Advertisement

Analytical and Bioanalytical Chemistry

, Volume 410, Issue 22, pp 5629–5640 | Cite as

Analysis of a variety of inorganic and organic additives in food products by ion-pairing liquid chromatography coupled to high-resolution mass spectrometry

  • Anton Kaufmann
  • Mirjam Widmer
  • Kathryn Maden
  • Patrick Butcher
  • Stephan Walker
Research Paper
Part of the following topical collections:
  1. Food Safety Analysis

Abstract

A reversed-phase ion-pairing chromatographic method was developed for the detection and quantification of inorganic and organic anionic food additives. A single-stage high-resolution mass spectrometer (orbitrap ion trap, Orbitrap) was used to detect the accurate masses of the unfragmented analyte ions. The developed ion-pairing chromatography method was based on a dibutylamine/hexafluoro-2-propanol buffer. Dibutylamine can be charged to serve as a chromatographic ion-pairing agent. This ensures sufficient retention of inorganic and organic anions. Yet, unlike quaternary amines, it can be de-charged in the electrospray to prevent the formation of neutral analyte ion-pairing agent adducts. This process is significantly facilitated by the added hexafluoro-2-propanol. This approach permits the sensitive detection and quantification of additives like nitrate and mono-, di-, and triphosphate as well as citric acid, a number of artificial sweeteners like cyclamate and aspartame, flavor enhancers like glutamate, and preservatives like sorbic acid. This is a major advantage, since the currently used analytical methods as utilized in food safety laboratories are only capable in monitoring a few compounds or a particular category of food additives.

Graphical abstract

Deptotonation of ion pair agent in the electrospray interface

Keywords

Anionic food additives Food safety Ion-pairing chromatography High-resolution mass spectrometry Orbitrap 

Introduction

Inorganic food additives are used for a number of purposes. Nitrite and nitrate prevent the growth of harmful organisms in meat products. Polyphosphates permit the additional uptake of water into fish or seafood products. The production of preserved food products frequently requires the use of a number of additives to achieve the desired sensorial and optical appearance. This may include the use of flavor enhancers like glutamate, artificial sweeteners like cyclamate, or preservatives like sorbic acid. Because of toxicological concerns, especially for infants, the maximum amount of several of these additives, including nitrite, nitrate, polyphosphates, and certain preservatives, is limited. No such restrictions exist for a number of other additives, yet their use must be clearly declared on food packaging. Frequently, consumers prefer the so-called natural products, which should be produced using as few food additives (also known by the term “e-numbers”) as is possible. Hence, there is an increasing interest in verifying the correctness of food labeling by the use of analytical methodologies.

Inorganic food additives are frequently detected by photometric, enzymatic, or ion chromatography-based methods. There are a number of enzymatic- or liquid chromatography-based methods for the detection of organic anionic additives like flavor enhancers or artificial sweeteners. Yet these methods detect and quantify only one or a few of the compounds of interest. The development of a liquid chromatography hyphenated to mass spectrometry (LC-MS) multimethod covering the detection of eight artificial sweeteners has been described [1]. A set of flavor enhancers was quantified by high-performance liquid chromatography [2]. An LC-MS method covering ten sweeteners in a variety of foods was recently published [3]. However, the detection of polyphosphates represents a particular problem. The conventional phosphate methods [4] do not permit the selective determination of exogenous polyphosphates (e.g., di- and triphosphate) but produces a cumulative value for P2O5. In other words, the high endogenous concentration of monophosphate prevents the detection of added polyphosphates. Until now, only the technique of suppressed gradient ion chromatography has been reported for the determination of polyphosphates in meat and fish products [5, 6, 7, 8, 9]. To the best of our knowledge, no work has been published which proposes an analytical method permitting the determination of several classes of food additives (inorganic salts and organic additives like sweeteners or preservatives) within a single analytical method.

It is a major analytical challenge to combine compounds with widely different pKa values into a single chromatographic method. Anions of strong acids, like nitrates or polyphosphates, are not retained on reversed-phase columns. However, they can be retained by ion exchange. Ion chromatography (IC) most often relies on the use of conductivity detection. Although only ions can give an analytical response, conductivity detection is frequently insufficiently selective. Unfortunately, ion chromatography eluents must contain ions in order to enable the elution of the ionic analytes. Hence, a high detector baseline is observed. This becomes even more difficult if ions of multiple charge states are to be analyzed. Decent peak shapes of single charged and triple or even higher charged analyte ions can only be obtained by gradient IC.

On-line ion suppression devices (suppressors) are used in order to remove the eluent ions before the eluent enters the detector. A combined IC detection of strongly dissociated inorganic ions and weakly dissociated organic compounds is not easy because weakly dissociated analytes produce only poor conductivity responses. Therefore, a promising approach is IC coupled to mass spectrometry.

It might be expected that analyzing ions by mass spectrometry is straightforward, because the already charged analytes do not need to be further ionized. This is certainly true, yet like in conventional IC, elution is only possible by the use of eluents with sufficient ion strengths. Such ion concentrations are incompatible with conventional electrospray interfaces. In addition, adducts are formed with the deprotonated analyte and the cations present in the mobile phase. Neutral adducts (e.g., a nitrate ion and a sodium ion) will produce no MS response. The few reported successful IC-MS methodologies used suppressors in order to remove the ions required for the elution of the analytes before the eluent is directed into the MS interface. Halogenated carboxylic acids were analyzed by suppressed sodium carbonate gradients [10], a set of inorganic and organic anions, as well as nitrate and glucose-6-phosphate by suppressed sodium hydroxide gradients [11, 12]. Commercial membrane suppressors are pressure sensitive devices; therefore, their use in front of an electrospray interface requires some care. In addition, suppressors are optimized for the suppression of strong anions. Weakly dissociated anions may cross the membrane wall, while some organic compounds may elute with tailing after interaction with the suppressor membranes. Finally, not all analytes survive in the high pH caused by the use of sodium hydroxide in the IC eluent.

