Comparison of Electrochemical Nitric Oxide Detection Methods with Chemiluminescence for Measuring Nitrite Concentration in Food Samples

Abstract

Nitrite is a naturally occurring species present in various food samples and also present in our bodies as a product of nitric oxide (NO) oxidation. Considering the ubiquity of nitrite, its determination is of great importance in both biological and food samples. Herein, a very facile indirect method of nitrite determination in meat samples via selective reduction to nitric oxide (NO) is presented. The resulting gaseous product is quantified via portable and cost-effective electrochemical sensors. Both a novel laboratory prepared Pt-Nafion based NO sensor and a commercially available amperometric NO sensor are compared. Excellent correlations between the nitrite amount found in tested samples using both of the electrochemical sensors and a reference chemiluminescence method are demonstrated (r = 0.997 and r = 0.999 for Pt-Nafion based and commercially available NO-B4 electrochemical sensors, respectively, n = 12). Moreover, the slope of the linear regression curves are very close to unity for the comparison of the three systems tested. The amperometric sensors compared within this work exhibit good precision and accuracy and are shown to be an attractive alternative to the costly chemiluminescence detection method for accurately determining nitrite levels in food samples.

Graphical Abstract

1. Introduction

There is a great interest in detecting nitrite since it plays an important role in many fields. Initially, nitrite, along with nitrate, was considered as an inert end-product of nitric oxide (NO) metabolism/oxidation or a redundant residue in the food chain. However, further discoveries have shown that nitrite can be reduced in blood and tissue via several enzymatic and nonenzymatic pathways to form NO, a pluripotent biological messenger [1–3]. Hence, nitrite has been recognized as a storage pool for endogenous NO and also a potential therapeutic agent with cytoprotective [4–6] and blood-pressure-lowering effects [7–9].

For many years, sodium nitrite has also been successfully employed as a corrosion inhibitor [10,11]. Moreover, nitrite and nitrate salts have been used as a food preservative to cure meats and fish since the antiquity [12]. Nitrite not only inhibits the growth of pathogens, most notably Clostridium botulinum [13,14], but also develops flavor and color in certain food products. At the same time, nitrite can also react with secondary amines to form carcinogenic N-nitrosamines [15]. Nitrite is also capable of oxidizing ferrous iron in hemoglobin to the ferric state forming methemoglobin that cannot bind oxygen and this can lead to the development of methemoglobinemia in infants [16]. Since too high of a nitrite intake can be harmful, strict regulations have been put in place to limit the amount of nitrite in our food. The U.S. Department of Agriculture has established regulatory limits for the addition of sodium nitrite at 120 ppm, 200 ppm, and 650 ppm for bacon, ham or pastrami, and dry-cured meat products, respectively [17]. Moreover, since ground waters contain inorganic nitrite (mainly from the fertilizers used) that may get into our drinking water, the maximum contaminant level of 1 ppm for nitrite (measured as nitrogen) in public water supplies has been set by the Environmental Protection Agency [18].

Several methods have been developed to quantitatively determine the concentration of nitrite in various biological, environmental and food samples. These include spectroscopic [19–21], chromatographic [22,23], electrophoretic [24], and electrochemical methods [25–28]. By far, the spectrophotometric methods are the most widely used for nitrite determination due to their low detection limits and simple protocols. The most common method is the Griess Assay which relies on the diazotization of a suitable aromatic amine (e.g., sulfanilamide) by acidified samples containing nitrite, combined with a subsequent coupling reaction resulting in a colored azo compound [29,30]. The presence of antioxidants and/or highly colored and heterogenous matrices of complex food or biological samples can significantly affect the accuracy of this method. Such issues can be overcome by employing electrochemical methods that utilize either an amperometric/voltammetric or potentiometric detection mode. Several Co(III) [31,32] and Rh(III)-ligand [25] complexes have been successfully applied as nitrite selective ionophores in ion-selective polymeric membrane electrodes that enable nitrite detection in various real samples [25,32]. It has been shown that more lipophilic anions such as salicylate, perchlorate, nitrate, acetate, and thiocyanate can be discriminated to a great extent.

