A rapid method for analysis of non-equilibrated ⁹⁰Sr/⁹⁰Y in infant formula

Abstract

A rapid analytical method for quantifying ⁹⁰Sr in infant formula prior to secular equilibrium is presented. The approach is dependent on the use of two separations of ⁹⁰Sr from ⁹⁰Y, with the first providing an ⁹⁰Y ingrowth start point and the second providing an ⁹⁰Y ingrowth end point. Data were obtained at activity concentrations of approximately 6 Bq/kg and 160 Bq/kg, the latter of which is representative of the US Food and Drug Administration (FDA) Derived Intervention Levels (DIL). Experiments were designed to collect data from ingrowth periods ranging from 16 h to 2 weeks. Activities obtained with a separation interval as low as 16 h ranged from 92.7 to 109.4% of the known value. When ⁹⁰Y ingrowth was allowed to occur for 24 h or longer, the activities ranged from 93.2 to 106.2% of the known value and the precision of this group improved from 5.2 to 3.1%. The limit of quantification (LOQ) was 0.5 Bq/kg using 250 g sample portions.

Introduction

Radioactive strontium isotopes, resultant from fresh fission contamination, are among the largest sources of dose to humans by ingestion. Both ⁹⁰Sr and ⁸⁹Sr are high yield products of ²³⁵U and ²³⁹Pu nuclear fuels which are commonly used in nuclear reactors [1]. Radiostrontium is of significant concern in milk products due to its tendency to follow calcium biological pathways, making it more likely to concentrate in the leaves and stalks of the hay and alfalfa used to feed cows, and subsequently transfer to the milk [2]. Human consumption of contaminated food products can result in bioaccumulation of radiostrontium in the bones where it may remain bound over the lifetime of the exposed individual [3]. This is particularly important for younger populations that may consume dairy as a disproportionate amount of their diet. The relatively long half-life of ⁹⁰Sr, 28.91 years [4], means that the retained strontium may deliver its dose for an extended period. Additionally, ⁹⁰Sr results in both a 546 keV beta particle due to its decay to ⁹⁰Y, and a higher energy 2.3 MeV beta particle due to the decay of ⁹⁰Y to ⁹⁰Zr, effectively increasing the overall dose [4]. Even ⁸⁹Sr, with a half-life of 50.6 days and a beta emission of 1.5 MeV [4], will potentially deliver its dose throughout an entire year. Ingestion of radiostrontium has been associated with increases in leukemia and bone cancer [5, 6]. Two isotopes of strontium, ⁸⁹Sr and ⁹⁰Sr, are particularly dangerous because they are not easily surveilled at levels of interest for food contamination. Neither isotope has a gamma-emission associated with its decay with an abundance of > 1% [4]. Furthermore, ⁸⁹Sr, which is of interest in the first few years following a reactor event, is more likely to be present with other fresh fission products making analysis by gamma-ray counting more difficult.

Efforts have been made to develop rapid methods for analyzing radiostrontium. A procedure for rapidly and simultaneously determining ⁸⁹Sr and ⁹⁰Sr in milk samples was reported in IAEA Analytical Quality in Nuclear Applications Series No. 27 [1]. Rapid screening analysis of ⁹⁰Sr in edible plant samples was performed by Amano et al. [7] The method is only applicable for analyzing samples in which ⁹⁰Sr and ⁹⁰Y have already reached secular equilibrium. The authors state that this method may not be used for milk samples in which equilibrium between ⁹⁰Sr and ⁹⁰Y is not established. A rapid method for analyzing radiostrontium in milk by resin extraction was presented by Saez-Munoz et al. [8]. This method was proposed to further shorten the analysis time of comparable methods and focused on the total activity of ⁸⁹Sr and ⁹⁰Sr. Swearingen and Wall used Bateman’s equation and successive daily liquid scintillation counts over a 5 day period to simultaneously quantify ⁹⁰Sr and ⁹⁰Y; however, the presented method can be used for samples that contain ⁹⁰Sr and ⁹⁰Y only because ⁸⁹Sr and other radionuclides were not considered [9].

