Detection of Toxic Contaminants in Alcohol-Based Hand Sanitizers Using Infrared Spectroscopy

Abstract

Recent fatalities due to methanol and 1-propanol toxicity in alcohol-based hand sanitizers has led the U.S. Food and Drug Administration (FDA) to ban 594 different hand sanitizer brands. The FDA also introduced methanol testing in alcohol-based hand sanitizers before allowing the products to enter the United States This requirement creates a need for inexpensive, rapid, and portable testing methods to measure methanol and 1-propanol concentrations in alcohol-based hand sanitizers. Here we study the performance of infrared (IR) spectroscopy for measuring methanol and 1-propanol concentrations in an alcohol-based hand sanitizer, and compare the performance of two portable spectrometers, Texas Instruments near-infrared (NIR) spectrometer (TI NIR) and NeoSpectra mid-infrared (MIR) spectrometer. The IR absorbance spectra were measured in transmission mode at different path lengths for 52 different hand sanitizer samples spiked with 0%–1% v/v concentrations of methanol and 1-propanol. A partial least-squares regression analysis shows ability to detect contaminant concentrations with a correlation coefficient of determination (r²) up to 0.99 and root mean square error of prediction as low as 0.34% v/v.

Graphical Abstract

This is a visual representation of the abstract.

Keywords

Hand sanitizers toxins methanol 1-propanol infrared spectroscopy partial least-squares regression PLSR

Introduction

The outbreak of the novel coronavirus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) brought a long-lasting coronavirus pandemic in 2019 (COVID-19) that affected every nation. Maintaining good hygiene standards can assist in preventing the spread of infections and counteract future pandemics.¹ Washing hands often with soap and water is essential for prevention. However, when this is not possible, the U.S. Centers for Disease Control and Prevention and the World Health Organization (WHO) recommend the use of alcohol-based hand sanitizers,² of which must contain at least 60% v/v of ethyl alcohol (ethanol).^3,4 The importance of hand hygiene in response to COVID-19 has resulted in unprecedented demand for hand sanitizers.

Due to the skyrocketing demand for raw materials and inadequate quality control standards for hand sanitizers, many commercial producers have been using prohibited chemicals as ingredients and ethanol substitutes. Some of these prohibited chemicals include methanol, 1-propanol, benzene, acetaldehyde, and acetal (1,1-diethoxyethane), which are genotoxic, carcinogenic, and/or neurotoxic when in direct contact with tissues.² These chemicals have found their way to the consumer market for making alcohol-based hand sanitizers. There are reports of more than 700 deaths caused by contaminants in hand sanitizers.^5–7 Among these contaminants, methanol, and 1-propanol are the most common due to their wide availability and low cost. In response, the FDA issued mandatory testing requirements for methanol and a warning for 1-propanol contamination in alcohol-based hand sanitizers in August 2020.⁸ This requirement came after the FDA detected toxic ingredients including methanol and/or 1-propanol in more than half of the imported hand sanitizers at dangerous levels between April and December 2020. This was the first time the FDA issued a countrywide import alert for any category of drug product. This demonstrates the need for the development of techniques and devices for rapid measurement of the methanol and 1-propanol contaminants in alcohol-based hand sanitizers.

The FDA regulations mandate that all lots of pharmaceutical alcohol be tested for contaminants prior to use in hand sanitizer manufacturing.⁹ However, the testing of manufactured hand sanitizers is necessary to verify their purity and ensure the safety of users. Government-recommended methods for contaminant detection in alcohol include gas chromatography flame ionization detection (GC-FID), specific gravity test, and ultraviolet absorption.¹⁰ A poster presentation explaining the implementation of GC-FID is available on the FDA website, which instructs the reader on the quantitation of methanol, ethanol, and isopropanol in gel hand sanitizers.¹¹ The FDA also developed a laboratory analytical procedure using gas chromatography-mass spectrometry (GC-MS) to detect methanol, benzene, acetaldehyde, and acetal in finished hand sanitizer products.¹² Additionally, recently reported techniques for the detection of toxic contaminants in hand sanitizers include GC-MS (Thermo Scientific ISQ 7000 single quadrupole mass spectrometer, coupled to a Thermo Scientific TRACE 1310 gas chromatograph) for acetaldehyde, methanol, acetone, isopropyl alcohol, 1-propanol, ethyl acetate, 2-methyl-1-propanol, benzene, 1-butanol, 1,1-diethoxy ethane, 3-methyl-1-butanol, and 1-pentanol¹³; GC-FID (Agilent 8860) for methanol and n-propanol¹⁴; and Fourier transform infrared spectroscopy (FT-IR; Thermo Scientific Nicolet iS50 FT-IR spectrometer) for 1-propanol and 2-propanol¹⁵ GC-MS and GC-FID are large instruments, require complex sample preparation and do not provide rapid results.^16,17 In addition, changes to certain properties or characteristics such as the stationary phase, changes in temperature, experimental conditions, or preparation of standards or chemicals, may cause variations in GC-MS analysis.¹⁸ These problems hinder GC-MS and GC-FID from being used to rapidly detect toxic contaminants in alcohol-based hand sanitizer in the field. The commercial FT-IR spectrometer from Thermo Scientific Nicolet iS50 has been advertised to detect contaminants (1-propanol and 2-propanol) in hand sanitizers.¹⁹ Although FT-IR spectroscopic instruments can provide rapid and simple testing, they are relatively bulky and expensive. A recent study demonstrated that a small and portable GC can detect 0.01%–100% v/v methanol in alcohol-based hand sanitizers.⁷ This device features a Tenax TA column and a chemo-resistive Pd-doped SnO₂ detector. Although this study is promising, scalability remains a drawback and further research is needed to explore potential to detect other contaminants.