Because of these incompatibilities, in this work, we investigated concepts based on ion-pairing reversed-phase chromatography. Permanently charged quaternary amines are frequently used for liquid chromatography separations utilizing UV detection techniques. Yet such compounds extensively suppress electrospray ionization (ESI)-based MS signals. Primary, secondary or ternary amines also work as ion-pairing agents when employed in a neutral or low pH environment, but signal suppression is still rather pronounced.

An interesting approach has been developed for the analysis of nucleotides [13, 14, 15]. The multiple negatively charged nucleotides are generally insufficiently retained on reversed-phase columns, yet are well retained when primary, secondary or ternary amines adjusted to a pH level of 7 to 9 are used (ion-pairing chromatography). Using hexafluoro-2-propanol (HFIP) rather than formic or acetic acid for the adjustment of the mobile phase pH significantly improves the sensitivity of detection. The slightly acid HFIP not only protonates the amine, which then functions as an ion-pairing agent. It is also thought that the volatile HFIP evaporates in the electrospray environment and de-charges the ion-pairing agent, improving the desorption conditions for the negatively charged analytes [14].

Our goal is to utilize this concept to develop a routine LC-MS method capable of detecting monovalent and polyvalent inorganic anions, together with anionic organic compounds possessing a wide range of pKa values and polarities. To the best of our knowledge, this is the first reported method using ion-paring chromatography coupled to mass spectrometry for the detection of such a wide variety of analytes in food matrices.

Material and methods

Chemicals

All compounds used—acesulfame K, adipinic acid, aspartame, succinic acid, citric acid, sodium bromate, sodium cyclamate, tetra-sodium diphosphate decahydrate, sodium l-glutamate monohydrate, sodium lactate, sodium phosphate, sodium nitrate, sodium propionate, saccharin, sorbic acid, sucralose, sodium triphosphate pentabasic, and tartaric acid—were obtained from Sigma-Aldrich (Buchs, Switzerland).

The stock standards and solutions were prepared using a dilution solution containing 1 g/l ethylenedinitrilotetraacetic acid (EDTA) disodium salt dehydrate in purified water. All solutions were prepared and stored in polypropylene vessels at 4 °C. Working solutions containing all analytes were prepared, diluted to appropriate concentrations, and stored in the same manner. The investigated analytes were present or added to food products at very different concentrations, which is why the mixed reference solution contained high concentration analytes at a ten times higher concentration than low concentration analytes. This led to the production of a spiking solution of 50/500 mg/l and four reference solutions of 20/200, 10/100, 1/10, and 0.1/1 mg/l. The concentration group to which a particular analyte belongs is listed in Table 1. All vessels were rinsed two times with purified water before use.
Table 1

Separation and detection information for the analytes covered by the method

Analyte

Elemental composition

Scan event

ESI-ion

Monoisotopic masse [m/z]

Retention time [min]

Acesulfame

C4H4NO4S

A

+e

161.98665

2.56

Adipinic acid

C6H10O4

A

-H

145.05063

1.4

Aspartame

C14H18N2O5

B

-H

293.11429

5.37

Benzoic acid

C7H6O2

A

-H

121.0295

2.71

Succinic

C4H6O4

A

-H

117.01933

1.38

Citric acida

C6H8O7

A

-H

191.01973

2.11

Cyclamate

C6H12NO3S

A

+e

178.05434

3.62

Diphosphatea

H4P2O7

A

-H

176.93595

2.53

Lactic acida

C3H6O3

A

-H

89.02442

1.01

Glutamate

C5H8NO4

A

+e

146.04588

1.01

Monophosphatea

H3PO4

A

-H

96.96962

1.35

Bromate

BrO3

A

+e

126.90363

1.01

Nitrate

NO3

A

+e

61.98837

1.05

Propionic acid

C3H6O2

B

-H

73.0295

1.38

Saccharine

C7H5NO3S

A

-H

181.99174

3.21

Sorbic acid

C6H8O2

A

-H

111.04515

3.10

Sucralose

C12H19Cl3O8

A

-H

395.00727

3.92

Triphosphatea

H5P3O10

A

-H

256.90228

3.61

Tartaric acid

C4H6O6

A

-H

149.00916

1.37

aAnalyte belongs to the high concentration group

EDTA, dibutylamine 99.5% (DBA), and 1,1,1,3,3,3-hexafluoro2-propanol 99% (HFIP) were obtained from Sigma-Aldrich. Acetonitrile HPLC grade was acquired from Baker (Deventer, Netherlands). The reference and extraction solutions were identical and consisted of 1 g/l EDTA in purified water.