In the event of a nuclear incident, samples need to be analyzed quickly to assess the impact on public health. Even in a non-emergency, a product such as milk, which has a short shelf-life, may need to be analyzed prior to the establishment of secular equilibrium. The assumption of secular equilibrium between ⁹⁰Sr and ⁹⁰Y cannot be confidently made because strontium and yttrium follow different chemical and biological pathways [10, 11]. When ⁹⁰Sr and ⁹⁰Y are not in secular equilibrium, they cannot be assumed to have the same activity, which makes an accurate determination of ⁹⁰Sr by measurement of ⁹⁰Y difficult.

Two primary analytical approaches exist for the determination of ⁹⁰Sr in food. In one, the strontium is chemically extracted and counted. The counting results have ⁹⁰Sr and ⁹⁰Y components as well as the potential for ⁸⁹Sr contribution. The strength of this approach is that strontium is assayed directly. However, because the beta emissions of ⁹⁰Sr, ⁸⁹Sr and ⁹⁰Y are not distinguishable by the instruments commonly used for this type of analysis, the method is typically dependent on attenuation curves for each of the three radionuclides and an ingrowth period followed by additional counts over at least two weeks. For analysis of food, especially if more than one year has passed since a known contamination event, this method is often abbreviated with the assumption that ⁸⁹Sr is not present. If this assumption is made in error, ⁹⁰Sr results may be biased high. In the second approach, yttrium is chemically extracted and counted. Typically, use of this single attenuation curve method requires the assumption that ⁹⁰Sr and ⁹⁰Y are in secular equilibrium. However, based on their relative half-lives, almost three weeks is required to reach equilibrium. Herein the authors present a single attenuation curve method that incorporates calculations to account for ⁹⁰Y ingrowth prior to equilibrium using just one count of 100 min following separation.

Strontium or yttrium separation procedures have been performed using ion-exchange resins or liquid–liquid extraction. Resins are effective and may also be applied to the schemes described in this paper. A modification of a liquid-liquid separation employed in our laboratory for routine analysis is also reported [12].

The primary goal of this study was to develop a method for ⁹⁰Sr analysis that could be performed rapidly following sample receipt, without the assumption of secular equilibrium between ⁹⁰Sr and ⁹⁰Y and without assuming the absence of ⁸⁹Sr. The method assumes that a rapid ashing procedure, for example, one that takes place overnight may be employed to initially decompose the sample, enabling faster digestion and solubilization. The work put forth in this submission is based on the approach described by Wei et al. [13] using the foundation established by Rutherford [14].

Experimental

Reagents and materials

Tri-n-butyl phosphate (TBP) and hydrofluoric acid were purchased from ACROS Organics (Carlsbad, CA). The Sr resin is obtained from Eichrom Technologies (Lisle, IL). All other reagents were obtained from Fisher Scientific (Waltham, MA). TBP (99+%) is used for extraction; all other chemicals used in the analysis are ACS Grade or better. NIST traceable radionuclide standards were obtained from Eckert and Ziegler. The lower concentration ⁹⁰Sr/⁹⁰Y solution, 1.618 Bq/mL (Reference Date: April 1, 2014), in a 0.1 M HCl solution containing approximately 20 μg/g Sr and 10 μg/g of Y, was purchased directly. The higher concentration ⁹⁰Sr/⁹⁰Y solution is prepared by dilution of an Eckert and Ziegler standard prepared in a 0.1 M HCl solution with 30 μg/g of both Sr and Y carriers. The standard is diluted with 0.1 M HCl gravimetrically to a final activity concentration of 40.63 Bq/g (Reference Date: April 8, 2019). The yttrium carrier used to evaluate chemical recovery is prepared by dissolving yttrium oxide in 1 M nitric acid to provide a Y³⁺ concentration of 20 mg/mL. The strontium carrier contains 90 mg/mL of Sr²⁺ in 0.016 M nitric acid.

Infant formula (Enfamil NueroPro, Milk-Based Powder with Iron) for analysis was purchased from a local grocery store.

