Low-level 40Ca determinations using nitrous oxide with reaction cell inductively coupled plasma–tandem mass spectrometry

In inductively coupled plasma mass spectrometry, the most abundant Ca isotope (40Ca) suffers from isobaric interference with argon, hindering the potential for low detection limits of Ca. A powerful approach is to remove the interference by using a reaction gas in a reaction cell. Ammonia (NH3) has proven to be an effective reaction gas by process of a charge transfer reaction. However, NH3 is highly corrosive and toxic and cannot remove isobaric 40 K. Therefore, this work proposes the use of nitrous oxide (N2O) to mass shift the target analyte 40Ca to 40Ca16O+ as a non-corrosive and non-toxic alternative. Instrument performance testing demonstrated that N2O was capable of reaching equivalent detection limits (0.015 ng g−1) and background equivalence concentrations (0.041 ng g−1) to that of NH3 and limited by the blank only. Further investigation of matrix interferences with synthetic standards highlighted that the N2O approach supports the separation of potassium (K) and magnesium (Mg)–based interferences at tested concentrations of more than 600 times and almost 800 times higher than Ca respectively, whereas NH3 was found to only support the removal of Mg. This work highlights a clear advantage of N2O for low-level Ca determinations with high matrix loads, as well as compatibility with other instrumentation sensitive to corrosion that supports reaction cell technology. Graphical abstract


Introduction
Inductively coupled plasma mass spectrometry (ICP-MS) is a widely used tool that boasts high sensitivity, low limit of detection (LOD), and high sample throughput. However, LODs of ICP-MS measurements can suffer due to spectral interferences from monatomic or polyatomic ions with the same mass to charge ratio (m/z). Calcium (Ca) is an example of an element that suffers from such interferences. In this case, the major isotope, 40 Ca, which has a natural abundance of 96.941% [1], shares an isobaric interference with argon (Ar), which is used as both a carrier gas and for the generation of the plasma. To separate these, the required mass resolution would be > 190,000 ( Table 1). As such, sensitivity and detection limits are greatly hampered by resorting to using less abundant Ca isotopes, such as 42 Ca (0.647% abundance), 43 Ca (0.135% abundance), and 44 Ca (2.086% abundance) [1]. One method commonly employed to overcome spectral interferences is the introduction of a reaction gas via a reaction cell. Recently developed instruments utilize an additional quadrupole as a mass filter in front of the reaction cell to avoid the formation of new interferences in the cell (ICP-tandem mass spectrometry (MS/MS)) [2,3]. Ammonia (NH 3 ) has been typically used as a reaction gas to remove interference of 40 Ar + for on-mass determination of 40 Ca + [4], as the charge transfer reaction (M + + NH 3 → NH 3 + + M) occurs at a much higher rate for Ar + than for Ca + [2,5,6]. However, NH 3 is a corrosive and toxic gas [7]. Therefore, it cannot be used in every instrument and its usage is sometimes limited in some laboratories. It is therefore of interest to assess suitable alternatives for greater sustainability regarding primarily the protection of the instrumentation.
Other reaction gases have also been used for on-mass determinations of 40 Ca, such as methane [8,9] and hydrogen [10]. However, mass-shift reactions of 40 Ca using a reaction gas (e.g. 40 Ca → 40 Ca 16 O + ) are less widely reported. Oxygen is the traditional reaction gas for analyte mass-shift determinations [2]. However, low formation of the CaO + product ion renders this approach unfavourable [11]. Nitrous oxide (N 2 O), on the other hand, is a more reactive alternative to oxygen [12,13] and has been explored recently for a number of elements using ICP-MS/ MS systems [14][15][16], highlighting a broad scope for use in routine multi-element analysis. While N 2 O shows high reactivity, it is notably much less corrosive than NH 3 and may serve as a suitable alternative. Apart from 40 Ar, other interferences have to be considered if Ca is determined at low levels in a complex matrix (see Table 1).
While sensitivity can be enhanced and the LOD can be decreased by using cell methodology, it is important to note that current LODs are limited by background levels of Ca. Therefore, it is important to be highly considerate of sources of contamination. Wu et al. suggested the use of a clean laboratory environment can allow for lower detection limits [9]. Retzmann et al. described in detail how to minimize the Ca background and reported that, e.g., the use of nitrile gloves and clean-room wipes were major sources of Ca contamination and should be avoided [17].
This work aims to evaluate the novel usage of N 2 O for quantification of Ca at low levels in complex matrices by ICP-MS/MS using 40 Ca. Optimization of both the cell gas flow rates and possible internal standards was carried out. Instrument performance parameters for both reaction gases (NH 3 and N 2 O) were compared with each other, as well as with the determination of 40 Ca under standard conditions (no cell gas). Additionally, the effectiveness of the removal of sample matrix interferences caused mainly by magnesium (Mg) and potassium (K) was also investigated for each cell gas.

Materials and methods
All preparations and measurements were made in a clean room (ISO class 8) to minimize the risks of contamination. Polyethylene gloves (Carl Roth GmbH, Karlsruhe, Germany) were used on top of nitrile gloves to avoid Ca contamination. The use of clean-room wipes and paper towels were avoided throughout.

