretreatment methods

The extraction recoveries of DLLME-MSPE were compared with DLLME or MSPE alone, and the results are shown in Figure 9. It was found that Fe3O4@GO nanocomposites adsorbed and separated ILs in the sample solution, greatly improving the separation and extraction efficiency, which solved the problem of long-time consumption by centrifugation or manual collection. The newly developed method possessed a wide linearity range, high accuracy, and short sample preparation time, which provides a powerful means for monitoring the trace amount of preservative compounds in cosmetics.

Figure 9 Comparison of pretreatment methods between DLLME-MSPE, DLLME, and MSPE.

Figure 9

Comparison of pretreatment methods between DLLME-MSPE, DLLME, and MSPE.

3.7 Comparison to reported methods

The comparison between the previously reported method and our proposed method for analyzing MP, EP, PP, and BP in different products is listed in Table 5. Identification of the four parabens in cosmetics, human secretions, and environmental and food samples were reported but the sample preparation using IL-DLLME-MSPE has rarely been reported for the analysis of preservatives in cosmetics. Alshana et al. applied capillary electrophoresis (CE) to determine the four parabens in breast milk and food samples, with an LOD of 0.1–0.2 μg/mL and recovery of 86.7–103.3% [29]. For the reported ultrasonic-assisted extraction (UAE) combining fast liquid chromatography (FLC) method, the LOD of the four parabens in cosmetics, cleaning agents, and pharmaceutical products ranged in 0.06–4.38 μg/mL [30]. Currently, for the analysis of preservatives, there is no universal sample preparation technique and analytical method.

Table 5

Comparison of the method proposed with the method developed earlier in the analysis of MP, PP, EP, and BP

Sample type Extraction and determination method Analysis time (min) Linear range (µg/mL) LOD (µg/mL) Precision (RSD%) Recovery (%) Reference
Water SPE-HPLC/MS-MS 20 0.08–500 0.012–0.024 <20 70–115 [31]
Saliva and toothpaste SPE-LC-UV 15 0.3–50 0.1–0.3 <4 86–113 [13]
Cosmetics, cleaning agents, and pharmaceuticals UAE-FLC-UV 27 0.25–10 0.06–4.38 <5 69–119 [30]
Water CPE-HPLC-MS/MS 8 0.009–0.073 0.013–0.038 <0.86 71.2–97.7 [32]
Marine sediments IL-MAE-HPLC 25 0.06–1.4 0.004–0.026 <17 87.9–104 [33]
Breast milk and food DLLME–CE 5 0.7–6.0 0.1–0.2 <3.5 86.7–103.3 [29]
Cosmetics DLLME-MSPE-HPLC/UV 8 0.5–100 0.064–0.135 <5.6 88.5–101 This study

4 Conclusion

In this experiment, two-step extraction techniques of IL-DLLME and MSPE were used, combined with HPLC method to carry out selective extraction enrichment and trace determination of four benzoate preservatives in different types of cosmetic samples. The optimized method was efficient, accurate, and environment-friendly. It can be applied to analyze preservatives in cosmetics samples without matrix interference. Compared with the traditional DLLME method, [C8MIM][PF6] of ILs is a relatively green extractant that offered higher sensitivity and linearity range. This method possessed accepted accuracy, linearity, precision, and detection limit for the determination of benzoic esters in real samples. This newly developed two-step extraction technology may have a broad application prospect in the analysis of trace preservatives in cosmetics.

    Funding information: This work was supported by the National Natural Science Foundation of China (No. 81503029) and Young and Middle-aged Backbone Personnel Training Program of Shenyang Pharmaceutical University (ZQN2016011).

    Author contributions: Kai Xu: writing – review and editing, conceptualization, funding acquisition, methodology; Xun Gao: conceptualization, formal analysis, funding acquisition, resources, supervision; Miaomiao Chi: data curation, writing – original draft, project administration; Jiaojiao Li: supervision, software; Lingzhe Suo: data curation, validation; Lin Zhu: software, validation; Kexin Chen: visualization; Jingqing Mu: investigation.

    Conflict of interest: The authors state no conflict of interest.

    Data availability: All data generated or analysed during this study are included in this published article.