Equipment and instrumentation

Ashing was performed using programmable furnaces, N450/G, N100/G, LT40 and N450/DB from Nabertherm GmbH. Samples were counted for 100 min on a Protean MPC-9604 gas proportional counter using Vista-2000 software. The ⁹⁰Y attenuation curve was prepared using linear regression of results from efficiency determination of prepared sources with varying amounts of yttrium oxalate solid deposition [12, 15].

A set of six calibration standards containing a similar amount of ⁹⁰Sr activity and varying amounts of yttrium carrier were prepared. Aliquots of yttrium carrier added were calculated to yield approximately 25 mg to 75 mg of yttrium oxalate representing a ~ 35% to ~ 110% yttrium recovery from the analytical procedure. Each aliquot was passed through a Sr resin column to separate yttrium from strontium before converting the yttrium to yttrium oxalate [15–17]. The yttrium oxalate was then prepared in the same geometry as was used in the analytical procedure and counted on a gas proportional counter, obtaining at least 10,000 counts [18]. The counting efficiency of each source was then calculated based on count rate, carrier recovery and known activity. An attenuation curve was created to calculate ⁹⁰Y counting efficiency as a function of yttrium oxalate weight.

Procedure

Due to the unavailability of non-equilibrated ⁹⁰Sr/⁹⁰Y systems, samples were created with various intervals of ⁹⁰Y ingrowth. Positive controls were made with a ⁹⁰Sr/⁹⁰Y standard and ⁹⁰Sr was separated from its progeny to establish a condition at which the ⁹⁰Y activity is known to be zero. Then, an interval of as short as 16 h to 1 day was allowed for ⁹⁰Y to ingrow. After which, the sample solutions were again separated and then analyzed for ⁹⁰Y using gas proportional counting. Accurate ⁹⁰Sr activities were calculated from the ⁹⁰Y analysis data using Eq. (1) without the need to wait for secular equilibrium.

To assess method performance, ashed samples of infant formula were spiked with ⁹⁰Sr to produce concentrations of approximately 6 Bq/kg and 160 Bq/kg. The 160 Bq/kg level is consistent with the US Food and Drug Administration (FDA) derived intervention level (DIL). Samples were analyzed at various intervals between separations 1 and 2 ranging from 16 h to 15 days.

The radiochemical analysis of ⁹⁰Sr is performed using a method based on Barrata et al. [12] which is modified to utilize a different approach to ⁹⁰Sr and ⁹⁰Y separation and to reduce the use of concentrated nitric acid. Rather than precipitating the strontium as a nitrate using fuming nitric acid and an ice bath, the yttrium portion is extracted immediately into equilibrated TBP. Acid portions are reduced from those described in the original method for digestion from 100 to 30 mL for the analysis of most foods (milk, fruits and vegetables, and fish), except for foods with a very high mineral content (i.e., > 2.5%). Sample portions of 250 g of food are used for this study, which typically correspond to less than 5 g of ash. Samples were ashed using ashing Procedure A from Radioassay Procedures for Environmental Samples [19], which is performed over a 5 day period. This ashing protocol may be modified to shorten the ashing time to less than 12 h. Data obtained using the rapid ashing protocol are discussed in the “Results and discussion” section. A 20 mL portion of 10.5 M nitric acid is added to the sample to dissolve and digest the ash, and filtered to remove intractable solids (most commonly silicates). The 10.5 M nitric acid solubilizes strontium and yttrium. Care should be taken to preserve or adjust this molarity prior to the filtration of the digest as strontium will start to precipitate at molarities over 12 M [20]. Acid portions are adjusted with fuming nitric acid to ensure a molarity of 14 M nitric acid for the extraction of yttrium into equilibrated TBP. Correspondingly, the amount of TBP used in each extraction has been reduced from 50 to 15 mL.

