Ionization Efficiency of Doubly Charged Ions Formed from Polyprotic Acids in Electrospray Negative Mode

Research Article


The ability of polyprotic acids to give doubly charged ions in negative mode electrospray was studied and related to physicochemical properties of the acids via linear discriminant analysis (LDA). It was discovered that the compound has to be strongly acidic (low pK a1 and pK a2) and to have high hydrophobicity (logP ow) to become multiply charged. Ability to give multiply charged ions in ESI/MS cannot be directly predicted from the solution phase acidities. Therefore, for the first time, a quantitative model to predict the charge state of the analyte in ESI/MS is proposed and validated for small anions. Also, a model to predict ionization efficiencies of these analytes was developed. Results indicate that acidity of the analyte, its octanol-water partition coefficient, and charge delocalization are important factors that influence ionization efficiencies as well as charge states of the analytes. The pH of the solvent was also found to be an important factor influencing the ionization efficiency of doubly charged ions.

Graphical Abstract


Electrospray Multiple charging Ionization efficiency 


Mass spectrometry with electrospray ionization source (ESI/MS) is increasingly the method of choice for determination of various biological substances [1]. Many analytes (e.g., peptides) form multiply charged ions in ESI source. The multiply charged ions are very beneficial for compounds which form singly charged ions that have too high m/z values for most mass analyzers. Multiple charging and its reasons have been studied; however, much is still unclear in this field. No studies have been done on ionization efficiency (IE) of multiply charged analytes to our knowledge, although its relevance has been pointed out [2].

It has been shown that ionization efficiency (ESI response) of large multiply charged molecules depends on the distance between chargeable sites [3] and on the structure of the molecule (how well the ionized form is stabilized) [4]. Nevertheless, it is still not fully clear how to predict the charged state of the molecule in ESI/MS analyses. Wong et al. [5] developed a model for predicting the maximum possible number of charges of polyethylene glycol depending on the number of monomers and their affinity towards the ion that is merging with it. Later, Schnier et al. [6] and Smith et al. [7] studied the number of protons that bind to the peptide in the gas phase and found that the number of protons is very similar to the number of basic amino acid residues in the peptide. They also noticed that close proximity of basic residues decreases the basicity of the observed residue. The same tendency was also observed by Felitsyn et al. [8] who studied native proteins. Furthermore, the availability of basic sites has been shown to be important [6, 7, 9, 10] for determining the charged state of the analyte.

It has been observed that the charge state of an analyte in solution does not always correlate with the charge state observed in MS analysis [11]. Iavarone et al. was the pioneer in studies of supercharging [12]. They observed an increase in the charge state for peptides if additives such as diethylamine, 2-methoxyethanol, ethylene glycol, glycerol, or m-nitrobenzyl alcohol were added to the solution [12, 13, 14]. Although the mechanism of supercharging is yet to be fully explained, the authors related this effect with the additives being less volatile than water and thus increasing surface tension. Supercharging has also been observed in negative ion mode [15, 16] where organic bases were used as supercharging reagents and a positive correlation was found between gas-phase basicities (GB) and charge state distributions. It was observed that in negative ionization mode the correlation was linear, and in positive mode the correlation had a maximum value at GB of approximately 800 kJ/mol [16].

To study ionization efficiency of different compounds, an ionization efficiency scale approach has been introduced for singly charged compounds. As the evaluation of absolute IEs is difficult if not impossible, the term relative ionization efficiency (RIE) has been introduced [17] (see Formula 1 in Experimental).

In contrast to absolute IEs, RIEs can be measured reliably. This approach has been previously used to develop a model for predicting IE of singly charged analytes both in positive [18] and negative mode [19]. In both sources, the RIEs were transformed to logIE values by anchoring the obtained values to one specific compound. In positive mode this compound was methyl benzoate, and in negative mode benzoic acid. Anchoring was used for the compounds to give comparable values for all compounds.