ICP-MS/MS measurements
All ICP-MS/MS measurements were carried out using a NexION 5000 (Perkin Elmer, Waltham, MA, USA), which is equipped with a dynamic reaction cell (DRC). Instrumental parameters are listed in Table 2. The sample was introduced to the ICP via a peristaltic pump. A mixed internal standard containing Sc (w = 4.7 ng g −1 ), Y (w = 2.2 ng g −1 ), and In (w = 2.2 ng g −1 ) was added online. Measurement parameters for the internal standards are displayed in Table 3. Calibrations in standard mode (both using single-quadrupole mode (Q3) and MS/MS mode) were carried out in the range of 5-1000 ng g −1 Ca and measured using the 44 Ca isotope. DRC calibrations were

Cell gas flow rate optimization
Optimization of the reaction gas flow rates for NH 3 and N 2 O was carried out on m/z 40 using a 10-ng g −1 single element Ca standard and a blank solution (w = 2% HNO 3 ). Tested flow rates ranged from 0.1 to 1.0 mL min −1 using N 2 O and from 0.4 to 1.5 mL min −1 using NH 3 .   1 3

Evaluation of matrix interferences
Evaluation of potential interference caused by Mg and K using N 2 O and NH 3 DRC was carried out by measurement of the Ca background equivalence concentration in each respective single element standard at dilutions between 0.5 and 5.0 µg g −1 . Determined Ca concentrations, made by means of external calibration, were used to evaluate the significance of each interference by comparing the mass fraction of Ca obtained using the non-interfered (by isobaric interference) isotope on m/z 44 (w(Ca) m/z 44 ) to that of the interfered isotope on m/z 40 (w(Ca) m/z 40 ). Where the difference was found to be significant, the magnitude of interference was calculated by subtraction of w(Ca) m/z 44 from w(Ca) m/z 40 .

Certified reference materials
River water certified reference materials (CRMs), SLRS-3 and SLRS-5 (both National Research Council Canada, Ontario, Canada), were used for validation. The materials were diluted to approximately 1 ng mL −1 Ca using dilute sub-boiled acid (w = 2% HNO 3 ). Analysis was carried out using N 2 O and NH 3 as reaction gases.

Statistics
Statistical testing and the generation of figures were carried out using RStudio (version 2021.9.2.382). LOD and LOQ were calculated respectively from 3 and 10 times the standard deviation of 5 repeated measurements of a blank solution (w = 2% HNO 3 ). Background equivalence concentration (BEC) was calculated from the calibration by division of the background (y-intercept) with the calibration slope. Equidistant standard concentrations of Ca between 0.2 and 1.0 ng g −1 were used when determining the LOD, LOQ, and BEC of the DRC methods. Evaluation of the effect of interferences was carried out using total least squares (Deming) regression on the concentrations obtained using the 40 Ca and 44 Ca isotopes, where standard errors were calculated using the jackknife method. The observed w(Ca) m/z 40 /w(Ca) m/z 44 ratio was statistically compared to the expected ratio of 1 using a Z test.

Cell gas flow rate optimization
The profile observed for mass shift using N 2 O (Fig. 1A) displayed optimum sensitivity for the formation of the 40 Ca 16 O + product ion at 0.4 mL min −1 . The background signal of 40 Ar + was found to decrease as the N 2 O flow rate increased.
Further observations of the blank signal conducted using the 44 Ca isotope at optimum N 2 O conditions found a similar signal ratio of 40 Ca/ 44 Ca to that of the natural abundance. This suggests that, while a small degree of 40 Ar 16 O + forms at lower N 2 O flow rates, Ar was successfully removed by N 2 O at optimum conditions and the observed blank signal is primarily due to the presence of background Ca.
On-mass determination of Ca was carried out using NH 3 flow rates between 0.4 and 1.5 mL min −1 , as preliminary tests (carried out using 38 Ar) suggested that lower flow rates would not remove enough Ar to prevent detector saturation on m/z 40. Within this range, the signal for Ca was observed to decrease with increasing NH 3 flow (Fig. 1B). Consideration of the signal to background ratio demonstrated a plateau at NH 3 flow rates greater than 0.7 mL min −1 (Fig. 1C); hence, this was determined to be the optimum condition. Similar to the use of N 2 O DRC, additional monitoring of the blank using 44 Ca highlighted that the blank signal obtained on m/z 40 at the optimum NH 3 flow rate was due to the presence of background Ca.