Extraction and clean-up procedures

Sample (5 g of either shrimp, cured meat sausage, or cooked sausage obtained from the local market in Zürich) was homogenized in a polypropylene centrifugation vessel with 150 ml of extraction solution (dilution solution). Spiking (A-spike) was done by adding appropriate volumes of spiking solution (e.g., 12 ml mixed spiking solution 50/500 mg/l). The centrifugation vessel was then placed into a boiling water bath for 20 min in order to deactivate native phosphatases. Afterward, the vessel was placed into an ice bath to cool down to room temperature. The cooled vessel was centrifuged at 14,000 rpm = 28,000×g for 5 min. The supernatant was filtered with a microfilter (Chromfil PET-20/25; 0.2 μm from Machery Nagel, Oensingen, Switzerland). 0.54 ml of the filtrate and 0.06 ml dilution solution was transferred into a plastic HPLC vial. B-spike solutions (to test for signal suppression) were prepared in the same way, using 0.54 ml of the filtrate and 0.06 ml of the mixed reference solution 5/50 mg/l. The A-spike is used for calculating the physical analyte recovery, while the B-spike is used to calculate and compensate for possible interface related signal suppression or enhancement effects. A B-spike was produced for each sample in order to determine sample specific signal suppression effects.

Liquid chromatography-high-resolution mass spectrometry

Liquid chromatography was performed on a Waters UPLC I-Class (Milford, USA) utilizing a BEH, C18 column of 2.1 × 100 mm internal diameter, 1.7 μm (Waters). The flow rate was 0.3 ml/min and the injection volume was 2 μl (partial loop). Mobile phase A consisted of a solution of 25 ml acetonitrile, 0.5 ml DBA, and 1.6 ml HFIP made up to 1 l with purified water. Mobile phase B consisted of a solution of 500 ml acetonitrile, 0.5 ml DBA, and 1.6 ml HFIP made up to 1 L with purified water. Separation was carried out in 9.5 min under the following conditions: 0 min, 5% B; 0.5 min, 5% B; 4 min, 25% B; 5.5 min, 80% B; 6.5 min, 80% B; 6.51 min, 5% B; 9.5 min, 5% B. The analyses were performed using a heated electrospray interface (HEIS) in negative ion mode at a spray voltage of − 2500 V on a single-stage Exactive instrument (Thermo, Bremen, Germany). The sheath gas flow (nitrogen) was set to 50 units, the aux gas flow to 10 units, the sweep gas flow to 0 units, applying a capillary temperature of 160 °C and a heater temperature of 250 °C. The spray voltage was set to − 37.5 V, the tube lens voltage to − 125 V, and the skimmer voltage to − 36 V. The utilized Orbitrap instrument does not contain a quadrupole. MS within the analyzer was acquired using full scan mode from 50 to 400 m/z based on a mass resolving power of 25,000 FWHM, an AGC target of 3,000,000 charges, and a maximum injection time of 50 ms. Acquisition mode A (scan event) did not utilize an in-source fragmentation voltage. Acquisition mode B applied 70 eV in-source fragmentation voltage. The analyte-specific detection parameters are listed in Table 1. Extracted were mass windows of 50 ppm.

Validation

Validation was based on spiking three different matrices (uncooked shrimp, cured meat sausage, and cooked sausage). The shrimp matrix was considered the most important; therefore, three independent analysis series were performed by different personnel on different days. The unfortified matrices and four independent repetitions for each of the four different fortification levels (levels A, B, C, and D) were analyzed. Depending on the compound, this translated into fortifications of 6, 30, 120, and 600, respectively, 60, 300, 1200, and 6000 mg/kg. Ten sample injections were followed by a calibration block. Hence, the produced calibration curve consists of several calibration blocks. A quadratic calibration curve without intercept and a weighting of the calibration curve by a factor of 1/x was selected. Signal suppressions were determined by spiking injection ready matrix extracts. The increase of the signal (peak area of the spiked minus peak area of the unspiked extract) was compared to the signal produced by the same concentration of analyte present in a pure standard solution limit of quantification were determined from standard solutions, since some of the analyzed analytes were present in any of the validated matrices.

Results and discussion

Extraction and sample processing

Our method was optimized to analyze polar anionic analytes. For this reason, an aqueous extraction was carried out. The inclusion of significantly less polar anionic analytes within the method was successfully tested during method development. However, their quantitative extraction from the matrix requires the use of a certain concentration of organic solvents in the extraction medium. It is this presence of organic solvent in the injection ready sample which negatively affects the peak shape of early eluting analytes. As a consequence, the finally developed method does not include apolar or medium polar analytes. The ratio between sample and extraction volume was chosen to obtain a diluted extract that minimizes signal suppression or enhancement effects in the electrospray interface. Extraction could be done with a significantly smaller extraction solution volume, in order to obtain a higher sensitivity of detection.

Extraction was done with EDTA in water. This additive was considered essential, as its absence significantly reduces the recoveries of polyphosphates.

An integral part of the extraction process is the heating of the extract. This step is intended to denature phosphatases, which are present in some matrices [6, 9]. Otherwise, di- and triphosphates are degraded to monophosphate within the extract or even while waiting for injection in the HPLC sample vial.

Another relevant observation was the presence of a strong chromatographic peak corresponding to citric acid peak in blank extracts. Investigations lead to the dishwasher used in our laboratory. The detergents used for the cleaning process contained high concentrations of citric acid. Therefore, relevant amounts of citric acids were still present in the washed glass- or plasticware. This was significantly reduced by the rinsing step as described in the method section. As discussed later, the method avoids the use of glassware. The reason for this is the reduction of cations (e.g., sodium and potassium) in the sample extract and probably even more important in the mobile phase (see the discussion below).

