Beryllium (Be) is an extremely useful but highly toxic element that is a key constituent in numerous materials such as high-performance ceramics, composites and specialty alloys. Be-containing components have myriad industrial, energy and defense uses including applications in microelectronic circuitry, medical devices, nuclear reactors, aerospace and even sports equipment. In the United States it is estimated that about 35,000 workers are exposed to Be during their work activities.¹ Exposure by inhalation or surface contact can cause immune sensitization in a percentage of those exposed. Once sensitized, succeeding exposure by inhalation can lead to chronic beryllium disease (CBD), a debilitating, incurable and potentially fatal lung disease. Exposure to airborne Be may also cause lung cancer.¹ Consequently there are efforts to prevent occupational exposures to Be through safety and health recommendations and regulations, which in turn raise challenges concerning measurement performance requirements for sampling and determination of trace amounts of Be in workplaces.

Accurate measurement of Be concentrations at ultra-trace levels in work air samples is necessary in consideration of the very low occupational exposure limits (OELs) that have been promulgated for this element.^{2, 3} It has been shown in previous work that a molecular fluorescence detection technique using hydroxybenzoquinoline sulfonate (HBQS) as fluorophore,⁴ after sample dissolution in dilute ammonium bifluoride (NH₄HF₂),⁵ offers method detection limits (MDLs) which are low enough to allow for accurate Be measurements at levels well below current OELs in the United States and Europe.^6,7 Adequately low Be MDLs for current OELs in the U.S. are also provided by graphite furnace atomic absorption spectrometric (GFAAS)⁸ and inductively coupled plasma mass spectrometric (ICP-MS)⁹ methods after digestion of samples in appropriate dissolution reagents.¹⁰

All of the previous work on the use of HBQS fluorescent dye was based on the report from Matsumiya et al.⁴ Those workers demonstrated that it is important that the pH of the dye solution and the sample solution be about 12.2 in order to obtain a high fluorescence signal. If the pH increased slightly there was a rapid decrease of fluorescent intensity. In addition, Matsumiya et al.⁴ also showed that the fluorescent intensity decreases below pH 12.2, and this decrease continues to a pH of about 4.5, below which there is no signal observed above the background. Thus the previous work using dilute ammonium bifluoride simplified the method by adding lysine to the dye solution (a buffer) so that when the dye solution was mixed with acidic solution (dilute ammonium bifluoride) containing the extracted beryllium, there was minimal pH shift, thus there was no need for extra titration steps to bring the pH back into the desired narrow range near 12.2. From this we postulated that there was no need of buffer if we could maintain a highly alkaline pH by adding more base. As is discussed in this paper, adding more base had many benefits: first, one could dissolve the samples in more acidic solutions, and the higher alkalinity of the dye solutions still preserved a high enough pH of the mixture to provide good fluorescence signal for beryllium quantification; second, one could mix higher amounts of the sample solution relative to the dye solution to obtain superior detection limits; and third, since lysine comes from natural sources and sometimes it can have naturally fluorescing organic contaminants, its elimination simplified the process of making the dye solution and lowered the background signal. The aim of this paper was to experimentally demonstrate that removing lysine and increasing alkalinity of the dye solution significantly benefits the method.

Since even further improvements in MDLs for airborne Be are desired in anticipation of decreasing OELs for this element, including significantly reduced short-term exposure limits (STELs), the above improvements, if borne out, can be of high importance. For instance, the US Occupational Safety and Health Administration (OSHA) has established a tenfold reduction in the permissible exposure limit (PEL), to 0.2 μg m⁻³ as an 8-hour time-weighted average (TWA),¹¹ which is in the neighborhood of the OEL for Be promulgated previously by the US Department of Energy (DOE).¹² OSHA has also established a STEL (15 min) that is on the order of its previous PEL. The American Conference of Governmental Industrial Hygienists (ACGIH) Threshold Limit Value (TLV^®) for Be is 0.05 μg m⁻³ as an 8-hour TWA (inhalable particles).² Outside of the U.S. the lowest Be OELs have been promulgated in Germany:³ 0.14 μg m⁻³ (inhalable particles) / 0.06 μg m⁻³ (respirable particles); these figures apply to 2-hour TWA sampling as well as a 15-minute STEL. In consideration of the above OELs for Be and air sample collection flow rates of 2–10 l min⁻¹ over time durations as little as 15 minutes, MDLs in the sub-nanogram per sample range are required to allow for quantitative measurements of Be in air at these very low levels.

