Traceable quantitation of cyanocobalamin (vitamin B₁₂) via measurement of cobalt and phosphorus: a comparative assessment using inductively coupled plasma atomic emission spectrometry (ICP-AES) and ion chromatography (IC)

Abstract

Traceable and precise quantitative measurements of cyanocobalamin (CN-Cbl) have been hampered by the lack of well characterized standards and pure materials of this bio-inorganic analyte that belongs to the water-soluble vitamins of the B-group known as vitamin B₁₂. Measurement of cobalt and/or phosphorus content of vitamin B₁₂ offer an approach for its quantitation that is traceable to the International System of Units (SI) with low measurement uncertainty. Cobalt and phosphorus measurements of CN-Cbl were carried out by Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES) and Ion Chromatography (IC). Use of a mixed bed ion exchange column coupled with post column reaction, IC provides a means to differentiate free cobalt from the cobalt complexed inside the corrin ring of the CN-Cbl molecule. In the case of ICP-AES and IC, a prerequisite for quality measurement is the purity of the starting vitamin B₁₂ material. The relative expanded uncertainties (% U) expressed at 95% confidence for these analyses range from 0.3 to 1%.

---

Introduction

Vitamin B₁₂, a tetrapyrrole complex which contains a cobalt atom in the molecule is an important nutrient. Vitamin B₁₂ is an essential nutrient for the maintenance of myelin sheath surrounding nerve cells in humans¹ that promotes growth, cell development and is essential for the rapid synthesis of DNA during cell division.² Vitamin B₁₂ deficiency may lead to excessive tiredness, and poor resistance to infection. Prolonged deficiency leads to nerve degeneration and irreversible neurological damage.³ The daily requirement of this vitamin is only as low as 1 to 2 μg.⁴ On the other hand, excessive consumption of vitamin B₁₂ may cause asthma and folic acid deficiency. So, the cyanocobalamin (CN-Cbl) content requires very precise determination at nano to pico gram levels in many areas of application like clinical, pharmaceutical and food products.⁵ Because of its availability and stability, the CN-Cbl is the form that is used to make primary calibration standards or as supplements in food and vitamin in preference to hydroxocobalamin, methyl cobalamin (Me-Cbl) and 5′-deoxyadenosylcobalamin (Ado-Cbl), which only differ with respect to the ligand attached to the cobalt atom at the centre of the corrin complex. Various analytical methods including microbiological assay,⁶ radioisotope dilution assay,⁷ spectrophotometry,⁸ chemiluminescence,⁹ atomic absorption spectrometry,¹⁰ matrix assisted laser desorption ionization-time of flight mass spectrometry,¹¹ high performance liquid chromatography with mass spectrometry,¹² and fluorescence detection¹³ have been proposed for the determination of vitamin B₁₂. Normally in many cases, the concentration of the stock solution for vitamin B₁₂ is measured using UV-visible spectrophotometry,¹⁴ which is based on the approximation that OD = 1 at 368 nm corresponds to 43.9 μg mL⁻¹ of analyte. However, this value being an approximation, the accuracy of such measurements is limited and the method uses highly toxic potassium cyanide. Although these methods are routinely used for many applications, they are not deemed appropriate for high accuracy and precise measurements to underpin materials that are meant to be used as primary standards and high-caliber certified reference materials of vitamin B₁₂.

Despite the availability of numerous vitamin B₁₂ assays,^14,15 there remains a continuing need for an SI (International System of Units) traceable reference method, as traceability is an important parameter of reference materials. Traceability ensures quantitation comparability between different laboratories, different instrumental techniques, and measurement made at different times. A detailed description of the importance of accurate and traceable quantitation with small uncertainty of CN-Cbl is described elsewhere.^5,12,15

Recently in a commercially available CRM of cyanocobalamin¹⁶ traceability has been established to reference materials (RMs) where CN-Cbl content has been measured via UV-visible spectrophotometry and hence not traceable to the SI. The stoichiometric existence of cobalt in the CN-Cbl is measured by AAS¹⁰ and capillary electrophoresis inductively coupled plasma mass spectrometry¹⁷ in order to determine the respective vitamin content in various vitamin tablets. Additionally cobalt measurements have previously made by coupling HPLC to element specific detector like ICP-AES¹⁸ or ICP-MS¹⁹ in the determination of CN-Cbl. Recently, rapid determination of CN-Cbl was carried out using silver nanoclusters in various pharmaceutical formulations.²⁰ Though these methods are used successfully in various matrices routinely for B₁₂ determination with accuracy between 4–10%, they are not suitable for providing relative expanded uncertainties of ≤1%. In the cyanocobalamin molecule, phosphorus is also present stoichiometrically and hence the measurement of phosphorus and cobalt content could provide different independent ways of traceable measurements using the available certified reference materials (CRMs) of phosphorus and cobalt. Small uncertainty is achieved through precision measurement approaches using the “high performance (HP)” methodology^21–24 developed at National Institute of Standards and Technology (NIST), for cobalt and phosphorus by both ICP-OES and IC.