An important additive (ascorbic acid) was originally included among the analytes. Poor reproducibility, perhaps due to oxidation of the analyte, led to its removal from the list of analytes.

Chromatography

Due to the polar character of the investigated compounds, no or only very weak chromatographic retention can be obtained when using conventional reversed-phase columns. Chromatographic separation is essential to move analytes away from the column void volume where matrix compounds can create irreproducible matrix-dependent signal suppression. In addition, it was observed that chemical rearrangements of phosphates take place in the heated electrospray interface. This subject will be discussed in further detail later in the article.

It is the dibutylamine (DBA) and not the HFIP that acts as an ion-pairing agent [13]. Based on common ion-pairing theories [13], the charged amine is distributed between the stationary and the mobile phase. The longer the carbon chain, the more this concentration ratio is expected to shift towards the stationary phase (reversed-phase column). This corresponds with our experiments, which showed that for a given ion-pairing agent, the retention power increased with the number of aliphatic carbons attached to the amine (carbon chain length). We found that very long aliphatic carbon chains led to distorted signals, probably due to slower column equilibration. Longer carbon chain amines also showed poorer solubility in the aqueous phase, which made it difficult to dissolve the amount of ion-pairing agent required for sufficient retention of the weakest retained compound. During method development, it was also observed that an aqueous mobile phase (mobile phase A) saturated with an ion pair agent can lead to instable separations (fluctuating retention times). This led to the selection of shorter carbon chain amines (e.g., DBA), as they were more soluble in mobile phase A. On the other hand, the concentration required for the sufficient retention of analyte is relatively high, which will reduce the sensitivity of detection.

Depending on the analyte, MS-based detection sensitivity may be significantly reduced in the presence of cationic ion pair agents. This is likely due to the formation of neutral ion pairs (anionic analyte ion and positively charged ion pair agent). It is desirable to maintain the ion pair agent in its pronated form only during chromatography. The charge should be removed as soon as the eluent enters the MS interface. Such a concept resembles the suppressed IC systems that remove the eluent ions (e.g., sodium hydroxide) after leaving the analytical column. A weakly acid, but volatile compound can be added to the eluent to ensure the protonation of the ion pair agent (primary, secondary, or ternary amine) during the chromatographic process. HFIP, which is sufficiently acid to protonate a secondary amine like DBA, is one such additive. The protonated DBA functions as an ion pair agent with the deprotonated HFIP acting as counter ion. However, this ion pair is disrupted in the electrospray interface. It is assumed that the volatile HFIP removes the proton from the charged amine and evaporates as neutral molecule [13]. The greatly reduced concentration of protonated DBA restricts the formation of neutral adducts (deprotonated analyte and protonated DBA). In addition, the proton depletion increases the pH of the evaporating droplets, which further promotes the deprotonation of weakly anionic analytes. Hence, the analytes can be more readily desorbed as negatively charged ions from the evaporating electrospray droplets.

The resulting pH of the ion pair agent protonated by the addition of HFIP was still relatively high for conventional reversed-phase columns (pH = 9–10). It should be noted that modern columns are more pH tolerant than previous generations. In addition, high pH values caused by inorganic bases seem to affect column life more drastically than the used organic ion pair agent.

Unlike in IC, the utilized gradient does not increase the ion strength of the eluent, but the increasing acetonitrile concentration forces the ion pair agent to migrate from the stationary into the mobile phase. This is clearly visible when acquiring data in the positive ionization mode and monitoring the protonated DBA across the chromatographic run time. Therefore, after the end of the gradient, a certain amount of time is needed until the original initial ion-pairing agent equilibrium is reestablished. For the same reason, three injections and gradient runs are needed before stable retention times can be obtained. Figure 1 shows the extracted ion chromatogram of selected compounds.
Fig. 1

Selected extracted ion chromatograms of analytes in an unfortified shrimp matrix (left) and a shrimp matrix fortified with 30/300 mg/kg (right)

Eluents containing non-quaternary amines and HFIP have become the commonly used additives for the separation of nucleotides [13, 14, 15]. There have been extensive investigations regarding the best-suited ion pair agent [13] and associated volatile acid [14]. One issue with this approach was the long-term stability of the separation [15]. This was linked to the production of nucleotide-cation adducts [15] and was addressed by the use of glass-free (plastic) materials as well the inclusion of a quick low pH step after the chromatographic gradient [15]. We did not include such a low pH rinsing step into the developed methods, since no visible benefits were obtained for the compounds of interest. In addition switching from a pH > 9 to a low pH regime and vice versa requires additional equilibration times.

