1 Introduction

Since a noticeable estrogenic potential was accidentally discovered by Krishnan et al. in 1993 [1], bisphenol A (BPA) was and remains in the spotlight of the scientific community as an infamous endocrine disruptor with severe implications on public health and the environment [2, 3]. Canada was the first country to list BPA as toxic [4], and although now banned in most countries from some consumer products, especially baby products [5], BPA is still being used in non-BPA free products as a base material, with an estimated global production of 7.2 million tons in 2020 and a forecasted 4% production increase per year for 2021–2025 [6].

Being a widely used xenobiotic chemical manufactured in large quantities as a monomer component in many products including plastics such as polycarbonates, epoxy and vinyl ester resins, and in multiple consumer products including thermal paper and food packaging materials, BPA has become a widespread environmental pollutant that poses risks to human health [7, 8].

BPA is now considered an ubiquitously distributed pollutant occurring in most surface waters, sediments and soils, as well as in the air [9], and harmful effects have been reported to many animals [10] and plants [11]. BPA analogues are equally treated as wide pollutants and might exhibit even enhanced endocrine activity [12]. In recent years, the use of BPA in some consumer products has decreased due to the awareness raised to BPA, but its analogues have been increasingly detected in humans [13], with some, such as bisphenol S, posing a higher risk compared to that of BPA [14]. Although BPA analogues are being increasingly used while still trailing behind in terms of regulations [15, 16], the negative effects of BPA on humans continue to be the object of numerous studies. These potentially harmful effects have been associated with both the toxicity of BPA and its role as an endocrine disruptor. A brief list of these dangers to humans and their health includes:

  1. (1)

    Having a structure similar to estrogens, BPA influences endocrine function [17]. These damages include developmental alterations, anti-androgenic activity, lower sperm development, miscarriage risk and development of polycystic ovarian syndrome.

  2. (2)

    Immune system effects include promoting autoimmunity, allergies, and immunosuppression [17, 18].

  3. (3)

    Cancers have been associated with BPA exposure and research has demonstrated its carcinogenic action [17, 19].

  4. (4)

    Maternal exposures may be linked to memory and cognitive impairment [20].

  5. (5)

    One endocrine disruption which has been widely reported is the obesogenic effect leading to our epidemic of weight gain and other metabolic syndrome issues such as Type II diabetes [21, 22].

  6. (6)

    Direct toxic effects have been reported, an example is pyroptotic cell death [23] as well as induced alterations of the microbiome [24], which is critical to almost all aspects of human health.

BPA, as well as other well-known plasticizers such as phthalate esters (PAEs), or additives like polychlorinated biphenyls (PCBs) and perfluorooctanoic acids (PFOAs) are highly persistent organic pollutants that tend to bio-accumulate [25,26,27,28]. Taking into consideration the massive varieties of impacts on human health and those on the environment, it is critical to consider the analytical methods for BPA as well, and assist with what was described as a “mindboggling complexity around BPA” [29]. Thus, this systematic review examines aspects of the most common and affordable analytical methodology of BPA to date, HPLC-UV. Also, a major focus of this work is on the absorbance properties of BPA as it was observed that different studies chose various points on its UV spectrum for its detection and quantification. Clarifications are made for the general public that might skeptically regard the wide variation of wavelengths found in the open-access literature.

2 Materials and methods

2.1 Literature search

Literature search was conducted on the Scopus (https://www.scopus.com/), Web of Science (https://www.webofscience.com/) and PubMed (https://pubmed.ncbi.nlm.nih.gov/) bibliographic databases based on a methodology (Fig. 1) conformed to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [30]. All the existing literature related to BPA determination by HPLC was collected, including papers from the inception of the previously mentioned databases until 16th December 2022. The identification of research papers was carried out using the specific search terms and Boolean logical operators as follows: (“bisphenol A” OR “BPA”) AND (“high performance liquid chromatography” OR “HPLC”) AND (“determination” OR “identification” OR “detection” OR “measurement” OR “quantification”). The query string used for Scopus also included the default field code TITLE-ABS-KEY, while for Web of Science and PubMed the All Fields (ALL) code was utilized, as they do not provide an equivalent setting. Using the abovementioned search strategy, a total number of 3665 peer-reviewed papers was retrieved by the three databases, 1114 originated from Scopus, 1414 from Web of Science and 1137 from PubMed, respectively.

