Volume 26, Issue 4 p. 565-576
Open Access

A novel protocol for quantitative determination of 1,4-dioxane in finished cleaning products

Brian Palumbo

Brian Palumbo

Impact Analytical, Midland, MI, USA

Search for more papers by this author
Deena Conrad-Vlasak

Deena Conrad-Vlasak

Impact Analytical, Midland, MI, USA

Search for more papers by this author
Kathleen Stanton

Corresponding Author

Kathleen Stanton

American Cleaning Institute, Washington, District of Columbia, USA


Kathleen Stanton, American Cleaning Institute, 1401 H Street, NW, Suite 700, Washington, DC 20005, USA.

Email: [email protected]

Search for more papers by this author
First published: 14 April 2023


A novel method to quantify 1,4-dioxane in finished cleaning products using headspace gas chromatograph with mass spectrometry (HS GC/MS), single ion monitoring, and a fully deuterated internal standard has been developed. The method generates very linear calibration curves with a R2 of at least 0.99, excellent accuracy with spike/recovery of 97%–102%, and effective precision of 1%–7%RSD for different cleaning products. The method also produces an instrument limit of quantitation (LOQ) of <20 ppb, and sample LOQ of <100 ppb. The method robustness was demonstrated with a ten-lab round-robin exercise that showed that even with some unexpected deviations from the method, excellent analytical results were obtained from the different labs. This new method will be valuable to evaluate consumer products impacted by government regulation to limit 1,4-dioxane concentrations in complex matrices.


1,4-Dioxane is a chemical impurity that can commonly be found in a variety of commercial products, especially those that use ethoxylated surfactants (Agency for Toxic Substances and Disease Registry (ATSDR), 2012). It is a substance of concern after being classified as a potential human carcinogen by multiple regulatory authorities and other organizations (U.S. Department of Health and Human Services (DHHS), 2016; EPA, 2013; European Chemicals Bureau (ECB), 2002; National Institute for Occupational Safety and Health (NIOSH), 2010). Product manufacturers are reducing the levels of 1,4-dioxane in household products by optimizing the production processes of ethoxylated surfactants, favoring base-catalyzed processes over acid catalysis, using vacuum stripping to remove volatile impurities, and derivatizing the surfactants.

The concerns of 1,4-dioxane are leading to increased scrutiny and regulation by a variety of government bodies. New York's legislature has passed limits on 1,4-dioxane in household cleaning, personal care, and cosmetic products starting in December 2022 which will be regulated under the Department of Environmental Conservation (DEC). These limits will range from 1 to 10 ppm (New York State Department of Environmental Conservation, n.d.). The California Department of Toxic Substances and Control (DTSC) is also considering consumer product regulations on 1,4-dioxane (California Department of Toxic Substances Control, n.d.).

To understand the amount of 1,4-dioxane in consumer products, sensitive analytical methods are required. There are many different analytical methods in the literature for 1,4-dioxane that can be used at different levels of analyte, different matrices, and using different kinds of instrumentation (Hayes et al., 2022). Most of these methods are focused on analyzing drinking water, a much simpler matrix than consumer cleaning products. Existing methods point to the benefits of different parts of the analysis, including gas chromatography (GC) (Cortellucci & Dietz, 1999), headspace (HS) sampling (Shin & Lim, 2011), mass spectrometry (MS) (Fuh et al., 2005), and using internal standards (IS) (Draper et al., 2000; Sun et al., 2016). However, the analysis of consumer products is complicated by the other components of these complex formulations. Previous methods have been developed to address these concerns using GC/MS, but many have employed either more complex sample preparation schemes or more complex MS techniques (Tanabe & Kawata, 2019; Zhou, 2019). The best and most useful of the previous work for these purposes was using static HS, GC/MS, and an isotopic IS by Tahara et al. (Tahara et al., 2013) These different published methods provide guidance when developing a new method to ensure sufficient sensitivity for sub-ppm quantitation, effectiveness across different cleaning product categories, relatively simple approaches using simpler testing techniques, and robustness across many different labs. The key learnings include the use of a chemically similar internal standard which negates any matrix effects, simple analyte extraction from complex formulations, use of chromatographic separation to isolate the analyte of interest, and highly sensitive detection that is specific to the analyte.

The method described below fulfills all these important criteria and addresses the needs as described in Hayes et al. (2022) We have developed the method with a deuterated internal standard which is chemically identical to the analyte under these conditions. We use headspace analyte extraction to optimize the delivery of 1,4-dioxane because it is applicable over the wide range of sample matrices present in the complex formulations of consumer products. Once in the instrument, we favor gas chromatography coupled with mass spectrometry with selective ion monitoring to provide specific and sensitive detection and quantitation of the 1,4-dioxane.