Currently, available portable and cost-effective infrared (IR) spectroscopic systems may offer a better alternative for rapid measurement of methanol and 1-propanol in alcohol-based hand sanitizers in the field. Near-IR (NIR) spectroscopy involves the measurement of higher harmonics and overtones in the wavelength range of 900–2500 nm.¹⁹ It has been used for the quantitative determination of a great variety of chemicals and materials.²⁰ NIR offers fast results, requires minimal to nonexistent sample preparation, is nondestructive, is easy to adapt to production lines, and is easy to operate in the field. A key advantage of NIR over FT-IR is that NIR instrumentation and components are inexpensive, small, and portable.^21–23 Moreover, existing off-the-shelf NIR instruments can be readily applied for this purpose without any hardware modifications.

For this paper, we compared two portable IR spectroscopy systems (TI NIR and NeoSpectra mid-infrared (MIR)) for the quantitative detection of methanol and 1-propanol in alcohol-based hand sanitizers. Methanol and 1-propanol were mixed with the hand sanitizer at concentrations ranging from 0.11 to 0.99% v/v. Partial least squares regression (PLSR), a common modeling technique for spectral analysis,²⁴ was used in this work for estimating the concentrations of the two compounds. The PLSR model was built using spectra of different concentrations as the dependent variables and IR absorbance at different wavelengths as independent variables. Values of r² and the root mean square error of prediction (RMSEP) are two critical result metrics in PLSR analysis.²⁵ The value of r² quantifies the proportion of variance in the response variable that can be explained by the predictor variables, indicating the model's goodness of fit to the training data. RMSEP is another essential metric because it measures the accuracy of the model's predictions on new, unseen data, providing insights into its generalization capability. A high r² value suggests a well-fitted model that captures a significant portion of the variability in the response variable. A low RMSEP signifies a PLSR model with accurate predictions on unseen data, indicating its reliability in real-world applications. In this study, we will use these two metrics to demonstrate the PLSR results. The r² and RMSEP values for methanol were found to be 0.97 and 0.41% v/v, respectively, for the TI NIR (1 mm pathlength) spectrometer, and 0.99 and 0.34% v/v, respectively, for the NeoSpectra MIR (100 µm pathlength) spectrometer. The r² and RMSEP values for 1-propanol were found to be 0.98 and 0.51% v/v, respectively, for the TI NIR (1 mm pathlength) spectrometer, and 0.99 and 0.40% v/v, respectively, for the NeoSpectra MIR (100 µm pathlength) spectrometer. We also performed interval partial least square (iPLS) regression analysis to determine the optimal narrow spectral ranges for both TI NIR and NeoSpectra MIR spectrometers for detecting methanol and 1-propanol in alcohol-based hand sanitizers. Our work demonstrates the applicability of portable IR spectroscopic devices as a rapid and simple tool for determining toxic contaminants in alcohol-based hand sanitizers.

Experimental

Materials and Methods

Absolute anhydrous grade ethyl alcohol (CAS: 64-17-5) was acquired from Pharmco. ACS reagent grade glycerol (CAS:56-81-5), methanol (CAS: 67-56-1), and 1-propanol (CAS:71-23-8) were acquired from Sigma Aldrich.

A stock hand sanitizer solution was prepared following FDA guidelines using 60% v/v ethanol and 40% v/v glycerol.⁸ Methanol and 1-propanol at different concentrations were added to the stock hand sanitizer solution. A total of 52 different solutions were prepared with varying contaminant concentrations. Table S1 and Figure S1 (Supplemental Material) show the different concentrations of methanol and 1-propanol used in each measured hand sanitizer solution. These values have been randomly generated for concentration ranges between 0.11 and 1.00% v/v.

