Indirect CE‐UV detection for the characterization of organic and inorganic ions of a broad mobility and pKa range in engine coolants

An alternative CE‐(indirect ultraviolet) method for the analysis of inorganic and organic anions in ethylene glycol‐based engine coolants is presented using a BGE with 4 mM pyromellitic acid and 3.4 mM 1,6‐hexamethylene diamine, pH 3. Baseline separation of six inorganic (e.g. nitrite, nitrate, and sulfate) and five organic anions (e.g. acetic and glycolic acid) was achieved. Quantification of 8 out of 11 specified anions was possible in stressed engine coolant samples after simple aqueous dilution. LODs between 0.8 and 15.1 mg/L with RSD values of peak areas between 2.6 and 11.9% were obtained. Some limitations due to matrix effects can be overcome with slight adaptations of the BGE. The flexibility of the method is vital regarding the increasing demands for the composition of engine coolants for pollution reduction.

(Waldbronn, Germany). Preconditioning of new capillaries: flushing 10 min each with MeOH, 1 M HCl, 1 M NaOH, H 2 O, and 20 min with the BGE. Between measurements, 2 min rinsing with BGE was sufficient. The length of the capillary was 60.5/76 cm with 50 µm id. The electric field strength was −395 V/cm. Injection used a pressure of 50 mbar for 5 s for aqueous standards and 100 mbar for 16 s for engine coolant samples (diluted 1:10 with water). Origin 9.1 (Origin-Lab, Northhampton, Massachusetts, USA) was used for data analysis.
Due to the similar absolute electrophoretic mobilities of some analyte pairs (e.g. fluoride and formate; glycolate and acetate; nitrite and nitrate, see Table 1), the pH of the BGE had to be adjusted to achieve the highest possible selectivity. We carefully checked the presence of system peaks [11] for all BGEs. The analysis of anions of a broad mobility range requires control and fine-tuning of the direction and magnitude of the EOF, which can be accomplished with different coatings [12,13]. We investigated coatings of high reversed EOF (statically adsorbed polybrene [14] and EOTrol HR [15,16]) and strongly suppressed EOF (static PVA [17,18] and statically adsorbed EOTrol LN [15,16]). Resolution with these coatings was insufficient. Using EOF reduction via the dynamic EOF modifier HMDA [19,20] was promising for further method development.
Advantageously, HMDA could be used for both EOF control and buffering serving as counterion. This strategy reduced the number of buffer components to a minimum and thus also the number of system peaks disturbing quantification when using indirect UV or C 4 D detection [11,21] (for an example, see Figure 1A).
The adjustment of the pH for highest selectivity was supported using simulations with PeakMaster [22]. Together with experiments we saw that pH values >7 resulted in comigration of acetic and glycolic acid due to very similar effective electrophoretic mobilities at full charge (see Table 1). Only at lower pH, differences in their pK a values and thus their effective electrophoretic mobilities were high enough to achieve baseline separation. At pH <3, acetate was insufficiently separated from the sample plug due to too low dissociation. The separation of sulfuric acid, nitric acid, and nitrous acid was challenging: comigration of nitrate and nitrite occurred at pH >5 (see Figure 1B). For the separation of nitrate and nitrite complexation with either cadium ions [23] or cyclodextrin [24] was successfully applied in the past. The addition of 40 mmol/L α-CD (BGE = 20 mM MES, 20 mM l-histidine, EOTrol LN, see Figure 1C) allowed to separate nitric acid from nitrous acid; however, only with concurrent comigration of the nitrite-α-CD-complex and sulfate. Unfortunately, the high concentration of hydronium ions at the advantageous pH <4, precluded conductivity detection [25] (see Figure 1D).
Best separation efficiency would be obtained with a probe ion of an electrophoretic mobility close to the analytes [26,27]. Due to the formation of disturbing system peaks, a combination of two probe ions (see Doble and Haddad [28]) to account for the broad mobility range of the analytes of interest (see Table 1) was not successful. Finally, PMA with an intermediate electrophoretic mobility and a suitable lowest pK a1 of 1.9 was chosen [29]. Baseline separation of all analytes was achieved after fine-tuning of the BGE to pH 3. A compromise had to be found for the separation of the most critical anions, nitrite, sulfate and nitrate, the sufficient dissociation of acetic and glycolic acid and the avoidance of disturbing system peaks. To the best of our knowledge, this is the first time that these five anions were included in one CE-(indirect)UV method.
The final method was: PMA (c = 4 mM) served as probe ion and HMDA (3.4 mM) as counter ion and EOF suppressor. Electropherograms were recorded at 220 nm using a separation voltage of −30 kV. All 11 inorganic and organic analytes and two additional internal standards, dichromic acid, and HIBA, were baseline-separated (see Figure 2, Trace A).
Some matrix effects were present: Figure 2, traces B and C show two different stressed cooling agent samples. Due to partial comigration of acetate and a matrix compound, it could only be identified but not quantified. Similarly, quantification of chloride and nitrate was hindered in matrix due to a system peak present only for stressed samples.
Standard addition was performed in triplicates with addition of 0.05, 0.10, and 0.15 mM analytes to samples (1:10 diluted with water). External calibration (0.07-0.3 mM) was performed in the linear range in duplicate. Determination of method precision was conducted with ten consecutive runs at analyte concentrations of 0.10 mM in a cooling agent sample. The electropherograms were illustrated and evaluated with Origin. Quantification was performed using the tool "Peak Analyzer" and LODs were determined according to DIN 32645. Table 1 shows validation data for precision, robustness, and LOD in matrix together with reference values from a method for the analysis of foodstuffs with a PMAbased BGE at pH 7 (without nitrite and glycolate so that better LODs were reached) [30]. Overall, good quantitative precision was observed except for succinate with its low dissociation at the chosen pH. Robustness was demonstrated by successfully transferring the method to a different instrument (Agilent 7100 CE, data not shown).
For quantification of analytes in cooling agent samples, it was necessary to increase the injection times in order to meet detection limits necessary for the quantification of all analytes but nitrite, chloride, and acetate. Quantification was achieved either via standard addition (Sample 1) or external calibration (Sample 2) dissolving the analytes in 50:50 v/v ethylene glycol/water to simulate the cooling agent matrix. Sulfate quantification needed a higher dilution ratio (1:60) and external calibration.
Further optimization of the method would mean to use separate methods for high and low mobility analytes. Here, the use of HMDA as a dynamic EOF modifier is advantageous as it allows to easily adapt the BGE when specific analytes are prioritized; for instance, suitable conditions to analyze phosphate, glycolate, acetate, and succinate were obtained using 20 mM benzoic acid as probe ion and increasing the pH to 4.5. Electropherograms for standard and sample measured under these conditions are shown in Figure 2, Traces D and E.
To our knowledge, this is the first CE-(indirect ultraviolet) method capable to analyze nitrate and nitrite as well as glycolate and acetate in one run, which was possible reducing system peaks having HMDA serving as buffering counterion and EOF modifier. The quantification of most analytes was possible in stressed cooling agents with sufficient LOD. Further optimization is only possible using separate methods for two sets of analytes. Using HMDA in both methods, no special rinsing procedures will be necessary and analysis time can still be kept clearly below 10 min per run, giving rise to analysis times comparable to ion chromatography, though at higher matrix tolerance.