Comparison of 624-type Capillary Columns II "Equivalent Columns" Deviation in the Quantitative Analysis

The present paper is the continuation of our "equivalent column" study. 624-type columns were investigated in a capillary gas chromatograph with a flame ionization detector (FID). In the preceding work, we used a homemade test mixture containing mainly acidic type compounds, which we complemented in the current study with additional basic test compounds to give a more precise description of the behaviour of the stationary phases. The columns had the same dimensions (30 m × 0.32 mm × 1.0 μm; length, internal diameter and film thickness, respectively) but were purchased from different manufacturers or the same manufacturer but from different batches. In addition, also their pre-life was different. These circumstances influence the condition of the stationary phase


Introduction
With quantitative gas chromatography measurements, the primary objective is to provide the necessary resolution and limit of quantitation (LoQ).After selecting the theoretically appropriate stationary phase for separation, analysts should decide whether to choose one from the laboratory column stock on hand or purchase a brand-new one from one of the manufacturers present on the market.Not only the history of a column but also different manufacturing technologies can affect the desirable results.
In the case of a used column, the history of the column needs also be reckoned with.The stationary phase preserves the imprint of the previously measured samples in the form of irreversibly bonded contaminations.These inevitably lead to an alteration in its quality.When buying a new column, differences may be expected due to different manufacturing technologies.The manufacturers provide a test chromatogram and the calculated chromatographic parameters such as selectivity, resolution, peak asymmetry factor, or height equivalent to a theoretical plate for some chosen compounds.However, the test mixture and the chromatographic circumstances could differ from producer to producer, which makes it difficult to compare the "equivalent columns".Even if the parameters are similar, sometimes it is difficult to tell the differences.In order to find the proper column for the analysis at hand, the analyst should test their columns with their specific test mixture in similar conditions [1].
By introducing the retention index system in 1958, Kováts [2] laid the foundations for the classification system of gas chromatographic stationary phases.This system was used by Rohrschneider [3], who chose five different test molecules to represent the second-order interactions between the analyte and the stationary phase.He determined the retention indices of the test molecules on 20 stationary phases and compared them to the indices measured on squalane chosen as reference.McReynolds [4] extended this method to a wider stationary phase range by changing the test compounds (gaseous alkanes were eliminated).For some cases, he suggested other five compounds for a more accurate determination of the average polarity of the phases.As the 0 value of the chromatographic polarity index scale, squalane (most apolar), and for the 100 value, 100% cyanopropyl siloxane was chosen.Using the McReynolds data with a pattern cognition method, Wold [5] classified 198 of 226 stationary phases into 16 clusters and distinguished 28 unique phases.In addition to the classification based on the retention index, grouping as per thermodynamic parameters such as sorption enthalpy is also widely used.Poole et al. [6] found that, on polar phases, the calculated selectivity based on the retention index is ridden with error due to the different dissolution mechanisms of the alkanes.By using the partial molar Gibbs free energy of solution for a homologue series of solutes they determined the polarity index based on the free energy of solvation of the methylene increment 'G CH s 0 2 .Abraham et al. [7] established a solvation equation between the retention volume and the different types of interactions with the stationary phase and calculated 'G CH s 0 2 for 77 phases.The classification of the phases is consistent with Wold's clusters [5].In 2000, this linear solvation energy model was corrected and revised, however, the results were found to be similar [8].These classification methods led to distinguishing the stationary phases based on polarity.However, not only polarity but also selectivity affects the results.In 1978, Grob et al. [9] introduced a 12-component test mixture to model all possible gas chromatographic interactions between analytes and stationary phases.The method required only one single injection to evaluate the information about the state of the stationary phase.Based on this technique it became possible to compile a database, which groups similar phases according to the chromatogram pattern observed.Three years later, they [10] complemented these results with new experiences and recommendations for the analysis.Although this is a single injection testing method, in some cases it is complicated to set the measurement conditions as per the length of the column or the quality of the carrier gas.In other cases, even the test mixture is not suitable for testing unique stationary phases.
In the literature, other test mixtures with different compositions can also be found.Horká et al. [11] tested a mixed Carbowax 20M and SE-54 phase with volatile fatty acids and amines, and calculated column efficiency and capacity factor with 2,6-dimethylaniline and 2,6-dimethylphenol.
Dresen et al. [12] applied only compounds containing nitrogen to test the performance of base-deactivated columns.Mehran et al. [13] carried out their column tests with alkyl halogenides, alkanes, toluene, and 2-methyl-1-butanol to compare different stationary phases based on the elution order and the calculated sorption enthalpies and entropies.