Infrared Spectroscopy

The IR absorbance spectra were collected using TI NIR and NeoSpectra MIR spectrometers. TI NIR was used over a spectral range of 1350–2450 nm with 1 nm resolution and 28 s (s) of acquisition time per spectrum. Five spectra were averaged for each solution. The IR transmission spectra were collected using 100, 150, 250, 500, and 1 mm pathlength cuvettes (model: 100-1-20, material code: OS, wavelength range: 320–2500 nm; Hellma Analytics). As cuvettes with smaller pathlengths than 1 mm were unable to be filled with the viscous hand sanitizer; only 1 mm pathlength cuvettes were used in the spectral collection for the hand sanitizer samples. All absorbance spectra were acquired at room temperature with the empty cuvette as a reference. The schematic illustration of the experimental setup of TI NIR is shown in Figure 1a, and the optical photograph of the TI NIR spectrometer is provided in Figure 1b.

Figure 1.

(a) Schematic representation of the measurement setup using the Texas Instruments near-infrared spectrometer (TI NIR) and (b) optical photograph of the TI NIR.²⁶ (c) Schematic representation of the measurement setup using the NeoSpectra MIR platform spectrometer and (d) optical photograph of NeoSpectra MIR.

IR absorbance spectra were also collected using NeoSpectra MIR spectrometer with a full spectral range of 1400–4800 nm, average resolution of 10.81 nm, 30 s acquisition time per spectra and averaging five spectra for each measurement. The spectra were collected using four different pathlengths (100, 150, 250, and 500 µm). The schematic illustration of the experimental setup of NeoSpectra MIR is shown in Figure 1c, and the optical photograph of the NeoSpectra MIR spectrometer is provided in Figure 1d. A comparison of the features and specifications between TI NIR and NeoSpectra MIR is given in Table 1.

Table 1.

Comparison of TI NIR and NeoSpectra MIR spectrometers.

Spectrometer features	Texas instruments (TI NIR)	NeoSpectra MIR
Manufacturer	Texas Instruments	NeoSpectra
Handheld/portable	Yes	Yes
Plug and play	Yes	Yes
Sample holder	Cuvette	Transmission flow cell
Mode of IR operation	Transmittance	Transmittance
Comparative sample size	Large	Small
Pathlength	<10 mm, pathlength is cuvette dependent	100, 150, 250, 500 µm achieved using different spacer thickness
Wavelength range	1350–2450 nm	1400–4800 nm
Resolution	1 nm	10.81 nm
No. of data point/datum (no. of points in each spectrum)	1100	313

The NeoSpectra MIR spectrometer provided flexibility for adjusting spacer size, allowing for variable pathlengths. In contrast, TI NIR employed cuvettes with predefined pathlengths. The tube-like inlet design of NeoSpectra MIR facilitated the seamless syringe insertion of highly viscous hand sanitizer samples. Conversely, introducing these viscous solutions into cuvettes proved difficult due to trapped air bubbles within the smaller cuvettes.

Data Processing

This study employs two data processing methods, PLSR and iPLSs, to model relationships between the spectral data obtained from the analytical instruments and the corresponding concentrations of contaminants in the samples. PLSR is a regression technique that seeks to identify combinations of predictor variables (spectra) explaining maximum variance in response variables (contaminant concentrations). While iPLS, an extension of PLSR, iteratively selects the most relevant variables, enhancing prediction accuracy by focusing on informative features and facilitating detection of contaminants within a narrower wavelength range. This approach has previously been used to identify optimal wavelengths for detecting chemical constituents.²⁷ This operation required 6–7 h of computation time using a local computer to complete. Prior to PLSR and iPLS, no preprocessing was conducted on the spectral data. Processing of spectra was done using Python 3.7 Jupyter software on a computer with Intel eleventh-generation Core i9 11900KF (8-core, 16 MB cache, 3.5 GHz to 5.3 GHz w/thermal velocity boost). The random-access memory on the computer was 32 GB and graphics processing unit was NVIDIA GeForce RTX 3080 10GB GDDR6X LHR.

Results and Discussion

Spectral Features and Interpretation

The IR spectra of 1-propanol, ethanol, glycerol, and methanol as acquired using the NeoSpectra MIR with 100, 150, 250, and 500 µm pathlengths shown in Figure S2 (Supplemental Material). IR peaks for different functional groups are summarized in Table II, and the spectra of 1-propanol, ethanol, glycerol, and methanol acquired using the TI NIR and NeoSpectra MIR are shown in Figure 2. The NIR spectra of glycerol in Figure 2 and Figure S2 (Supplemental Material) show two OH peaks at 1500 and 2080 nm. The 1500 nm peak is due to the C–O–H bending vibration, while the 2080 nm peak is due to the presence of three –OH groups in glycerol. The CH and CH2 peaks are at 1710, 2270, and 2320 nm, respectively. Methanol has a CH overtone at 1690 nm, an OH combination band at 2080 nm, and CH₃ peaks at 2275 and 2340 nm. Ethanol has a similar hump around 2350 nm, along with peaks at 1690, 1740, 2080, 2285, and 2305 nm; 1-propanol has similar peaks to ethanol, with a CH peak at 2300 nm, a CH₂ peak at 2440 nm, and a CH₃ peak at 2390 nm. These spectra are consistent with the spectra reported in the literature.