For singly charged compounds, IEs of various analytes in different solvents with varying equipment have been studied. Most studies have been made in positive mode. Tang and Kebarle [20] found that the signal of analyte is affected by surface tension of the solvent and ion evaporation rate constant of the analyte. Cech and Enke [2] found that in the case of selected tripeptides, the nonpolar area affects their signal in mass-spectrometer. They concluded that analytes need to have a large enough nonpolar area to move to the surface of the droplet and also a structural element that allows charging (protonation, deprotonation, adduct formation, etc.). This tendency was also confirmed by Leito et al. [17] based on a study of esters and aromatic amines. Ehrmann et al. [21] tried to predict analyte signal in ESI/MS and observed the best predicting power while using the solution phase basicity of analyte (pK b). The effect of gas-phase proton affinities was smaller than previously stated for the studied compounds [22]. Oss et al. [18] calculated six physicochemical properties and found that ionization efficiency correlates best with pK a value and molecular volume of the analyte. Chalcraft et al. [23] modeled response factors for polar metabolites and found that important descriptors are molecular volume, octanol-water distribution coefficient (logD ow), and absolute mobility of the ion. Nguyen et al. [24] found a positive correlation between ESI signal and adjusted mass. In negative ionization mode, Henriksen et al. [25] found that for phenols and phenoxy alkanoic acids ionization was more efficient in methanol than in acetonitrile and that analytes logP ow (octanol-water partition coefficient) was in better correlation with ESI/MS signal than acidity of the analyte (pK a). Huffman et al. [4] found that in negative mode, compounds that are more acidic and hydrophobic ionize better. Kruve et al. [19] discovered in the negative ionization mode that ionization efficiency can be best predicted by using degree of dissociation for the analyte and charge delocalization (WAPS parameter) that describes the delocalization of the charge in the anion. However, high ionization efficiency of the anions in media where acidic compounds are expected to be present as neutrals has been observed. This phenomena, called wrong-way-round ionization, has been described by Zhou et al. [26].

As the ionization efficiency of multiply charged ions has not been studied, the aim of this paper is to study multiple charging in negative ionization mode in acidic and basic solvent. In the literature, there is much ambiguity of the charge states of the compounds and therefore we aim to reveal the factors that affect the ability to give multiple charging in ESI/MS based on analysis of small molecules in negative ESI ionization mode. We aim to propose a quantitative model that can be used to predict the charge state. The compounds included in this study cover a wide acidity range (pK a values from –3 to 11 in water) and wide hydrophobicity range (logP ow from –0.5 to 7.5). Small molecules allow determining conditions and parameters that affect multiple charging based on a relatively simple system. This is a good starting point to move on to bigger systems such as proteins.



The measurements were carried out using Agilent XCT ion trap mass spectrometer. Default settings for ESI source were used: nebulizer gas pressure 15 psi, drying gas flow rate 7 L/min, drying gas temperature 300 °C. The capillary voltage between the MS and nebulizer was –3500 V. All remaining ion transport parameters were determined by the Target Mass (TM) parameter, set by the operator. The TM was optimized for each component and the optimal value used for during measurements of ionization efficiency [27].

The ion trap parameters were: Smart Target (parameter characterizing the number of ions accumulated in the trap) 100,000, maximum accumulation time 300 ms. Each spectrum was scanned from m/z 30 to 1000. Depending on the compound one ion [M – H] or two ions [M – H] and [M – 2H]2- were observed in mass spectra.

For every RIE measurement of two compounds, two syringe pumps connected with a “T-piece” were used. The concentration ratio was varied by changing the infusion rates of two pumps from 1.7 μL/min to 6.7 μL/min so that the sum of the infusion rates of two pumps would always be 8.4 μL/min. Concentrations of compounds in the sprayed solutions were between 0.2 and 60 μmol/L depending on the compounds. During every measurement, the linearity of signal to concentration graph was checked and thereby it was verified that the signal was not saturated in any of the measurements [see example Supplementary Figure S1 in Supplementary Information (SI)].

Two solvents were used: (1) acidic solvent with the addition of formic acid (80/20 MeCN/0.1% formic acid solution), and (2) basic with the addition of ammonia (80/20 MeCN/0.1% ammonium hydroxide solution). The solvent was prepared from acetonitrile (HPLC grade, J. T. Baker, Deventer, The Netherlands), ultrapure water (18.2 MΩ·cm, TOC 1–2 ppb, prepared by a Millipore Milli-Q Advantage A10 water purification system), concentrated aqueous ammonia (25% solution; Lach:Ner, Czech Republic), and formic acid (≥98%, Sigma, Steinheim, Germany).