In our previous studies [14,15] we investigated polyethylene glycol and 5% diphenyl -95% dimethyl polysiloxane stationary phase gas chromatographic columns produced by different manufacturers.We modified the Grob test mixture to the isothermal measurement conditions to provide enough retention and avoid co-elutions.We demonstrated that the chromatogram of the test mixture on "equivalent columns" shows deviations.We determined sorption enthalpies and entropies to present the strength of the interactions between the different types of analytes and the stationary phase.We calculated the height equivalent to a theoretical plate (HETP) at 8 linear velocity (u) levels.With HETP -u curves, we visualized the deviations in the efficiency change of the columns to be compared.
The 624-type columns contain 6% cyanopropylphenyl -94% dimethylpolysiloxane stationary phase which is commonly used for volatile compounds separation.This phase has strong dipole and hydrogen-bond base interactions, although the interactions with the analytes' lone electron pair are weak [16].Becerra et al. [17] compared the OV-101 (dimethyl polysiloxane) and OV-105 (5% cyanopropyl -95% dimethyl polysiloxane) stationary phases using the specific retention volumes, solute-polymer interaction parameters, and the partial molar enthalpies of solution and mixing.They found that the increasing cyanopropyl content of the stationary phase decreased the contribution of the methylene group to the partial molar free energy of the n-alkanes solution and increased the polarity of the phase.Fernández-Sánchez et al. [18] established, using Kováts' coefficient, that the polarity of the dimethyl polysiloxane stationary phase increases by increasing the ratio of the phenyl and/or cyanopropyl groups.Kiridena et al. [19] investigated the difference between the DB-608 (50% diphenyl -50% dimethyl polysiloxane) and DB-624 columns using the solvation parameter model.They also confirmed the equivalent selectivity of the DB-624 and DB-1301 columns by the regression of the retention factors of a wide range of test compounds.Caulfield et al. [20] compared the strength of interactions between nitroaromatic explosives and 4 different stationary phases.Based on the calculated enthalpy of the solution, the strength of the interactions in the following order was established.The weakest interactions were formed with the 20% diphenyl -80% dimethyl polysiloxane, which was followed by 14% cyanopropylphenyl -86% dimethyl polysiloxane, then 50% diphenyl -50% dimethyl polysiloxane, and finally polyethylene glycol phase.
In the FID, due to the hydrogen flame burning, a complex pool of radicals, ions and electrons is formed.The charged particles resulting from the radical mechanism chain reaction are set in motion by two electrodes between which there is a potential difference of 150-300 V.As a result of this movement the electrical circuit closes and the generated electrical signal is measurable as current or voltage.If only detector gases (hydrogen, air, and nitrogen) enter the flame only a low electrical current is generated, however, in the case of carbon-containing compounds it increases by orders of magnitude.For hydrocarbons, the signal is proportional to the carbon content of the molecule.This results from the formation of the ground state CH radicals and the proportionally generated electrons to them.[21].However, heteroatom-containing molecules break this proportionality.Atoms, other than carbon and hydrogen, modify the signal depending on their quality and the type of the functional group in which they take place [21,22].Data are found in the literature for the different effects on the FID response specific to functional groups [22][23][24][25][26].
Sternberg [22] used Eq. ( 1) to define the effective carbon number ( ECN ) to describe the detector responses of the heteroatom-containing molecules: where s index is for the standard molecule, i is the index for the investigated molecule, n is the carbon number, A is the area of the peak, m is the mass of the molecule, M is its molar mass and f is the relative response.
Although the different molecule classes have their own characteristic effective carbon number increments, as we earlier demonstrated, the various chromatographic parameters have an effect on ECN [27,28].
In this study, we complement our latest results [1] on seven 624-type gas chromatographic columns.The selected test compounds were split into two test mixtures based on their characteristics and also their retention to avoid overlapping.In the previous study, we investigated mainly oxygenated (aldehyde, alcohol, acid, ester) compounds, furthermore chloroalkane, an aromatic hydrocarbon and some nitrogen-containing molecules (amide, sulphonamide).However, in the present study, we complemented the test substances with aromatic and aliphatic ethers, a diol and a polycyclic compound and a nitroaromatic, as well as a heterocyclic compound and amines to receive a more basic test mixture.
The pattern of the chromatogram provides us with an opportunity to compare the equivalent stationary phases from different manufacturers and/or with different histories.The strength of the reversible interactions with the stationary phases can be illustrated with the peak asymmetry factor.The irreversible interactions are manifested as decreased peak areas.These differences are visualized with the calculated ECN and limit of detection.Our column test method described in the previous and present paper, is capable of indicating even the little differences that can be critical during method development and method adaptation.
The standards were purchased from Sigma-Aldrich. 25 mg of each compound was measured into a 25 mL volumetric flask and diluted with methanol.1 mL of the stock solution was diluted to 20 ml with methanol.The concentration of the test molecules was approximately 50 µg mL −1 .