This paper evaluates the HP methodology by two instrumental techniques (ICP-AES and IC) to provide traceable measurement of CN-Cbl mass with small uncertainties via measurement of cobalt and phosphorus content. The effectiveness of microwave assisted UV digestion for the mineralization of CN-Cbl and simultaneous conversion of phosphate moiety to orthophosphate ion and the release of cobalt ion from CN-Cbl molecule have been investigated prior to cobalt and phosphate measurement by IC. Ion chromatography offers a route to determine extraneous free cobalt which provides a correction route to obtain better accuracy in the calculation of CN-Cbl mass fraction. Based on the experimental results, the proposed measurement methods will be suitable for characterizing primary standards and certified reference materials of CN-Cbl with relative expanded uncertainties of less than 1%.

Experimental section

Reagents and chemicals

All reagents were of analytical grade and contain very low concentrations of trace metals as found in the specification levels provided by the manufacturer. Hydrochloric acid used was GR grade (Merck, India). Sodium perborate (NaBO₃·4H₂O), pyridine-2.6-dicarboxyclic acid (PDCA) and 4-(2-pyridylazo) resorcinol (PAR) were obtained from Sigma-Aldrich (Bengaluru, India). Potassium hydroxide, potassium sulfate, formic acid, 2-dimethyl amino ethanol, ammonium hydroxide and sodium bicarbonate were of analytical reagent grade (Alfa Aesar – Thermo Fisher, USA). A stock solution of sodium hydroxide (50% w/w) from Fisher Chemical (USA) was used to prepare eluent for IC separation of phosphate. Ultra pure water with resistivity 0f 18 MΩ cm was obtained from a Milli-Q system (Millipore, Bedford, USA). The CRM from which calibration solutions for phosphorus was prepared for the ICP-AES was the CertiPUR® reference material (phosphorus standard solution, lot no. HC934479). The certified P mass fraction is 1.00 mg g⁻¹, which was traceable to NIST SRM® 3139a, lot 060717. The cobalt CRM (CERTISOL®, batch no E082/0408/2104/96, traceable to NIST SRM 3113) was obtained from SD Fine Chem (India) and the certified Co mass fraction is 1.00 mg g⁻¹ and this was used for the calibration of both ICP-AES and IC measurements. For the calibration of IC, a CRM of phosphate (CertiPUR®, lot no. BCBM077V, traceable to SI unit kg), was used whose certified mass fraction is 1.00 mg g⁻¹. Nitrate CRM (CertiPUR®, lot no. BCBP0387V) was used for making matrix matched calibrant used for IC analysis. All the CertiPUR® standard stock solutions were procured from Sigma-Aldrich (Bengaluru, India). The CN-Cbl used in this study was from Sigma-Aldrich (Bengaluru, India) bearing lot no. MKBR6786V, with a stated purity level of >98%. Before making the stock solution of CN-Cbl, the mass fraction of water content in it was determined using a Karl-Fisher auto titrator, where the titrator was calibrated with 1-octanol solution. Appropriate quantity (25 mg) of CN-Cbl was weighed using a calibrated four decimal balance. The CN-Cbl was dissolved in a volume of aqueous solution (pH = 4.3, using 0.1 M HCl) so as to obtain a mass fraction of 500 μg of CN-Cbl g⁻¹. The density of CN-Cbl solution at this mass fraction was found to be essentially same as water (0.998 g cm⁻³ versus 0.996 g cm⁻³) for DI water at room temperature. A dark colour borosilicate glass bottle was used to prepare the CN-Cbl solution to protect from light and stored at 4 °C.

Instrumentation

A Multiwave-3000 pressurized microwave digestion device, equipped with quartz vessels (80 mL capacity) from Anton Paar GmbH (Graz, Austria) was used for digestion. Cadmium low-pressure discharge microwave lamp (part no. 16846; Anton Paar GmbH) was used as the UV source. Each quartz vessel and its corresponding lamp were treated like a unit to avoid any contamination. Each reaction vessel included an electrode less Cd lamp, a PTFE based stand and ring in order to maintain the UV lamp in the vertical position in order to avoid damage, an unit arrangement, which is shown in the ESI Fig. 1.† The maximum pressure and temperature were set at 75 bar and 240 °C respectively. An axial-view ICP-AES (Agilent Technologies 725 series, USA) with solid state CCD detector having a free running 40 MHz RF generator was used for the measurement of phosphorus and cobalt. The instrumental details and setting parameters are presented in Table 1. These parameters produced robust plasma giving Mg(II) 280.270 nm to Mg(I) 285.213 nm ratios consistently greater than 8.

Table 1 ICP-OES instrument details

Parameter	Setting
Plasma Ar flow rate	15 L min^−1
Auxiliary Ar flow rate	0.5 L min^−1
Nebulizer Ar flow rate	0.8 L min^−1
Sample uptake rate	1 mL min^−1
Nebulizer	Concentric with cyclonic spray chamber
Analytes wavelengths	P (213.618 nm); Co (238.892 nm)
Internal std. wavelengths	Ge (265.117 nm); Sc (361.383 nm)
Measurement time	10 s
Replicate measurement	6

---

Phosphate, total cobalt and free cobalt ions analyses were performed using an ICS-3000 ion chromatography system (Dionex, Sunnyvale, CA, USA) equipped with a quaternary gradient pump; detector compartment containing a conductivity detector (ICS-3000 series) with suppressor, ICS-3000 diode array detector (DAD) and post column reagent delivery module. The chromatography oven was utilized to help prevent baseline drift caused by temperature variation. The instrument control, data acquisition and processing were performed with Chromeleon® software (version 6.80). The detailed chromatographic conditions are listed in Table 2. The residual carbon content (RCC) of the CN-Cbl digest was measured using a TOC analyzer (TOC-V_CPN 5000A, Shimadzu, Japan).