Nitrite can also be directly oxidized or reduced at a working electrode surface. The oxidation of nitrite is usually preferred over reduction since the latter one is more prone to interference from other readily reducible species such as nitrate ion or molecular oxygen. Various electrodes materials have been exploited for nitrite reduction including platinum, gold, copper, transition metal oxides, boron-doped diamond (BDD), and glassy carbon (GC); however, the kinetics of the charge transfer is slow. This results in poor sensitivity and reproducibility mainly due to cumulative electrode passivation effects. In addition, the need to apply large overpotentials can considerably affect the selectivity of this approach. To solve these problems, working electrode surfaces have been modified with suitable catalysts such as metallophthalocyanines and metalloporphyrins [33–35], enzymes with appropriate electron donors [36], metal and metal oxide nanoparticles [37,38] and carbon nanotubes [27,39]. Direct electrochemical determination of nitrite is very facile; however, it can still be susceptible to interferences from components in complex samples.

In contrast, indirect detection of nitrite by reductive conversion to NO gas that can be purged into a gas phase detection system can offer greater scope for the elimination of matrix effects. In this configuration, non-volatile interfering species are effectively separated from the analyte of interest. Moreover, tedious and complex sample pretreatment is not required, since the sample’s color and turbidity do not affect the measurement. Nitrite can be efficiently converted to NO using either ascorbic acid or iodide-based reducing solutions. The resulting NO can be then determined using a number of sensitive methods. Chemiluminescence detection is the most often employed to measure NO levels in the gas phase. It is highly sensitive, selective, and provides a wide dynamic linear range, typically from 0.5 ppb to 500 ppm NO. In addition, the commercially available chemiluminescence-based nitric oxide analyzers offer real-time monitoring essential for in-situ experiments. However, these NO analyzers are very expensive, and require an ozone generator as well as frequent equipment maintenance. Thus, we have decided to focus on the use of simple and cost-effective electrochemical gas-phase NO sensors that have been previously developed in our group [40–42], and this attractive approach of nitrite detection is demonstrated in this paper. The levels of nitrite in food samples are assessed using a laboratory prepared amperometric Pt-Nafion gas phase sensor [41] and a commercially available electrochemical gas phase NO sensor (i.e., Alphasense, NO-B4). Additionally, the values measured with both electrochemical sensors are compared to those obtained for the same food samples using the gold-standard chemiluminescence NO detection method [43].

2. Experimental

2.1. Materials and reagents

Nitric acid (65%), sulfuric acid (95–98%), sodium hydroxide (≥98%), sodium borohydride (≥98%), potassium iodide (≥99%), tetraammineplatinum(II) chloride hydrate (98%), sodium nitrite (≥99%), sodium phosphate monobasic monohydrate (98–102%) and sodium phosphate dibasic heptahydrate (98–102%) were purchased from Sigma-Aldrich (St. Louis, MO) and used as received. Nafion 117 sheets were obtained from FuelCellsEtc (College Station, TX). Ultrapure water from a MilliporeMili-Q system (Millipore, Bedford, MA) was used to prepare all the aqueous solutions. The homemade gas sensor cell assembly was prepared by Glassblowing Services, Department of Chemistry, University of Michigan (Ann Arbor, MI). Food samples were selected for their known presence of NO₂⁻ and obtained from local grocery markets.

2.2. Gas phase sensors

The homemade amperometric gas phase NO sensor was fabricated according to the protocol included in our previous publication [41]. Briefly, the Pt-Nafion working electrode was prepared using the impregnation-reduction method. A 1.6 cm dia. circles of Nafion 117 were cut out from the sheet and cleaned of impurities by successive boiling in 3 M nitric acid for 1 h and deionized water for 1 h. Next, the Nafion membrane was mounted between two glass cells with 0.94 cm dia. openings and one side of the ionomer membrane was exposed to 2 mM Pt(NH₃)₄Cl₂ at 37°C for 24 h. After that time, the platinum solution was removed, and 50 mM NaBH₄ in 1 M NaOH was placed at the same side of the membrane. The platinum reduction onto/into the Nafion membrane proceeding according to the equation below was carried out at 37°C for 60 min.