Figure 2.

(a) TI NIR absorbance spectra of hand-sanitizer stock solution, glycerol, methanol, ethanol, and 1-propanol using a 1 mm pathlength cuvette. Glycerol spectra were taken from a wet film and created by spreading a thin glycerol layer on the outer surface of the cuvette using the motion of a blade positioned 1 mm from the cuvette surface.²⁸ (b) NeoSpectra MIR absorbance spectra of the stock solution, methanol, ethanol, and 1-propanol using 250 µm pathlength (IR spectra of glycerol was not recorded as the high viscosity of glycerol prevents its introduction into the narrow flow cell).

Table II.

Peaks and responsible functional groups of glycerol, methanol, ethanol, and 1-propanol.

Chemical compound	Peaks	Functional group
Glycerol	2080 nm	–OH
Glycerol	1500 nm	C–O–H
Glycerol	1710, 2270, and 2320 nm	CH, CH₂
Methanol	1690 nm	CH
Methanol	2080 nm	OH
Methanol	2275–2340 nm	CH₃
Ethanol	2976–3086 nm	OH
Ethanol	2350 nm	CH
1-Propanol	2857–3125 nm	OH
1-Propanol	2300 nm	CH
1-Propanol	2440 nm	CH₂
1-Propanol	2390 nm	CH₃

Analysis and Prediction: Full Spectra

The flowchart of PLSR model development for the estimation of methanol and 1-propanol concentrations in hand sanitizer solutions is shown in Figure 3. The spectral data (n = 52) was used as two sets for NeoSpectra NIR: One set as it is, and the second with removed data points for wavelength region with high absorbance (>1.00). This high absorbance means that very little light reaches the detector, thus mostly noise. For TI NIR spectra, only one set of spectra was used as no region of high absorbance was found in the spectra. For each of these three sets of spectra from two spectrometer, 60% of these spectra (n = 31) were randomly selected to make up the training set. The remaining 40% (n = 21) were used as the test set. The model was used to generate estimated concentrations for the test data set and calculate RMSEP. The two metrics – r² value of the fitted model and RMSEP, demonstrate the effectiveness of the PLSR model for the determination of the concentration of contaminants.

Figure 3.

A flowchart of PLSR analysis used in this study.

The PLS regression equation to develop the model is as follows in Eq. 1:

y = X B + e

(1)

where the response variable y (I × M) is a vector of contaminant concentration, X (I × K) is the spectral data for the samples, b (K × 1) is a vector of regression coefficients, B (K × M) is a matrix of regression coefficients for the M responses, and e (I × M) is a vector of residuals, which contains the differences between the predicted concentrations and the actual concentrations.

To measure out-of-sample performance of the model, the test data set was tested with the PLSR model and RMSEP was calculated as follows in Eq. 2:

RMSEP = \sqrt{\frac{\sum_{i = 1}^{n} {(y_{i} - \hat{y_{i}})}^{2}}{n}}

(2)

where y_i and

{\hat{y}}_{i}

represent the measured and the estimated concentration for sample i, respectively, and n is the number of samples in the data sets (n = 52).

The PLSR results of spectra acquired from TI NIR using the 1 mm pathlength are shown in Figure 4. Figure 4a shows the values of the estimated methanol concentration (% v/v) versus actual methanol concentration (% v/v), with r² value of 0.97 and RMSEP of 0.41% v/v. Figure 4b shows the values of the estimated 1-propanol concentration (% v/v) versus actual 1-propanol concentration (% v/v), with r² value of 0.98 and RMSEP of 0.51% v/v. These PLSR results show that TI NIR can estimate methanol and 1-propanol concentration in ethanol–glycerol-based hand sanitizers with high (>0.9⁶) r² value.

Figure 4.

Correlation of (a) actual methanol concentration versus estimated methanol concentration and (b) actual 1-propanol concentration versus estimated 1-propanol concentration after removal of high absorbance section of TI NIR spectrum of hand sanitizers with different concentrations of methanol and 1-propanol.