Apart from air samples, there is a need for accurate measurements of trace levels of Be in other occupational and environmental matrices such as surfaces, dusts, soils, etc.¹³ Regulatory limits for Be on surfaces (of equipment, workrooms, etc.) have been established by DOE, with 0.2 μg Be per 100 cm² sampling area as the lowest limit for demonstration of a contamination-free work environment and for purposes of equipment release.¹² Consequently, methods for sampling Be in surface dust have been developed, evaluated, validated and standardized.^14∓16 Measurement of trace Be in soils is also important as this enables assessment of potential anthropogenic sources of suspected Be-contaminated environments.^17–19 Molecular fluorescence detection after dilute NH₄HF₂ extraction with heating has been shown to yield Be trace measurement results that are comparable to data obtained from atomic spectrometric determination following strong acid digestion.^17,20

This work describes further improvements to the dilute NH₄HF₂ extraction / HBQS fluorometric detection method, with demonstrated successful applications to ultra-trace Be measurements in air filters, surface wipe samples, soils and newly-available occupational air sampler inserts. It is shown that the use of highly alkaline HBQS dye solution for Be determination by fluorescence results in significant improvement in MDLs for occupational hygiene and environmental samples.

Ethylenediaminetetraacetic acid disodium salt dehydrate (EDTA, ≥99%), ferric chloride hexahydrate (reagent grade) and L-lysine monohydrochloride (lysine; ≥98%) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Ammonium bifluoride (NH₄HF₂, ≥98%) was purchased from Fisher Scientific (Hampton, NH, USA). Hydroxybenzoquinoline sulfonate (HBQS) was prepared and purified after the procedure of Matsumiya et al.⁴ from HBQ precursor obtained from Sigma-Aldrich. Aqueous beryllium standard solution (1000 μg ml⁻¹) came from Spex Certiprep (Metuchen, NJ, USA). Sodium hydroxide solution (2.5 M), plastic centrifuge tubes (15-ml and 50-ml), and plastic syringes with Luer-lock apparatus (5-ml and 10-ml) were obtained from Fisher Scientific. GH Polypro™ (GHP) hydrophilic polypropylene Acrodisc^® syringe filters (25-mm dia., 0.2-μm pore size) and nylon bottletop filters (500-ml capacity, 0.2-μm pore size) were purchased from Pall Corporation (Port Washington, NY, USA). Mixed-cellulose ester (MCE) filters (37-mm dia., 0.8-μm pore size) were obtained from SKC (Eighty-Four, PA, USA), and cellulosic cassette inserts (Solu-Serts^®) were provided by Zefon International (Ocala, FL, USA). Whatman 541 filters (47-mm diameter) were obtained from Sigma-Aldrich. Typically, MCE filters are used for air sampling and Whatman filters have been used for surface sampling. Certified MCE filters²¹ spiked with known quantities of aqueous beryllium standards (diluted from 1000 μg ml⁻¹ standard solution) and National Institute of Standards and Technology Standard Reference Material (NIST SRM^®) 1877 beryllium oxide (NIST, Gaithersburg, MD, USA) were provided by High-Purity Standards (North Charleston, SC, USA). Certified reference material (CRM) soils with established beryllium contents were obtained from NIST and from the Canadian Certified Reference Materials Project (CCRMP, Ottawa, ON, Canada). Mechanical pipets (with plastic pipet tips to fit) of various sizes, used in carrying out most experiments, were purchased from Eppendorf (Hamburg, Germany).