As it is complicated to measure logIE values, we measured the RIE values between compounds 1 and 2:
$$ RIE\frac{A_1}{A_2}=\frac{IE\left({A}_1\right)}{IE\left({A}_2\right)}=\frac{K_1^{\prime }}{K_2^{\prime }}=\frac{R_1{C}_2}{R_2{C}_1} $$
where R1 and R2 are the responses of the compounds and C1 and C2 the respective concentrations of the compounds in the spray. For each compound pair, the RIE was measured on five concentration ratios and the obtained RIE values were averaged. First, in both solvents the logIE values were temporarily assigned to compounds by minimizing the sum of squares (SS) of differences between measured logRIE values and the assigned logIE values [18]:
$$ SS={\displaystyle \sum_{k=1}^{n_m}{\left\{ \log RI{E}_k\left({A}_i,{A}_j\right)-\left[ \log IE\left({A}_i\right)- \log IE\left({A}_j\right)\right]\right\}}^2}\to \min $$
where n m is the number of measurements and logRIE k(Ai, Aj) is the result of k-th measurement that has been conducted between compounds Ai and Aj. For this step, the logIE value of benzoic acid was temporarily taken as zero in both solutions.
Second, to obtain the logIE values that could be compared between two solvents the logIE values from Formula 2 were anchored to benzoic acid in solution (2), for which the logIE value of zero has been previously assigned [19]. To anchor the scale obtained in solvent (1) to the solvent (2), the MS signal intensities for benzoic acid solution (concentration in both solutions approximately 60 μmol/L) were measured on the same day and the logIE for benzoic acid in solvent (1) was calculated as:
$$ \log I{E}_{\mathrm{benzoic}\ \mathrm{acid},\mathrm{solvent}\ (1)}= \log \frac{R_{\mathrm{solvent}\ (1)}\cdot {C}_{\mathrm{solvent}\ (2)}}{R_{\mathrm{solvent}\ (2)}\cdot {C}_{\mathrm{solvent}\ (1)}} $$

The logIE benzoic acid,solvent(1) = 0.84 was observed by averaging over 3 day measurements (one measurement per day, mean standard deviation of measurements was 0.085 logarithmic units). Thereafter, the logIE benzoic acid,solvent(1) was used to obtain logIE values for other compounds in solvent (1).

In the following discussion and results, the logIE values are used, as these provide the possibility to compare ionization efficiencies of all the compounds without having to measure each relative ionization efficiency value for all of the compounds. Also, the anchoring procedure between two solvents [Formula 3 followed by the re-anchoring for solvent (1)] allows comparing the logIE values for different solvents.

Consistency of the scale is expressed as consistency standard deviation:
$$ s={\left(\frac{SS}{n_m-{n}_c}\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.} $$
where n c is the number of assigned logIE values [i.e., number of analyzed substances – number of anchors (here: 1)].


Studied compounds were (structures can be found in SI) bought from Reakhim, Russia: 3-nitrophthalic acid, adipic acid, eosin B, bromophenol blue, bromothymol blue, phenol-2,4-disulfonic acid, phenolphthalein, cresol red, m-cresol purple, tiron, thymolphthalein, benzoic acid; from Chemapol, Czech Republic: bathocuproinedisulfonic acid, SPADNS; from Schering AG, Berlin, Germany: bromocresol purple; from Sigma-Aldrich, USA: bromocresol green, eosin Y; from E. Merck, Darmstadt, Germany: maleic acid, thymol blue; from Lach-ner, Czech Republic: 5-sulfosalicylic acid; and from Aldrich, USA: glutaric acid, pimelic acid, suberic acid. Obtained as a kind gift from the Institute of Pharmacology, Estonia were phthalic acid, fumaric acid, isophthalic acid, itaconic acid, succinic acid, mesaconic acid, and terephthalic acid.


COSMO-RS method [28] was used for calculating various parameters: aqueous pK a, logP ow (octanol-water), charge delocalization parameters (WAPS) and Klamt parameters. Degree of dissociation α of the compounds was calculated from the computed pK a values and the water phase pH.

First, full geometry optimization and energy calculation were carried out at the DFT BP TZVP level with the RI approximation and applying the COSMO continuum solvation model for all compounds using Turbomole, ver. 6.4 [29]. For most compounds, several conformers corresponding to different local energy minima were found. All of these were taken into account by statistical weighing inherent in the COSMO-RS procedure. The default convergence criteria of Turbomole were used: wave function convergence max difference 10–6 Hartree, geometry convergence max gradient |dE/dxyz| 10–3 Hartree/Bohr. This first computation step yields for every conformer the following data: the geometry of the conformer, detailed data on the shape of molecular cavity, the polarization charge densities mapped onto the cavity surface, the total electronic energy of the species submerged into a virtual conductor (ε = ∞), and molecular surface area and volume. Molecular cavity refers to the cavity constructed for the particular conformer within the COSMO solvation theory—constructed utilizing smoothed spheres using atomic radii ~20% larger than van der Waals radii. This cavity was later used as the molecular volume. The cavity surface refers to the so-called sigma-surface–polarization charge density on the molecular surface. For further information about the COSMO-RS theory, see reference [28].