The importance of HFIP as an additive was tested by using identical eluents, but the alternative HFIP-free eluent was neutralized with formic acid and adjusted to the same pH value (pH eluent A = 9.0 and pH eluent B = 10) as used in the HFIP-containing mobile phase. All analytes with the exception of propionic acid showed significantly higher peak areas. The intensity ratio of each analyte is given in Table 2. The results show that not only nucleotides [13, 14, 15] but also ionic food additives benefit from the use of HFIP instead of formic or acetic acid as a mobile phase additive.
Table 2

Relative increase of analyte peak when using HFIP rather than formic acid to protonate DBA

Analyte

Signal enhancement factor

Acesulfame K

2.6

Adipinic acid

5.1

Aspartame

1.8

Benzoic acid

14.2

Succinic acid

4.4

Bromate

5.3

Citric acid

1.5

Cyclamate

5.1

Diphosphate

4.7

Glutamate

3.4

Lactic acid

6.0

Monophosphate

15.4

Nitrate

3.1

Propionic acid

0.4

Saccharin

4.0

Sorbic acid

39.1

Sucralose

5.5

Triphosphate

5.2

Tartaric acid

3.8

To the best of our knowledge, the chromatographic system we developed is capable of separating more anionic food additives than any other published method. There are, however, two limitations. First, as observed in most IC or ion pair reversed-phase methods, retention times shift slightly at high analyte concentrations. This is probably due to the local saturation of the ion exchanger sites. At high concentration, some of the ions move deeper into the column until they encounter unsaturated ion exchange sites. This leads to shorter analyte retention times. Fortunately, this is less relevant than when using conductivity-based IC or ultra-violet detection-based ion-paring reversed-phase chromatography. The high selectivity of HRMS-based detection prevents the wrong identification of a slightly shifted chromatographic peak. A more pertinent issue is the fact that conventional column rinsing is incapable of removing the used ion pair agent from the LC. The presence of the remaining ion pair agent will not only produce the base peak within the spectrum but can also reduce the sensitivity of other reversed-phase-based assays. Therefore, before switching to another ion pair free method, the LC has to be thoroughly purged for a prolonged time. This should not be done with conventional reversed-phase LC-MS solvent (e.g., acetonitrile, water, and formic acid) but with a buffer containing solution (e.g., 1 mol ammoniumformiate and 1 mol ammoniumhydroxide in 50% methanol). Purging should include the autosampler and the space behind the piston seals (seal wash).

Mass spectrometry

The utilized HRMS mass resolving power in combination with narrow mass windows produced a significantly higher selectivity than the obtainable selectivity when using conventional conductivity-based IC or UV detection. A limitation of the proposed method is the incapability of monitoring nitrite. This is due to the fact the mass of this ion is 46. This is below the mass range of the utilized instrument (m/z = 50).

We observed the existence of chemical transformation products produced within the electrospray interface. Monophosphate produced diphosphate, while diphosphate was partially converted into triphosphate. This is visible in Fig. 2, where the extracted ion chromatograms of the three phosphate species are shown. The blank shrimp matrix (left) contains a high endogenous level of monophosphate but no exogenous di- or triphosphate. The retention times of the three analytes are indicated in Fig. 2 with vertical arrows. Note the diphosphate trace (middle, left) shows no diphosphate at the expected retention time, yet another chromatographic peak appears within the diphosphate trace at 1.56 min. This corresponds exactly with the retention time and the peak shape of monophosphate (top trace). The intensity of this peak within the m/z = 177 trace is around 10% of the signal obtained within the m/z = 97 trace. The appearance of this diphosphate signal at the retention time of monophosphate is due to thermal rearrangement within the electrospray interface. Spiking the “blank” matrix (right side) with a standard containing all three phosphate species produces di- and triphosphate peaks at 2.73 and 3.65 min. In addition, the triphosphate trace shows a wrong triphosphate peak appearing exactly at the retention time of diphosphate. This is also explained by thermal rearrangement (a portion of the diphosphate ions are converted into triphosphate). The interface settings (temperatures) were adjusted so that less than 10% of monophosphate and diphosphate underwent transformations (see Fig. 2). Further minimization would go along with a drastic loss of sensitivity. However, such transformations do not truly affect quantification, because of the good chromatographic separation of the phosphate species.
Fig. 2

Extracted ion chromatograms of mono-, di-, and triphosphates. The unfortified shrimp matrix (left) and the matrix fortified with 300 mg/kg (right). Note the appearance of diphosphate at the retention time of monophosphate and the appearance of triphosphate at the retention time of diphosphate

Validation

The method was validated for three different matrices (shrimp, cured meat sausage, and cooked sausage). Validation was based on the addition of a single mixed spiking solution containing all investigated analytes. Due to technological and sensorial reasons, the investigated food additives are used and therefore have to be quantified at a wide concentration range (mg/kg to g/kg). Depending on these concentrations, a compound was placed into a low or high concentration group. This is stated in Table 1 and also indicated in Tables 3, 4, and 5. The validation data are given in Tables 3, 4, and 5. Multilevel external calibration was used for all compounds. Due to the high dilution of the extracts, signal suppression was not very pronounced (see Tables 3, 4, and 5). This simplifies quantification. Sensitivity was sufficient for all the investigated compounds. In other words, the limit of quantification (LOQ) was not only below the maximum permitted levels but also significantly below the technological required concentrations of the investigated food additives. It is important to notice that the validated (unfortified) matrices already contained a relatively high level of a few of the analyzed analytes. These concentrations are stated in Tables 3, 4, and 5. As a consequence, spiking a small amount of monophosphate into a sample that already contains a high endogenous level creates an environment where calculated recoveries and coefficient of determination become unreliable. More realistic recovery rates, which reflect the performance of the method, were obtained when using higher spiking levels. On the other hand, spiking a high amount of phosphate into a sample that already contains a high endogenous concentration of that analyte can lead to concentrations that saturate the separation and/or detection system (e.g., lactate and phosphate). Therefore, no recovery was calculated for the highest monophosphate spiking level into the shrimp matrix.
Table 3