Table 2 Ion chromatography operating conditions of ICS-3000 for analysis of phosphate and cobalt

Column for phosphate and bromide (internal standard)analysis	Dionex IonPac AS20, (250 mm × 4 mm)
Guard column	Dionex IonPac AG20, (50 mm × 4 mm)
Detection	Suppressed conductivity; conductivity cell set at T = 35 °C
Injection loop volume	25 μL
Eluent, flow rate	NaOH (20 mM), 1 mL min^−1
Suppressor	ASRS-300 (4 mm), current: 60 mA
Column for total cobalt and nickel (internal std.) analysis	Dionex IonPac CS5A, (250 mm × 4 mm) (50 mm × 4 mm)
Guard column	Dionex IonPac CG5A, (50 mm × 4 mm)
Eluent	7 mM PDCA + 66 mM KOH + 5.6 mM K2SO4 + 74 mM HCOOH, 1 mL min^−1
Flow rate	(pH = 4.3)
Post-column reagent (PCR)	0.4 mM PAR + 1 M 2-dimethylaminoethanol + 0.5 M NH4OH + 0.3 M NaHCO3 (pH = 10.4)
Flow rate (PCR)	0.5 mL min^−1
Detection	Visible absorbance at 530 nm
Injection volume	50 μL (total cobalt); 400 μL (free cobalt)
Oven and column temperature	30 °C

---

Sample preparation

Sample and calibrant preparation following HP methodology. The sample and calibrant preparations were carried out as per high performance methodology for ICP-AES and IC.²³ Briefly, HP methodology²⁵ employs a careful experimental comparison of a sample with a calibration standard, that is prepared to mimic the expected nature of the sample in terms of both matrix and analyte mass fraction. Germanium, scandium and bromide were used as internal standards for the ICP-AES and IC measurements respectively to compensate for short term noise and a drift correction procedure^22,23 to correct for long term noise. In the case of determination of free cobalt by IC, HP methodology was not used. All sample handling dilutions and addition of internal standards were carried out gravimetrically and the amount fraction is determined as mass fraction, thus eliminating the uncertainty associated with density and its temperature dependence.

Microwave-assisted UV-digestion (IC analysis). Each quartz vessel with UV lamp unit was cleaned by introducing 6 mL of concentrated nitric acid and subjecting to microwave heating for 2 min at 1000 W followed by 0 watt for 15 min, and air dried. As the weight of the quartz MW vessel (380 g) lies outside the range of analytical weighing balance, required quantities of the aliquots of CN-Cbl sample (P mass fraction: 2.0–5.0 μg g⁻¹; Co mass fraction: 4.0–8.0 μg g⁻¹), required amounts of internal standards (Ni²⁺ and Br⁻ for Co²⁺ and PO₄³⁻ respectively) and other reagents (2 mL of 1 g L⁻¹ of sodium perborate, 50 μL of 1 M HCl) were weighed into PFA vial (30 mL capacity, Cole Parmer, USA) and gravimetrically made up to 10 g with DI water, mixed well and transferred into the quartz microwave vessel containing the lamp. Once the solutions are added and mixed well, the amount ratio is fixed and thus potential bias associated with the quantitative transfer and subsequent dilutions is eliminated. The microwave heating programme was 1000 watt for 20 min followed by 0 watt for 40 min and then allowed to cool to room temperature. The digestion solution was transferred into a PFA vial for IC measurement. Three blanks were prepared containing only the reagents required for photolysis (sodium perborate and hydrochloric acid solution). The matrix matched (sodium perborate, hydrochloric acid and nitrate ions) calibrant and CN-Cbl digests were injected in randomised block sequence.

Conventional microwave assisted digestion (ICP-AES analysis). The conventional microwave digestion method using dilute nitric acid was used to digest the CN-Cbl. A required quantities of the aliquots of CN-Cbl sample (P mass fraction: 4–5 μg g⁻¹; Co mass fraction: 6–9 μg g⁻¹), internal standard (Ge & Sc at a mass fractions of 0.7 and 0.2 μg g⁻¹ respectively) and other reagents (0.4 mL of conc. nitric acid) were weighed into PFA vial (30 mL capacity, Cole Parmer, USA) and gravimetrically made up to 10 g with DI water, mixed well and transferred into the quartz microwave vessel. The microwave heating programme was 1000 watt for 10 min and was allowed to be cooled to room temperature and the intensity ratios were measured by ICP-AES.

Sample preparation for free cobalt determination. CN-Cbl stock (mass fraction 500 μg g⁻¹) was directly injected into IonPack column without any sample pretreatment.