$$4{\left[\text{Pt}{\left({\text{NH}}_3\right)}_4\right]}^{2^{+}}+{\text{NaBH}}_4+8{\text{OH}}^{-}\to {\text{NaBO}}_2^{-}+4{\text{Pt}}^{\mathrm{o}}+16{\text{NH}}_3\uparrow +6{\text{H}}_2\text{O}$$ (1)

To remove any remaining Pt complex and the reducing agent, the Pt-Nafion membrane was boiled in DI water for 1 h. The final prepared membrane was mounted in a glass sensor assembly with the metallic side of the electrode facing the gas phase, and the other side was exposed to 0.5 M sulfuric acid inner filling solution. A reference electrode consisting of a single junction Ag/AgCl saturated with KCl, and a platinum wire that served as the counter electrode were placed within the IFS chamber. The lead of the working electrode was made of a 10 mm × 2 mm piece of 50 μM thick gold foil secured between the Pt-Nafion based electrode and the gas inlet/outlet section of the sensor. Prior to measurements, to clean the electrode/electrolyte interface, the electrode’s potentials were scanned between −0.3 and 1.5 V for 20 cycles at a scan rate of 20 mV/s under a nitrogen atmosphere at a flow rate of 200 mL·min⁻¹. To measure NO levels resulting from nitrite reduction in acidic conditions, 1 V vs. Ag/AgCl_sat’ed was applied to the working electrode and the sensor’s output current was recorded using a CHI800 potentiostat (CH Instruments, Austin, TX).

A commercially available Alphasense (Alphasense, Great Notley, UK) electrochemical gas phase NO sensor (model NO-B4) was also evaluated for nitrite in food measurements. This amperometric sensor is described in a series patents [44,45] and requires application of 0.2 V to the working electrode vs. and onboard reference electrode.

2.3. Determination of nitrite in food samples

For the nitrite determination in various meat samples, the extraction procedure adopted from Pietrzak and Meyerhoff was used [25]. Ten grams of a representative sample of a finely chopped meat were added to 80 mL of 50 mM phosphate buffer (pH 7.4) and homogenized with a hand blender. The mixture was then transferred to a beaker and heated to 80°C in a water bath for 30 min. Subsequently, the solution was cooled at room temperature and filtered using filter paper of medium texture. The resulting sample solution was injected into the reaction cell containing an equal volume of 0.6 M potassium iodide and 0.25 M sulfuric acid. The solution in the reaction cell was continuously purged with nitrogen at a flow rate of 50 mL min⁻¹ using MC-200SCCM mass flow controller (Alicat Scientific, Tuscon, AZ). The gaseous sample was analyzed downstream via the homemade Pt-Nafion based amperometric sensor or the commercially available Alphasense NO-B4 gas sensor equipped with individual sensor board, or a Sievers chemiluminescence Nitric Oxide Analyzer® (NOA) 280i (GE Analytical, Boulder, CO). The paired Student’s t-test (two-tailed) was used to compare nitrite levels in the sample solutions obtained using each sensor and the chemiluminescence reference method. Values of p < 0.05 were considered statistically significant for all tests.

3. Results & discussion

The interfering species present in the real samples pose a challenge to all analytical methods, including those typically employed for nitrite measurements. The selective analyte conversion to the gaseous species (NO in this case) and the subsequent determination in a gas phase instead of in the liquid phase is one approach that can minimize interferences. Herein, we have decided to utilize this facile approach in combination with the electrochemical detection method to determine nitrite in food samples. Processed meat products i.e., ham, hot-dog, salami, and bologna sausage were chosen as samples. The samples were prepared according to the previously described procedure [25], and the concentration of nitrite was assessed based on the amount of NO released to the gas phase. In order to do so, nitrite present in standards and samples was quantitatively reduced to NO by potassium iodide under acidic conditions. The reducing mixture employed is considered to be selective for nitrite reduction. It reduces NO₂⁻ to NO, but it is not able to reduce nitrate (NO₃⁻) to NO, and since nitrate is also present at high levels in the meat products, the approach yields high selectivity for nitrite over nitrate. Sample solutions were injected in triplicate to the reaction vessel, and the reducing solution was continuously purged with a carrier nitrogen gas to facilitate the NO transfer to the gas phase. The schematic of the configuration employed in this study using the homemade electrochemical Pt-Nafion gas phase sensor is presented in Figure 1. An analogous configuration was employed when using the commercial electrochemical Alphasense NO-B4 sensor or the chemiluminescence analyzer as the gas phase NO detector.