Validation data for shrimp

Analyte

Concentration of analyte in matrix (mg/kg)

Coefficient of determination [r2]

Matrix effect [%]

LOQ [mg/kg]

Recovery [%]

RSD [%] within (between) days

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Acesulfame K

0

0.998

107

6

98

87

93

89

8.5 (11.5)

1.8 (3.4)

4.3 (6.0)

4.5 (18.4)

Adipinic acid

0

0.998

120

6

92

98

93

83

15.7 (88.1)

10.0 (13.6)

3.7 (5.5)

4.2 (14.6)

Aspartame

0

0.998

94

30

4

71

69

60

48.4 (346.8)

10.9 (4.6)

5.0 (4.5)

1.6 (20.6)

Benzoic acid

0

0.998

95

3

88

82

96

92

16.0 (24.3)

5.1 (15.1)

4.4 (16.1)

4.3 (10.5)

Succinic acid

39

0.998

97

3

82

88

99

96

9.7 (21.7)

5.9 (18.4)

3.9 (10.8)

3.2 (9.1)

Bromate

0

0.999

90

6

39

65

69

72

16.2 (134.1)

3.7 (21.9)

3.2 (10.1)

2.6 (29.2)

Citric acid

105

0.996

95

30

93

93

89

75

4.8 (30.1)

2.2 (14.4)

2.5 (3.4)

5.0 (32.1)

Cyclamate

0

0.998

105

3

70

90

92

79

47.5 (66.7)

2.6 (12.7)

3.3 (5.7)

3.0 (11.2)

Diphosphate

0

1.000

92

30

119

99

96

87

4.9 (20.3)

3.0 (7.4)

1.8 (4.6)

1.8 (1.3)

Glutamate

561

0.987

76

3

52

129

97

106

3.1 (4.3)

2.2 (2.6)

3.2 (4.2)

3.1 (10.7)

Lactic acid

1096

0.769

86

593

1023

400

339

3.1 (10.5)

1.9 (5.7)

4.0 (8.2)

2.2 (17.5)

Monophosphate

5113

0.978

93

3

234

123

115

*

3.1 (2.4)

2.6 (5.1)

2.2 (3.8)

*

Nitrate

134

0.989

72

3

76

75

95

5.4 (9.2)

4.2 (13.1)

4.1 (6.5)

8.1 (5.1)

Propionic acid

30

0.999

76

3

90

88

87

92

10.1 (26.6)

4.6 (23.3)

3.9 (0.8)

2.9 (10.4)

Saccharin

0

0.998

93

6

94

90

98

88

37.8 (24.6)

3.9 (14.6)

3.4 (13.8)

3.4 (10.5)

Sorbic acid

0

0.998

95

15

81

76

88

10.3 (346.4)

17.0 (28.6)

6.1 (9.6)

3.4 (14.4)

Sucralose

0

0.994

96

15

87

96

84

228.3 (346)

20.6 (31.3)

6.2 (20.4)

*

Triphosphate

0

0.998

87

30

15

40

65

86

3.9 (118.1)

4.3 (45.6)

5.0 (17.8)

1.8 (16.7)

Tartaric acid

0

0.997

74

6

124

99

95

83

24.2 (186.4)

4.9 (20.8)

3.2 (4.1)

4.5 (5.7)

High concentration level compounds (e.g., 6000 mg/kg for the D spiking level) are shown in italic, while low concentration level compounds (e.g., 600 mg/kg for the D spiking level) are in normal fonts

*Observed signal exceedes the linear dynamic range

Table 4

Validation data for salami (cured meat sausage)

Analyte

Concentration of analyte in matrix (mg/kg)

Coefficient of determination [r2]

Matrix effect [%]

LOQ [mg/kg]

Recovery [%]

RSD [%] within days

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Acesulfame K

0

0.998

127

6

70

88

83

72

5.3

1.9

1.6

3.6

Adipinic acid

0

0.997

72

6

99

104

89

104

10.2

8.5

2.9

4.4

Aspartame

0

0.999

88

6

41

88

89

83

37.6

10.5

3.3

2.9

Benzoic acid

0

0.999

98

6

93

88

86

87

5.7

2.0

2.4

3.4

Succinic acid

71.6

0.979

103

1

147

65

81

93

11.0

9.7

3.6

10.4

Bromate

0

0.990

35

20

3

56

61

70

146.8

17.0

11.2

9.6

Citric acid

34.2

0.999

111

10

103

89

85

86

4.3

2.7

2.7

2.5

Cyclamate

0

1.000

88

6

115

87

91

92

7.1

2.5

3.3

1.6

Diphosphate

0

1.000

97

30

67

87

84

89

1.1

2.8

2.2

1.0

Glutamate

89.1

0.967

129

3

316

57

47

53

5.1

7.1

9.3

2.6

Lactic acid

194.6

0.093

141

359

0

23

4

4.1

8.7

4.1

6.8

Monophosphate

2345.2

0.993

106

3

302

168

133

106

1.9

1.1

5.2

3.3

Nitrate

9.6

0.995

37

3

74

103

89

90

15.0

3.7

8.2

6.6

Propionic acid

54.3

0.990

87

6

122

68

79

86

7.9

7.3

2.7

2.8

Saccharin

0

1.000

89

6

98

88

86

88

6.6

3.7

2.2

2.0

Sorbic acid

0

0.998

94

6

114

79

85

94

39.9

6.3

5.6

3.3

Sucralose

0

0.998

85

30

95

88

95

7.4

2.0

4.3

Triphosphate

0

0.996

73

30

24

36

59

85

10.5

8.4

3.9

1.9

Tartaric acid

0

0.999

56

6

51

88

94

105

5.3

6.1

5.0

1.0

High concentration level compounds (e.g., 6000 mg/kg for the D spiking level) are shown in italic, while low concentration level compounds (e.g., 600 mg/kg for the D spiking level) are in normal fonts