Once developed, we sought to validate the methodology through a ten-lab round robin exercise using contract, industrial and academic analytical laboratories. Representative cleaning product formulations were produced to simulate laundry and hand dish detergents used in both consumer and industrial, institutional, and commercial settings as these are some of the product categories regulated under NYS DEC and are categories under consideration for regulations by CA DTSC.


A certified reference standard of 99.8% pure 1,4-dioxane was purchased from Sigma Aldrich (CRM48367). A 99% pure, fully deuterated 1,4-dioxane-d8 standard was purchased from Sigma-Aldrich (catalog number 186406) or Alfa Aesar (catalog number 36516). This is used as the IS.

Finished cleaning materials were modeled from common off-the-shelf consumer products. Each has a specified amount of surfactant in a water base. The relative concentrations of the components are included in Appendix A (Tables A1 and A2):
  • Hand dish soap 1 (30% surfactant in water)
  • Hand dish soap 2 (15% surfactant in water)
  • Consumer laundry detergent 1 (15% surfactant in water)
  • Consumer laundry detergent 2 (50% surfactant in water)
  • Industrial and institutional (I&I) laundry detergent (50% surfactant in water)

This protocol uses HS GC coupled to MS (GC/MS) to quantify 1,4-dioxane in finished cleaning products. The protocol has been shown to be robust in terms of specific instrument and the fine details of the experimental method. The original method featured an Agilent 7679A HS unit coupled with an Agilent 6890 GC with a Zebron, ZB-624, 60 m × 0.25 mm × 1.4 μm column and 1.1 mL/min helium carrier gas flow. Samples were analyzed with a split ratio of 3:1 in an Agilent 5973 single quadrupole MSD with electron ionization (EI) using selective ion monitoring. Both total ion chromatograms (TIC) and specific extracted ion chromatograms (EIC) are shown. The full instrumental experimental conditions from the protocol are included as Appendix B.

Calibration standards of 1,4-Dioxane in DI water were prepared at concentrations ranging from 50 ppb to 2 ppm, and used in the same day as the preparation. Dimethyl sulfoxide (DMSO) has also been determined to be a suitable solvent for standard and sample preparation. A 1 mL aliquot of each standard solution was transferred to a 10 mL headspace vial, to which, 0.1 mL of a 5 ppm 1,4-dioxane-d8 internal standard solution was added prior to analysis.

Six preparations of each sample were made to determine the method precision. Accuracy and recovery were determined by making three preparations of each sample that were spiked with 0.02 ppm of 1,4-dioxane, driven by the LOQ of the method. After sample solutions were prepared, a 1 mL aliquot of each solution was transferred to a 10 mL headspace vial, to which, 0.1 mL of a 5 ppm 1,4-dioxane-d8 internal standard solution was added prior to analysis.

Under these experimental conditions, the 1,4-dioxane and 1,4-dioxane-d8 internal standard will coelute at retention time of 9.6 min. EIC were utilized to differentiate between the 1,4-dioxane and 1,4-dioxane-d8 present in sample and standard chromatograms. 1,4-Dioxane was quantified by integrating the peak area of the 88 u EIC peak, using 58 u as a qualifying peak. 1,4-Dioxane-d8 was quantified by integrating the peak area of the 96 u EIC peak. Figure 1 shows the results for a 1 ppm 1,4-dioxane standard with the TIC at the top, the 88 u EIC for the standard in the middle, and the 96 u EIC for the internal standard at the bottom. The standard and internal standard are easily identified and quantified using this method with no meaningful interferences from the water or introduced by the method. Calibration standards were analyzed in the middle and at the end of the analysis sequence.

Details are in the caption following the image
Total ion chromatogram (TIC, top), 88 u EIC (middle), and 96 u EIC (bottom) of a 1 ppm 1,4-dioxane standard with internal standard (IS)

A response factor (RF) for 1,4-dioxane was determined by dividing the 88 u peak area for standards and samples by the 96 u peak area of the internal standard. The calibration curve was created by plotting the concentration of 1,4-dioxane (ppm) against the RF for each standard solution. Figure 2 shows the resulting calibration curve from 50 ppb to 2 ppm. The calibration curve shows excellent linearity with an R2 of nearly 1. Analyzing the data generated for this calibration curve demonstrates that the method has a method detection limit (MDL) of <10 ppb and a limit of quantitation (LOQ) of <20 ppb.