Figure 1.

The sensing configuration employed for the determination of NO obtained by nitrite reduction. All three detection methods used for this purpose are disclosed.

Both electrochemical sensors utilized in this work exhibit fast response time, t₉₀ < 5 s and t₉₀ < 45 s (from 0 to 2 ppm) for the homemade Pt-Nafion and Alphasense NO-B4 devices, respectively. The limits of detection (based on 3xSD obtained for the blank/slope) of 3.8 ± 0.8 ppb (n = 5) for the Pt-Nafion, and 9.8 ± 0.4 ppb (n = 3) for NO-B4, were determined while the sensors were operated with a 200 sccm gas flow rate, and these values are in good agreement with the literature data presented for amperometric gas phase NO sensors [40,46]. The Pt-Nafion sensor, however, has better sensitivity compared to the NO-B4 sensor. The typical slope of the calibration curve for the Pt-Nafion sensor operating in the diffusion/mass transport-controlled region was in the range of 1100 – 1250 pA/ppb while for Alphasense NO-B4 sensors slope values from 500 to 850 pA/ppb have been reported. The origin of this difference is related to the difference in area of the working electrode and the presence of a gas diffusion barrier above the working electrode in the case of the commercial NO-B4 sensor. In the presence of the diffusion barrier, the effect of the flow rate on the sensitivity of the amperometric senor is generally limited [47,48]. On the other hand, for the Pt-Nafion sensor, the working electrode directly faces the gas sample stream; hence, essentially no diffusion barrier exists. In such an arrangement, there is a significant dependence of the sensor sensitivity on the gas flow rate [40,47]. Consequently, this feature can be used to improve the senor’s response to some extent. Nonetheless, the potential value applied to the working electrode of both sensors is similar; thus, we presume that they should exhibit comparable selectivity over interfering species such as CO, NH₃, CO₂, NO₂, VOCs, etc. that could be oxidized at the high potentials required for NO oxidation.

The Pt-Nafion and Alphasense NO-B4 sensors responses to the hot-dog sample solution and nitrite standard solution injections are presented in Figure 2.A and Figure 3, respectively. As can be seen, fully reversible and reproducible signals were obtained using both electrochemical sensors. Moreover, a stable baseline during the course of the experiments was observed suggesting that the electrode surface was not poisoned by volatile sample components in either case. The area under the electrochemical response peak was linearly proportional to the nitrite concentration (see Figure 2.B), and this dependence was used to calculate the nitrite content in the unknown sample solutions. For each sample type tested, three trials were conducted using both electrochemical sensors, and the values of nitrite determined based on prior calibration of each sensor with nitrite standards are summarized in Table 1. Additionally, to determine the nitrite concentration in the selected meats using a reference method, the chemiluminescence method was employed, and the obtained results are also listed Table 1. For the purpose of the comparison with previous literature data, the nitrite concentration determined in food samples were also converted to mg of NO₂⁻ per kg of meat (ppm).

Figure 2.

A) Pt-Nafion based amperometric sensor response to 250 μL of (a) sample solution (hot dog), and nitrite standard solutions (b) 10⁻⁵ M, (c) 10⁻⁴ M, and (d) 5 × 10⁻⁵ M injected to the reaction cell. B) Corresponding calibration curve used to calculate nitrite content in the hot-dog sample.

Figure 3.

Alphasense NO-B4 response to 250 μL of (a) 10⁻⁵M nitrite standard solution and (b) hot dog sample solution injected into the reaction cell.

Table 1.

The results of nitrite determinations in meat samples from local grocery stores using two electrochemical detection methods and chemiluminescence reference method. The reported results are average ± SD for n = 3 injections for each sample.

