Introduction

Vegetables are an excellent source of vitamins, minerals, and biologically active compounds (Favell 1998; Kidmose et al. 2001). However, vegetables are one of the main sources of entering nitrite to our bodies (Favell 1998; Kidmose et al. 2001). Nitrates are commonly abundant in food primarily because plants take up nitrogen from the soil in this ionic form. Then, nitrates can be reduced to nitrite in foods because of the action of some bacteria (Meah et al. 1994).

Nitrite ion is commonly monitored in agriculture and food control. Nitrite is an important intermediate in the biological nitrogen cycle and is present in soils and surface waters. Nitrite leads to the production of carcinogenic nitrosamines when it reacts with secondary amines in the stomach. In addition, nitrite is known to cause methemoglobinemia (oxygen deficiency) in infants (Seike et al. 2004). Excess concentration of nitrite in drinking water is hazardous to health, especially to pregnant women (Kazemzadeh and Ensafi 2001). Therefore, simple and sensitive methods are required to provide a way for monitoring the legal recommended limits of this ion in agricultural samples and foods.

The most commonly applied methods for the determination of nitrite in addition to spectrophotometry (López Pasquali et al. 2007, 2010; Aydın et al. 2005; Abdul Galil et al. 2007; Afkhami et al. 2005; Borcherding et al. 2000; Ridnour et al. 2000; Chen et al. 2000; Abbas and Mostafa 2000; Miro et al. 2000; Barzcgar et al. 2000; Burakhama et al. 2004; Suvardhana et al. 2005; Yue et al.; 2004) include spectrofluorimetry (Chen et al. 2007; Huang et al. 2000; Wang et al. 2000; Geetha and Balasubramanian 2000; Helaleh and Korenaga 2000a; b; Meininger et al. 2000; Masserini and Fanning 2000; Biswas et al. 2004; Zhang et al. 2002; Argϋelles et al. 2004); flow injection analysis (FIA)-kinetic spectrophotometric method (Nouroozi and Mirshafian 2009); electrochemical methods such as differential pulse polarography (Yilmaz and Somer 2008), amperometry (Kalimuthu and John 2009), and potentiometry (Hassan et al. 2003); chemiluminescence (Paolo et al. 2001; Aoki et al. 1997; Wen and Kang 2004; Gao et al. 1990; Markušová and Fedurco 1991; Lu et al. 1992; Butt et al. 2001; Connolly and Paull 2001; He et al. 2011; Akyuz and Ata 2009; Tsikas et al. 1994; Helaleh and Korenaga 2000a; b); electrometric methods (Butt et al. 2001; Connolly and Paull 2001; He et al. 2011; Akyuz and Ata 2009; Tsikas et al. 1994; Helaleh and Korenaga 2000a; b); and chromatography (Butt et al. 2001; Connolly and Paull 2001; He et al. 2011; Akyuz and Ata 2009; Tsikas et al. 1994; Helaleh and Korenaga 2000a; b; Stefanović et al. 2001; Binghui et al. 2006).

Spectrophotometric methods for the determination of nitrite (AOAC 1997; Kawakami and Igrashi 1996) suffer from poor sensitivity and interference from some anions. However, spectrofluorimetric methods for the determination of nitrite have been developed (Damiani and Burini 1986; Ohta et al. 1986; Jiao et al. 2004), and both sensitivity and selectivity have been improved. In the reported spectrofluorimetric methods, nitrite ion reacts with a reagent or acts as a catalyst for various types of the chemical reactions.

In this work, a simple, selective, and sensitive spectrofluorimetric method is presented for the determination of trace amounts of nitrite in complex vegetable samples. The method is based on the quenching of the fluorescence of 4-amino-3-hydroxynaphthalene-1-sulfonic acid by nitrite in acidic medium.

Experimental

Apparatus and Software

A Jasco spectrofluorimeter (fp 6200) equipped with a xenon discharge lamp and a 1-cm pathlength quartz cell was used for all measurements. A Jenway ion meter model 3345 was employed for pH measurements. Design and analysis of the experiments were carried out by MINITAB (Minitab Inc. Release 16.0) statistical package.