For ensuring ⁹⁰Y ingrowth correction, the aqueous portion of the first TBP extraction containing strontium and not yttrium is retained, and the start of the shake is the separation time, which is recorded as separation (1). Following separation, yttrium carrier is added to the aqueous portion and the molarity is adjusted to 10.5 M. After some period of ingrowth ranging from 16 h to 15 days, fuming nitric is added to the retained aqueous portion to ensure a molarity of 14 M. This portion is again extracted into another freshly equilibrated TBP portion, and the time is recorded as separation (2). This time, the yttrium portion extracted into the TBP layer is back extracted with water and 3 M nitric acid. Actinides are removed with a fluoride precipitation. Excess hydrofluoric acid is removed with boric acid and a subsequent hydroxide precipitation. Finally, yttrium is precipitated as an oxalate on a filter that is mounted on a disk and counted on a gas proportional counter. Yttrium carrier recovery is determined by gravimetric comparison to an established yttrium carrier standardization with yttrium oxalate.

Calculations

⁹⁰Sr activity calculation

$$A_{\text{Sr}-{90}_{t_{\text{ref}}}}=A_{\text{Y}-90t_{\text{m}}}\frac{{\lambda }_2-{\lambda }_1}{{\lambda }_2\left[e^{-{\lambda }_1\Delta t_1}-e^{-{\lambda }_2\Delta t_1}\right]}{e}^{{\lambda }_2\Delta t_2+{\lambda }_1\Delta t_3}$$ (1)

where

$$A_{\text{Y}-90t_{\text{m}}}=\left[\frac{C_{\text{s}}-C_{\text{b}}}{\left({ϵ}_{\text{s}}\right)\left(Y_{\text{Ys}}\right)\left(Y_{\text{Srs}}\right)\left(T_{\text{s}}\right)\left(W_{\text{s}}\right)}\right]-A_{\text{tmb}}$$ (2)

For comparison, the equation for calculating ⁹⁰Sr activity with the assumption of equilibrium is provided

$$A_{\text{Sr}-{90}_{t_{\text{ref}}}}=A_{\text{Y}-90t_{\text{m}}}{e}^{{\lambda }_2\Delta t_2+{\lambda }_1\Delta t_3}$$ (3)

$A_{\text{Sr}-{90}_{t_{\text{ref}}}}$ = Activity concentration of Sr-90 at the reference time (t_ref).

$A_{\text{Y}-90t_{\text{m}}}$ = Activity concentration of Y-90 at the midpoint of the count (t_m).

C_s = Sample counts

C_b = Background counts

ϵ_s = Counting efficiency for the sample filter = (S * W_yoxs + I)

S = Slope of counting efficiency attenuation curve (cpm/dpm/mg)

W_yoxs = Net yttrium oxalate weight in sample (mg)

I = Intercept of counting efficiency attenuation curve (cpm/dpm)

Y_Ys = Fractional yield of yttrium oxalate = $\left(\frac{W_{\text{yoxs}}}{W_{\text{yoxc}}}\right)$

T_s = The length of the sample count (s).

W_yoxc = Net yttrium oxalate weight in carrier (mg)

Y_Srs = Fractional yield of strontium

W_s = Sample weight (kg)

A_tmb = Typical activity of ⁹⁰Sr in method blank (Bq); (a historical average blank value)

λ₁ = Decay constant ⁹⁰Sr (days⁻¹)

λ₂ = Decay constant ⁹⁰Y (days⁻¹)

Δt₁ = Interval between separation 1 and 2 (the 1st and 2nd separations of ⁹⁰Sr and ⁹⁰Y) (days)

Δt₂ = Interval between separation 2 and midpoint of count (days)

Δt₃ = Interval between reference time and separation 2 (days)

Results and discussion

In this study, infant formula is used as a matrix to investigate a method that uses two separations with a short ⁹⁰Y ingrowth period and calculations described in Wei et al. [13] as a means to reduce the time required to generate accurate ⁹⁰Sr results following a nuclear incident. Dairy products are among the samples most likely to be collected to assess foodborne contamination without certainty that ⁹⁰Sr and ⁹⁰Y are in equilibrium. Milk-based infant formula was chosen as a matrix because it contains the same primary macromolecules as milk, but additionally includes mineral fortification that challenges the extraction method.