Second, COSMO-RS calculation was carried out on all compounds using the above-listed data as input data with the COSMOtherm, ver. C3.0, release 14.01 [30]. COSMO-RS calculations take into account the interactions between species and the solvent/medium molecules, as well as between the solvent molecules themselves (implicit solvation model). The solvent composition is a required input parameter for COSMO-RS calculations, and in these calculations water was used as a solvent. Zero concentrations were used for the studied molecules. This way the interaction between the studied compounds and the solvent is taken into account but not the interaction between the molecules of studied compounds themselves. This situation corresponds well to the reality of very low concentrations used in the experiments and is common practice for carrying out such calculations. Both van der Waals interactions (electrostatic interactions: dipole–dipole, ion–dipole, etc. forces as well as dispersion forces) and hydrogen bonds (implicitly) are taken into account. These interactions are quantified via statistical counting and averaging of energies of pairwise interactions of molecular surface segments using polarization charge density maps of compounds created in the first step, taking into account the concentrations of the respective species in the solution. Terms accounting for vibrational contributions to the G tot are also added in this step. This is done implicitly, as these are represented through the experimental data used for parameterization of the method. The energetics of these interactions are calculated at the 298 K, using statistical thermodynamics procedure whereby also the conformers of all the interacting molecules are taken into account and statistically weighted based on their relative stabilities. This way, entropy effect of the same species present in multiple conformers is also accounted for. As a result, G tot value is found for every compound.

Model Development

To find out why certain substances give multiply charged species in mass spectrum and why others do not, it is necessary to know the physicochemical properties of all of the substances. In addition, the properties of the solvent need to be taken into account. It is known that some processes that lead to ionization of the analyte occur in the solvent phase and some in the gas phase [1]. Since it is difficult to account for processes in the spray that lead to the change in solvent properties, a simplification is usually done and the properties of analyte in the water phase are used [18]. In the ESI spray during droplet evaporation water content increases as the more volatile organic component vaporizes. Also, it is significantly less complicated to measure water phase pH-s. The same assumption is used in the current work. For model development, physicochemical properties obtained by COSMO-RS calculation were used. All statistical tests were carried out at 95% confidence level. Linear discriminant analysis (LDA) was carried out with statistical program R using the package Mass.


The ionization efficiency scale was compiled for both basic and acidic solvents, and for both doubly and singly charged ions. Altogether 29 compounds were studied, nine of them gave multiply charged ions in addition to singly charge ions in mass spectrometer in both solvents (given in Table 1 with the corresponding logIE values). In addition to the compounds brought in Table 1, also phenol-2,4-disulfonic acid gave doubly charged ions, but due to spectral interferences it was not possible to measure the corresponding logIE value. Benzoic acid was a reference substance, its logIE value taken to be zero in basic solvent as defined earlier in literature [19]. Altogether, 23 measurements in basic and 17 measurements in acidic solvent were made. In basic solvent, the ionization efficiency scale range was about 3 logarithmic units for doubly charged ions and 4 logarithmic units for singly charged ions. In acidic solvent the corresponding values were 4 and 3 logarithmic units. For singly charged ions the ionization efficiency values did not statistically differ in two solvents according to the t-test (95% confidence level was used). For doubly charged ions the logIE difference in two solvents was statistically significant for all compounds, except for tiron and SPADNS.
Table 1

LogIE Values in Negative Ionization Mode for Molecules that Gave Doubly Charged Species


Doubly charged

Singly charged

pK a1 a

pK a2 a

logP ow a

α 2 a pH=10.74

α 2 a pH=2.68

logIE 2 b pH=10.74

logIE 2 b pH=2.68

WAPS ·105a

logIE 1 b pH=10.74

logIE 1 b pH=2.68

WAPS ·105a

Bromophenol blue












Bathocuproinedisulfonic acid












Bromocresol green












Bromothymol blue












Eosin Y
























Sulphosalicylic acid
























s c








aCalculated pK a values, degrees of dissociation (α 2), logP ow values for neutrals and WAPS parameters for singly and doubly charged species can be found.

bTwo log ionization efficiency values can be found in the table: logIE 1 is the logIE of the molecule via formation of singly charged ion; logIE 2 of the molecule via formation of doubly charged ion. Reference substance is taken to be benzoic acid.

c s – consistency standard deviation, see Formula 4.