		Nitrite concentration (μM)			Nitrite content (mg of NO2−/kg of meat)
		NOA	Pt-Nafion	NO-B4	NOA	Pt-Nafion	NO-B4
Hot-dog (wiener)	sample 1	30.75 ± 0.69	30.16 ± 0.43	28.56 ± 0.44	11.32 ± 0.25	11.10 ± 0.16	10.51 ± 0.16
	sample 2	31.87 ± 0.42	31.68 ± 0.18	30.42 ± 2.76	11.73 ± 0.15	11.66 ± 0.07	11.19 ± 1.02
	sample 3	28.56 ± 0.61	28.64 ± 0.50	27.98 ± 0.17	10.51 ± 0.22	10.54 ± 0.18	10.30 ± 0.06
Ham	sample 4	8.16 ± 0.09	6.21 ± 0.06	7.84 ± 0.21	3.00 ± 0.03	2.29 ± 0.02	2.89 ± 0.08
	sample 5	6.58 ± 0.06	7.13 ± 0.14	6.69 ± 0.09	2.42 ± 0.02	2.62 ± 0.05	2.46 ± 0.03
	sample 6	8.26 ± 1.60	9.64 ± 0.45	7.85 ± 0.13	3.04 ± 0.59	3.55 ± 0.17	2.89 ± 0.05
Salami	sample 7	10.73 ± 0.03	10.89 ± 0.19	10.80 ± 0.14	3.95 ± 0.01	4.01 ± 0.07	3.97 ± 0.05
	sample 8	7.61 ± 0.16	6.80 ± 0.61	7.81 ± 0.32	2.80 ± 0.06	2.50 ± 0.22	2.87 ± 0.12
	sample 9	7.61 ± 0.09	6.98 ± 0.16	7.62 ± 0.04	2.80 ± 0.03	2.57 ± 0.06	2.80 ± 0.01
Bologna	sample 10	27.58 ± 0.36	26.38 ± 0.37	26.45 ± 0.02	10.15 ± 0.13	9.71 ± 0.14	9.73 ± 0.01
	sample 11	28.99 ± 0.12	27.92 ± 1.10	28.36 ± 0.26	10.67 ± 0.04	10.27 ± 0.40	10.44 ± 0.10
	sample 12	27.18 ± 0.68	26.41 ± 0.16	26.79 ± 0.36	10.00 ± 0.25	9.72 ± 0.06	9.86 ± 0.13

As can be seen, slight variations in the nitrite level were observed for a given sample type in different trials; however, similar concentration values were obtained using all three detection methods. Thus, those variations are most likely attributed to the nitrite extraction procedure or a heterogeneous composition of the meat samples employed. For this reason, to compare the performance of the electrochemical sensors, we decided to analyze each sample individually. Good correlation was found for measurements with the Pt-Nafion based sensor vs. NO-B4 results as evidenced by the high value of the Pearson correlation coefficient, r = 0.996 (n = 12). Furthermore, when the nitrite levels obtained using Pt-Nafion based and NO-B4 sensors were compared to the concentrations determined with the reference method, even higher correlation was found, r = 0.997 and 0.999 (n = 12), respectively (see Figure 4) demonstrating an excellent precision for the amperometric sensors. The Student’s t-test revealed that statistically there is no difference between the results obtained with both electrochemical sensors and between the values determined with the Pt-Nafion gas sensor and the chemiluminescence detection method. The corresponding calculated t-values were 0.458 and 1.651, and those values are clearly smaller than t-critical = 2.201 (two-tailed test with 11 degrees of freedom, and the significance level of α = 0.05). However, when the nitrite concentration in the sample solutions determined with the commercially available NO-B4 sensor were compared to the nitrite amounts assessed with the reference method, a higher t-value of 2.706 was found. Moreover, the p-value was smaller than the chosen significance level. On this basis, the null hypothesis that there is no difference between the nitrite levels obtained using those methods had to be rejected. This is mainly caused by the differences in nitrite content obtained for the hot-dog samples. When the values assessed for this sample type were not taken into account, the requirements for accepting the null hypothesis of the Student’s t-test were met.

Figure 4.

Comparison of nitrite concentrations obtained using A) Pt-Nafion based and Alphasense NOB4 amperometric sensors, B) Pt-Nafion based sensor and chemiluminescence method, and C) Alphasense NO-B4 sensor and chemiluminescence method using least-square linear regression analysis. The nitrite levels found in () hot-dog, () ham, () salami, and () bologna samples. Error bars represent SD obtained for 3 injections.