In an effort to shorten the sample preparation period and keep the overall analytical time to less than a few days, a rapid ashing procedure was tested to demonstrate the ashing process could be completed overnight (an 8 h linear ramp from room temperature to 600 °C). Twenty different dairy samples, including four samples each of whole milk, skim milk, 2% milk, half and half, and infant formula, were ashed by both the 5 day protocol [19] and the 8 h protocol. The quality of ash obtained from the overnight ashing cycle was compared to those of the 5-day ashing protocol both in terms of visual appearance of the ash and the percent of ash remaining. All samples ashed overnight yielded white or very light ash that was not discernable from the ash obtained from a 5 day ashing cycle and had an average percent ash, between 0.7 and 0.8% for all types of milk and half-and-half, which is consistent with the known value for milk samples (0.7%) [19]. The infant formula ashed by this 8 h protocol resulted in an ash percent of 3.0%, which is comparable to the 3.4% ash content obtained from the samples ashed by the 5 day protocol. These data indicate that the 8 h protocol will provide adequate ash for this application.

Samples consisting of low-level and DIL-level positive controls, negative controls, and reagent blanks were run at intervals between separation 1 and 2 varying from 16 h to 15 days. Data for the higher-level spike, presented in Table 1, show a trueness of + 7.1% for samples with only 16 h between the first and second separation times, and a trueness of − 1.7% for the samples with a separation interval of greater than 1 day. After 24 h, we observed a consistent trueness for all intervals. Conversely, if secular equilibrium were assumed, in error, between ⁹⁰Sr and ⁹⁰Y, and the simple exponential decay law were applied by using Eq. (3), trueness values of − 83.6% and − 37.2% would be obtained for the 16 h and greater than 24-h analyses, respectively.

Table 1.

Measured activity concentrations of ⁹⁰Sr in non-equilibrated systems calculated using Eq. (1) at derived intervention level (DIL) concentration (160 Bq/kg); Spike level is 162.5 Bq/kg

Δt1 (days)	Measured 90Sr (Bq/kg)	Uncertainty ± 2σ (Bq/kg)	% Recovered activitya
0.665	176.2	18.5	108.4
0.667	177.8	18.6	109.4
0.667	173.0	17.5	106.4
0.667	173.6	17.6	106.8
0.667	170.3	18.0	104.8
0.668	177.9	17.7	109.4
0.670	169.5	17.8	104.3
1.000	167.3	13.4	102.9
1.001	172.6	13.4	106.2
1.002	167.8	13.3	103.3
3.719	159.9	10.4	98.4
3.733	159.5	10.3	98.1
3.740	156.5	10.3	96.3
3.802	159.4	10.6	98.1
5.692	156.5	9.9	96.3
5.753	155.2	10.4	95.5
5.754	158.7	10.4	97.6
5.774	161.3	10.8	99.2
7.710	153.9	9.8	94.7
7.752	154.5	10.0	95.1
7.759	159.8	10.4	98.3

^a $\%\text{Recovered}\text{activity}=100\%\frac{\text{Measured}\text{Sr-90}\text{Mactivity}}{\text{Spiked}\text{activity}}$

Measured values for low-level positive controls are provided in Table 2. Data for the low-level spike show a trueness of 1.4% for samples with only 16 h between the first and second separation times and a trueness of − 1.9% for samples with a separation interval of greater than 1 day. When using the assumption of equilibrium on lower-level activity samples with an ingrowth period of 1 day or less, the calculation will result in a 2-sigma uncertainty that is greater than the reported value, which may be considered not detected with 95% confidence.

Table 2.

Measured activity concentrations of ⁹⁰Sr in non-equilibrated systems calculated using Eq. (1) at low-level concentration; Spike level is 6.5 Bq/kg

Δt1 (days)	Measured 90Sr (Bq/kg)	Uncertainty ± 2σ (Bq/kg)	% Recovered activitya
0.667	6.6	0.9	102.1
0.667	6.4	0.9	98.4
0.667	6.7	0.9	103.8
0.667	6.0	0.8	92.7
0.667	6.2	0.8	94.9
0.668	6.4	0.8	99.4
1.000	6.7	0.7	103.0
1.001	6.8	0.6	104.6
1.002	6.4	0.6	99.4
3.073	6.5	0.4	99.7
3.073	6.3	0.4	97.6
3.079	6.3	0.4	97.0
3.081	6.3	0.4	97.8
5.022	6.2	0.4	96.1
5.037	6.0	0.4	93.2
5.037	6.0	0.4	93.3
5.040	6.3	0.4	97.6
6.982	6.3	0.4	97.8
6.982	6.2	0.4	96.1
7.029	6.2	0.4	96.4
7.033	6.5	0.4	100.9
12.985	6.4	0.4	99.4
13.995	6.2	0.4	95.7
14.925	6.5	0.4	99.5

