MIT (≥ 94%) and BIT (97%) were purchased from Sigma-Aldrich (Steinheim, Germany). CMIT (99.9%), OIT (99.9%) and IPBC (99.7%) were purchased from Ultra scientific (North Kingstown, USA). Two deuterated compounds of 2-octyl-4-isothiazolin-3-one-d17 (OIT-d17, Toronto Research Chemicals, North York, Canada) and 3-(4-isopropylphenyl)-1,1-dimethylurea (Isoproturon-d6, Dr. Ehrenstorfer, Augsburg, Germany) were used as internal and recovery standards, respectively. The detailed information of target compounds was presented in the Table 1. Used organic solvents including acetone (Ac), acetonitrile (ACN), ethyl acetate (EtAC), methanol (MeOH), methyl tert-butyl ether (MTBE) were of HPLC grade while toluene (Tol) and methylene chloride (MC) were pesticide grade solvents that purchased from J.T. Baker (Philipsburg, NJ, USA). Supelclean ENVI-carb, ENVI-Chrom P cartridges (500 mg, 6 cc) and filtration tubes (empty cartridge, 6 cc) were purchased from Supelco (Bellefonte, USA). Oasis HLB cartridges (200 mg, 6 cc) were purchased from Waters (Milford, USA). Aluminum oxide (98% purity) was purchased from Sigma-Aldrich. The polytetrafluoroethylene (PTFE) syringe filter (0.45 μm pore size, 13 mm diameter) was purchased from Whatman (Maidston, UK). Standard solution for all target compounds was prepared in MeOH.

Target compounds were analyzed by high performance liquid chromatography (Agilent 1200 series HPLC; Agilent Technologies, Santa Clara, CA, USA) coupled with mass spectrometry (1100 MS; Agilent Technologies, Santa Clara, CA, USA) with C18 column (Hypersil gold; 4.6 mm × 250 mm, 5 μm; Thermo Fisher Scientific, Waltham, MA, USA) and Eclipse XDB C18 guard column (4.6 mm × 12.5 mm, 5 μm; Agilent Technologies). The injection volume was 10 μL and positive electrospray ionization mode was used. The details of instrumental conditions, mobile phase, fragment masses for selective ion, and energy for quantification and identification are presented in Table S2.

The type and brand of consumer product samples were chosen based on published market analysis report of products with containing high content of biocidal active ingredients and domestic market survey of reporting approximate usage and production volume. The items selected for testing analytical method in this study were 21 wet wipes (WW), three powder type (PD) and five liquid type (LD) laundry detergents (including two fabric softeners, Table S3). To compare and optimize the extraction and purification conditions per sample type, one each powder and liquid detergent (PD 1 and LD 2), and three wet wipes (WW 1–3) were analyzed.

To optimize the analytical condition of biocides from consumer products, direct injection after dilution, solid phase extraction (SPE) and ultrasonic assisted extraction (UAE) were tested. All of the tested samples were initially fortified with known concentration of internal standard (OIT-d17) before extraction. The recovery standard (Isoproturon-d6) was spiked into the final extractant prior to inject into HPLC/MS.

A nine-point standard curve covering concentration range of 20–1,500 ng/mL of the five target biocides was prepared in MeOH. All calibration standards contained 500 ng/mL of the internal standard (OIT-d17) and 200 ng/mL of the recovery standard (Isoproturon-d6). The middle stage (250 ng/mL) calibration standard was checked every 12 h as an initial calibration verification during analysis. The same standard solution was regularly injected for checking the signal stability from HPLC/MS to avoid any inter-batch difference of sensitivity during this study. The extraction efficiency of each tested method was estimated by calculating recovery rate of tested consumer product samples. The method detection limit (MDL) and accuracy were verified by analyzing matrix spike samples which contain 100 ng/mL and 250 ng/mL of target biocides, respectively. The method detection limits (MDLs, Table S1) were determined by multiplying by 3.14 and the standard deviation of the seven matrix spike replicates per sample type. To assess potential background contamination, procedural blanks (n = 4 per batch) and instrument blanks (n = 7–10 per batch) were analyzed in every experimental batch.

C18 column was chosen for this study on the basis of the previously reported analytical conditions of targeting MIT and CMIT [6, 18, 19]. Based on the preliminary chromatographic separation (SI 1) with solvent A (50% MeOH in water) and B (MeOH), gradient program condition (i) was developed that started with 50% B for 14 min (flow rate; 0.35 mL/min until 11 min then increased as 0.4 mL/min until 14 min), increased to 90% B for 25 min, and increased to 100% B until 35 min (flow rate; 0.5 mL/min). All target compounds were clearly separated as shown in Fig. 1(a). There were no critical variations in the peak abundance during the five sequential injections, showing a standard deviation within 2%. The optimal ionic strength of polar solvent (solvent A in the gradient program) was investigated by comparing the peak abundances between pure water and 0.5% formic acid for the condition (i). The peak abundance of MIT and CMIT was 2–3 times higher with pure water than the 0.5% formic acid introduced condition as shown in Fig. 2.