Figure S2 (Supplemental Material) shows the IR absorbance spectra of stock solution, 1-propanol, ethanol, and methanol, measured at different pathlengths (100, 150, 250, and 500 µm) using the NeoSpectra MIR and the TI NIR. There is a region of high absorbance (above 1.0) in the wavelength range between 2750 and 3700 nm for all four pathlengths, similar to the spectra in Figure 2b. This high absorbance is due to the strong IR light interaction with the molecules, resulting in a very small amount of light reaching the detector, thus displaying noise in this wavelength range. Shorter pathlength would enable the proper measurement of the spectral peaks in this region. For methanol, using PLSR model across the entire wavelength (including this region of high absorbance), r² and RMSEP values are found as 0.80 and 0.80% v/v for 100 µm pathlength, 0.78 and 0.84% v/v for 150 µm pathlength, 0.90 and 0.59% v/v for 250 µm pathlength, 0.87 and 0.65% v/v for 500 µm pathlength, respectively. Using the same PLSR model for 1-propanol, r² and RMSEP values are found as 0.84 and 1.07% v/v for 100 µm pathlength, 0.91 and 0.80% v/v for 150 µm pathlength, 0.90 and 0.87% v/v for 250 µm pathlength, 0.88 and 0.93% v/v for 500 µm pathlength. These results are shown in Figure S3 as the correlation between actual and estimated concentration of methanol and 1-propanol for the four pathlengths.

Another PLSR model excluding the wavelength range of 2750–3700 nm to avoid the high absorbance region was conducted. Those results are shown in Figure 5 as estimated methanol concentration versus actual methanol concentration and estimated 1-propanol concentration (% v/v) versus actual 1-propanol concentration (% v/v) for pathlengths of 100, 150, 250, and 500 µm. For methanol, the corresponding r² and RMSEP values have been found as 0.99 and 0.34% v/v for 100 µm pathlength, 0.97 and 0.46% v/v for 150 µm pathlength, 0.97 and 0.51% v/v for 250 µm pathlength, and 0.97 and 0.68% v/v for 500 µm pathlength, respectively. For 1-propanol, using the same PLSR model, r² and RMSEP values have been found as 0.99 and 0.40% v/v for the 100 µm pathlength, 0.97 and 0.60% v/v for the 150 µm pathlength, 0.98 and 0.71% v/v for the 250 µm pathlength, and 0.97 and 0.79% v/v for the 500 µm pathlength, respectively. Thus, omitting wavelength regions of high absorbance can help estimate methanol and 1-propanol concentration more accurately.

Figure 5.

Correlation between actual and estimated spectra and the effect of pathlength–actual methanol and 1-propanol concentrations versus estimated concentrations for (a,b) 100 µm pathlength, (c,d) 150 µm pathlength, (e,f) 250 µm pathlength, and (g,h) 500 µm pathlength. Spectra were collected using the NeoSpectra MIR. Spectra ranges with regions of high absorbance as indicated by a detector signal with absorbance greater than 1.0 were removed.

Comparing the two spectrometers, the PLSR model showed better performance for estimating methanol concentration more accurately using spectra acquired using the TI NIR, while for 1-propanol, spectra acquired using NeoSpectra MIR provided better results. However, the differences in performance are not large.

Analysis and Prediction: iPLS Variable Selection

Measuring spectra over a narrow wavelength range rather than a wide one could potentially lower instrument costs and reduce the difficulty of manufacturing. This is both because components for a narrower wavelength can be less expensive and because a narrower wavelength needs cheaper optical components and detectors. iPLS was used to find a narrow wavelength range that allows one to determine concentrations of methanol and 1-propanol. The iPLS study revealed the optimal narrow wavelength ranges for the detection of methanol and 1-propanal concentrations to be 1703.42–1858.46 nm and 1679.00–1767.97 nm, respectively. Each of these ranges contains the band of wavelengths for vibration of CH and OH,²⁹ and it is expected that the most significant regions for methanol and 1-propanol concentration determination rely on these CH and OH vibrations. For the TI NIR, the r² values for methanol and 1-propanol have been calculated as 0.98 and 0.99%, respectively, in these optimal narrow wavelength ranges. RMSEP was found to be 0.42% v/v and 0.32% v/v, respectively. These RMSEP values show that the spread of the estimated versus actual concentrations is very small in this optimal narrow wavelength range. The same PLSR model was used for analyzing spectra acquired by NeoSpectra MIR for different pathlengths show the optimal narrow wavelength ranges for methanol and 1-proponal detection are 3600–4200 nm and 2400–2500 nm, respectively, for the 100 µm pathlength. C–H and O–H stretching, as well as C–O and CH₃ rocking, bending, and stretching vibrations fall within these wavelength ranges.²⁹ The r² and RMSEP values have been found as 0.90 and 0.43% v/v and 0.90 and 0.63% v/v, respectively, for methanol and 1-propanol concentrations estimation for a 100 µm pathlength. The optimal narrow wavelength range for the 150 µm pathlength has been found as 2400–2500 nm (r² = 0.96 and RMSEP = 0.35% v/v) and 2245–2325 nm (r² = 0.95 and RMSEP = 0.61% v/v) for methanol and 1-propanol, respectively. Notably, the wavelength span of 2245–2325 nm encompasses the CH₃ scissoring and C–O scissoring vibrations, while the range of 2400–2500 nm includes CH₂ vibrations. For 250 µm, optimal narrow wavelength bands for methanol and 1-propanol have been found to be 2390–2500 nm (r² = 0.93 and RMSEP = 0.48% v/v) and 2247–2353 nm (r² = 0.87 and RMSEP = 0.68% v/v), respectively. Similarly, for the 500 µm pathlength the optimal narrow wavelength ranges for methanol and 1-propanol have been found as 2100–2700 nm (r² = 0.86 and RMSEP = 0.50% v/v) and 2398–2426 nm (r² = 0.94 and RMSEP = 0.64% v/v). In the wide wavelength range of 2100–2700 nm, C–O, CH₃, and O–H vibrations are present. However, for the wavelength range of 2398–2426 nm, only the C–O and CH₃ stretching vibrations exist. Changes in the r² value with changes in pathlength occur due to a combination of different absorbances for different functional groups in different wavelength ranges. These optimal narrow wavelength ranges demonstrate that 1-propanol concentration determination is highly dependent on C–O and CH₃ stretching while methanol concentration determination is dependent on C–O, CH₃, and O–H vibrations. The above-mentioned results of regression analysis are summarized below in Table III. Detailed results of regression analysis are given in Table S2 (Supplemental Material).

