2.1 Reagents

All chemicals were of analytical reagent grade unless otherwise specified. Deionized water (ASTM type I) was obtained from a Millipore (Milestone, Sorisole, Italy) Direct-Q 5 water purifier.

Standard atomic stock solutions (Merck, Darmstadt, Germany) of 1000 mg L⁻¹ of Cd and As were employed.

Pb stock solution was prepared from Pb(NO₃)₂ 99.99 % (Sigma Aldrich, USA).

Working standard solutions were prepared daily by dilution of the stock solution with 4:21 HNO₃: H₂O.

Palladium modifier solution (500 mg L⁻¹) was prepared by dilution with 0.1% v/v HNO₃ of palladium matrix modifier for graphite furnace, 10 g L⁻¹ (Merck, Darmstadt, Germany).

Ammonium dihydrogen phosphate modifier solution (5 g L⁻¹) was prepared from (NH₄)H₂PO₄ matrix modifier (Merck, Darmstadt, Germany) for graphite furnace, 99.99 % in deionized water. Nickel nitrate modifier solution 10 g L⁻¹ was prepared from Ni(NO₃)₂ (Fisher Scientific, Hampton, USA) in deionized water.

Sodium tetrahydroborate solution (3.0 % (w/v) in sodium hydroxide 0.05 mol L⁻¹) was prepared from sodium tetrahydroborate (Hydride-Generation grade Fluka, NY, USA).

Certified reference material (CRM) 1568a Rice flour was obtained from NIST.

High purity argon (Ar) (Praxair, Montevideo, Uruguay) was employed.

2.2 Instrumentation

Dried and grinded samples were digested in 100 mL high-pressure closed vessels EasyPrep Plus, using a microwave oven (Mars 6, CEM Corporation, Matthews, NC, USA).

Cd and Pb determinations both in potato and sweet potato samples, and As determination in sweet potato samples were performed in a longitudinal heated, electro-thermal atomic absorption spectrometer (Perkin Elmer HGA 900, Shelton, USA) fitted with deuterium background corrector. Atomization took place on solid pyrolytic graphite L'vov platform inserted in standard pyrocoated graphite tubes. Cd electrodeless discharge lamp (210 mA, 228.8 nm, Perkin Elmer precisely, Shelton, CT USA), As electrodeless discharge lamp (400 mA, 193.7 nm, Perkin Elmer precisely, Shelton, CT USA) or Pb hollow cathode lamp (10 mA, 283.3 nm, Perkin Elmer Lumina, Singapore) were used as appropriate.

For As determination in potatoes a nitrogen microwave induced plasma atomic emission spectrometer 4210 (MIP OES, Agilent Technologies, Santa Clara, USA) equipped with a standard torch and a multimode spray chamber (MSIS, Agilent) was used. Nitrogen (99.5 %) was supplied by a nitrogen generator model 4107 (Agilent Technologies, Santa Clara, USA) coupled with an air compressor model KK70 TA-200 K (DürrTechnik, Bietigheim-Bissingen, Germany).

2.3 Sampling and sample preparation

Sampling was carried out at Metropolitan Agrifood Market²³ (UAM, formerly Mercado Modelo of Montevideo). The UAM is the main wholesale logistic/distribution center for fruit and vegetables trade and distribution in Uruguay, supplying more than 60% of the vegetables consumed in the country²⁴. The fresh vegetables enter to the market directly from the producers and from there they are distributed to greengroceries, warehouses, self-services, supermarkets, hospitals, etc. Sampling was performed with randomly selected traders in coordination with UAM management before the opening of daily operations (between 4:00-6:00 AM). In this way, the completely random sampling was representative of the distribution of the products to the country. This allowed the monitoring of contaminants in vegetables that will be offered in a wide range of retail sales and human consumption points. Approximately 1 kg (5 pieces) of each vegetable were placed in plastic bags, sealed and labeled. Samples were transported to the processing lab within 1 hour of collection. Samples were collected between 2018 and 2020 at two times of the year, in February and between July and August. The sweet potatoes were from Salto and Canelones departments, while the potatoes were from San José and Canelones.

Potatoes and sweet potatoes were chopped with a ceramic knife and dried in an air-forced oven, Model DN93 (Yamato, Tokyo, Japan) set at 70 °C. After 96 hours, the weight was recorded for water content calculations and samples were grinded in a Model 4 Wiley mill (Thomas Scientific, NJ, USA), with a 1 mm sieve, and later placed in 50 mL Falcon tubes and kept in dried conditions until analysis. Three pooled samples of whole potato and sweet potato, 5 pooled samples of peeled potato and 11 pooled samples of peeled sweet potato, 2 of them of orange pulp, were obtained to be analyzed. All the samples of potato collected were of white/cream pulp and red/purple skin.

The number of pooled samples obtained was determined by the availability of each product on the sampling date. The traceability of the samples was assured in order to analyze the individual samples if a result above the MAL was obtained.

Prior to analytical measurements, 0.5 g of dried and grinded samples were treated with 8 mL of HNO₃ 1:1, let stand for 10 min and digested with microwave assistance. The heating program consisted of a 20 min ramp up to 120 °C, holding at that temperature for 20 min, followed by a 20 min ramp from 120 °C to 170 °C and remaining 15 min at that temperature. The remaining solution was diluted with ultrapure water to 25 g. For each element, the determinations (including reagent blanks) were carried out by triplicate.