Results and discussion

Optimization of MW-UV digestion for IC

For the determination of cobalt in CN-Cbl, complete mineralization of the molecule is required. When spectroscopic techniques like ICP-AES or GFAAS etc. are used for the determination of cobalt in pharmaceutical products, nitric acid (6 mL) and hydrogen peroxide are used in the microwave assisted acid digestion.²⁶ However considering the basic eluent that is normally used for the separation of phosphate in IC, conventional MW assisted acid digestion could not be used, as the low pH of the resulting digest can cause disruption of the multiple ion exchange equilibrium existing between the eluent species and the column, leading to severe base line perturbations. Further, for the separation of cobalt in IonPack CS5A column, the resulting highly acidic digest need to be diluted to a pH range of 4–5, which would impact adversely for the precise determination of cobalt. Additionally, for the digestion the organo-phosphorus compound, hydrochloric acid is necessary²⁷ and photo-oxidative mineralization^26,28 requires additional oxidants like hydrogen peroxide. So a MW based UV digestion²⁴ was optimised where mineralization is usually achieved using much lesser quantity of dilute acid.

Multiple digestion experiments were carried out with intentionally low quantity of hydrochloric acid (50 μL of 1 M HCl) with increasingly varying amount of hydrogen peroxide as the later reagent decomposes into water and as a result the digest is better suited for suppressed conductivity detection of phosphate by IC. However, it was seen that with the increase of hydrogen peroxide, the phosphate peak height increased correspondingly. Two millilitre of AR grade hydrogen peroxide contributes to blank values of 5 μg of phosphate (ESI Fig. 2†). Even using suprapur grade H₂O₂, the blank values for phosphate were in the range of 1–2 μg of phosphate. Phosphates and pyrophosphates are often added as stabilizers²⁹ to various grades of hydrogen peroxide, which contributes to the process blank. To circumvent this blank problem, sodium perborate was used that produced in situ hydrogen peroxide and the blank levels for phosphate were reduced by a factor 50 times, compared to the blank values obtained with AR grade hydrogen peroxide. The in situ production of hydrogen peroxide is explained in the following reaction:

[NaBO(OH)₂·3H₂O]₂ ↔ 2NaBO₂ + 2H₂O₂ + 6H₂O (1)

Hydrogen peroxide produced in eqn (1) degrades to active oxygen and water as follows:

2H₂O₂ → 2H₂O + 2O˙ (2)

Further, H₂O₂ absorbs UV light and undergoes O–O bond cleavage from its electronically excited state, leading to hydroxyl radical (OH˙) production and these hydroxyl radicals initiate the degradation of organics into carbon dioxide and water.²⁸ Though nitric acid was not used in the digestion process, nitrate ions originated due to the presence of nitrate in the sodium perborate solution and the conversion of nitrogen atoms present in the CN-Cbl due to oxidative photolysis. Consequently, nitrate ions are additional major source of OH˙ free radicals which further initiates radical chain reactions with organic substance as follows:

NO₃⁻ + H₂O + hv → NO₂ + OH⁻ + OH˙ (3)

Multiple MW-UV experiments were carried out using varying amounts of sodium perborate (0.5–3.0 mL, 0.1% w/v); time duration and the residual carbon content (%) of the vitamin digest (mass fraction ∼ 500 μg g⁻¹) was determined to evaluate mineralization efficiency. A combination of 2 mL of perborate with 15 min of MW assisted photolysis provided a lowest achievable residual carbon content of 0.2%. The fast mineralization of the CN-Cbl may be attributed to: (a) high efficiency of UV photons due to 4π geometry and the attainment of temperature of 230 °C. Further, the oxidizing power of hydrogen peroxide is enhanced in the presence of UV light²⁸ and most importantly the digestion is carried out using only 50 μL of dilute hydrochloric acid as a result of which the digest is amenable for the IC analysis of Co and P without further dilution. The complete decolourization of CN-Cbl and water like solution obtained after microwave assisted photolysis indicates the efficacy of mineralization (ESI Fig. 3†).

Quantitation of phosphorus by IC

An assumption is made in this experiment that the proportion of standard to unknown used to calculate CN-Cbl mass fraction is equivalent whether phosphorus or phosphate is measured. Phosphorus is measured in IC as phosphate. So for the calibration of IC, a CRM of phosphate anion was used. In the CN-Cbl photo-digest, the chloride and nitrate mass fractions were ∼150 and 300 μg g⁻¹ respectively. The peaks of phosphate (analyte) and bromide (internal standard) were base line resolved (Fig. 1). However, Brennan et al.²³ had advised that quantitation of anions through “high performance” methodology by IC containing significant anionic matrix requires in-depth investigation relating to matrix matching of the calibrant to ensure that the analytical sensitivity is consistent throughout the analysis. Therefore experiments were carried out to evaluate the effect of nitrate and chloride ions on the ratio of phosphate/bromide peak height. Set of three solutions were prepared each of that contained identical phosphate and bromide mass fractions in the presence of chloride (150 μg g⁻¹) and nitrate (300 μg g⁻¹), mass fractions. These ratios of phosphate to bromide were compared with the results obtained when no chloride and nitrate were present. The presence of chloride and nitrate at their maximum mass fraction found in the digest, a 2% reduction in the phosphate to bromide peak height ratio was observed. An error of 2% could not be neglected, when concentration uncertainties of 0.5% or better are required for the reference measurement. Therefore, in the present experiment, close matrix matching of the calibrant was carried out to ensure that the analytical sensitivity is consistent throughout the analysis. For chloride matrix, exact amount of hydrochloric acid was added to the calibrant and the nitrate was added from an aqueous solution of nitrate CRM and additionally, that way the pH of the calibrant was close to the UV digest of cyanocobalamin.