Measurements of pH were conducted using an Orion 290A+ meter (Thermo Electron, Beverly, MA, USA) calibrated using buffer standards of pH 4.0, 7.0, 10.0 and 13.0. Standard solutions of pH 4, 7 and 10 were obtained from ACROS (Geel, Belgium) and that for pH 13 was obtained from Ricca Chemical (Arlington, TX, USA). Deionized (18 MΩ-cm resistivity) water, used in all experiments, was prepared using a Barnstead^® purification system (Thermo Fisher Scientific, Waltham, MA, USA). Where necessary for sample agitation, a Labquake^® rotator (Thermo Fisher) was used for this purpose.

HBQS fluorometric dye solution containing lysine was made following the procedure outlined in ASTM D7202:²² Lysine, 19.508 g (intended for high-pH buffering),²³ 2.208 g of EDTA, and 0.0382 g of HBQS were added to 1800 ml of deionized water. This mixture was stirred at room temperature until all constituents had dissolved; the pH of the resultant solution was 4.65. For adjustment of the pH to alkaline conditions, to this solution was added 150 ml of 2.5 M NaOH; the measured pH of the resulting mixture was 12.77. An additional 17 ml of 2.5 M NaOH was added stepwise to bring the pH to 12.86; finally to this 33 ml of DI water was added and the final pH remained at 12.86. The solution was then filtered through a 2-μm GHP polypropylene filter. A highly basic dye solution without lysine was prepared as follows: 1.104 g of EDTA, 0.019 g of HBQS and 900 ml of DI water were stirred at room temperature until a clear yellow solution was obtained. To this 114.5 ml of 2.5 M NaOH was added. After mixing, a clear yellow solution was obtained; this had a measured pH of 13.17. The resultant solution was then filtered through a 2-μm filter as above. The concentration of the HBQS dye in solutions with and without lysine was the same, at 61 μM. These dye solutions were found to be stable (by pH and fluorescence measurements) for at least twelve months.

For fluorescence measurements, 2–4 ml of analyte solution was placed into 10-mm path length plastic cuvettes with transmittance >330 nm (Sarstedt, Nümbrecht, Germany). Fluorescence measurements with excitation at λ of 365 nm or 384 nm were carried out using a Modulus fluorometer (Turner Biosystems, Sunnyvale, CA, USA). Some experiments using excitation at λ = 384 nm were conducted on a RF 5301PC Spectrofluorometer (Shimadzu Scientific, Columbia, MD, USA). Fluorescence emission was monitored at the maximum for the HBQS-Be adduct of 480 (±5) nm.⁴

To investigate optimized analytical conditions, a suite of experiments was carried out in order to characterize the performance of the ultra-trace Be fluorescence measurement methodology in lysine-free and lysine-containing dye solutions.^22,24,25 Influence on background fluorescence signal was studied and a battery of analytical tests on representative samples of interest in occupational and environmental hygiene was carried out. Of particular interest were comparisons of the analysis results obtained between HBQS dye with and without lysine described above, as well as performance studies on a new workplace air sampling apparatus (i.e., a cellulosic sampler insert).²⁶

Solutions were made up in deionized water with varying NH₄HF₂ contents ranging from 1 – 20% (w˖w⁻¹). In an initial experiment, these NH₄HF₂ solutions were added to the two dye solutions prepared above in a volumetric ratio of 1:19 (20× dilution) and the resulting pH was measured (Figure 1). It can be seen that the pH of the solution mixtures (measurement solutions) made using dye solution with lysine fell below 12 when these were mixed with 6% NH₄HF₂ solution, whereas the pH for lysine-free dye solution remained above 12 for solutions containing up to 17% NH₄HF₂. The lysine-free dye solution was able to tolerate more added acid (i.e., NH₄HF₂) as compared to the solution containing lysine, which is a triprotic amino acid (pK₁ ≈2.2; pK₂ ≈9.0; pK₃ ≈10.5).²⁷ The results of Figure 1 show that lysine does not provide effective high-pH buffering properties, as has been proposed;²³ on the contrary, better acid tolerance is observed and high pH is maintained when lysine is absent. Both the NH₄HF₂ and lysine neutralize the hydroxide used in the preparation of the dye solutions, such that more NH₄HF₂ can be added to lysine-free solutions before there are any significant changes in pH. These results are important analytically as it is desired to maintain very high pH for optimal fluorescence of the HBQS-Be adduct.⁴ It is also shown that the addition of lysine is not only unnecessary but can be disadvantageous when the dye solution is challenged with acidic sample extract.