Reagents and Solutions

All chemicals were of analytical reagent grade, and double-distilled water was used for preparation of all solutions. Sodium nitrite, ethanol, and hydrochloric acid were supplied by Merck (Darmstadt, Germany). The reagent 4-amino-3-hydroxynaphthalene-1-sulfonic acid (AHNSA) was supplied by BDH Chemicals Ltd., Poole, England. Stock 100.0 mg L−1 nitrite solution was prepared in double-distilled water. Stock 1.0 × 10−3 mol L−1 solution of AHNSA was prepared in double-distilled water and ethanol 50:50 % (v/v). Working solutions were prepared by diluting the stock solutions to an appropriate volume with double-distilled water whenever required. Acidic solution (pH 2.3) was prepared by adding hydrochloric acid and sodium hydroxide to double-distilled water.

Calibration Curve

No significant changes in the fluorescence values were observed within the temperature range of 10–35 °C, and the product was stable for about 2 h in the solution.

Volumes equivalent to 0.5 mL of the stock solution of AHNSA (1.0 × 10−3 mol L−1) were transferred to 5.0-mL volumetric flasks, and after diluting to the volume of 5.0 mL by addition of acidic solution with optimal pH (pH 2.3), 0.25–25 μL of the stock nitrite solution (100.0 mg L−1) was added. It must be mentioned that the maximum fluorescence quenching was reached at 40 min after mixing. After this time, while the solutions have been maintained at room temperature, their fluorescence spectra were recorded. The calibration graph was obtained by plotting the fluorescence changes of the reagent against the concentration of nitrite with excitation at 340 nm and using emission at 442 nm.

Experiments were performed for evaluating the effect of time (30 and 45 min). In most cases, the results showed that the average of the signal at 30 min is about 5.4 % less than the signal at optimum time (40 min). Moreover, for 45 min as the time for obtaining signal, the signal in the average rises only about 0.7 %.

Sample Preparation for the Determination of Nitrite in Vegetable Samples

To analyze dietary products of vegetables and fruits, a volume equivalent to 100 mL of double-distilled water was added to 10 g of the cut, milled, and homogenized sample of each vegetable. The resulted mixture was heated under stirring for approximately 10 min in 75 °C using a water bath. Then, it was allowed to cool and was filtered through a Whatman filter paper. After addition of an excess of EDTA for masking divalent and trivalent interferent cations, it was analyzed within the same conditions for acquiring calibration data. First, pH of the filtrate was adjusted to 2.3. A volume equivalent to 4.5 mL of the resulting sample without dilution was transferred to a 5-mL volumetric flask containing 0.5 mL AHNSA (1.0 × 10−3 mol L−1), and after shaking well and standing for optimal time, its fluorescence spectrum was recorded.

Results and Discussion

Central Composite Experimental Design and Optimization of the Factors

With design of experiment (DoE), the maximum amount of information of the system is extracted in an economical way (Keeney et al. 1970). Among this information is the interaction between the factors. For this purpose, all factors are changed from one experiment to the next, simultaneously. The reason for performing this type of experiment is that variables can influence each other and the optimal value for one of them may be dependent on the values of the others (Williams 1979).

Central composite design (CCD) is an efficient technique for optimization. CCD provides almost the same information that multilevel full factorial design gives with fewer experiments (West and Ramachandran 1966; Norwitz and Keliher 1978). In CCD, it is assumed that the central point for each factor is 0 and the design is symmetrical around it. Optimal values of the factors for determination of nitrite based on the reaction with AHNSA were obtained by analysis of the results of experiments collected in Table 1.

Table 1 Experiments designed based on central composite design with two factors
Full size table

The possible reaction between nitrite and AHNSA is shown in Scheme 1. Because of the lower rigidity of the product relative to AHNSA, its fluorescence intensity is lower. Moreover, the electron withdrawing group of N=N can be mentioned as another reason for the decrease in the fluorescence intensity of the product.