Fig. 1
figure 1

PRISMA flowchart displaying the number of papers at each step of the systematic literature analysis

Full size image

2.2 Paper screening and eligibility criteria used

The first step in selecting articles was represented by applying the databases built-in filters as follows: open-access articles published in scientific journals in English language. After these filters were applied, the number of articles from the three databases narrowed down to 725. The number of papers was further reduced to 484 by the removal of duplicates (241). Finally, these 484 articles were retained, and their titles and abstracts were independently screened based on an eligibility criteria checklist. The full-text scanning was performed only if the relevance of a paper was not clearly reflected from its title or abstract.

Papers that conformed to our criteria were: (i) articles that referred to BPA (papers referring only to bisphenol S, F or derivatives were marked as non-eligible); (ii) articles that used HPLC methods and results to analyze the presence of BPA in different samples; (iii) also, articles that used an HPLC device coupled only with an absorption detector (papers using HPLC, but mass spectrometry or fluorescence detection were marked as non-eligible) in order to focus our analysis on the issues that might arise when BPA is spectrophotometrically detected. The number of articles that satisfied the before mentioned criteria was 57, but we decided to exclude 2 of them because of errors that lower their reliability. All the 55 studies included in our systematic analysis are listed in Table S1 together with the targeted information.

2.3 Data extraction and visualization

The list of included articles was used to determine the temporal distribution of research on BPA, the main research directions in which HPLC-UV equipment was involved, and some features of the methods used, focusing on absorbance wavelength. Data extraction from papers was needed for the previously mentioned analyses. Extraction of information from the articles was done independently by two of the authors. Information extracted from articles regarded: papers’ purpose, the wavelength used, and composition of the mobile phases. Moreover, for a clear analysis, records were grouped based on their purpose as follows: (i) articles using HPLC to identify BPA release from different products, raising concerns regarding food and environmental contamination or human health risks; (ii) articles using HPLC to determine administered BPA in plants/animals for laboratory toxicology research; (iii) articles dealing with novel methods to identify/extract BPA (based on HPLC); (iv) others. Extracted data were converted into charts/diagrams using GraphPad Prism software (version 8; GraphPad Software Inc., San Diego, CA, USA).

2.4 Materials and methods

BPA ≥ 99% (Sigma-Aldrich, Darmstadt, Germany) solutions were quantified by the HPLC method using a L-3000 system (Rigol Technologies Inc., Beijing, China) consisting of a quaternary pump, a diode-array detector, and a Kinetex C18 Evo (150 × 4.6 mm i.d.; 5 μm particle size; Phenomenex, Torrance, CA, USA). Operating conditions were: isocratic elution of the mobile phase (pH 2.27) composed of eluent A represented by 0.1% trifluoroacetic acid (TFA) in acetonitrile respectively methanol; eluent B (0.1% TFA in water) = 50:50, flow rate was set at 1.0 mL/min, injection volume 10 μL, detection was monitored at: λmax = 200–282 nm, column temperature was set at 30 ℃.

A cuvette spectrophotometer (Implen NP80, München, Germany) with wavelength accuracy of ± 0.75 nm and wavelength reproducibility of ± 0.2 nm was used to perform wavelength scans in the 200–400 nm range for each of the solutions of BPA which were solubilized in methanol (MeOH) and acetonitrile (ACN). All BPA solutions were prepared using only glassware and were kept in glass vials at − 20 ℃ prior analysis.