Parameters calculated with COSMO-RS method were correlated with measured logIE values in basic solvent. Correlation was poor (R 2 < 0.55) with Klamt parameters (polarity, polarizability, hydrogen bond accepting capability, hydrogen bond donating capability; see SI for details). It was not possible to correlate α 2 to ionization efficiency values obtained in basic solvent because all the analytes were fully deprotonated and degree of dissociation therefore equal to one in the case of all analytes. With the rest of the parameters, the correlation was good. Best correlation was observed between pK a1 and ionization efficiencies corresponding to singly charged analytes (R 2 = 0.79, correlation was positive). Good correlation was also obtained with WAPS and hydrogen bond donating capability for neutral (R 2 is 0.68 and 0.72 accordingly). For doubly charged ions, good correlation with WAPS parameter, molecular volume, and hydrogen bond donating capability was observed. In acidic solvent the correlations were similar; however, there was no correlation between doubly charged analytes ionization efficiency and WAPS parameter.

It was also of interest whether a predictive model for logIE values can be obtained based on physicochemical properties of analyzed substances. Best quantitative model (both for basic and acidic solvent) for predicting singly charged analytes ionization efficiency was obtained by using WAPS values for singly charged ions:
$$ \log I{E}_1=\left(-1.14\pm 0.23\right)\cdot WAPS+\left(3.36\pm 0.51\right) $$

Prediction precision of this model can be estimated by root mean error of 0.71 logIE units and square of correlation coefficient (R 2) of 0.64. The only substance to deviate significantly from the model was tiron. However, for doubly charged ions the obtained models for predicting logIE values quantitatively gave inacceptable results.

Linear discriminant analysis (LDA) was carried out to find properties to predict whether an analyte gives or does not give doubly charged ions in negative ionization mode. LDA model was found using training set composed of 18 randomly chosen compounds. Among training set there were five substances that gave doubly charged ions and 13 that did not give doubly charged ions (see SI, Supplementary Table S3). Different combinations of molecular parameters calculated with COSMOtherm were used (logP ow, WAPS parameters for anions and dianions, molecular area, pK a values). A model was chosen so that it would give the highest prediction precision for training set. This model was based on pK a2 and logP ow values:
$$ F=-0.48\cdot p{K}_{a2}(A)+0.60\cdot \log {P}_{ow}(A) $$

If F > 0, then analyte A gives doubly charged ions,

If F < 0, then analyte A does not give doubly charged ions.

This model has a prediction precision of 94%. An only slightly worse model (89%) was obtained with a model that contained molecular volume or WAPS instead of logP ow value. This model applies in solvent (1) as well as in solvent (2) since the same compounds gave doubly charged ions in both solvents, although with different ionization efficiency. The prediction model gave false positive result in the case of eosin B. This model was tested on a validation set containing 11 substances of which four gave doubly charged ions. The prediction precision was 82% for validation set. The model was incorrect for bathocuproinedisulfonic acid (false negative) and bromocresol purple (false positive). Possible reasons for incorrectness of the model for these substances are explored in the Discussion section.


It was observed that analytes with certain functional groups tended to give doubly charged ions. All substances that gave doubly charged ions had at least one sulfo group; also, in most cases (except bathocuproinedisulfonic acid) one or several electronegative groups (hydroxyl or bromo) and were aromatic. Dicarboxylic acids did not give doubly charged species in the MS independent of the pK a values and distance between carboxyl groups. Maximum distance between carboxylic groups in the case of dicarboxylic acids was about 10 Å in the case of suberic acid (octane diacid). In the case of tiron or sulfosalicylic acid, for example, which formed doubly charged ions, the distance between charged groups was about 5 Å.

From the LDA model it can be seen that the main properties that determine multiple charging are acidity and hydrophobicity. One important parameter is pK a2 that describes dissociation of the second acidic group. However, several substances that according to pK a value of the substance and pH of the solvent should be charged, did not give doubly charged ions (for example eosin B, phthalic acid, pimelic acid). Only analytes that had substantially lower pK a than solvent pH gave doubly charged ions. A similar result was also obtained by Felitsyn et al. [8], who studied multiple charging of peptides. They showed that obtaining a second charge in ESI process is hindered by electrostatic repulsion of charges and influenced by the hydrophobicity of the compound. Dissociation makes acids more hydrophilic, whicht in turn makes them stay inside the droplet. This means that only substances with substantial hydrophobic character can move to the surface of the droplet in ESI process. As a result, only substances with reasonable acidity (in this study pKa2 value below 5.8) combined with remarkable hydrophobic character can give doubly charged ions in ESI process.