[1] Martins I, Carreira FC, Canaes LS, Junior F, Cruz L, Rath S. Determination of parabens in shampoo using high performance liquid chromatography with amperometric detection on a boron-doped diamond electrode. Talanta. 2011;85(1):1–7. 10.1016/j.talanta.2011.04.047. Search in Google Scholar

[2] Nowak K, Ewa J, Wioletta RW. Controversy around parabens: alternative strategies for preservative use in cosmetics and personal care products. Environ Res. 2020 in press. 10.1016/j.envres.2020.110488. Search in Google Scholar

[3] Lee MR, Lin CY, Li ZG, Tzu TF. Simultaneous analysis of antioxidants and preservatives in cosmetics by supercritical fluid extraction combined with liquid chromatography-mass spectrometry. J Chromatogr A. 2006;1120(1–2):244–8. 10.1016/j.chroma.2006.01.075. Search in Google Scholar

[4] Fransway AF, Fransway PJ, Belsito DV, Warshaw EM, Sasseville D, Fowler JF. Parabens. Dermatitis. 2019;30(1):3–31. 10.1097/DER.0000000000000429. Search in Google Scholar

[5] Fei T, Li H, Ding M, Ito M, Lin JM. Determination of parabens in cosmetic products by solid-phase microextraction of poly(ethylene glycol) diacrylate thin film on fibers and ultra high-speed liquid chromatography with diode array detector. J Sep Sci. 2015;34(13):1599–606. 10.1002/jssc.201100225. Search in Google Scholar

[6] Núnez LE, Turiel A, Martin E, Tadeo JL. Molecularly imprinted polymer for the extraction of parabens from environmental solid samples prior to their determination by high performance liquid chromatography-ultraviolet detection. Tal anta. 2010;80(5):1782–7. 10.1016/j.talanta.2009.10.023. Search in Google Scholar

[7] Aschenbeck KA, Erin MW. Allergenic ingredients in facial wet wipes. Dermatitis. 2017;28(6):353. 10.1097/DER.0000000000000268. Search in Google Scholar

[8] Nowak K, Wioletta RW, Maria GB, Ewa JA. Parabens and their effects on the endocrine system. Mol Cell Endocrinol. 2018;474:238–12. 10.1016/j.mce.2018.03.014. Search in Google Scholar

[9] Ocana G, Juan A, Mercedes VN, María RP, Rut FT, Miguel-Angel BL. New developments in the extraction and determination of parabens in cosmetics and environmental samples: a review. Anal Chim Acta. 2015;858:1–15. 10.1016/j.aca.2014.07.002. Search in Google Scholar

[10] Chang CH, Wang PW, Liang HW, Huang YF, Huang LW, Chen HC, et al. The sex-specific association between maternal paraben exposure and size at birth. Int J Hyg Environ Heal. 2019;222(6):955–10. 10.1016/j.ijheh.2019.06.004. Search in Google Scholar

[11] Labat LE, Kummer PD, Dubost JP. Comparison of high-performance liquid chromatography and capillary zone electrophoresis for the determination of parabens in a cosmetic product. J Pharmaceut Biomed. 2000;23(4):763–7. 10.1016/S0731-7085(00)00358-7. Search in Google Scholar

[12] Núez L, Tadeo JL, García-Valcárcel AI, Turiel E. Determination of parabens in environmental solid samples by ultrasonic-assisted extraction and liquid chromatography with triple quadrupole mass spectrometry. J Chromatogr A. 2008;1214(1):178–82. 10.1016/j.chroma.2008.10.105. Search in Google Scholar

[13] Zotou A, Sakla I, Tzanavaras PD. LC-determination of five paraben preservatives in saliva and toothpaste samples using UV detection and a short monolithic column. J Pharmaceut Biomed. 2010;53(3):785–9. 10.1016/j.jpba.2010.05.018. Search in Google Scholar

[14] Zhong HN, Li ZC, Chen S, Zeng Y, Zheng JG, Zeng Y, et al. Simultaneous quantitative analysis of six isothiazolinones in water-based adhesive used for food contact materials by high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS). Molecules. 2019;24(21):3894. 10.3390/molecules24213894. Search in Google Scholar

[15] Lecce R, Luca GR, Carlo M, Giampaolo I, Rita P, Alessia P. Screening of preservatives by HPLC-PDA-ESI/MS: a focus on both allowed and recently forbidden compounds in the new EU cosmetics regulation. J Pharmaceut Biomed. 2016;125:260–9. 10.1016/j.jpba.2016.03.044. Search in Google Scholar

[16] Shen HY, Jiang HL, Mao HL, Pan G, Zhou L, Cao YF. Simultaneous determination of seven phthalates and four parabens in cosmetic products using HPLC-DAD and GC-MS methods. J Sep Sci. 2007;30(1):48–7. 10.1002/jssc.200600215. Search in Google Scholar

[17] Beshare H, Shamsipur M, Fattahi N. Solid-phase extraction followed by dispersive liquid-liquid microextraction based on solidification of floating organic drop for the determination of parabens. J Chromatogr Sci. 2015;53(8):1414–9. 10.1093/chromsci/bmv011. Search in Google Scholar

[18] Peng Y, Ren H, Qiu H, Xia L, Jiang S. Determination of four trace preservatives in street food by ionic liquid-based dispersive liquid-liquid micro-extraction. Chem Pap. 2011;65(6):747–53. 10.2478/s11696-011-0071-9. Search in Google Scholar