3 Results and discussions

3.1 Temporal distribution and purpose of the selected articles

Over the time period examined, we observed that the number of open-access articles dealing with detection of BPA with HPLC-UV generally increased, even though the tendency was inconsistent. The oldest article we found was from 1997 [31]. The number of papers started to significantly increase after 2018 and reached its peak in 2021 (Fig. 2).HPLC equipment coupled with an UV detector is probably more prevalent in research facilities from around the world than others (e.g., HPLC–MS), due to cost reasons, but its affinity is generally lower. Therefore, we grouped the selected papers into different sub-domains for a better understanding of the use of HPLC-UV equipment in practice. Our results showed that nearly half of the studies were performed mainly for developing new approaches for BPA identification and quantification, whether there are considered new purification, extraction or preconcentration methods. HPLC-UV was also utilized as a reference by some authors that aimed to develop novel alternative methods for the determination of BPA [32,33,34]. Moreover, approximately a quarter of studies (23.64%) utilized HPLC-UV as a technique for quantification of BPA in environmental samples [35,36,37,38,39,40,41,42,43,44,45], mostly from aquatic habitats. 12.73% of the analyzed papers studied the leaching of BPA from consumer products, mostly dentistry resins [31, 46,47,48,49,50,51,52], but also from plastic packaging and bottles [53, 54]. The migration of BPA from food packaging could lead to food contamination, but only a few studies (3.64%) directly investigated this phenomenon [55, 56]. In addition, 5.45% of the studies were mainly directed to health risks of BPA in humans, quantifying the pollutant from human biological specimens (e.g., saliva, blood, breast milk) [47, 57,58,59], and only a few papers were designed for the investigation of BPA effects on plants or animal models [60, 61], (Fig. 3A, B). We noted the first studies (chronologically) were directed toward HPLC-UV detection of BPA leached from dentistry materials and then, the equipment started to be used in studies dealing with environmental or food contamination.

Fig. 2
figure 2

Temporal distribution of open-access scientific articles on HPLC-UV determination of BPA for different purposes

Full size image
Fig. 3
figure 3

A Main purpose of the analyzed studies as revealed by the publications in Table S1. B Overlapped main purposes with other secondary objectives of the analyzed papers

Full size image

The worldwide distribution of BPA has led to several research works and surveys on BPA determination and quantification. As previously stated, we screened a total of 55 open-access articles published between 1997 and 2022 that reported the use of different mobile phases and wavelengths (Table 1). The variety of these publications represents a reflection of the growing concerns in regard to BPA contamination. The literature search revealed that BPA was mostly identified in water samples [35, 62] and sediment [63], but it could also contaminate soil via plastic trash and through dump leachates [64] or by air pollution near polycarbonate molding plants [65], bee pollen [66], mussels [67] and brine shrimp [68].

Table 1 Mobile solvent phase and the wavelength used
Full size table

Food contamination was determined in meat samples [53], beverages [55, 69, 70], milk [59, 63, 71, 72], oils [56, 63], and canned food products [59, 71, 73] with most studies revealing that BPA could be leached from the plastic can linings, packaging, and bottles [53, 55, 56, 71, 74,75,76].

Contamination of organisms with BPA was assessed by analyzing the saliva [31, 47, 77, 78], blood [47, 57], urine [63], and breast milk [59]. The literature search yielded evidence suggesting that certain dental and orthodontic products may contribute to BPA contamination [31, 46, 49,50,51,52, 77,78,79].