LDA gave incorrect results for three substances (eosin B, bathocuproinedisulfonic acid, and bromocresol purple). The false positive result for eosin B and bromocresol purple may be caused by additives in the substance that could suppress signal and make mass spectrum noisy. Therefore, it is possible that eosin B could give doubly charged ions but its intensity remains below the limit of detection. On the other hand, it is possible that some other parameter is affecting the formation of doubly charged species that has not been considered in this study. The logP ow value of bathocuproinedisulfonic acid is very similar to substances that did not give doubly charged species and probably this is the main reason that the prediction is false negative. In calculation of physicochemical properties of bathocuproinedisulfonic acid, zwitterionic conformers were included and observed to be the most favored conformers in the liquid phase. It has been shown by Teesch et al. that the structure of gas-phase ions and solvent-phase ions are; however, different [31]. Consequently, it can be presumed that logP ow is not the best parameter to describe distribution of zwitterionic compounds such as bathocuproinedisulfonic acid on the surface of the droplet.

Span of the logIE scale and order of the substances in the scale was found to depend on solvent pH. In acidic solvent, the range of the logIE scale for doubly charged species is one logarithmic unit wider than in basic solvent. Therefore, the pH is a suitable tool to increase or decrease MS sensitivity for doubly charged species. The order of the substances is also dependent on solvent pH. Generally the ionization efficiencies for doubly charged species are about one logarithmic unit lower in basic solvent than for acidic solvent. The ionization efficiency values for singly charged species are less influenced by solvent pH (see Figure 1).
Figure 1

Ionization efficiencies of analytes in acidic and basic solvents, where logIE 1 is the logIE of the molecule via formation of singly charged ion; logIE 2 of the molecule via formation of doubly charged ion

In basic solvent, more doubly charged species are formed in the solution phase (pK a2 is substantially lower than solvent pH allowing the second protonation step) and, therefore, the signal corresponding to doubly charged species and corresponding logIE values are significantly higher. In acidic solvent, the formation of doubly charged ions is not favored and, therefore, more singly charged ions are formed. As a result, the IEs of doubly charged ions are lower than in basic solvent. Only for SPADNS and tiron the logIE values in basic and acidic solvent do not change, which can be explained by their very low pK a2 values (negative) and degrees of dissociation (α 2 value) of 1 in both solvents.

In basic solvent only bromothymol blue and eosin Y give higher IE for singly charged species than for doubly charges species. As mentioned before, the changes between acidic and basic solvent for singly charged species are, according to t-test, statistically insignificant.

In both solvents, bromothymol blue has the highest logIE value for singly charged species. This is probably because it has the highest hydrophobic character and, therefore, the singly charged species will move to the surface of the droplet easily, even if not a lot of them are formed in solution. The lowest logIE value belongs to tiron, which is a relatively small molecule with high hydrophilic character compared with other studied compounds. Also, another similar small and hydrophilic molecule, sulfosalicylic acid, has a low logIE value.

In acidic solvent, higher logIE 1 for singly charged species were observed for analytes that (1) have higher molecular volume, and (2) give doubly charged species with lower logIE 2 (bromothymol blue, bromocresol green, and bromophenol blue). Among the previously mentioned compounds, the highest logIE 1 value belongs to bromothymol blue, which also has the highest hydrophobicity. Substances with lower logIE values either give doubly charged ions with high logIE 2 value or are relatively hydrophilic substances, such as sulfosalicylic acid and tiron. This phenomenon can be explained that if the compound forms extensively doubly charged species, there is only a limited amount of compound left for forming singly charged species. Relatively large eosin Y has medium ionization efficiency and is different from others in hat it does not contain sulfo groups.

To formulate a model for predicting logIE 1 via formation of singly charged ions, different parameters were tested that had the best correlation with logIE 1 (pK a1, WAPS, and hydrogen bonding capacity for neutral substance) and also their combinations. The best parameter to describe ionization efficiencies (logIE 1) was charge delocalization parameter WAPS. It seems that charge delocalization is most important.

All in all, it can be seen that logIE 2 values in basic and acidic solvents differ for doubly charged species, which shows that logIE values depend on pH values. Solvent pH is therefore an important parameter to take into account to form a universal logIE prediction model and also to reach higher selectivity in analysis. Also, it can be seen from previous the discussion that formation of singly and doubly charged species are strongly related processes.