Details are in the caption following the image
Calibration curve for 1,4-dioxane from 50 ppb to 2 ppm generated from the RF determined from the internal standards

Each of the five finished cleaning goods samples produced clean and readily quantifiable data for 1,4-dioxane. Figure 3 shows representative HS GC/MS 88 u EIC data for the five samples, along with a 1 ppm 1,4-dioxane standard. Figure 4 shows representative HS GC/MS data for the hand dish soap 1 30%. Both the 1,4-dioxane and the IS are easily identified and quantified in this sample.

Details are in the caption following the image
(From top to bottom) Representative 88 u EIC a 1 ppm 1,4-dioxane standard, a prepared Hand Dish 30% sample, a prepared Consumer Laundry 15% sample, a prepared Hand Dish 15% sample, and a prepared Consumer Laundry 50% sample, and a prepared I&I Laundry 50% sample
Details are in the caption following the image
Representative TIC (top), expanded EIC m/z 88 (middle), and expanded EIC m/z 96 (bottom) of a prepared Hand Dish 30% sample


Impact analysis analytical summary

To determine the concentration of 1,4-dioxane in the samples, Equation (1) was used.
Sample ppm = RF S B M × V W (1)
where RFS is the response factor of a specific sample, B is the Y-intercept of the calibration curve, M is the slope of calibration curve, V is the total volume of the prepared sample solution (10 mL), and W is the weight of sample in the prepared sample solution (~2 g).

Each sample showed significant 1,4-dioxane, and the concentration could be readily determined using this method. Table 1 shows the results of the quantitation of 1,4-dioxane for the five samples. Furthermore, the spike/recovery results demonstrate that the method is quite robust with results ranging from 97% to 102% recovery. The results are shown in Table 2.

TABLE 1. Concentrations of 1,4-dioxane determined in each of the five finished cleaning products
Sample Prep 1,4-Dioxane (ppm) Average (ppm) Standard Deviation %RSD
Hand dish soap 2 30% 1 3.20 3.2 0.21 6.67
2 2.89
3 3.23
4 3.25
5 3.18
6 3.56
Consumer laundry detergent 1 15% 1 0.94 0.93 0.01 0.85
2 0.94
3 0.94
4 0.92
5 0.92
6 0.94
Hand dish soap 1 15% 1 1.88 1.9 0.02 1.06
2 1.86
3 1.82
4 1.85
5 1.84
6 1.85
Consumer laundry 2 50% 1 2.77 2.8 0.04 1.44
2 2.78
3 2.73
4 2.68
5 2.79
6 2.77
I&I laundry 50% 1 0.93 0.93 0.01 0.75
2 0.93
3 0.93
4 0.92
5 0.93
6 0.94
TABLE 2. Spike/recovery data for the five samples using triplicate 0.02 ppm spikes.
Sample Prep % Recovery Average
Hand dish soap 2 30% 1 100 100
2 100
3 100
Consumer laundry 1 15% 1 97 97
2 98
3 98
Hand dish soap 1 15% 1 98 100
2 101
3 101
Consumer laundry 2 50% 1 101 102
2 102
3 103
I&I laundry 50% 1 100 99
2 99
3 99

Round robin analytical testing

To further test the utility of the method, a round-robin experiment was designed involving ten different labs. The method was documented as a step-wise standard operating procedure (SOP). The SOP allowed similar, but not identical analytical instrumentation. The SOP followed the experimental conditions described here, except that the finished cleaning samples were prepared in triplicate with duplicate analyses of each sample. Both the method and the five finished cleaning good samples were distributed to the participating labs. The results of the round-robin are shown in Table 3.

TABLE 3. Summary of the round-robin testing on the five finished cleaning products by testing lab
Lab (average ppm)/sample Hand dish soap 1 30% Consumer laundry detergent 1 15% Hand dish soap 2 15% Consumer laundry detergent 2 50% I&I detergent 50%
Lab 1 3.5 0.92 1.8 2.6 0.81
Lab 2 3.6 1.1 2.0 2.9 1.1
Lab 3 3.6 1.0 1.9 2.7 0.93
Lab 4 3.5 0.95 1.8 2.9 0.87
Lab 5 4.5 1.2 2.2 3.1 1.1
Lab 6 3.5 1.0 1.9 2.8 1.0
Lab 7 3.5 0.98 1.9 2.8 1.0
Lab 8 3.8 1.0 2.0 2.9 0.99
Lab 9 3.1 0.99 1.8 2.4 1.1
Lab 10 3.6 1.0 1.9 2.7 0.94
Average 3.6 1.0 1.9 2.8 1.0
Standard deviation 0.36 0.08 0.13 0.19 0.10
% RSD 9.8 7.9 6.7 7.0 10.4

The round-robin results across the ten labs are very consistent. Labs 1–3, and 10 followed the written SOP closely, and the agreement between their results demonstrate the strength and robustness of the method. Labs 1–3, and 10 used slightly different approaches to the headspace analysis including static headspace, gas tight syringe, and a pressure balanced approach. Labs 4–9 had some deviations to written SOP, ranging from modifications to the method to lack of spike/recovery tests, to lack of duplicate analysis.