The optimized mobile phase condition for MIT and CMIT was further expanded to analyze BIT, IPBC, and OIT simultaneously after a minor modification to produce HPLC condition (ii); 50% B for 11 min (0.35 mL/min), increased to 70% B until 15 min (0.4 mL/min), and then held until 38 min (0.6 mL/min). The total ion chromatogram (TIC) of condition (ii) was shown in Fig. 1(b). All target compounds were well separated, with showing capacity factor (α) of 1.02–3.31 between each compound. Therefore, all test samples for optimizing pre-treatment methods were analyzed by condition (i) for MIT/CMIT and continued to be analyzed by condition (ii) to conduct simultaneous analysis of five biocides. The typical HPLC/MS ion chromatogram of each type of sample was shown in Fig. S1.

The experimental methods were firstly screened out by comparing reported extraction recovery from previous literatures. Mostly, liquid samples (i.e., tap and river water, sewage and wastewater) were pre-treated by SPE [3, 4, 15], while solid samples (i.e., sludge, sediment, cosmetics, and paper for food packaging) were pre-treated by UAE [15, 17, 18]. Therefore, these two methodologies were primarily considered and optimized by manipulating various experimental parameters to achieve high extraction efficiency and purification of target biocides from various types of consumer products. All test samples were spiked with 1,000 ng/mL of MIT/CMIT and 500 ng/mL of internal standard (OIT-d17) prior to extraction and were spiked with 200 ng/mL of recovery standard (Isoproturon-d6) before injection. The final extractants of samples were filtered with a polytetrafluoroethylene (PTFE) syringe filter (0.45 μm), concentrated under a gentle stream of nitrogen, and reconstituted with 1 mL of MeOH. The detailed procedure, parameters, and results were described below per each sample type.

To confirm the applicability of the optimized analytical method of biocides developed in this study, 29 consumer products from the Korean market were analyzed for five biocides and their concentration profile was shown in Table 4 and Table S5. Among three powder type detergents, OIT (38.6 mg/kg) was detected in only one sample. In case of five liquid type detergent samples, MIT, BIT, and OIT were detected in the range of 2.59–99.0 mg/kg and their detection frequency (DF) was less than 40%. Target biocides except OIT were found lower than MDL (0.0005–0.0054 mg/kg) in wet wipes used for personal care. However, MIT (100%), CMIT (69%), and IPBC (77%) were detected with higher frequency in wet wipes that were collected from local restaurants, which are usually given to customers as sales promotion. The observed concentrations of MIT (0.03–70.2 mg/kg) and CMIT (ND-11.3 mg/kg) in complementary wet wipes without brand labels were still lower than the Korean guideline value (MIT, 100 mg/kg; CMIT, 15 mg/kg, [14]) but it found to be possible that human exposure for biocides through use of the complementary products even after the banning of production and sales of major consumer products. The samples analyzed in this study are not cosmetics but we can consider the leave-on effect from wet wipes as customers usually use wet wipes to clean their hands instead of washing hands with water. The maximum concentration of MIT/CMIT allowed in Europe was 15 ppm in cosmetics while in the United States, the concentrations allowed are set to 15 ppm and 7.5 ppm in rinse-off products and leave-on cosmetics, respectively [25]. Also, considering these compounds are widely used preservatives in many different types of consumer products which could be another source of biocides to human exposure [26], it should not be underestimated that the potential exposure risk derived from the observed amounts of MIT and CMIT in the consumer products off the regulation to human.

From previous study in Spain, the levels of MIT were reported in laundry detergent and softener that were similar to the levels observed in this study: 1.18 and 2.71 ppm (w/w), respectively [5]. However, the level of BIT (average concentration, 150 ppm) in cleaning products was higher than our study. In the US, the level of MIT/CMIT in cosmetic products, such as shampoo/conditioner and skin care products, was in the range of ND-89.6 mg/kg and ND-10.3 mg/kg, respectively, and the detection frequencies of MIT and CMIT were 95% and 29%, respectively [20].

In light of widespread usage of MIT/CMIT and other biocides in personal care products, skin sensitization in the general population is drawing rising attention. Recently, analytical method for targeting metabolites of isothiazolinones in human urine recently reported [27–29]. To fill the gap between wide occurrence of isothiazolinones and human health risk derived from exposure to biocides, further studies of monitoring human specimens with matched information about life style (e.g. consumer product usage and frequency) questionnaires will be recommended.