A model for predicting whether a compound gives or does not give doubly charged ions in negative ionization mode was developed based on pK a2 and logP ow values with the prediction precision of 82%. Also, a model was developed for predicting ionization efficiencies (logIE 1) corresponding to singly charged species of multiply chargeable substances based on charge delocalization (WAPS) parameter. It was also shown that double charging is most influenced by acidity of a substance, its hydrophobicity (octanol-water partition coefficient), and charge delocalization. It was observed that solvent pH affects the ionization efficiencies of analytes and its optimal choice is a good tool to enhance selectivity and sensitivity. Therefore, the factors influencing the ionization efficiency of doubly charged ions are similar to the ones that have been reported to influence ionization efficiencies of singly charged analytes. However, it is shown that the solution phase degree of dissociation alone does not relate to the appearance of doubly charged ions in ESI/MS spectra. This demonstrates the complexity of ESI processes and the need to further investigate ESI mechanism.



The authors acknowledge support for this work by Personal Research Funding Project 34 from the Estonian Research Council.

Supplementary material

13361_2016_1384_MOESM1_ESM.docx (140 kb)
ESM 1 (DOCX 139 kb)


  1. 1.
    Cech, N.B., Enke, C.G.: Practical implications of some recent studies in electrospray ionization fundamentals. Mass Spectrom. Rev. 20, 362–387 (2001)CrossRefGoogle Scholar
  2. 2.
    Cech, N.B., Enke, C.G.: Relating electrospray ionization response to nonpolar character of small peptides. Anal. Chem. 72, 2717–2723 (2000)CrossRefGoogle Scholar
  3. 3.
    Douglass, K.A., Venter, A.R.: Investigating the role of adducts in protein supercharging with sulfolane. J. Am. Soc. Mass Spectrom. 23, 489–497 (2012)CrossRefGoogle Scholar
  4. 4.
    Huffman, B.A., Poltash, M.L., Hughey, C.A.: Effect of polar protic and polar aprotic solvents on negative-ion electrospray ionization and chromatographic separation of small acidic molecules. Anal. Chem. 84, 9942–9950 (2012)CrossRefGoogle Scholar
  5. 5.
    Wong, S.F., Meng, C.K., Fenn, J.B.: Multiple charging in electrospray ionization of poly (ethylene glycols). J. Phys. Chem. 92, 546–550 (1988)CrossRefGoogle Scholar
  6. 6.
    Schnier, P.D., Gross, D.S., Williams, E.R.: On the maximum charge state and proton transfer reactivity of peptide and protein ions formed by electrospray ionization. J. Am. Soc. Mass Spectrom. 6, 1086–1097 (1995)CrossRefGoogle Scholar
  7. 7.
    Smith, R.D., Loo, J.A., Loo, R.R.O., Busman, M., Udseth, H.R.: Principles and practice of electrospray ionization-mass spectrometry for large polypeptides and proteins. Mass Spectrom. Rev. 10, 359–451 (1991)CrossRefGoogle Scholar
  8. 8.
    Felitsyn, N., Peschke, M., Kebarle, P.: Origin and number of charges observed on multiply-protonated native proteins produced by ESI. Int. J. Mass Spectrom. 219, 39–62 (2002)CrossRefGoogle Scholar
  9. 9.
    Loo, J.A., Edmonds, C.G., Udseth, H.R., Smith, R.D.: Effect of reducing disulfide-containing proteins on electrospray ionization mass spectra. Anal. Chem. 62, 693–698 (1990)CrossRefGoogle Scholar
  10. 10.
    Loo, J.A., Udseth, H.R., Smith, R.D.: Solvent effects on the charge distribution observed with electrospray ionization-mass spectrometry of large molecules. Biol. Mass Spectrom. 17, 411–414 (1988)CrossRefGoogle Scholar
  11. 11.
    Cole, R.B.: Electrospray and MALDI mass spectrometry: fundamentals, instrumentation, practicalities, and biological applications. Wiley, Hoboken (2010)CrossRefGoogle Scholar
  12. 12.
    Iavarone, A.T., Jurchen, J.C., Williams, E.R.: Effects of solvent on the maximum charge state and charge state distribution of protein ions produced by electrospray ionization. J. Am. Soc. Mass Spectrom. 11, 976–985 (2000)CrossRefGoogle Scholar
  13. 13.
    Iavarone, A.T., Jurchen, J.C., Williams, E.R.: Supercharged protein and peptide ions formed by electrospray ionization. Anal. Chem. 73, 1455–1460 (2001)CrossRefGoogle Scholar
  14. 14.