Columns
We tested the same seven 624-type columns that were investigated in our previous paper (Table 2); however, with a different test solution containing a larger number of basic compounds than acidic ones.
The columns' dimensions were 30 m × 0.32 mm × 1.0 µm.The columns, supplied by the same manufacturer but coming from different batches, were marked with "a" and "b" to be distinguished.All columns were used for different samples.Before testing conditioning was applied in every case.

Apparatus
A Shimadzu GC-2014 gas chromatograph was applied with an AOC-20i autosampler.The injector was held at 250 °C, and 1 µL test solution was injected at each measurement condition five times.To prevent excess sorption in the injection port, liners were silylated before use.The test solution was measured at 50, 100, 150, and 200 cm s −1 linear velocity.Hydrogen (purity 99.98%) was applied as a carrier gas.The split ratio was 20:1.The column temperature program started at 40 °C, then programmed to 220 °C at a rate of 5 °C min −1 .The flame ionization detector was held at 250 °C.The hydrogen, air, and nitrogen (make-up, purity 99.996%) gas pressures applied in the FID were 75 kPa, 50 kPa, and 60 kPa, respectively.

Effective carbon number
The theoretical ECN values of the target compounds are listed in Table 1.These values are based on our earlier and other authors' ECN studies.In case no experimentally determined ECN was available for the specific compounds, we indicated a calculated value based on the sum of the ECN increments characteristic for the molecule functional groups.No data was found in the literature for the -NO 2 group.

Results and discussion 3.1 Retention order
Chromatograms evaluated on the tested columns are shown in Fig. 1.Except for the Elite Volatiles column, the retention order was the same for every column.Although the Elite Volatiles column is listed as a 624 column in the cross-reference lists, it has a slightly different stationary phase.The apolar alkanes had similar retention times on every column, but the polar test compounds had less retention on the Elite-Volatiles.This means that this stationary phase is less polar than the 624 columns.The retention order of pyridine (3) and 1,2-propanediol (4) changed on Elite-Volatiles to 1,2-propanediol (4) and pyridine (3), In addition, n-undecane (8) co-eluted at 50 cm s −1 with nitrobenzene (9).
Although as on every column, the peak maximum of heptylamine appeared between anisole and n-undecane on the Rtx-624 b column, its retention time was significantly higher, and eluted partially under the n-undecanone.However, this retention time increase was not observed on the other Rtx-624 column (Rtx-624 a).

Peak shape
The asymmetry factor (A s ) was used to describe the peak distortion due to the stronger interaction between the stationary phase and the test molecules (Fig. 2).These values were calculated with Eq. ( 2): where b is the distance between the peak maximum and the endpoint of the peak at 10% peak height, and a is the distance between the start point and the maximum of the peak at 10% peak height.The maximum point is the retention time of the compound.The results show that almost all peaks have tailing (Fig. 3 (a)).However, significant  differences are evidenced in comparison with the nearly symmetric (A s ≈ 1) peaks for pyridine, 1,2-propanediol, and heptylamine.Although the heptylamine peak has a tailing on every column there are significant differences between the extents of the tailing (The asymmetry factor of heptylamine changes in the range of 2,6 -38,3 at 100 cm s −1 ).The difference in asymmetry for the pyridine peak at 100 cm s −1 measured on the CP-Select 624b (1.3) and the Elite Volatiles (4.3) column is shown in Fig. 2.These asymmetric peaks let us conclude the stationary phases are slightly acidic, which causes stronger reversible interactions with the amino and hydroxyl function groups.To show the effect of the stationary phase on the peak shape, we also represented the test solutes' average asymmetry factor for each column in Fig. 3 (b).Active points causing tailing resulted from the different manufacturing techniques and the history of the columns.