As shown in Figure 4.C, the slope of the linear regression curve was 0.9497 ± 0.0128 indicating that the commercial NO-B4 sensor, despite exhibiting exquisite precision, on average yields slightly lower values than the chemiluminescence analyzer. The homemade Pt-Nafion gas sensor also provided slightly lower values (slope = 0.9829 ± 0.0246). Nonetheless, the slope values of the linear regression curves for both sensors are close to 1, demonstrating high accuracy of both amperometric sensors relative to the gold-standard chemiluminescence method. Furthermore, the data also proves that there were no volatile interfering species in the test samples that contribute to the measured oxidation current for the electrochemical sensing methods. The very small deviations could be attributed to the higher detection limit of the electrochemical sensors, and their slower response time compared to the chemiluminescence analyzer. This further supports the idea that the conversion of the analyte of interest into the volatile product and its detection in the gas phase, instead of the liquid phase, can greatly eliminate matrix effects and potential interferent species.

The agreement between the electrochemical and the reference method was also verified by Bland Altman analysis plots. The difference of the nitrite levels in meat samples obtained with either Pt-Nafion or NO-B4 amperometric sensor and the chemiluminescence detection method were plotted against the mean of the two paired measurements (Figure 5). Normal distribution of the differences was confirmed using Shapiro-Wilk test with a significance level α = 0.05. As shown in Figure 5.A, a random variability with a bias of 0.42 μM and 0.56 μM was obtained for Pt-Nafion based and NO-B4 sensors, respectively. At the same time, a wider agreement range was obtained in the first case. The difference of +2.15 or −1.31 units is more important for lower nitrite levels, while it is not that significant for higher concentrations. This is better displayed in Figure 5.B where the differences are plotted as a percentage of the concentration. The bias is 2.95%, and the 2xSD agreement range is ± 20.33%, predominantly caused by the measurements of lower nitrite concentrations. For the samples containing higher nitrite levels, the 2xSD agreement range is < 3%. On the other hand, the difference plot obtained for the commercial NO-B4 sensor displays a moderate positive trend of differences, proportional to the magnitude of the measurement. The narrower agreement range from −0.84 to 1.96 units is equally significant for high and low nitrite levels. This can be clearly seen once the percentage of the concentration is evaluated. The bias is 2.16%, and the 2xSD agreement range is ± 5.92%. The above analysis indicates that there is a good agreement between both of electrochemical methods and the chemiluminescence detection method.

Figure 5.

Bland Altman plots for nitrite concentrations obtained using the reference chemiluminescence method and the values obtained with the amperometric senor. Plots of differences expressed A) in units of concentrations i.e., μM, and B) as a percentage of the values. The solid line represents the bias of the method and the dotted line the limits of agreement from −1.96 SD to +1.96 SD. () Pt-Nafion gas sensor () commercial NO-B4.

4. Conclusion

Indirect detection of nitrite level in food samples via selective chemical reduction to NO and subsequent determination of the gaseous product by electrochemical NO sensors is a viable approach to eliminate the effect of the sample matrix and achieve high selectivity. It has been demonstrated that amperometric NO sensors are a competitive alternative to chemiluminescence detection method for such in-situ NO monitoring. A very high correlation between the results obtained with all three detection methods used within this work was shown. The best correlation was found for the nitrite concentrations determined using the commercially available Alphasense NO-B4 sensor and the nitrite levels obtained with the chemiluminescence detection method. The data show that the Pt-Nafion based senor, despite somewhat lower precision than the commercial NO-B4 sensor, exhibited a higher degree of accuracy for the nitrite measurements in food samples. Additionally, a good agreement between the nitrite concentrations found with amperometric sensors and the reference method was confirmed by Bland Altman analysis. The presented comparison of results with the gas phase NO sensors proves that they can serve as a very useful detection method for indirect measuring of nitrite in food samples such as processed meats, thus broadening the ever-expanding list of applications for such electrochemical devices.

Highlights.

• A facile method of nitrite detection in meat samples via selective reduction to NO.

• Excellent correlation between the amperometric sensors and the chemiluminescence method.

• Good precision and accuracy of the gas phase electrochemical NO sensors.

• Detection in the gas phase efficiently eliminates the interferent species.