    Iavarone, A.T., Williams, E.R.: Mechanism of charging and supercharging molecules in electrospray ionization. J. Am. Chem. Soc. 125, 2319–2327 (2003)CrossRefGoogle Scholar
  15. 15.
    Ganisl, B., Taucher, M., Riml, C., Breuker, K.: Charge as you like! Efficient manipulation of negative ion net charge in electrospray ionization of proteins and nucleic acids. Eur. J. Mass Spectrom. 17 (2011)Google Scholar
  16. 16.
    Douglass, K., Venter, A.: On the role of a direct interaction between protein ions and solvent additives during protein supercharging by electrospray ionization mass spectrometry. Eur. J. Mass Spectrom. 21, 641 (2015)CrossRefGoogle Scholar
  17. 17.
    Leito, I., Herodes, K., Huopolainen, M., Virro, K., Künnapas, A., Kruve, A., Tanner, R.: Towards the electrospray ionization mass spectrometry ionization efficiency scale of organic compounds. Rapid Commun. Mass Spectrom. 22, 379–384 (2008)CrossRefGoogle Scholar
  18. 18.
    Oss, M., Kruve, A., Herodes, K., Leito, I.: Electrospray ionization efficiency scale of organic compounds. Anal. Chem. 82, 2865–2872 (2010)CrossRefGoogle Scholar
  19. 19.
    Kruve, A., Kaupmees, K., Liigand, J., Leito, I.: Negative electrospray ionization via deprotonation: predicting the ionization efficiency. Anal. Chem. 86, 4822–4830 (2014)CrossRefGoogle Scholar
  20. 20.
    Tang, L., Kebarle, P.: Dependence of ion intensity in electrospray mass spectrometry on the concentration of the analytes in the electrosprayed solution. Anal. Chem. 65, 3654–3668 (1993)CrossRefGoogle Scholar
  21. 21.
    Ehrmann, B.M., Henriksen, T., Cech, N.B.: Relative importance of basicity in the gas phase and in solution for determining selectivity in electrospray ionization mass spectrometry. J. Am. Soc. Mass Spectrom. 19, 719–728 (2008)CrossRefGoogle Scholar
  22. 22.
    Amad, M.H., Cech, N.B., Jackson, G.S., Enke, C.G.: Importance of gas-phase proton affinities in determining the electrospray ionization response for analytes and solvents. J. Mass Spectrom. 35, 784–789 (2000)CrossRefGoogle Scholar
  23. 23.
    Chalcraft, K.R., Lee, R., Mills, C., Britz-McKibbin, P.: Virtual quantification of metabolites by capillary electrophoresis-electrospray ionization-mass spectrometry: predicting ionization efficiency without chemical standards. Anal. Chem. 81, 2506–2515 (2009)CrossRefGoogle Scholar
  24. 24.
    Nguyen, T.B., Nizkorodov, S.A., Laskin, A., Laskin, J.: An approach toward quantification of organic compounds in complex environmental samples using high-resolution electrospray ionization mass spectrometry. Anal. Methods 5, 72–80 (2013)CrossRefGoogle Scholar
  25. 25.
    Henriksen, T., Juhler, R.K., Svensmark, B., Cech, N.B.: The relative influences of acidity and polarity on responsiveness of small organic molecules to analysis with negative ion electrospray ionization mass spectrometry (ESI-MS). J. Am. Soc. Mass Spectrom. 16, 446–455 (2005)CrossRefGoogle Scholar
  26. 26.
    Zhou, S., Cook, K.D.: Protonation in electrospray mass spectrometry: wrong-way-round or right-way-round? J. Am. Soc. Mass Spectrom. 11, 961–966 (2000)CrossRefGoogle Scholar
  27. 27.
    Kruve, A., Kaupmees, K., Liigand, J., Oss, M., Leito, I.: Sodium adduct formation efficiency in ESI source: sodium adduct formation efficiency in ESI source. J. Mass Spectrom. 48, 695–702 (2013)CrossRefGoogle Scholar
  28. 28.
    Klamt, A.: COSMO-RS from quantum chemistry to fluid phase thermodynamics and drug design. Elsevier Science Ltd., Amsterdam (2005)Google Scholar
  29. 29.
    TURBOMOLE V6.4, 2012, a development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989–2007, TURBOMOLE GmbH, since 2007; available at Accessed 2013
  30. 30.
    Eckert, F., Klamt, A.: COSMOtherm, Version C3.0, Release 13.01; COSMOlogic GmbH&CoKG, Leverkusen, Germany, 2013; available from Accessed: 2013
  31. 31.
    Teesch, L.M., Orlando, R.C., Adams, J.: Location of the alkali metal ion in gas-phase peptide complexes. J. Am. Chem. Soc. 113, 3668–3675 (1991)CrossRefGoogle Scholar

Copyright information

© American Society for Mass Spectrometry 2016

Authors and Affiliations

  1. 1.Institute of Chemistry, Faculty of Science and TechnologyUniversity of TartuTartuEstonia

Personalised recommendations