Fig. 1 Chromatogram showing phosphate in the MW-UV digest of CN-Cbl (mass fraction: 505.76 μg g⁻¹). For details, see Experimental section.

Quantitation of free cobalt by IC

Mostly, cyanocobalamin is produced via biosynthetic fermentation process where substantial amount of cobalt salt solution is added to the culture media.³⁰ Hence it becomes imperative to determine the extraneous cobalt ion in CN-Cbl that is exogenous to the corrin ring and thus offers a route to correct for the total cobalt mass fraction. This is an important aspect as the stoichiometric ratio between CN-Cbl to cobalt is very large, (23 times) that will in turn affect the accuracy of CN-Cbl mass fraction. Notwithstanding the development of sensitive analytical techniques for the determination of extremely trace levels of cobalt, difficulties still lie in the speciation of free cobalt as outlined by Baars et al. and the references there in.³¹ Recently, Berton et al.³² have successfully carried out the determination of vitamin B₁₂ and inorganic cobalt which involves a number of steps like chelation of inorganic cobalt, extraction into ionic liquids (IL) and dissolution of the enriched phase (IL) into methanol followed by quantitation using GFAAS. One major requirement for the speciation is the minimal sample preparation and maintaining the integrity of species and preventing their inter-conversion. Considering these aspects, ion chromatographic method³³ is ideally suitable for speciation of transition metal ions, where improved detection limits are achieved by large volume direct injection without any sample pretreatment. The CN-Cbl molecule is highly stable in the pH range of 4.0–4.5 in the absence of any oxidizing/reducing agents.³⁴ Taking these facts into account, the CN-Cbl stock solution was prepared in an aqueous solution where the desired pH was brought in using a non-oxidizing dilute hydrochloric acid. Further, the complexing PDCA eluent used had a pH of 4.2 as required for the mixed bed IC column for the separation of transition metal cations and hence is in compliance with the stability requirements of CN-Cbl.

The validation for the determination of free cobalt was established by spike recovery test which was found to be in the range of 99.6–99.8%. Representative chromatogram of the free cobalt determination in CN-Cbl is presented in Fig. 2. The CN-Cbl molecule possessing reverse phase properties is not retained in the mixed bed ion exchange column. Repeated injection of the CN-Cbl at a mass fraction of 500 μg g⁻¹ does not change the retention time of the free cobalt ion.

Fig. 2 Chromatogram showing presence of free cobalt in CN-Cbl samples when injected directly.

Matrix matching for ICP-AES

Initially it was assumed that the residence time in the ICP-plasma would be sufficient for the mineralization of CN-Cbl. To verify the validity of the assumption, the intensity ratios of analytes to their internal standard ratios were measured and compared between digested and undigested CN-Cbl for at a mass fraction of 250 μg g⁻¹. It was observed that the P/Ge and Co/Sc intensity ratios were higher by 2.67% and 0.5% respectively in the digests, compared to the undigested CN-Cbl. This observation indicated that ICP-AES measurement of phosphorus is slightly influenced by matrix effects and in the present case the matrix happens to be nitric acid and the organic matrix of the CN-Cbl. The nitric acid matrix could be matched closely in the calibrant by adding the same amount of the acid in the calibrant as added before digestion. The carbon matrix of CN-Cbl was mineralized after microwave digestion (2–4 ppm of residual carbon) compared to the carbon concentration of 110 μg g⁻¹ of the undigested cyanocobalamin sample and thus in a way closed matrix matching with the calibrant was achieved after microwave digestion. However, higher intensity ratio of 2.67% of the digested sample does not contribute significantly to the targeted relative expanded uncertainties (U_rel) of 1%.