^a $\%\text{Recovered}\text{activity}=100\%\frac{\text{Measured}\text{Sr-90}\text{activity}}{\text{Spiked}\text{activity}}$

Table 3 shows typical relative uncertainties for Eqs. (1) and (2) parameters and the final 1 and 2 s activity concentrations of the lower and higher activity standard with minimal intervals between separation 1 and 2. In addition to the yttrium yield uncertainty calculated as part of the procedure, strontium recovery uncertainty was added to address the possibility of a potential minimal loss of ⁹⁰Sr activity in the separation. For the activity concentration calculation, the strontium yield is assumed to be 100%. An associated uncertainty is assigned at 1.0% (1σ), which is based on the studies of Tinker et al., to account for this issue [21].

Table 3.

Uncertainty budget for the ⁹⁰Sr activity concentration

Source of uncertainty	Standard uncertainty (A) Statistical method (B) Other method	Relative uncertainty (%)
		Low level standard*‡	High level standard**‡
Gross count rate of sample, Rs	σRs, estimated (A)	2.14	0.64
Background count rate, RB	σRB, estimated (A)	23.57	12.13
Counting efficiency of Y-90, eY	σeY, estimated (A)	3.32	3.24
Chemical yield of Y, YY	σYY, estimated (A)	1.62	1.38
Chemical yield of Sr, YSr	σYSr, estimated (B)	1.00	1.00
Duration of the count, TS	σTS, estimated (B)	0.00	0.00
Sample weight, ws	σWS, estimated (B)	0.33	0.33
Typical matrix blank activity, Atmb	σAtmb, estimated (A)	61.21	61.21
Net Y-90 activity at measurement, AY-90	σAY-90, estimated (A)	4.63	3.79
Sr-90 decay constant, l1	σl1, estimated (B)	0.20	0.20
Y-90 decay constant, l2	σl2, estimated (B)	0.30	0.30
Interval between sep 1 and 2, Dt1	σDt1, estimated (B)	0.96	2.94
Interval between sep 2 and midpoint of count, Dt2	σDt2, estimated (B)	6.76	5.86
Interval between ref time and sep 2, Dt3	σDt3, estimated (B)	0.00	0.00
Relative total combined standard uncertainty of AY-90, %		9.82	4.12
Relative expanded uncertainty of AY-90 at k = 2, %		19.6	8.2

^*Values are used for a Dt₁ of 16 h (L01)

^**Values are used for a Dt₁ of 1 day (L41)

^‡σ terms provided here do not include sensitivity factors. For full uncertainty calculation please refer to supplementary information

The minimum detectable activity concentration (MDC) and LOQ are determined using the average gross counts for the negative controls (μ_B ± σ_B). The uncertainty in the counts is determined as a standard deviation and is not equivalent to the square root of the average counts. The LOQ calculated using Currie’s equation [22], whether using the positive or negative control samples, is 0.5 Bq/kg

Conclusions

The data presented demonstrate that this method may be used to provide accurate and precise results for ⁹⁰Sr within a few days of sample receipt without the assumption of equilibrium or the assumption of the absence of ⁸⁹Sr. This is critically important because rapid analysis and dissemination of results to decision-makers are crucial for consequence management following a nuclear incident. The modified extraction procedure is more environment friendly, and the results obtained were within ± 10% accuracy. The dual separation scheme and mathematical solutions could be employed with other extraction methods.

Disclaimer

The views expressed in this document are those of the authors and should not be interpreted as the official opinion or policy of the U.S. Food and Drug Administration, Department of Health and Human Services, or any other agency or component of the U.S. government. The mention of trade names, commercial products, or organizations is for clarification of the methods used and should not be interpreted as an endorsement of a product or manufacturer.