[19] Ali I, Suhail M, Sanagi MM, Aboul-Enein HY. Ionic liquids in HPLC and CE: a hope for future. Crit Rev Anal Chem. 2017;47(4):1–8. 10.1080/10408347.2017.1294047. Search in Google Scholar

[20] Bystrzanowska M, Pena PF, Marcinkowski L, Tobiszewski M. How green are ionic liquids? – a multicriteria decision analysis approach. Ecotox Environ Safe. 2019;174:455–8. 10.1016/j.ecoenv.2019.03.014. Search in Google Scholar

[21] Wang L, Zang X, Wang C, Wang Z. Graphene oxide as a micro-solid-phase extraction sorbent for the enrichment of parabens from water and vinegar samples. J Sep Sci. 2014;37(13):1656–62. 10.1002/jssc.201400028. Search in Google Scholar

[22] Elham T, Yadollah Y, Ali M, Fateme R. Polyaniline-coated Fe3O4 nanoparticles: an anion exchange magnetic sorbent for solid-phase extraction. J Sep Sci. 2015;35(17):2256–65. 10.1002/jssc.201200345. Search in Google Scholar

[23] Ye N, Shi P, Li J, Wang Q. Application of graphene as solid phase extraction absorbent for the determination of parabens in cosmetic products by capillary electrophoresis. Anal Lett. 2013;46(13):1991–2000. 10.1080/00032719.2013.784916. Search in Google Scholar

[24] Zhai Y, Na L, Lei L, Xiao Y, Zhang H. Dispersive micro-solid-phase extraction of hormones in liquid cosmetics with metal–organic framework. Anal Methods. 2014;6(23):9435–11. 10.1039/C4AY01763C. Search in Google Scholar

[25] Ebrahimpour B, Yamini Y, Seidi S, Tajik M. Nano polypyrrole-coated magnetic solid phase extraction followed by dispersive liquid phase microextraction for trace determination of megestrol acetate and levonorgestrel. Anal Chim Acta. 2015;885:98–105. 10.1016/j.aca.2015.05.025. Search in Google Scholar

[26] Huang Y, Lin H, Song R, Tian Y, Zhang Z. Optimization of dispersive liquid–liquid microextraction for analysis of levonorgestrel in water samples using uniform design. Anal Methods. 2011;3(4):857–8. Search in Google Scholar

[27] Deng X, Shi S, Li S, Yang T. Magnetic ligand fishing combination with high-performance liquid chromatography-diode array detector-mass spectrometry to screen and characterize cyclooxygenase-2 inhibitors from green tea. J Chromatogr B. 2014;973:55–60. 10.1016/j.jchromb.2014.10.010. Search in Google Scholar

[28] Zhao H, Yin YX. Interpretation of “cosmetics safety technical specification (2015 Edition)”. Daily chemical science; 2017. p. 1. 10.13222/j.cnki.dc.2017.01.002. Search in Google Scholar

[29] Alshana U, Ertas N, Goger NG. Determination of parabens in human milk and other food samples by capillary electrophoresis after dispersive liquid-liquid microextraction with back-extraction. Food Chem. 2015;181:1–8. 10.1016/j.foodchem.2015.02.074. Search in Google Scholar

[30] Irena B, Iwona W, Natalia S, Ewa K. Determination of preservatives in cosmetics, cleaning agents and pharmaceuticals using fast liquid chromatography. J Chromatogr Sci. 2014;51(1):88–7. 10.1093/chromsci/bms210. Search in Google Scholar

[31] Marta-Sanchez AV, Caldas SS, Schneider A, Cardoso S, Primel EG. Trace analysis of parabens preservatives in drinking water treatment sludge, treated, and mineral water samples. Environ Sci Pollut Res Int. 2018;25(15):1–11. 10.1007/s11356-018-1583-4. Search in Google Scholar

[32] Noorashikin MS, Raoov M, Mohamad S, Abas MR. Cloud point extraction of parabens using non-ionic surfactant with cylodextrin functionalized ionic liquid as a modifier. Int J Mol Sci. 2013;14(12):24531–48. 10.3390/ijms141224531. Search in Google Scholar

[33] Delgado B, Pino V, Anderson JL, Ayala JH, Afonso AM, González V. An in-situ extraction-preconcentration method using ionic liquid-based surfactants for the determination of organic contaminants contained in marine sediments. Talanta. 2012;99:972–83. 10.1016/j.talanta.2012.07.073. Search in Google Scholar

Received: 2020-11-22
Revised: 2021-03-21
Accepted: 2021-04-07
Published Online: 2021-05-19

© 2021 Xun Gao et al., published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.