3.2 Wavelengths used for the identification of BPA in the UV spectrum

Many phenolic compounds have a strong absorbance peak at λmax = 210 nm and two lower absorbance peaks at around λmax = 230 nm and λmax = 280 nm [80]. As its UV spectrum reveals, BPA is no exception to this physical property (see Sect. 3.4). However, we found that, over the three decades of publications, different methods utilizing various wavelengths (Fig. 4) for the identification of BPA after HPLC separation were performed without knowledge regarding their reliability. Our results show that 16.36% of the papers determined BPA based on its absorbance at λmax = 280 nm, while only 9.1% of the papers identified by us quantified the absorbance of BPA at λmax = 230 nm and 7.27% at λmax = 210 nm, respectively. However, the wavelengths used varied around the specific peaks of BPA: 34.55% of the analyzed studies quantified BPA in a spectrum range between λmax = 200 and λmax = 220 nm, 25.45% between λmax = 224 and λmax = 230 nm, and a significant number of papers representing 32.73% of the entire literature survey determined BPA at wavelengths between λmax = 270 and λmax = 280 nm. Interestingly, 3 groups of authors [41, 71, 73] chose to set the detection of BPA by HPLC-UV at the wavelength of λmax = 254 nm. Selection of this wavelength value is very popular in HPLC analysis although in most cases the absorption maxima is not even close. The reason for that is represented by older detector with mercury lamps which emit the higher intensity light at 254 nm. Thus, a large number of analysts automatically chose to select 254 nm in spite of improvements brought to the detectors in last years. Diode array detectors (DAD) change gradually this perception and allow to choose a suitable wavelength for each compound. However in case of BPA the discrepancy between absorbance values of the two maxima that appear in the spectrum make difficult to select a specific wavelength for different concentration ranges.

Fig. 4
figure 4

Wavelengths used in BPA determination and quantification

Full size image

3.3 HPLC-mobile phases

BPA is a compound with very poor solubility in water, the solubility in pure water at 25 ℃ was reported at 300 mg/L [81, 82], and also reported at 120 mg/L [83]. Shareef et al. [81] also found that BPA solubility does not vary significantly with pH values: 298 mg/L (pH = 4), 300 mg/L (pH = 7) and 319 mg/L (pH = 10) respectively. In spite of relatively low solvation in water these levels surpass the current regulations regarding endocrine disruptors, hence water can be an efficient vector for environmental dispersion. BPA also can be easily dissolved in polar organic solvents, such as methanol, acetonitrile, etc. [84], and it is recommended to be filtered only through PTFE, as filter composition (especially cellulose and nylon) significantly affect the analyte concentration after filtration [83]. Based on its solubility in polar organic solvents, BPA can be determined by HPLC using a mobile phase containing different amounts of water and acetonitrile. We observed that this method was utilized by the majority of studies, with slight modifications regarding the proportion between water and acetonitrile (50–50; 40–60; 70–30). Only a few papers described HPLC methods utilizing a water–methanol mobile phase [36, 57, 63, 65, 75, 76, 85, 86]. Moreover, Gamonchuang et al. [70] separated BPA from carbonated beverages in a 57% acetonitrile—43% acetic acid mobile phase.

3.4 Technical notes

BPA comprises two hydroxyphenyl and two methyl groups attached by single bonds to a central C atom. BPA is a π-conjugated system in which the HOMO and LUMO molecular orbitals extend over the phenyl moieties [87]. These give rise to the characteristic UV absorption at λmax = 280 nm [88]. In addition, O atoms lone pairs are delocalized in the π-system [87]. These will give rise to the π–π* peak at ~ 228 nm [89].

It is important to note that the exact position and intensity of BPA absorption peaks vary depending on the solvent used. The UV–VIS spectra of samples with various concentrations of BPA in methanol are presented in Fig. 5. Two BPA absorption peaks are located at λmax = 228 nm and λmax = 279 nm regardless of concentration. A third absorption peak at λmax = 203 nm in the most diluted sample is redshift at increasing concentrations to λmax = 207 nm in the case of the most concentrated sample. The relationship between absorbance and concentration was initially addressed in the case of the peaks at λmax = 228 and λmax = 279 nm. As can be seen in Fig. 6, there is a linear increase of absorption with concentration, suggesting that the two wavelengths are suited for quantitative analysis of BPA.