Table 5

Validation data for Wienerli (cooked sausage)

Analyte

Concentration of analyte in matrix (mg/kg)

Coefficient of determination [r2]

Matrix effect [%]

LOQ [mg/kg]

Recovery [%]

RSD [%] within days

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Level A (6/60 mg/kg)

Level B (30/300 mg/kg)

Level C (120/1200 mg/kg)

Level D (600/6000 mg/kg)

Acesulfame K

0

1.000

108

6

79

88

80

81

4.0

2.2

1.7

1.1

Adipinic acid

0

0.997

87

6

100

81

91

92

11.9

4.9

3.8

5.1

Aspartame

0

0.999

99

15

27

82

86

74

38.6

4.9

2.8

1.0

Benzoic acid

0

0.999

98

3

88

78

83

91

5.4

1.0

1.0

3.2

Succinic acid

53.6

0.999

114

3

112

82

84

84

5.2

3.2

3.2

2.7

Bromate

0

0.999

72

6

39

59

72

79

21.3

2.2

2.2

2.1

Citric acid

23.2

1.000

125

10

91

78

79

77

4.6

2.6

1.7

2.2

Cyclamate

0

0.998

100

3

96

79

89

84

9.7

8.0

4.0

3.6

Diphosphate

0

0.999

102

30

68

84

87

86

4.3

1.3

1.9

2.4

Glutamate

34.5

0.998

45

6

125

78

92

87

8.3

3.5

6.1

3.3

Lactic acid

2574.4

0.294

266

541

207

84

18

3.2

1.7

2.1

6.6

Monophosphate

1261.1

0.991

107

1

182

85

108

121

2.3

1.1

1.1

7.1

Nitrate

30.9

0.998

51

1

88

84

86

81

2.7

3.5

2.1

4.4

Propionic acid

45.4

0.999

90

1

85

78

85

86

2.2

1.5

2.7

2.1

Saccharin

0

1.000

97

3

90

77

86

85

7.5

4.4

2.0

1.2

Sorbic acid

0

0.996

89

10

25

82

91

93

83.5

9.0

5.5

5.9

Sucralose

0

0.999

94

10

29

86

90

86

89.6

1.1

2.4

2.1

Triphosphate

0

0.997

92

10

22

35

63

77

3.2

2.5

4.3

4.0

Tartaric acid

0

0.999

76

6

59

87

100

89

14.0

6.7

1.9

1.3

High concentration level compounds (e.g., 6000 mg/kg for the D spiking level) are shown in italic, while low concentration level compounds (e.g., 600 mg/kg for the D spiking level) are in normal fonts

Attempts were made to estimate and consequently to reduce the uncertainty of the analytical procedure [16]. The sample preparation and the extraction step introduces only a small degree of uncertainty. Important is the thermal treatment of the sample in order to prevent the degradation of analytes by endogen enzymes (e.g., polyphosphates present in frozen raw shrimps). Extraction, as mentioned above is nearly exhaustive. Signal suppression are clearly the most relevant source of uncertainty. Due to the non-availability of isotopically labeled internal standards, external standards had to be used. Signal suppression should be mathematically compensated by evaluating the extent of signal suppression for a particular analyte and matrix pair. Care has to be taken that a similar matrix is utilized. Alternatively, positive samples can be fortified. This approach is feasible if the fortification at least doubles the peak area but not yet causes the response to become non-linear. Samples containing high analyte concentrations may require a dilution. This provides two benefits. First, the peak area to be calibrated stays within the linear dynamic range. Second, signal suppression becomes a much less relevant issue.

Conclusions

Our proposed method permits the detection and quantification of many anionic food additives by using a single extraction and detection method. This comprehensive approach is an alternative to conventional methods, in which only a single compound or a set of compounds belonging to a particular food additive group can be analyzed. The resulting reduction of analysis time, manpower, dedicated instrumentation (e.g., suppressed IC), and consumables justifies the use of HRMS instrumentation. A significant advantage of the proposed method is the very high selectivity as provided by HRMS. As compared to conductivity or UV-based IC separations, HRMS permits the reliable detection of low analyte concentrations in heavy matrix extracts.

The development of this method revealed that even a high-end detection technique like HRMS requires chromatography for reliable quantitative and qualitative results. The observed phosphate rearrangements within the electrospray interface would lead to false-positive findings if the compounds were to elute in the column void volume.

This method will likely be a valuable tool in verifying the correctness of food additive declarations. The proposed separation technology may also be suitable for the detection of organic acids present in fruit juice or wine. Initial tests indicate that the potential number of analyzable compounds of interest is much larger. They were not included in the method because they possess lower polarities and therefore require more non-polar extraction solvents for their exhaustive extraction from the matrix, which would negatively affect the peak shape of early eluting inorganic ions.

Notes

Compliance with ethical standards

The authors adhered to ethical standards.

Conflict of interest

The authors declare that they have no conflicts of interest.