Drift correction

The ratios of the analyte intensities to the internal standard intensities are calculated and the resulting ratios are corrected for drift using a drift correction approach described by Salit et al.^21,22,35 In the analytical method, the set of calibration standards and the set of CN-Cbl sample solutions were gravimetrically prepared, so that all standards and samples are nominally the same with respect to the analyte mass fractions and in the internal standard mass fractions. Such close matching of calibrant and sample results in better accuracy and precision. Five preparations were made gravimetrically each for the calibration standard and the cyanocobalamin sample. Two blanks were prepared containing the reagents required for respective digestion approaches for the ICP-AES and IC. All the preparations were run five times in a randomized complete block sequence, until each solution has been passed once, then again in a randomized sequence, until each solution has been passed a second time, and so forth, until each preparation has been passed five number of times. So, in all there were 60 readings, 12 per block times 5 blocks. The analyte/internal standard signal ratios were calculated and corrected for drift after the subtraction of the blank values. The observed cobalt to scandium intensities for all injections (60) of the randomized block sequence is plotted against the solution run sequence. The replicate signal ratios pertaining to each solution were normalized to their mean value in this plot to allow the data for all solutions to be plotted on a common ordinate scale. A polynomial is fitted to the plotted data. The equation for the fitted polynomial is then used to correct the signal ratios for the drift. Fig. 3 represents the ICP-AES instrumental drift for cobalt measurement. The numerical model fitted to the data is also depicted in the figure. The effectiveness of the drift correction was evident by comparing the RSD values of replicate measurements with and without drift correction applied for both the techniques. The RSD with drift correction for a cobalt mass fraction of 8 μg g⁻¹ was smaller (0.95%) compared to an RSD value of 1.0, which was obtained without drift correction.

Fig. 3 ICP-OES instrumental drift for Co²⁺ measurement with Sc as the internal standard using peak height. The polynomial equation depicted inside the figure is used to correct the signal ratios for the drift.

Cyanocobalamin mass fraction

The analyte/internal standard signal ratios were calculated and corrected for drift³⁵ and the blank intensities were subtracted. The high performance methodology is a relative method that compares the analyte-to-internal standard signal ratios measured in an unknown sample to those ratios measured in a calibrant whose amount ratio is well known. The differences between the analyte (Co, P) mass fraction of the calibration standard and the CN-Cbl digest observed instrumentally are used to calculate the Co and P mass fraction of the cyanocobalamin stock. Eqn (4) (ref. 22) is used to calculate the phosphorus/cobalt mass fraction in the CN-Cbl from the measured signal and mass ratios of the calibrant and the CN-Cbl solution or its digest:where I is the signal intensity and m is the mass of the phosphorus or cobalt in the calibrant and CN-Cbl digest.

A set of different mass fraction levels of CN-Cbl prepared from the purified CN-Cbl stock (500 μg g⁻¹) were digested by MW-UV, the P and Co mass fractions obtained from IC and ICP-AES analyses are presented in Table 3. Consistent results were obtained with both the techniques over a range of starting mass fractions when back calculated to the original CN-Cbl stock mass fraction. The calculated CN-Cbl mass showed relative expanded uncertainties that were in the range of 0.3 to 1.0%. There is statistically good agreement between the IC and ICP-AES values for CN-Cbl mass fraction, although the IC values had comparatively higher uncertainties. This agreement further underpins that digestion and conversion of phosphorus of CN-Cbl to orthophosphate ion by MW-UV photolysis is quantitative. IC measurements require much smaller sample size than ICP-AES. However, it may be noted that in the present experiment, digestion of the CN-Cbl was carried out in a larger solution volume of ∼10 mL, mainly due to the requirement of the available microwave (MW) lamp-quartz vessel combination. In other way, IC provides a route for the quantitation of the free cobalt ion, which improves the accuracy for the determination of CN-Cbl mass fraction. The mass fraction of CN-Cbl obtained from total phosphorus is higher than the same obtained by total cobalt measurements. The higher mass fraction obtained via phosphorous measurements could be attributed to the normally much larger quantity (2000 times) of phosphate salts compared to the cobalt salt used during the biosynthesis of CN-Cbl, where higher traces of extraneous phosphorous as impurity might remain after purification of CN-Cbl. Thus cyanocobalamin mass fraction obtained through the quantitation of phosphorus has systematic error by both ICP-AES and IC techniques. The mass fractions of cyanocobalamin obtained via cobalt measurement by both the techniques are in close agreement. However, as IC provides a route for the quantitation of free cobalt, a more accurate cyanocobalamin mass fraction is obtained by subtraction of cyanocobalamin mass fraction corresponding to the free cobalt. In the investigated sample, the free cobalt mass fraction was measured to be 0.076 μg g⁻¹ which corresponds to a CN-Cbl mass fraction of 1.75 μg g⁻¹. Thus the corrected mass fraction of cyanocobalamin (Co_total − Co_free) was found to be 496.22 μg g⁻¹ obtained from IC against 497.32 μg g⁻¹ of ICP-AES (Table 3).

Table 3 Quantitation of the mass fraction of phosphorus, cobalt and free cobalt in cyanocobalamin by ICP-OES and IC

	Mass fraction^a (μg g^−1)	Corrected^b mass fraction (μg g^−1)	Cyanocobalamin mass fraction (μg g^−1)
a The values in parenthesis represent expanded uncertainties.b Mass fraction corrected to the original CN-Cbl stock from which all the samples were prepared.
ICP-OES
Phosphorus	4.610 (0.021)	11.56 (0.05)	505.76 (2.4)
Cobalt	8.648 (0.03)	21.62 (0.08)	497.32 (1.9)
IC
Phosphorus	2.263 (0.02)	11.32 (0.09)	505.30 (3.8)
Cobalt (total)	4.335 (0.04)	21.65 (0.19)	497.97 (4.4)
Cobalt (free)	0.076 (0.006)	—	1.75 (0.138)