Fig. 5
figure 5

UV absorbtion spectra (200–400 nm) of different concentrations (1–25 μg/mL) of bisphenol A in methanol (a) with highlight on the specific peaks, b λmax = 228 nm and c λmax = 279 nm

Full size image
Fig. 6
figure 6

Linearity of the curve described by the absorption of bisphenol A in methanol, measured spectrophotometrically at a λmax = 228 nm and b λmax = 279 nm. The correlation coeficient (R2) is displayed on the graph

Full size image

The spectra of increasing concentrations of BPA in acetonitrile are presented in Fig. 7. The spectra present two peaks at λmax = 228 nm and λmax = 279 nm whose locations do not change with BPA concentration. A third peak appears at concentrations larger than 10 µg/mL, but its location shifts with concentration with no apparent rule. As in the case of methanol solvent, the samples prepared in acetonitrile present a linear increase of absorption with concentration at λmax = 228 nm and λmax = 279 nm (Fig. 8).

Fig. 7
figure 7

UV absorbtion spectra (λmax = 200–400 nm) of different concentrations (1–25 μg/mL) of bisphenol A in acetonitrile (a) with highlight on the specific peaks, b λmax = 228 nm and c λmax = 279 nm

Full size image
Fig. 8
figure 8

Linearity of the curve described by the absorption of bisphenol A in acetonitrile, measured spectrophotometrically at a λmax = 228 nm and b λmax = 279 nm. The correlation coeficient (R2) is displayed on the graph

Full size image

Since the studies reviewed here used different wavelengths to detect BPA, we questioned whether the linearity of the absorbance with concentration is preserved at the other wavelengths summarized in Table 1. The line intercept and slope, as well as the statistical parameters of the linear fit—residual sum of squares, Pearson’s r and Adj R2 are presented in Table 2. In the case of BPA in methanol and in acetonitrile, starting with λmax = 210 nm, the linearity of absorption with concentration is well preserved (R2 > 0.99, Pearson’s correlation coefficients > 0.99). A deviation from linearity can be seen in the case of measurements performed at λmax = 200 nm, λmax = 202 nm and λmax = 205 nm, especially at the two lower wavelengths. This suggests that in the case of BPA in methanol or in acetonitrile, these low wavenumbers are not suited for quantification applications. The analysis of linearity was extended to the wavelength range of λmax = 200–300 nm. Results show that for BPA in methanol samples, linearity does not characterize the absorptions in the 200–206 nm and 293–300 nm interval (R2 < 0.99). For BPA in acetonitrile, the intervals not characterized by linearity are narrower than in the case of methanol: λmax = 200–206 nm and λmax = 295–300 nm (R2 < 0.99).

Table 2 Parameters of the lines that fit the UV absorptions of BPA in methanol and acetonitrile at different wavelengths
Full size table

HPLC chromatograms are given in Supplementary Fig. S1. Rt and Area values can be found in Supplementary Table S2.

In the Figs. 9 and 10 it can be observed that the steepness of the linear calibration curves slopes varies over a wide range of values regardless the solvent used. The slope generally can be computed from the correlation factor (r) and the standard deviations of explanatory variable (Sx) and response variable (Sy). (Eq. 1):

$$ b = r\frac{{S_{y} }}{{S_{x} }} $$
(1)
Fig. 9
figure 9

BPA in methanol calibration curve based on area value. The symbols are used to represent the measurements at diffent wavelenghts. The data are fitted with lines that are represented with the color of the corresponding symbols. The line that fits the results at 200 nm was calculated for concentrations up to 20 µg/mL

Full size image
Fig. 10
figure 10

BPA in acetonitrile calibration curve based on area value. The symbols are used to represent the measurements at diffent wavelenghts. The data are fitted with lines that are represented with the color of the corresponding symbols. The line that fits the results at 200 nm was calculated for concentrations up to 20 µg/mL

Full size image

A stronger correlation between the two variables lead to a steeped slope. In case of BPA analysis, the optimum analysis can be performed using ACN mobile phase at lower wavelength values, taking into consideration the ACN UV cutoff. At higher wavelength values the slope presents lower values hence a poor correlation between variables and conversely a source of significant errors. For λmax = 200 nm the linearity range is very narrow but for polynomial fitting the correlation is strong and the data are consistent.