The above solutions with varying percentages of NH₄HF₂, but containing 4 μg l⁻¹ of beryllium in the final mixture (for analysis), were prepared. These solutions were subjected to fluorescence measurements, with results (uncorrected for background) shown in Figure 2. It can be seen that the fluorescence signal remains high for lysine-free solutions at much higher NH₄HF₂ content vs. HBQS dye solutions containing lysine. Figure 3 shows the dependence of fluorescence intensity on the pH of dye solutions with and without beryllium. The results from both lysine-containing and lysine-free dye solutions do not show a rapid decrease in fluorescence intensity above a pH of about 12.6. A rapid drop in fluorescence intensity is observed at a pH value of about 11 and below (Figure 3). For comparison, the change in fluorescence with pH data from Matsumiya et al.⁴ is shown in Figure 3. (Since this is relative fluorescent intensity, the data of Matsumiya et al.⁴ were normalized so that peak intensity of their results coincide with the maximum of our data shown in this figure.) When we carried out more detailed analysis we observed a very different trend between data from this study and the results of Matsumiya et al. First there was no sharp drop-off the fluorescent intensity at higher alkalinity (i.e., higher than 12.2), and second, that there was no discernable fluorescent signal below pH 10, and thus the method is not practically usable below pH 11; preferably the pH should be at or greater than 11.5. As a practical guideline the pH of the mixture of the acidic extraction solution and the dye solution should be about 12 or higher for effective determination of beryllium by this method. It is noted (Figure 2) that the pH drop in lysine-free solution occurs around 19% NH₄HF₂, whereas the same happens at about 7% NH₄HF₂ for solution containing lysine. This demonstrates that lysine-free HBQS dye solution can be used with highly acidic starting solutions, a decided advantage for certain samples requiring stronger NH₄HF₂ concentrations for effective extraction of Be prior to fluorescence analysis. In all cases, the fluorescence intensity resulting from the HBQS-Be adduct when using excitation at 384 nm was about double that obtained with excitation at 365 nm.

Dye solutions described above (i.e., with and without lysine) were mixed with 1% and 3% NH₄HF₂ solutions at 20× and 5× dilution, and the resultant pH values were measured (Table 1). The reason for the choice of these dilutions and concentrations of NH₄HF₂ solutions is that such conditions have been used to analyze surface wipe and air filter samples for Be content, which are typically extracted in 1% NH₄HF₂,^5,6 as well as soil samples, where 3% NH₄HF₂ is used for Be extraction.^17,20 Usage of highly alkaline solutions allows for the use of 5× dilution with 3% NH₄HF₂ and hence enables Be to be detected with higher sensitivity when these solutions are used for extracting Be into an aqueous phase. One may even increase the alkalinity of the dye solutions further for applications with yet higher concentration of NH₄HF₂ and/or lower dilutions. When the fluorescence at 480 nm of the above solutions was monitored as a function of [Be] over the range 0 – 20 μg l⁻¹, it was found that the sensitivity of the method was greater when using lysine-free solution vs. solution containing lysine, only because in the former a higher pH could be maintained. This is a benefit of the use of highly alkaline conditions sans lysine throughout the measurement process.