---

Uncertainty evaluation

The recommendations from the ISO Guide to the Expression of Uncertainty in measurement,³⁶ were used to estimate the expanded uncertainties expressed at the 95% level of confidence taking into consideration all systematic and random sources of uncertainty. Eqn (5)–(7) is used to calculate the cobalt/phosphorus mass fractions from the measured signals from ICP-AES or IC. The standard uncertainties of the measured ratios (analyte/internal standard) were calculated as the standard deviation from the six replicate measurements. The variation of replicate analyses was estimated by using the standard deviation of the mean of the five replicate readings. Variability in sample dilution/digestion is evaluated with replicate dilution/digestion; variability in calibrant preparation is quantified with replicate calibrant preparation. Uncertainties in the known value for the cobalt/phosphorus CRM standards were accounted for the calculation of the expanded uncertainty. All weighing were performed on a four figure analytical balance. The standard uncertainty on each mass was determined from the repeatability of calibrated weight measurements and the balance certificate. In particular, expanded uncertainties were determined for phosphorus and cobalt measurements using the following equations:

U = ku_c (6)

% U = U × 100/x (7)

where u_i (i = 1, 2, 3…) represents the individual component of uncertainty, u_c is the combined uncertainty, k is the coverage factor (2), U is the expanded uncertainty, and x is the observed measurement of cobalt mass. The uncertainty reported is the expanded uncertainty and is dominated by the standard deviation of replicate measurements. A typical relative expanded uncertainty for cobalt mass fraction for 4.335 μg g⁻¹ of CN-Cbl by IC is given in Table 4.

Table 4 Cobalt mass fraction and uncertainty components

	Type of uncertainty (A/B)	IC
Cobalt mass fraction (μg g^−1)		4.335
Uncertainty due to replication variability	A	0.01223
Uncertainty in cobalt CRM	B	0.00005
Uncertainty due to instrument sensitivity variability	A	0.01382
Uncertainty due to blank variability	A	0.00567
uc (combined uncertainty)		0.019306
k (expansion factor)		2
U (expanded uncertainty)		0.038612
% Urel (relative expanded uncertainties)		0.8906

---

Comparison with other methods: cost, complexity, uncertainty and accuracy

The described method is intended to be used as a reference method for the characterisation of cyanocobalamin CRM. In order not to boost the uncertainty in routine analysis, the uncertainty of the CRM should be as small as possible. The uncertainty budget of a CRM usually consists of three main components, which are contributed from homogeneity study, stability study and the measurement study carried out by a reference or primary method of measurement, where the major source of uncertainty is from reference measurement method. So this necessitates that the analytical reference procedure should have a relative expanded uncertainties of 1–2% or preferably below 1% expressed at 95% confidence interval and traceable to SI units. Further not to boost the price of CRM, the reference method should be cost effective and less complex to be used by the participating laboratories in an Inter Laboratory Comparison Exercise (ILCE) for precise measurement. In the described ICP-AES and IC methods, all significant components of uncertainties like variability in measurement, variability in sample and calibrant preparation etc. have been accounted for. Traceability to SI units is established using an inorganic CRM of cobalt. Possible advantages of the proposed methods over other methods pertaining to method complexity, cost, uncertainty and accuracy are compiled in Table 5. The methods described provide a relative expanded uncertainty ≤1% compared to other methods^18,37 where only the measurement uncertainties are between 4–5% and the complete uncertainty budgets are not provided and method like HPLC-ICPMS¹⁹ is costly and complex.

Table 5 Comparative assessment with other methods

Method name (cobalt measurement)		Measurement precision	Relative expanded uncertainty	Traceability to SI units	Operation cost	Method complexity
ICP-AES^18		5–10%	Not mentioned	Not established	Moderate	Easy
HPLC-ICPMS^19		5–10%	Not mentioned	Not established	Costly	Complex
Ag-nano cluster^20		2–4%	Not mentioned	Not established	Very low	Easy
Present method	IC	0.3–0.7%	0.5–1%	Established	Low	Easy
	ICPAES	0.3–0.6%	0.5–1%	Established	Moderate	Easy

---

Conclusions

Both instrumental techniques provide the traceable measurements of total phosphorus and cobalt values and the trueness of mass fraction of CN-Cbl obtained by such approaches depends upon the purity of the CN-Cbl stock available. The quantitation of extraneous phosphorus, exogenous to the CN-Cbl, could not be carried out by both techniques. However, correction for the presence of free cobalt could be obtained by ion chromatography method. Hence the combined measurement of total cobalt and free cobalt by IC which provides accurate mass fraction and lower measurement uncertainty should be a preferred approach for quantitation of CN-Cbl. This approach will be an improved alternative for the traceable reference measurements of CN-Cbl compared to the conventional UV-vis measurements. The relative expanded uncertainties (% U) expressed at 95% confidence for these analyses range from 0.3 to 1.0%. The described reference method for the traceable quantitation of cyanocobalamin for the production of a CRM and/or primary standard using ICP-AES and IC is new. In order to obtain relative expanded uncertainty ≤1%, a high performance methodology described for other matrices was followed. The estimation of the total cobalt and the free cobalt present as impurity in CN-Cbl has not been published so far in literature.