Ethical approval

No studies involving human participants or animals were conducted.

Informed consent

Not applicable.

References

  1. 1.
    Kubica P, Namiesnik J, Wasik A. Sweeteners and common Stevia rebaudiana glycosides in non-alcoholic and alcoholic beverages by reversed-phase liquid chromatography coupled with tandem mass spectrometry. J Anal Bioanal Chem. 2014;406:1505–12.  https://doi.org/10.1007/s00216-014-8355-x.CrossRefGoogle Scholar
  2. 2.
    Lorenzo-Fernandez M, Lopez-Hernandez J, Pineiro-Sotelo M, Rodriguez-Bernaldo A. High performance liquid chromatography for joint determination of the flavour enhancers monosodium glutamate, disodium guanylate and disodium inosinate in dehydrated bouillons. Dtsch Lebensm Rundsch. 2002;98:93–895.Google Scholar
  3. 3.
    Chang CS, Yeh TS. Detection of 10 sweeteners in various foods by liquid chromatography/tandem mass spectrometry. J Food Drug Anal. 2014;22:318–28.  https://doi.org/10.1016/j.jfda.2014.01.024.CrossRefPubMedGoogle Scholar
  4. 4.
    Goncalves AG, Rech TB, Rodrigues PM. Quality evaluation of frozen seafood (Genypterus brasiliensis, Prionotus punctatus, Pleoticus muelleri and Perna perna) previously treated with phosphates. Pan-Am J Aquat Sci. 2008;3(3):248–58.Google Scholar
  5. 5.
    Ruiz-Calero V, Galceran MT. Ion chromatography separations of phosphorus species: a review. Talanta. 2005;66:376–410.  https://doi.org/10.1016/j.talanta.2005.01.027.CrossRefPubMedGoogle Scholar
  6. 6.
    Kaufmann A, Maden K, Leisser W, Gude T. Analysis of polyphosphates in fish and shrimps tissues by two different ion chromatographyy methods: implications on false-negative and positive findings. Food Addit Contam. 2005;22:1073–82.  https://doi.org/10.1080/02652030500239565.CrossRefPubMedGoogle Scholar
  7. 7.
    Baluyot E, Hartford C. Comparison of polyphosphate analysis by ion chromatography and by modified end-group titration. J Chrom A. 1996;739:217–22.CrossRefGoogle Scholar
  8. 8.
    Iammarino M, Di Taranto A. Determination of polyphosphates in products of animal origin: application of a validated ion chromatography method for commercial analyses. Eur Food Res Technol. 2012;235:409–17.CrossRefGoogle Scholar
  9. 9.
    Dafflon O, Scheurer O, Gobet H, Bossed J. Polyphosphate determination in seafood and processed cheese using high-performance anion-exchange chromatography after phosphatase inhibition using microwave heat shock. Mitt Lebensm Hyg. 2003;94:127–35.Google Scholar
  10. 10.
    Gallidabinio MD, Hamdan L, Murphy B, Barron L. Suspect screening of halogenated carboxylic acids in drinking water using ion exchange chromatography–high resolution (Orbitrap) mass spectrometry (IC-HRMS). Talanta. 2018;178:57–68.CrossRefGoogle Scholar
  11. 11.
    Gilchrist E, Nesterenko P, Smith N, Barron L. Organic solvent and temperature-enhanced ion chromatography-high resolution mass spectrometry for the determination of low molecular weight organic and inorganic anions. Anal Chim Acta. 2015;865:83–91.  https://doi.org/10.1016/j.aca.2015.01.031.CrossRefPubMedGoogle Scholar
  12. 12.
    Saccani C, Tanzi E, Cavalli S, Rohrer J. Determination of nitrite, nitrate, and glucose-6-phosphate in muscle tissues and cured meat by LC/MS. J AOAC. 2006;89:712–9.Google Scholar
  13. 13.
    Studzinska S, Rola R, Buszewski B. The impact of ion-pairing reagents on the selectivity and sensitivity in the analysis of modified oligonucleotides in serum samples by liquid chromatography coupled with tandem mass spectrometry. J Pharm Biomed Anal. 2017;138:146–52.  https://doi.org/10.1016/j.pba.2017.02.014.CrossRefPubMedGoogle Scholar
  14. 14.
    Basiri B, van Hattum H, van Dongen WD, Murph MM, Barlett MG. The rule of fluorinated alcohols as mobile phase modifiers for LC-MS analysis of oligonucleotides. J Am Soc Mass Spectr. 2017;28:190–9.  https://doi.org/10.1007/S13361-016-1500-3.CrossRefGoogle Scholar
  15. 15.
    Birdsall RE, Gilar M, Shion H, Yu YQ, Chen W. Reduction of metal adducts in oligonucleotide mass spectra in ion-pair reversed-phase chromatography/mass spectrometry analysis. Rapid Commun Mass Spectr. 2016;30:1667–79.  https://doi.org/10.1002/rcm.7596.CrossRefGoogle Scholar
  16. 16.
    Konieczka P, Namiesnik J. Estimation uncertainty in analytical procedures based on chromatographic techniques. J Chrom A. 2010;1217:882–91.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Anton Kaufmann
    • 1
  • Mirjam Widmer
    • 1
  • Kathryn Maden
    • 1
  • Patrick Butcher
    • 1
  • Stephan Walker
    • 1
  1. 1.Official Food Control AuthorityZürichSwitzerland

Personalised recommendations