Internal standards
Sc, Y, and In were tested as possible internal standards, as they have a narrow range of ionization energies (5.78-6.56 eV) close to that of Ca (6.11 eV) [18]. Variation of the internal standard response to the DRC gas flow rate is shown in Fig. 2. Initial tests using N 2 O DRC highlighted that In could not be measured in mass-shift mode, as no formation of the InO + product ion was observed. However, detection of In + on mass with N 2 O was found to be feasible, as an adequate and stable signal was obtained. Given that the sensitivity of the In signal increased with N 2 O flow rates up to 0.7 mL min −1 , it could be interpreted that the signal observed may be enhanced by collisional focussing. Sc and Y, on the other hand, were observed to form the oxide product ion. Additionally, the sensitivity profiles closely matched that of Ca, with both internal standards displaying an optimum N 2 O flow rate of 0.4 mL min −1 . The use of Sc and Y may then prove more advantageous in comparison to In, as they display similar cell reaction characteristics to that of the analyte.
In contrast, Sc and Y were observed to be unfit for use as internal standards when using NH 3 as a reaction gas (Fig. 2B), as both elements were effectively removed when measuring on mass. The profile of In showed a collisional focussing effect, where a sufficient and stable signal was obtained for use as an internal standard at the previously determined optimum NH 3 flow rate of 0.7 mL min −1 . Given that, of the three internal standards considered, In was the only one found to be suitable for measurements, all subsequent data reported for NH 3 DRC only includes internal normalization to In.

Performance data
The instrument performance for the two DRC methods, as well as standard mode, was assessed and is presented in Table 4. Between the three internal standards used for N 2 O DRC, minimal differences in sensitivity (slope = 128,000-129,000 cps/ng g −1 ) and BEC (0.41-0.42 ng g −1 ) were observed. However, stability data indicated a lower RSD when using Sc (1.6%) as an internal standard compared to Y (2.2%) and In (2.9%). This was also reflected in the slightly lower LOD (of 0.015 ng g −1 ) and LOQ (of 0.049 ng g −1 ) values obtained using Sc. Therefore, while all three internal standards can be used for determination of Ca, Sc has been shown here to be the optimal in this case.
While the calibration slope observed for 40 Ca using NH 3 was 1.75 times greater than that of the N 2 O, the LOD (of 0.015 ng g −1 ) and LOQ (of 0.049 ng g −1 ) remained consistent. Additionally, determined BEC values (of 0.37 ng g −1 and 0.41 ng g −1 for NH 3 and N 2 O DRC respectively) were similar between the two methods. This further indicates that the background signal for both DRC methods is likely due to background Ca levels. In comparison to the measurement of 44 Ca using the Q3 mode with no cell gas (which is more sensitive than the MS/MS mode with no cell gas), the application of both NH 3 and N 2 O DRC methods showed marked improvement, with approximately 10 times lower LOD (of 0.015 ng g −1 ) and LOQ (of 0.049 ng g −1 ), and approximately 45 times lower BEC (of 0.42 ng g −1 ).

Removal of interferences
Mg and K were identified as the primary interferences of concern for measurements using the 40 Ca isotope, as the polyatomic 24 Mg 16 O + interference formed in the plasma and isobaric interference from 40 K + cannot be removed by the Q1 mass filter. Selenium, as 80 Se ++ , was not considered to be of concern due to the high ionization energy required to form such interfering species [18] (Table 5).
Native Ca concentrations in the measured single element standards of both Mg and K ranged between 1 and 10 ng g −1 . For measurements of Ca with high Mg load, the w(Ca) m/z 40 /w(Ca) m/z 44 ratio did not significantly differ from 1 for both N 2 O and NH 3 , which indicates that both gases can achieve successful removal of 24 Mg 16 O + interference on m/z 40. However, measurements of native Ca in high-level K standards using NH 3 DRC showed a significantly greater w(Ca) m/z 40 /w(Ca) m/z 44 ratio than expected, indicating significant bias introduced due to the presence of interfering 40 K that was not removed. Conversely, no such significant bias was observed for measurements of K single element standards using N 2 O DRC, suggesting that N 2 O can be used to successfully overcome the interference of K.
The level of interference from 40 K using NH 3 DRC was determined by the difference of the observed w(Ca) m/z 40 and w(Ca) m/z 44 and plotted against the prepared K concentration (Fig. 3). The slope of the plot indicated that the level of interference on 40 Ca appears as 0.0128% of the matrix K concentration. To obtain < 1% interference on 40 Ca, matrix K concentrations cannot exceed about 80 times that of the analyte Ca concentrations. Further variation of the NH 3 flow rate was not able to overcome the isobaric interference from K, as the Ca/K signal ratio only decreased with increasing flow rate (Fig. 4).

Method validation
Validation of the total determination of Ca using N 2 O DRC was carried out using two river water CRMs that were certified for Ca. Results are shown in Table 6. Excellent recoveries between 99.2 and 103% were obtained for the two CRMs using Sc and Y as internal standards. Slightly lower recoveries of 95.4-96.9% were obtained using In as an internal standard, though these still fall within the stated uncertainty of the CRM. Recoveries of 97.9-98.5% were obtained using NH 3 DRC, indicating that the N 2 O DRC method for determination of total Ca performs as well as existing methodology for real sample matrices.

Conclusions
The data presented within this study suggests that N 2 O is not only a suitable replacement for NH 3 for total Ca determinations, but can also ensure matrix-free determinations of 40 Ca, especially in K-rich matrices with low Ca content. Despite apparent lower sensitivity of the N 2 O approach (by a factor of 1.75), detection limits and BEC were found to be similar between the two cell gases. We therefore propose that N 2 O should be used in place of NH 3 for routine measurements of Ca, as well as incorporating this approach into wider multi-element analysis.