Scheme 1
scheme 1

The reaction between the AHNSA and nitrite in pH 2.3

Full size image

The values of emission difference for experiments 10 and 12 in Table 1 are negative. In these experiments, in the presence of nitrite, emission has increased to some extent. As can be seen in Table 1, experiments 10 and 12 have been performed in higher pH values and with relatively high concentrations of AHNSA. In these pH values, (1) in the absence of nitrite, deprotonation of the sulfonic (pKa ≈ 2.81) and amine (pKa ≈ 3.92) groups of AHNSA increases which results in the increase in the fluorescence intensity of AHNSA, and (2) in the presence of nitrite, formation of HNO2 with a pKa value of 3.40 which is the active agent for producing the product in Scheme 1 is incomplete and nitrite facilitates the two above deprotonations. Therefore, after addition of nitrite, fluorescence intensity of the solution increases.

In Fig. 1, the variation of the emission difference as the response with pH and concentration of the reagent (AHNSA) is shown. As can be seen from Fig. 1, in the lower values of the concentration of the reagent, the emission difference is higher.

Fig. 1
figure 1

Variation of the response with concentration of the reagent (Cr) multiplied by 104 and pH for the reaction of nitrite with AHNSA

Full size image

Analysis of variance (ANOVA) of the experiments in Table 1 was performed. The resulted model for response (emission difference) can be shown as follows:

Emission difference = −81 + 103 × pH + 234,271 × (concentration of AHNSA) − 22 × (pH × pH) − 263,510,625 × (concentration of AHNSA × concentration of AHNSA) − 97,929 × (pH × concentration of AHNSA)

$$ {R}^2\left(\mathrm{a}\mathrm{d}\mathrm{j}\right) = 78.34\ \%,\ F = 8.96,p=0.009 $$

ANOVA showed that in the studied system, pH with probability value (p) of 0.001 is a significant factor at 95 % confidence level (p < 0.05). Quadratic term of pH (pH × pH) is also a significant term for modeling the response at 95 % confidence level (with p = 0.001). This is responsible for the curvature of the response surface with variation of pH in Fig. 1. Probability value for the regression model is 0.009 which indicates that the model is robust and can be used for accurate prediction of the response (emission difference).

Optimal values of the factors for the determination of nitrite based on the reaction with AHNSA were obtained by ANOVA. The optimal values for the concentration of AHNSA and pH were 1.0 × 10−4 mol L−1 and 2.3, respectively.

Fluorescence Spectra

In the presence of nitrite, the fluorescence intensity of AHNSA decreases. The amount of decrease in the intensity was correlated with the concentration of nitrite. This was the basis of the proposed method for the determination of nitrite.

Analytical Data

Under the optimal experimental conditions, a calibration curve was constructed for nitrite which was linear in the concentration range of 0.005–0.500 mg L−1. The fluorescence changes of the reagent with concentration of nitrite in the optimal conditions are shown in Fig. 2. The statistical data of the calibration curve are reported in Table 2. The limit of detection (LOD) and quantitation (LOQ) were calculated according to the ICH guidelines using the formulae LOD = 3.3 × S/b and LOQ = 10 × S/b, respectively (where S is the standard deviation of the blank emission signal with optimal excitation and emission wavelengths and b is the slope of the calibration plot). The linearity of the calibration curve was validated by the high value of correlation coefficient of the regression equation. The high values of molar absorptivity and low values of LOD indicate that the proposed method is very sensitive.

Fig. 2
figure 2

Fluorescence changes during the reaction between 0.005 and 0.500 mg L−1 nitrite solutions and 1.0 × 10−4 mol L−1 solution of AHNSA in pH 2.3 after 40 min while the solution has been maintained at room temperature

Full size image
Table 2 Analytical data of the constructed calibration curve
Full size table

Effect of Foreign Ions

As the method was developed for the analysis of vegetable samples, the interference from foreign ions commonly present in such samples was studied by adding known amounts of foreign ions to a solution containing nitrite with a concentration of 0.100 mg L−1. The tolerance limit of a potentially interfering ion was taken as its maximum amount causing an error of ≥±5 % during the determination of nitrite in water samples. The tolerance limits for the ions studied are given in Table 3. From Table 3, it is clear that the interference from divalent and trivalent cations may be important. Therefore, in the analysis of the real samples, EDTA was used to mask them.