Statistical analysis of results

The results for the round robin study showed that the method provided effective precision with replicate sample preparations and analyses having %RSD of <5%. The method displayed sufficient accuracy, with average spiked recoveries of the samples falling between 97% and 107%. Nine of the participating laboratories demonstrated excellent linearity with calibration curves having correlation coefficients R2 > 0.99, while one lab reported an R2 = 0.9895. Despite the alterations made to the provided method by some of the participating laboratories, the use of the deuterated internal standard and extracted ions for quantitation was proven to be effective to determine 1,4-dioxane in finished cleaning products. This indicates that this novel method is robust, both against the details of the protocol, and against different labs with different equipment and personnel. The limits of detection and quantitation for the method were not consistent across the testing laboratories. It appears that these results are dependent on some combination of the instrumentation used and the details of the sample preparation, but most labs could achieve an instrument limit of detection of <10 ppb with instruments in proper working condition.

Labs 1–3 provided a subset of data which was well structured for performing variance components analysis to partition the total observed variation into three quantifiable contributing sources: lab, preparation, and measurement error. This allows an assessment of the relative importance of these error sources in terms of their impact on total variation, as shown in Table 4, where df is degrees of freedom, a statistically focused measure of useful sample size. This overall analysis focuses on the lab testing results and provides highly useful general guidance for future method improvement efforts. It does not include any variance substructure differences from each of the five soaps/detergents studied in the round-robin. The process of dividing Total Variation into % Total is dependent on the use of the standard deviation squared, or variance, and does not total to 100% in standard deviation terms. The key learning is that lab to lab variation is the dominant source of the observed variation at 92.5% of total observed variation. Neither measurement error variation nor preparation variation within a given lab has substantial impact on the total variation.

TABLE 4. Overall variance components analysis (labs 1–3)
Source df Variance %Total SD
Total 85 0.0787 100% 0.281
Lab 10 0.0728 92.5% 0.270
Prep 30 0.0039 5.0% 0.063
Error 45 0.0020 2.5% 0.045

Variance component analysis was then performed at the level of each sample studied, as shown in Figure 5, where each datum is based on 1/5 the degrees of freedom (df) in Table 4. This graphic illustrates that for every sample studied, lab to lab variation remained the largest variance component. Further production variance reduction in 1,4-dioxane variation, if desired, needs to be focused on reducing lab to lab variation. Figure 5 additionally illustrates that measurement error variation and preparation variation has an increasing trend component related to 1,4-dioxane level.

Details are in the caption following the image
Variance components analysis by sample level (Labs 1-3)

Findings and recommendations

Calibration data were provided by Labs 1–5 (see Appendix C). It was studied from multiple statistical contexts. The analyses indicated that calibration variation is likely a meaningful component of the observed lab to lab variation. As noted previously, one Lab had an R2 less than 0.99, and this is impactful. A 24% reduction in lab-to-lab variation in the variance component analysis resulted if this lab's results were not included in the variance components analysis. Additionally, among those labs which provided calibration data:
  1. Some while having an R2 above the nominal target level, 0.99, were still weaker than desirable
  2. Consistently high or consistently low (across all samples) results were sometimes observed among the Labs providing calibration data
  3. Calibration data strongly showed systematically increased variation as the level of 1,4-dioxane increased
These observations indicate that improvements beyond the acceptable results reported here could be achieved by modifying future calibration protocols. Improvement can be achieved by performing one or more of the following:
  1. Improvement of calibration data collection protocols
  2. Collection of more calibration data than used in the round-robin
  3. Routinely checking the current calibration's validity
  4. Fitting calibration by weighted least squares (WLS) rather than ordinary least squares (OLS or LS) analysis

The absence of a WLS-based calibration with the observed increasing pattern in variation has the consequence of overfitting the high values and underfitting the low values. This impacts both estimation of the calibration model's intercept as well as detection limit estimation.