Table III.

Summary of regression analysis, method, chemical compound, coefficient of determination (r²), and RMSEP (% v/v).

Method	Chemical compound	r ²	RMSEP (% v/v)
1 mm pathlength, TI NIR spectra (1350–2450 nm)	Methanol	0.97	0.41
1 mm pathlength, TI NIR spectra (1350–2450 nm)	1-propanol	0.98	0.51
500 µm pathlength, NeoSpectra MIR spectra (1400–4800 nm), omitting high-absorbance section	Methanol	0.97	0.68
500 µm pathlength, NeoSpectra MIR spectra (1400–4800 nm), omitting high-absorbance section	1-Propanol	0.96	0.79
1 mm pathlength, optimal narrow detection range (1703.42–1858.46 nm), TI NIR	Methanol	0.98	0.42
1 mm pathlength, optimal narrow detection range (1679.07–1767.97 nm), TI NIR	1-Propanol	0.99	0.32
500 µm pathlength, optimal narrow detection range (2100–2700 nm), 500 µm pathlength, NeoSpectra MIR spectra	Methanol	0.86	0.50
500 µm pathlength, optimal narrow detection range (2398–2426 nm), 500 µm pathlength, NeoSpectra MIR spectra	1-Propanol	0.94	0.64

Conclusion

IR spectroscopy requires little to no preparation for sampling. In this study, we applied PLSR analysis to the IR spectra of alcohol-based hand sanitizer solutions doped with methanol and 1-propanol. The IR spectra were collected using two portable spectrometers, the TI NIR and NeoSpectra MIR, to show the effectiveness of using IR spectroscopy to detect contaminants in hand sanitizer. Without removal of the high absorbance section of the NIR spectra in the range of 1350–2450 nm using the TI NIR, methanol and 1-propanol PLSR showed r² and RMSEP values of 0.94 and 0.65%, and 0.73 and 1.16%, respectively. With the removal of the high-absorbance section of IR spectra acquired using the NeoSpectra MIR in the spectral range of 1400 to 4800 nm, the r² and RMSEP values were found to be 0.99 and 0.34% and 0.99 and 0.40% for methanol and 1-propanol, respectively, for a 100 µm pathlength. It is also observed that with an increase in pathlength, the RMSEP increases, and r² value decreases slowly for NeoSpectra MIR. Also, it has been shown that methanol and 1-propanol concentration in hand sanitizer can be estimated by using a smaller wavelength range; these wavelength ranges are 1703.42–1858.46 nm for methanol and 1679.07–1767.97 nm for 1-propanol using the TI NIR spectrometer. For NeoSpectra MIR, these wavelength ranges are 3600–4200 nm for methanol for a 100 µm pathlength, 2400–2500 nm for 1-propanol for a 100 µm pathlength, 2245–2325 nm for methanol for a 150 µm pathlength, 2390–2500 nm for 1-propanol for a 150 µm pathlength, 2247–2353 nm for methanol for a 250 µm pathlength, 2379–2409 nm for a 1-propanol for 250 µm pathlength, 2100–2700 nm for methanol for a 500 µm pathlength, and 2398–2426 nm for 1-propanol for a 500 µm pathlength. Thus, a dedicated spectrometer to detect contaminants in hand sanitizer can be built focusing on these wavelength ranges. Overall, we have demonstrated that both the TI NIR and NeoSpectra MIR can be used for methanol and 1-propanol detection in alcohol-based hand sanitizers. Further optimization with respect to optical pathlength, preprocessing of data, and analysis of the spectra should be employed for developing practical solution for measuring contaminants in alcohol-based hand sanitizers using IR spectroscopy.