2.4 Analytical determinations

The determination of Cd and Pb in potatoes and sweet potatoes and of As in sweet potatoes was performed by electrothermal atomic absorption spectrometry (ETAAS).

Working conditions are described in Table 1.

Table 1: Temperature programs for As, Cd and Pb determinations by ETAAS 

μL digest / μg matrix modifier: As, 20/ 100 Ni(NO₃)₂; Cd, 10/ 5 Pd; Pb, 20/ 50 NH₄H₂PO₄ (a cool down step until 200 °C holding for 30 s was added).

The As determination in potatoes was performed by hydride generation nitrogen microwave induced plasma atomic emission spectrometry (HG - MIP OES). Prior to the determination, a pre-reduction step with 20 % (w/v) KI and 37 % HCl in the ratio, digested sample, or standard (10): HCl (1): KI (1) at room temperature for 1 hour had to be carried out. Both the reduced digested sample and sodium tetraborate entered to the multimode spray chamber at 0.90 mL min^-1 (pump speed 30 rpm). After 20 s of stabilization of the system, the hydridegenerated was carried towards the atomizer by a 0.7 L min^-1N₂ flow. The determination was performed at 193.695 nm with automatic background correction. The viewing position was 10 (respect to the centre of the torch) and the reading time was 10 s (read by duplicate).

Analytical methods were validated according to the Commission Decision 2002/657/EC criteria and their figures of merit were compared to the AOAC requirements²⁵⁾⁽²⁶.

2.5 Validation

The following figures of merit were evaluated: linear range, precision under repeatability conditions, trueness, and detection (LOD) and quantification limits (LOQ).

Linearity was evaluated by means of a 6-point calibration curve. Quality of the linear fit was verified by both visual inspection and assessment of the value of the coefficient of determination (R²)²⁶. Precision (repeatability) was evaluated, for As and Cd determination, by analyzing a certified reference material (CRM NIST 1568a Rice flour; n=6). For Pb it was assessed by measuring a sample previously spiked with Pb, at the MAL level of 0.1 mg kg^-1 (n=6).

Detection and quantification limits (3.3 s and 10 s criteria, respectively) were estimated by measuring 10 blanks²⁷. Trueness was evaluated by analyzing a certified reference material (CRM) NIST 1568a Rice flour for As and Cd, and as this CRM did not inform a certified value for Pb concentration, trueness in this case was evaluated with 6 samples previously spiked at the 0.1 mg kg ^-1 level.

3.1 Analytical determinations

The AOAC requirements for an analytical method to be consider a reference method for heavy metals monitoring purposes establishes that LOQ should be ≤ 0.010 mg kg^-1(25. This requirement was not fulfilled in the case of As (0.011 mg kg ^-1) and Pb (0.018 mg kg ^-1). Nonetheless, it was considered that the limits of quantification achieved widely satisfy the needs for their application in potatoes and sweet potatoes. The LOQ´s are more than 5 times below the MAL for Pb and almost 20 times below the MAL for As, and this was considered appropriate. Besides this, the obtained LOQ values are also better than those reported in several published studies^28)(29)(30.

In the case of As determination, when it was determined by ETAAS, matrix incidence in the response was different for potatoes and sweet potatoes. Interference abatement was unsuccessful in potato samples, thus for As determination in that matrix, HG-MIP OES was used instead, reaching suitable limits of detection and quantification.

With regard to the other figures of merit, all of them fulfilled the performance requirements established by the AOAC, as it can be observed in Table 2.

Finally, the simple instrumentation required by these methods is an important advantage that makes the wide application of them possible. Moreover, the use of MIP OES does not need expensive gases such us argon or acetylene as other atomic spectrometric techniques, making it a suitable option for monitoring purposes.

Following the fully described methods, any laboratory with the adequate instrumentation could implement a reliable analytical platform for food safety monitoring taking advantage of the suggested aspects, that are alternatives to standard methods, which make them simpler, more economical and in accordance with the principles of green chemistry. These characteristics make them a powerful tool for monitoring the presence of these contaminants in developing countries where these products are increasingly important.

3.2 Inorganic contaminants concentration

The inorganic contaminants content was in most of the cases more than three times below the MALs.

As was not quantified in any sample; Pb was found both in peeled and whole samples, and, as it was mentioned before, it could be related to environmental contamination, and Cd was present in all potato samples. Cd accumulation in potato has been reported due to the use of phosphate fertilizers³¹. This kind of fertilizers is widely used worldwide and their misuse may be one of the main causes of the presence of Cd.

However, in sweet potatoes, Cd and Pb are present in the whole samples analyzed but not in the peeled ones, and the same behavior appears in most of the analyzed samples, indicating that the determination in the whole product could be a better indicator of their safety. As it was mentioned before, the revalorization of the skin of these vegetables contributes to the tendency of consuming them unpeeled. Due to this fact, we consider that the analysis of the whole food gives valuable information that needs to be considered to perform a risk assessment which contemplates the trend of consumption of these foods unpeeled.

Regarding the sampling plan, despite our study involved few samples, each pooled sample represents a random sampling of several producers in the country. We propose that pooling samples from wholesales market site, when traceability is assured (site of production, irrigation conditions, soil type), could be used as an alternative way to sampling in farms or retail shops. In this way, if any contaminant is found above the MAL in the pool, the individual samples could be analyzed, and using survey information more tests could be performed at production site (soil and irrigation water).