Method detection limits (MDLs) for Be were estimated by analyzing ten blank media and reporting the MDL as three times the standard deviation of the mean blank signal.²² Calibration solutions were made with known amounts of Be in 1% NH₄HF₂ solutions and were mixed with either dye solution using 20×, 5× or 3× dilution. Batches of filters were prepared and analyzed in the presence of soluble Be⁵ at low concentrations (0.05–2 μg l⁻¹). Also, filters spiked with high-fired beryllium oxide (BeO) were subjected to sample preparation and fluorescence analysis. The certified values for Be on BeO-spiked filters were within ±15% for all spiking levels. The filters were placed into 15-ml plastic centrifuge tubes with 5 ml of 1% NH₄HF₂ solution, capped, and heated for one hour at 90 °C to extract Be into the dilute aqueous NH₄HF₂ solution. Solutions were analyzed by mixing with highly alkaline lysine-free dye solution and also with dye solution which contained lysine. Analyses of Be-spiked media were done at least in triplicate using dilutions of 5× and 20× (and in some cases 3×) in accordance with procedures delineated in ASTM D7202.²² Analysis results for Whatman 541 filters are shown in Table 2 and data for MCE filters are presented in Table 3.

In another set of experiments, 37-mm diameter cellulosic sampler inserts consisting of MCE filters melded to cellulose acetate housings (for insertion into 37-mm air-sampling cassettes: Solu-Serts)²⁸ were tested. The Solu-Serts were spiked using pre-determined quantities of NH₄HF₂ solutions containing soluble beryllium and dried. Blank media as well as spikes were evaluated in a similar manner as that described earlier for surface sampling media. These cassette inserts were folded and inserted into 15-ml centrifuge tubes (in a similar fashion as was done for Whatman and MCE filters) and extracted in 5 ml of 1% NH₄HF₂ solution at 90 °C for 60 min before being allowed to cool to room temperature. Fluorescence at 480 nm was measured after mixing these extract solutions with the lysine-free HBQS dye by a 3× dilution factor. The fluorescence values were evaluated against two calibration curves for low-level calibration standards (“Low Cal-1” which used 0, 0.05, 0.1, 0.2 and 0.8 μg l⁻¹ of Be as calibration standards after mixing the dye and the standard solution and “Low Cal-2” which used 0, 0.2, 0.5, 0.8 and 4 μg l⁻¹ of Be); the results are shown in Table 4. The fluorescence values were corrected for any background contributed by the blank sampling media. Optimal results were found using the “Low Cal-2” calibration curve (Table 4) and show that the method may be used to quantify to as low as 0.5 ng Be per sample in cellulosic sampler inserts.

Table 5 shows results on MCE filters spiked with high-fired BeO. The spiked filters were obtained commercially. Beryllium was extracted in 5 ml of 1% NH₄HF₂ solutions as described earlier. The table shows the nominal values of Be (values supplied by the commercial supplier) and the expected nominal concentration of beryllium in solution (assuming 100% recovery) after mixing a portion of the extracted NH₄HF₂ solution with the dye solution in a dilution ratio of 3×. For calibration, solutions were made with soluble beryllium in 1% NH₄HF₂ solutions after mixing them with the dye solutions in a ratio of 1:2 (3× dilution). Beryllium in the calibration solutions was 0, 0.15, 0.3, 0.6 and 2.4 μg l⁻¹. The results on the spiked samples in Table 5 show excellent agreement with the expected values.