Phenols generally exhibit two absorption bands in the UV spectral region, the band at longer wavelength is referred to as the C-band, and the band at shorter wavelength as the B-band [97], both being slightly altered with change of solvent [98]. C-band having a finer structure, it is frequently reduced if the phenol is associated with a solvent molecule by means of an intermolecular hydrogen bound [97]. Being considered valid for phenols and polyphenols in general [80], the dedicated wavelengths (λmax = 230 nm and λmax = 280 nm, in our case λmax = 228 nm and λmax = 279 nm) are preferred, as well as their ratio (if the UV detector permits it). Using lower wavelengths may lead to increased interference and more noise from the solvents and other chemicals. As the λmax = 228 nm and λmax = 279 nm (respectively λmax = 230 and λmax = 280 nm) are both useful, the peak at λmax = 279 is particularly important as few other non-phenolic compounds absorb at this wavelength. Still, data suggest that BPA analysis can also be performed in the range of 203–210 nm, firstly if the BPA spectra is confirmed, as well as if the low concentration range and UV cut-off value of the solvent used justifies it. In our analysis, due to methanol having a UV cut-off value at λmax = 205 nm [99], it is unsuitable for detection under 205 nm.

Another reason for selecting acetonitrile and methanol in this work is related to the general procedure for BPA analysis [59, 95, 100,101,102,103,104]. Thus, regardless the source, BPA is extracted in most cases in water (at pH = 2–4) and, in the next step, solid phase extraction (SPE) or disperse SPE techniques are used. BPA retained on solid phase is eluted with 100% acetonitrile or methanol.

3.5 Statistical analysis

3.5.1 Methods

HPLC-UV data recorded on BPA samples in methanol and acetonitrile were analyzed by statistical methods in order to identify the best conditions to perform BPA concentration measurements in terms of best wavelength and best solvent. Also, the consistency of data was addressed by performing correlation analysis, calculations of coefficients of variation and estimation the errors for concentration calculation using our data. These were performed with OriginLab and ChatGPT [105], through the Advanced Data Analysis facility [106]. The conversation with ChatGPT is given in Supplementary material 1. The analysis performed with ChatGPT involved performing linear regressions [107], T-tests [108], Analysis of variance (ANOVA) [109], correlation analysis and calculation [110] of coefficients of variation [111]. The analysis performed by ChatGPT involved using Python [112] and the libraries NumPy [113], pandas [114], SciPy [115], Matplotlib [116], Seaborn [117] and scikit-learn [118].

3.5.2 Identification of the best conditions to measure BPA

Linear regression analysis was performed using the measurements on increasing BPA concentrations at all wavelengths, using methanol or acetonitrile in the mobile phases. In the case of samples with methanol, as can be seen in Supplementary Table S2, the highest R2 value (0.99658) was obtained at 203 nm. In the case of samples with acetonitrile, the best R2 value (0.99949) was obtained for measurements at 205 nm. When comparing results on the experiments with methanol and acetonitrile, it appears that the usage of the last solvent leads to better results, meaning that acetonitrile could be more suited for concentration calibration of BPA.

The line that fits BPA in methanol absorption at 203 nm presents the slope of 81.82076 ± 1.51447 and an intercept of − 9.0452 ± 22.40392. In the case of BPA in acetonitrile, the line that fits the results at 205 nm presents a slope of 67.54442 ± 0.48419 and an intercept of − 3.22624 ± 7.16274. The lines are presented in Fig. 11. A t-test applied on the slopes and intercepts of these lines showed that the parameters are statistically different (p = 1.87 × 10−8 for slopes and p = 5.52 × 10−13 for intercepts), thus the calibration curves are statistically different. In addition, an ANOVA test showed that there are no statistically significant differences in concentration measurements between the two solvents (F-statistic = 0.3483, p = 0.5617).