Effective carbon number
Since our previous studies show that ECN values depend on chromatographic parameters, the aim of our present study was not to reach the theoretical ECN values listed in Table 1.They served as a starting point for the approximate expected ECN values.In the present paper, we used the experimentally determined ECN values to illustrate the effect of the stationary phase on the quantification of different types of compounds.
The ECN values determined are shown in Table 3.For each compound, we marked the lowest ECN (ECN min ) and the highest ECN (ECN max ) values in italics and in bold, respectively.The difference between these two values displays the effect of the stationary phase on measuring the test compounds.The reduced ECN value is due to the irreversible interactions with the stationary phase, and the high Δ(max-min)% means significant differences in the stationary phase.These values let us differentiate the columns and help us choose the best column for the future target compounds.In our test molecules, the aromatic rings interact the same way with the stationary phases; however, with the non-hindered hydroxy and amino groups, some stationary phases form strong irreversible, instead of reversible, bonds.The highest Δ(max-min)% values were obtained for 1,2-propanediol and heptilamine.
Like in our previous work, we used the ΔECN column value to establish the order between the columns.The ΔECN column gives the average influence of the stationary phase on the ECNs.
In Eq. ( 3) ECN i is the ECN value of the i-th solute on the stationary phase and ECN i,max is the highest ECN value of the i-th compound on the 7 investigated columns.The theoretically achievable maximum score is 8 in the case of the 8 compounds.The established order of the columns is shown in Table 3.
These results enable us to choose the best column in our laboratory for the measurements specific to diols, amines, heterocycles, or aromatic compounds.

Limit of quantitation
We determined the noise level on each column at each linear velocity to calculate the limit of quantitation with Eq. (4): where N i is the mean area of the noise in mVs for the i-th solute.On every chromatogram, three different baseline sections were integrated.One-point calibration was used to convert the LoQ i values into concentrations.Fig. 4. shows the limit of quantitation at 100 cm s −1 for the test components in µg mL −1 units.These differences can be attributed to two factors.As we concluded in our previous article, on the one hand, variance mainly derives from the different  detector responses manifesting in the ECN values of the components and is caused by irreversible interactions.
On the other hand, column bleeding increases the number of substances reaching the detector, consequently resulting in higher noise levels.This effect is related to the manufacturer's technology.In addition, at higher linear velocities an increased noise level was observed due to more molecules from the carrier gas entering the flame.The increased background noise levels also result in higher LoQ i values.

Conclusion
The recent study is the continuation of our previous paper and concludes our investigation of the stationary phase equivalency research of seven 624-type gas chromatographic columns.In the previous study, ten different molecules were tested, while in the present paper, eight more were used to cover the gaps with more aromatic, heterocyclic, and nitrogen-containing compounds as well as basic molecules and a diol.By adding these molecules to the test mix, we can select more specifically the optimal column for our measurements.
We found remarkable differences in the retention order, peak shape, and peak area.
The retention order correlated with the manufacturers and enabled us to differentiate between "equivalent columns".
According to our results, the condition of the column stationary phase affects quantitative determination too.
The reduction of the measurable peak area results from irreversible sorption, which we quantified by the decrease of the ECN value and increase of the LoQ value.The differences in area can be imputed both to different manufacturing technologies and the history of the columns.Impurities left over from previous measurements modify the surface of the stationary phase.Also, storage conditions can modify the stationary phase by humidity.The LoQ also contains information about column bleeding, which is also subject to manufacturing technology.
The ECN was an indicator of the irreversible sorption of non-hindered amines and diols on the stationary phase.Consequently, ECN loss is also a help in the proper column selection specific to the target compound characteristics.
Based on our results, even columns from the same manufacturer may not always be interchangeable.
In light of the presented results, a homemade test mixture and a couple of measurements in similar conditions can help the analyst select the optimal columns for the analytical task at hand.By means of tests repeated from time to time, the decomposition of the stationary phase can be monitored, ruling out measurements effected on a damaged, unsuitable column.

Fig. 3
Fig. 3 (a) Asymmetry factor of the test compounds at 100 cm s −1 (Asymmetry factor of Heptylamin on Rtx-624 a column is 38.3);(b) Sum of asymmetry factors of the investigated columns at 100 cm s −1

Fig. 4
Fig. 4 Limit of quantitations of the test compounds for all investigated columns

Table 1
Test compounds

Table 2
The tested columns

Table 3
The ECN values at 100 cm s−1The lowest ECN (ECN min ) and the highest ECN (ECN max ) values are marked in italics and in bold, respectively.