Supplemental Material

sj-docx-1-app-10.1177_27551857231204630 - Supplemental material for Detection of Toxic Contaminants in Alcohol-Based Hand Sanitizers Using Infrared Spectroscopy

Supplemental material, sj-docx-1-app-10.1177_27551857231204630 for Detection of Toxic Contaminants in Alcohol-Based Hand Sanitizers Using Infrared Spectroscopy by Aminur Rashid Chowdhury, Umar Burney, David King, Tse-Ang Lee, Dan Hutter and Tanya Hutter in Applied Spectroscopy Practica

Footnotes

Acknowledgments

the authors thank Erik R. Deutsch for providing the NeoSpectra MIR spectrometer for testing.

Author Contributions

Conceptualization was handled by T.H. Methodology was designed by T.H., A.R.C., D.H. Investigation was performed by A.R.C., U.B., T.-A.L. Visualization was done by A.R.C. and U.B. Supervision was done by T.H. Writing (original draft) was done by A.R.C.. Writing (review and editing) was done by T. H., U.B., D.H., D.K., and A.R.C..

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

All supplemental material mentioned in the text is available in the online version of the journal.

References

Cavanagh

Wambier

C.G.

. “Rational Hand Hygiene During the Coronavirus 2019 (COVID-19) Pandemic”. J. Am. Acad. Dermatol. 2020. 82(6): E211. 10.1016/j.jaad.2020.03.090

U.S. Food and Drug Administration Center for Drug Evaluation and Research. “Temporary Policy for Preparation of Certain Alcohol-Based Hand Sanitizer Products During the Public Health Emergency (COVID-19). Guidance for Industry”. http://resource.nlm.nih.gov/9918231203506676 [accessed September 14 2023].

Hilburn

Hammond

B.S.

Fendler

E.J.

Groziak

P.A.

. “Use of Alcohol Hand Sanitizer as an Infection Control Strategy in an Acute Care Facility”. Am. J. Infect. Control. 2003. 31(2): 109–116. 10.1067/mic.2003.15

U.S. Food and Drug Administration Center for Drug Evaluation and Research. “Policy for Temporary Compounding of Certain Alcohol-Based Hand Sanitizer Products During the Public Health Emergency”. http://resource.nlm.nih.gov/9918227353906676 [accessed September 14 2023].

Overbeek

D.L.

Watson

C.J.

Castañeda

N.R.

Ganetsky

. “A Geographically Distinct Case of Fatal Methanol Toxicity from Ingestion of a Contaminated Hand Sanitizer Product During the COVID-19 Pandemic”. J. Med. Toxicol. 2021. 17(2): 218–221. 10.1007/s13181-021-00824-9

Yip

Bixler

Brooks

D.E.

Clarke

K.R.

, et al. “Serious Adverse Health Events, Including Death, Associated with Ingesting Alcohol-Based Hand Sanitizers Containing Methanol: Arizona and New Mexico, May–June 2020”. Morb. Mortal. Wkly. Rep. 2020. 69(32): 1070–1073. 10.15585/mmwr.mm6932e1

Güntner

A.T.

Magro

Van Den Broek

Pratsinis

S.E.

. “Detecting Methanol in Hand Sanitizers”. iScience. 2021. 24(2): 102050. 10.1016/j.isci.2021.102050

U.S. Food and Drug Administration. “FDA Updates on Hand Sanitizers Consumers Should Not Use”. https://www.fda.gov/drugs/drug-safety-and-availability/fda-updates-hand-sanitizers-consumers-should-not-use [accessed September 14 2023].

U.S. Food and Drug Administration. “Policy for Testing of Alcohol (Ethanol) and Isopropyl Alcohol for Methanol, Including During the Public Health Emergency (COVID-19)”. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/policy-testing-alcohol-ethanol-and-isopropyl-alcohol-methanol-including-during-public-health [accessed September 14 2023]

10.

U.S. Pharmacopeia. “Benzyl Alcohol”. https://www.usp.org/harmonization-standards/pdg/excipients/alcohol [accessed September 14 2023].

11.

Staake

M.D.

Goodson

F.S.

Howe

Hudson-Davis

M.L.

, et al. “Quantitation of Methanol, Ethanol and Isopropanol in Gel Hand Sanitizer Products by GC-FID”. Poster. CARTS Project No. IR01828. https://www.fda.gov/media/148871/download [accessed September 14 2023].

12.

U.S. Food and Drug Administration. “Direct Injection Gas Chromatography Mass Spectrometry (GC-MS) Method for the Detection of Listed Impurities in Hand Sanitizers”. https://www.fda.gov/media/141501/download [accessed September 14 2023].

13.

ThermoFisher Scientific. “Rapid GC-MS Assessment of Hand Sanitizers According to the U.S. FDA Method”. https://appslab.thermofisher.com/app/4505/rapid-gcms-assessment-hand-sanitizers-according-us-fda-method [accessed September 14 2023].

14.