Fig. 11
figure 11

Results on BPA in methanol at 203 nm and on BPA in acetonitrile at 205 nm

Full size image

3.5.3 Assessment of consistency in concentration measurements across wavelengths

The analysis involved a correlation analysis of the concentrations at different wavelengths considering the same solvent. According to the heatmaps in Fig. 12, the results present strong positive correlations, with values close to 1. The less strong positive correlations are seen in the case of both solvents, for 200 nm with all the other wavelengths.

Fig. 12
figure 12

Half heatmap representations of the correlation matrices of different wavelengths within methanol (a) and acetonitrile (b) measurements. The color scale used red to present the maximum positive correlations and blue to present the less strong positive correlations between concentrations obtained at different wavelengths

Full size image

Results at all wavelengths in each solvent were analyzed by calculating the coefficients of variation (CV), taking into account that lower CVs indicate better repeatability and more reliable data for concentration measurements. Results in Fig. 13 show that methanol samples present CVs ranging from 68 to 82% across the wavelengths, while results in acetonitrile present CVs of 65% to 72% across the wavelengths. It is worth mentioning that extreme values in CVs were obtained for measurements at 200 nm (0.35–81.5% for samples in methanol or 0.1–71.6% for samples in acetonitrile). Also, measurements at 275 nm and 279 nm present large CVs, as labelled in Fig. 13.

Fig. 13
figure 13

Coefficients of variation (CV) for each considered wavelength in methanol and acetonitrile, excluding the Retention Time (RT)

Full size image

The average errors in concentration determinations based on the linear fits for each wavelength are plotted in Fig. 14. The figure shows that samples in methanol present lower error values for the wavelengths between 203 and 232 nm, with the largest errors at the extremes. Samples in acetonitrile present a large error at 200 nm, but otherwise, the errors are below 2% at considered wavelengths. The errors of concentration estimation for samples in acetonitrile are smaller than those of samples in methanol, suggesting that acetonitrile is better suited for BPA calibration curves.

Fig. 14
figure 14

The relative average errors in concentration determinations for samples in methanol and acetonitrile across various wavelengths. The relative error values (calculated as Average Error Mean Concentration × 100) are labelled directly on the image

Full size image

4 Conclusions

The various unspecific wavelengths used in BPA HPLC-UV detection and quantification might be indicative of the use of less selective and possibly misleading analytical methods. In the best practice of spectrophotometric detection, wavelengths where the specific compound of interest peaks are preferred, while taking into consideration the UV cut off of the solvent used, pH, and temperature. However, depending on the circumstances different wavelengths could be used only for the quantification of the compound. In these cases, experimenters must ensure in advance that there is a strong linear relation between concentration and absorbance at the chosen wavelength, considering also the possible noise resulted from other compounds within the analysed sample.

When considering the R2 values from Table 2, it appears more suitable to utilize the lower peaks of BPA (λmax = 205 and λmax = 210 nm) for both solvents, acetonitrile and methanol. A compromise could be made in certain conditions for the detection at the particular high absorbance in the region of 205–210 nm, when the absorbance peak is still visible whereas the other two dedicated peaks (corresponding to λmax = 228 nm and λmax = 279 nm) are not visible anymore; this is due to the delection of small concentration of BPA (at ≈ 1 µg/mL), that can vary depending on the sensitivity and quality of the equipment used. Even so, such a compromise limits the quality of the analysis. Therefore, we concluded that BPA detection and quantification using HPLC-UV requires a more careful method calibration, as well as full transparency in method settings, in order to limit the misinterpretation of other phenols as BPA.

The majority of studies that we analyzed were in accordance with our technical validation. However, the few studies using questionable wavelengths for the quantification of BPA indicate that more attention is needed in the future to strengthen the scientific literature’s accuracy.