Acknowledgements

Grateful acknowledgement is made to Dr Sunil Jai Kumar, Head, for his keen interest and encouragement throughout this work. We acknowledge C-MET, Hyderabad, for HP-ICP-OES measurements.

References

1. E. E. Ziegler, L. J. Filer and I. L. S. I.-N. Foundation, Present Knowledge in Nutrition, ILSI Press, 1996.
2. G. F. Combs, The Vitamins, Elsevier Science, 2012.
3. V. Herbert, Am. J. Clin. Nutr., 1987, 45, 671–678.
4. B. V. R. C. Gopalan and S. C. Balasubramanian, Nutritive Value of Indian Foods, National Institute of Nutrition, Hyderabad, India, 2000, p. 2000.
5. S. S. Kumar, R. S. Chouhan and M. S. Thakur, Anal. Biochem., 2010, 398, 139–149.
6. G. F. M. Ball, Bioavailability and Analysis of Vitamins in Foods, Springer, US, 2013.
7. K. Muhammad, D. Briggs and G. Jones, Food Chem., 1993, 48, 431–434.
8. H. Karşilayan, Spectrochim. Acta, Part A, 1996, 52, 1163–1168.
9. P. Viñas, N. Campillo, I. López-García and M. Hernández-Córdoba, Chromatographia, 1996, 42, 566–570.
10. F. R. Adolfo, P. C. do Nascimento, D. Bohrer, L. M. de Carvalho, C. Viana, A. Guarda, A. Nunes Colim and P. Mattiazzi, Talanta, 2016, 147, 241–245.
11. X. Fei, G. Wei and K. K. Murray, Anal. Chem., 1996, 68, 1143–1147.
12. X. Luo, B. Chen, L. Ding, F. Tang and S. Yao, Anal. Chim. Acta, 2006, 562, 185–189.
13. H.-B. Li, F. Chen and Y. Jiang, J. Chromatogr. A, 2000, 891, 243–247.
14. A. P. De Leenheer and W. Lambert, Modern Chromatographic Analysis Of Vitamins: Revised And Expanded, Taylor & Francis, 2000.
15. V. Ollilainen, P. M. Finglas, H. van den Berg and I. de Froidmont-Görtz, J. Agric. Food Chem., 2001, 49, 315–321.
16. http://www.sigmaaldrich.com/catalog/CertOfAnalysisPage.do?symbol=PHR1234%26LotNo=P500234%26brandTest=SIAL.
17. S. A. Baker and N. J. Miller-Ihli, Spectrochim. Acta, Part B, 2000, 55, 1823–1832.
18. E. Svantesson, J. Pettersson and K. E. Markides, J. Anal. At. Spectrom., 2002, 17, 491–496.
19. H. Chassaigne and R. Łobiński, Anal. Chim. Acta, 1998, 359, 227–235.
20. F. Qu, Q. Song and J. You, Anal. Methods, 2016, 8, 4324–4327.
21. M. L. Salit, R. D. Vocke and W. R. Kelly, Anal. Chem., 2000, 72, 3504–3511.
22. M. L. Salit, G. C. Turk, A. P. Lindstrom, T. A. Butler, C. M. Beck and B. Norman, Anal. Chem., 2001, 73, 4821–4829.
23. R. G. Brennan, T. A. Butler and M. R. Winchester, Anal. Chem., 2011, 83, 3801–3807.
24. K. Dash, L. Rastogi and J. Arunachalam, Analyst, 2012, 137, 668–674.
25. M. J. Holden, S. A. Rabb, Y. B. Tewari and M. R. Winchester, Anal. Chem., 2007, 79, 1536–1541.
26. A.-I. Stoica, M. Peltea, G.-E. Baiulescu and M. Ionica, J. Pharm. Biomed. Anal., 2004, 36, 653–656.
27. V. Ruiz-Calero and M. T. Galceran, Talanta, 2005, 66, 376–410.
28. J. Golimowski and K. Golimowska, Anal. Chim. Acta, 1996, 325, 111–133.
29. J. Kerth and D. Jensen, J. Chromatogr. A, 1995, 706, 191–198.
30. A. Kośmider, W. Białas, P. Kubiak, A. Drożdżyńska and K. Czaczyk, Bioresour. Technol., 2012, 105, 128–133.
31. O. Baars and P. L. Croot, Mar. Chem., 2015, 173, 310–319.
32. P. Berton, E. M. Martinis, L. D. Martinez and R. G. Wuilloud, Anal. Chim. Acta, 2012, 713, 56–62.
33. H. Small, Ion Chromatography, Springer, US, 2013.
34. P. Lawrance, Report no. LGC/R/2011/171, 2011, pp. 1–12.
35. M. L. Salit and G. C. Turk, Anal. Chem., 1998, 70, 3184–3190.
36. Guide to the Expression of Uncertainty in Measurement, International Organisation for Standarization (ISO), Geneva, Switzerland, 1st edn, 1993.
37. J. Vogl, M. Koenig, W. Pritzkow and G. Riebe, J. Anal. At. Spectrom., 2010, 25, 1633–1642.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16637g

---