Zhang

. “Hand Sanitizer Analysis Using the Agilent 8860 GC Configured with a Flame Ionization Detector”. https://www.agilent.com/cs/library/applications/application-hand-sanitizer-analysis-8860-fid-5994-2089en-agilent.pdf [accessed September 14 2023].

15.

ThermoFisher Scientific. “How to Use FT-IR to Ensure the Safety and Efficacy of Alcohol-Based Hand Sanitizers: From Initial Research to Production”. https://assets.thermofisher.cn/TFS-Assets/MSD/Application-Notes/EB53441-ftir-safety-efficacy-alcohol-based-hand-sanitizer.pdf [accessed September 14 2023].

16.

Tiscione

N.B.

Alford

Yeatman

D.T.

Shan

. “Ethanol Analysis by Headspace Gas Chromatography with Simultaneous Flame-Ionization and Mass Spectrometry Detection”. J. Anal. Toxicol. 2011. 35(7): 501–511. 10.1093/anatox/35.7.501

17.

J.K.

Tay

F.H.

Wray

P.S.

Mohd Saberi

S.S.

, et al. “Inexpensive Portable Infrared Device to Detect and Quantify Alcohols in Hand Sanitizers for Public Health and Safety”. Anal. Chem. 2021. 93(45): 15015–15023. 10.1021/acs.analchem.1c02652

18.

Amigo

J.M.

Skov

Bro

Coello

Maspoch

. “Solving GC-MS Problems with PARAFAC2”. TrAC, Trends Anal. Chem. 2008. 27(8): 714–725. 10.1016/j.trac.2008.05.011

19.

Ruru

Yingqing

Yan

Qidong

Lei

. “Measuring Pure Water Absorption Coefficient in the Near-Infrared Spectrum (900–2500 nm)”. Yaogan Xuebao [J. Remote Sens.]. 2012. 16(1): 192–206. 10.11834/jrs.20121188

20.

Ozaki

Genkawa

Futami

. “Near-Infrared Spectroscopy”. In: Lindon

J.C.

Tranter

G.E.

Koppenaal

D.W.

, editors. Encyclopedia of Spectroscopy and Spectrometry. Oxford: Academic Press, 2017. Pp. 40–49. 10.1016/B978-0-12-409547-2.12164-X

21.

Fernandes

H.L.

Pasquini

I.M.R.Jr. Rohwedder

J.J.R.

. “Simultaneous Determination of Methanol and Ethanol in Gasoline Using NIR Spectroscopy: Effect of Gasoline Composition”. Talanta. 2008. 75(3): 804–810. 10.1016/j.talanta.2007.12.025

22.

Pasquini

“Near Infrared Spectroscopy: Fundamentals, Practical Aspects, and Analytical Applications”. J. Braz. Chem. Soc. 2003. 14(2): 198–219. 10.1590/s0103-50532003000200006

23.

Lai

Zheng

Søndergaard

Josephsen

, et al. “Determination of Aluminium Content in Aluminium Hydroxide Formulation by FT-NIR Transmittance Spectroscopy”. Vaccine. 2007. 25(52): 8732–8740. 10.1016/j.vaccine.2007.10.026

24.

Wold

Sjöström

Eriksson

. “PLS-Regression: A Basic Tool of Chemometrics”. Chemom. Intell. Lab. Syst. 2001. 58(2): 109–130. 10.1016/s0169-7439(01)00155-1

25.

Kooistra

Wanders

Epema

G.F.

Leuven

R.S.E.W.

, et al. “The Potential of Field Spectroscopy for the Assessment of Sediment Properties in River Floodplains”. Anal. Chim. Acta. 2003. 484(2): 189–200. 10.1016/s0003-2670(03)00331-3

26.

Texas Instruments. “DLP NIRScan EVM Evaluation Module”. https://www.ti.com/tool/dlpnirscanevm#tech-docs [accessed Oct 4 2021].

27.

Borin

Poppi

R.J.

. “Application of Mid Infrared Spectroscopy and iPLS for the Quantification of Contaminants in Lubricating Oil”. Vib. Spectrosc. 2005. 37(1): 27–32. 10.1016/j.vibspec.2004.05.003

28.

Cherrington

Liang

. “Materials and Deposition Processes for Multifunctionality”. In: Goodship

Middleton

Cherrington

, editors. Design and Manufacture of Plastic Components for Multifunctionality: Structural Composites, Injection Molding, and 3D Printing. Oxford: William Andrew Publishing, 2016. Chap. 2, Pp. 19–51. 10.1016/b978-0-323-34061-8.00002-8

29.

Beć

K.B.

Futami

Wójcik

M.J.

Ozaki

. “A Spectroscopic and Theoretical Study in the Near-Infrared Region of Low Concentration Aliphatic Alcohols”. Phys. Chem. Chem. Phys. 2016. 18(19): 13666–13682. 10.1039/C6CP00924G

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.68 MB