Table 3 Tolerance limit of different ions in the determination of nitrite with a concentration of 0.100 mg L−1
Full size table

Application

Under the optimal experimental conditions, validity of the proposed method was checked by determining nitrite in vegetable samples. Samples were also spiked with a known amount of nitrite. The difference emission spectra for different real samples are shown in Fig. 3. As can be seen, the spectra in Fig. 3 are very similar to the spectra shown in Fig. 2. This shows that the species from the analyzed real samples do not interfere in the determination of nitrite by the proposed method. The results of the analysis of the real samples are reported in Table 4. The standard deviations of the results are very low which shows the precision of the proposed method.

Fig. 3
figure 3

Difference emission spectra obtained for a cucumber, b cabbage, c lettuce, and d tomato in optimal conditions for five replicates

Full size image
Table 4 Determination of nitrite in vegetable samples based on the proposed method
Full size table

In order to approve the accuracy of the proposed method, real samples were also analyzed by the Griess method as a reference method. The results of the application of the reference method are also reported in Table 4. For comparison of the results obtained by the proposed method and the reference method, t statistics was calculated. As can be seen from Table 4, all of the calculated t values are below the critical t value of 2.31 for 95 % confidence level and 8 degrees of freedom. Therefore, the accuracy of the method for the determination of nitrite is confirmed.

Comparison of Reported Methods with the Proposed Method for the Determination of Nitrite

Some of the published spectrofluorimetric methods for the determination of nitrite are collected in Table 5. The methods have been compared based on the linear range, detection limit, time of the analysis, and percent relative standard deviation (RSD%). Based on the RSD% values reported in Table 5, the spectrofluorimetric method introduced here is the most precise method. It must be noted that the reported RSD% value for the proposed method is the highest value obtained for the analyzed real samples.

Table 5 Published results for the spectrofluorimetric determination of nitrite
Full size table

In the published work by Geetha and Balasubramanian (2000), the reaction step of the analysis requires 5 min; however, the extraction process may take a long time. Masserini and Fanning (2000) introduced a method which possesses a throughput rate of 18 samples per hour (about 3.3 min for each sample). However, their method is based on the flow injection analysis (FIA). In the spectrofluorimetric method proposed by Helaleh and Korenaga (2000a, b), though the overall reaction time is 10 min, it takes place in two steps which may delay more than 10 min. Lee and Field (1984) introduced the reaction of nitrite with Ce(IV). The reaction time was reported to be 1.5 min. They used ion chromatography for the separation and fluorescence for detection. Wang et al. (2000) synthesized a reagent for the spectrofluorimetric determination of nitrite in which its reaction with nitrite is completed in 25 min. However, synthesis of the reagent is time consuming.

As can be seen from the first column of Table 5, most of the published spectrofluorimetric methods for the determination of nitrite have been applied to real samples with simple matrices like water. The dynamic linear range of the present method is wide. Considering the detection limit and lower limit of the linear range, the proposed method is very sensitive. The low LOD obtained by Buldt and Karst has been achieved by HPLC with fluorescence detection (Buldt and Karst 1999). Very low LOD reported by Wang et al. (2000) has been calculated by LOD = 3 × S. Masserini and Fanning have calculated LOD based on the formula LOD = 2 × S (Masserini and Fanning 2000).

Conclusions

A simple spectrofluorimetric method for the sensitive determination of nitrite in vegetable samples was proposed and validated. Excellent precision and recoveries were obtained when the proposed method was applied to complex vegetable samples. The proposed method is very sensitive with low detection limit and lower limit of the linear range. This method offers a simple and cost-effective alternative to existing methods for nitrite determination. The proposed method requires no control of temperature.