A novel HS GC/MS method to quantify 1,4-dioxane in finished cleaning products has been developed. The method makes use of a fully deuterated 1,4-dioxane-d internal standard. The method has been shown to produce excellent accuracy, precision, and spike/recovery. The initial method was shown to produce a very linear calibration curve between 50 ppb and 2 ppm, and to generate a LOQ of 13 ppb. To further test the method, a written SOP was developed and ten labs were recruited to test the method through a round-robin exercise. The round-robin demonstrated that the method is robust when followed closely, and even preforms well when some significant modifications to the method were employed. Statistical analysis of the data show that improvements can be made in the calibration process to make the application of this method even more useful when applied across many different labs (Ketkar & Bzik, 2000). This SOP can now be trusted to produce effective quantitation of 1,4-dioxane in consumer goods. Additional testing is planned to extend the method to other consumer goods to further test the presence of 1,4-dioxane in the goods intended to be used at home by consumers.


Brian Palumbo, Deena Conrad-Vlasak and Kathleen Stanton conceived and designed the study. Kathleen Stanton recruited test laboratories. Brian Palumbo and Deena Conrad-Vlasak oversaw the analysis of the data. All authors contributed to and approved the final draft of the manuscript.


The authors would like to acknowledge Impact Analytical for the development of this novel method. We would also like to acknowledge the support of the American Cleaning Institute for supporting this project and the ten labs that participated in the round-robin exercise. We would also like to acknowledge Scott D. Hanton, PhD at Hanton Consulting LLC, and Tom Bzik for their support in the creation of the manuscript.


    No human or animal subjects were used in this research.


    A.1 Relative concentrations in the finished cleaning material models

    TABLE A1. Concentrations of the components of the laundry formulations
    Component Consumer laundry detergent 1 (15% surfactant) (%) Consumer laundry detergent 2 (50% surfactant) (%) Industrial and institutional (I&I) laundry detergent (50% surfactant) (%)
    Linear alkylbenzene sulfonate (LAS) acid 9.00 30.00 17.40
    Sodium lauryl ether sulfate (SLES)—3 EO (30%) 4.50 15.00
    Tomadol® 91-6 3.00 10.00 20.00
    Tomadol® 91-2.5 2.00 6.66 13.30
    50% caustic 1.50 5.00 3.00
    Sodium benzisothiazolin 20% in dipropylene glycol-water (BIT Proxel GXL) 0.10 0.10 0.10
    Water 79.90 33.24 46.20
    Total 100.00 100.00 100.00
    pH 8–8.5 8–8.5 8–8.5
    TABLE A2. Concentrations of the components of the hand wash formulations
    Component Hand dish soap 2 (15% surfactant) (%) Hand dish soap 1 (30% surfactant) (%)
    LAS acid 10.00 20.00
    SLES—3 EO (30%) 16.00 32.00
    Lauryl dimethyl amine oxide (30%) 2.00 4.00
    Sodium xylene sulfonate (SXS) (40%) 2.00 5.00
    50% caustic 2.50 5.00
    BIT Proxel GXL 0.05 0.05
    Water 67.45 33.95
    Total 100.00 100.00
    pH 6.5–7.5 6.5–7.5
    • Note: In all formulations, the BIT Proxel GXL is used as a preservative to prevent microbial spoilage.


    B.1 1,4-Dioxane HS GC/MS experimental conditions

    GC system Agilent 6890
    Capillary column Zebron, ZB-624, 60 m × 0.25 mm I.D. × 1.4-μm film
    Inlet 250°C
    Carrier gas (He) flow 1.1 mL/min, constant flow
    Injection 1 mL loop, split injection, split ratio 3:1
    Temperature program 50°C (2-min hold) increased at 15°C/min to 250°C (7-min hold)
    Transfer line 270°C
    5973 MSD Electron ionization (EI)
    Single quadrupole MS
    Selective ion monitoring (SIM) mode
    Dwell time of 10s
    Dioxane quantitation using 88 u
    Dioxane qualification using 58 u
    Dioxane d8 standard quantitation using 96 u
    Solvent delay None
    Headspace conditions Agilent 7697A
    Zone temps Oven 90°C
    Loop 140°C
    Transfer line 150°C
    Event times GC cycle time 30 min
    Vial eq. time 12 min
    Injection time 1 min
    Loop fill mode: Default
    Loop eq. time default (0.05 min)
    Loop fill time default (0.1 min)
    Pressurize eq. time default (0.1 min)
    Vial fill pressure 15 psi


    C.1 | Individual calibration curves from labs participating in the round-robin testing