Experiments were carried out to investigate if there would be any effect of interference from other metals (besides Be) when using highly alkaline solutions. Iron (Fe) was chosen for these trials as it has been shown in the past that this element (and also to some extent titanium) can cause interference to the fluorescence measurement of the HBQS-Be adduct by imparting a yellow color to the measurement solution.⁵ This interferes with fluorescence measurement and biases the results towards lower amounts of measured beryllium.²⁹ The yellowness in the solutions may be removed by immediate filtering through GHP hydrophilic polypropylene filters of pore size 0.2 μm or finer; alternatively, it is an option to wait for at least two hours for these metal impurities to settle out before filtering. Solutions were made up with beryllium and ferric chloride hexahydrate (as a source of Fe) in dilute NH₄HF₂ (1% and 3% NH₄HF₂ in water) and then mixed with the appropriate dye solution (lysine-free or with lysine) using 5× and 20× dilution. In one set the solutions were immediately filtered through Acrodisc GHP 25-mm syringe filters with hydrophilic polypropylene 0.2 μm membrane. Another set of solutions was left standing for two hours and then filtered using similar filters. Control samples with the same concentration of beryllium but without Fe were also measured. In each series (i.e., for a specific dye solution used, NH₄HF₂ concentration and dilution factor) the fluorescence intensity was normalized to the sample without Fe within the same series. In all of the solutions (mixtures of the dye and the sample solutions) the Be concentration was 0.1 μg l⁻¹. The Fe concentration in solutions with 20× dilution was 1.1 mg l⁻¹ and for 5× it was 4.4 mg l⁻¹. In each case the Fe concentration was more than 10,000 times the Be concentration. The results of these experiments are shown in Table 6, and demonstrate that Fe interference to Be fluorescence measurement can be effectively eliminated in lysine-free (as well as lysine-containing⁵) HBQS dye solutions.

Experiments were conducted on representative CRM soils to investigate the performance of the two dye solutions (i.e., with lysine and lysine-free) as made up above and in accordance with ASTM D7458.³⁰ NIST SRM 2710, Montana soil ([Be] = 2.5 μg g⁻¹), and CCRMP Till-1 soil ([Be] = 2.4 μg g⁻¹) were analyzed for beryllium content in the following manner. The extraction solutions (using 3% aqueous NH₄HF₂) used in this analysis were the same as used in previously-published studies.^17,20 Each sample (0.5 g) was extracted in 50 ml 3% NH₄HF₂ at 90 °C for 40 hours. Three aliquots of each the extracted solutions were mixed with the respective dye solutions at a 5× dilution ratio. The pH of the dye solution with lysine mixed with the sample solution was 9.4, while that for highly alkaline, lysine-free solutions was 12.4. Measured [Be] values (n=3) using the high-pH mixture were 2.49 ± 0.014 μg g⁻¹ and 2.04 ± 0.046 μg g⁻¹ (99.5% and 84.9% recovery, respectively), while the low-pH (lysine-containing) dye solution yielded Be measurement results for both soils that were below the MDL. These trials demonstrate that the use of highly alkaline conditions can enable the measurement of low levels of beryllium in bulk environmental samples such as soils.

The experiments described herein show that there are several advantages in using highly alkaline, lysine-free HBQS dye solutions for trace and ultra-trace measurement of Be in occupational and environmental samples. Some of the important advantages include: (a) facile preparation of HBQS dye solutions sans lysine, which is not only superfluous but can be deleterious to high-pH ultra-trace Be fluorometric analysis; (b) solutions of higher acidity can be used to extract Be from refractory and silicate materials, without affecting analytical sensitivity; (c) standard acidic solutions used for analysis of surface wipes and air samples containing 1% NH₄HF₂ can be mixed with the dye solutions in 3× dilution to enhance the MDLs of Be to below 0.1 ng (and quantification from <0.5 ng); (d) these dye solutions may be used to enhance the method detection limits of Be in bulk sample analysis such as soils by using 5× dilution where 3% NH₄HF₂ solutions are used for extraction; and (e) interference from Fe is effectively eliminated. The results are important in consideration of very low occupational exposure limits for Be that have been established globally and in newly-promulgated regulations in the US. The extremely low method detection limits for Be will enable short-term airborne workplace exposures to be reliably monitored by means of a field-portable technique. The method offers promise for applications to a wider range of acidic extraction solutions, which could potentially be used to extract Be from challenging metallic samples such as aluminum and specialty alloys as well as ceramic materials.