The most common approach to separate isomers with similar spatial conformations is the use of mono- and divalent metal cations, which form distinguishable adduct complexes with the target analytes. This can lead to successful separations of isomeric lipids and recognition of different ion conformations when using ESI in positive mode (Figure 2A). The cations (X) are added to the working solutions in the form of commercially-available salts (e.g., acetate and nitrate salts), allowing their interactions with the analytes and forcing the formation of monomers (e.g., [M+X]⁺, [M+2X]²⁺) and multimers (e.g., [2M+X]⁺ and [3M+X]⁺). The most commonly used cations for adduction are: 1) alkali metals (e.g., monovalent cations of sodium (Na⁺), potassium (K⁺), lithium (Li⁺), caesium (Cs⁺) and rubidium (Rb⁺)); 2) alkaline earth metals (e.g., divalent cations of magnesium (Mg²⁺), calcium (Ca²⁺), strontium (Sr²⁺) and barium (Ba²⁺)); and 3) transition metals (e.g., monovalent cations of silver (Ag⁺) and divalent cations of iron (Fe²⁺), cobalt (Co²⁺), nickel (Ni²⁺), copper (Cu²⁺) and zinc (Zn²⁺)).

Investigations into cation adducts have led to interesting separation results, especially with those involving alkali metals. One of the most commonly used cation adducts is sodium, which was used to successfully resolve isomeric molecules among different lipid classes. Thus, monomers of sodiated isomeric species of GL (i.e., MG and DG) have been tested in DI-DTIMS-MS (May et al., 2020) using high resolution demultiplexing (HRdm), and in LC-TIMS-MS (Fouque et al., 2019) to increase the instrument’s resolution. For instance, peaks of sn-positional MG isomers (i.e., 1-lineloyl glycerol (1-LG) and 2-LG) (May et al., 2020), and DG sn-regioisomers (e.g., DG 22:1/22:1/0:0 and DG 22:1/0:0/22:1) could be successfully differentiated (Fouque et al., 2019). While double-bond isomerism can be resolved using LC alone, IMS has a demonstrated ability of distinguishing acyl chain isomers.

A great deal of interest has been placed on PC species, on account of the abundance of isomeric species in biological samples. The position of the acyl chain on PC species plays an important role in their function, as phospholipase A2 produces lipid messengers upon cleavage of the fatty acid in the sn-2 position. High-resolution and multiplexing DI-DTIMS-MS has been used to separate db-positional isomers of PC with the same double bond geometries (Groessl et al., 2015). The [M+Na]⁺ adducts for PC 18:1(6Z)/18:1(6Z) and PC 18:1(9Z)/18:1(9Z) could be unequivocally detected and PC 18:1/16:0 and its acyl positional isomer could be quantified in complex extracts. Contrastingly, LC-TIMS-MS was unable to resolve the same PC molecules, as the resolution was lower for sodium adducts than for their protonated species (Fouque et al., 2019).

Steroid-hormone lipids, which comprise a wide variety of structurally related molecules, have been extensively studied in terms of their isomeric resolution because of their importance in physiological processes. Structural isomers of this group exhibit differences in the presence and position of ketone/hydroxyl groups, double bonds, and A- and D-ring functional groups. Some baseline resolved examples include dimers (i.e., [2M+Na]⁺) of corticosterone and 11-deoxycortisol by DI-TWIMS-MS (Rister et al., 2019) and, additionally, testosterone and dehydroepiandrosterone (DHEA), 17-hydroxyprogesterone and 11-deoxycorticosterone by DI-DTIMS-MS (Chouinard et al., 2017a). In the latter study, dimers showed lower resolution than their monomeric counterparts, whereas monomers showed a higher resolution for aldosterone and cortisone. Analysis of sodiated monomers in the same instrument also revealed isomer separation in human urine — for example, 7-keto-DHEA and methyl-1-testosterone (Velosa et al., 2022b).

Bile acids (BA) are steroids synthesised in the liver from cholesterol. The primary BA cholic acid, and its derivatives, exhibited measurable ion mobility differences upon investigation by DI-TWIMS (Hadavi et al., 2022). A representative isomeric pair from this study is glycodeoxycholic acid (GDCA) and glycochenodeoxycholic acid (GCDCA), whose structural differences are related to the position of the hydroxyl group on C7 and C12, respectively. The CCS values for their sodium monomers showed a difference of 8.37 Ų, owing to the adoption of a more planar conformation in GDCA. Conversely, multiple sodium monomers presented a bulky conformation with a poorer separation between isomers. Thus, the presence of multiple adducts does not necessarily increase their ΔCCS value (Hadavi et al., 2022).

A glycosphingolipid pair of disialoganglosides, GD1a and GD1b, which differ in the localization of sialic acid residues in their oligosaccharide head group, could only be resolved in standard mixtures with IMS-MS as doubly sodiated species [M+2Na]²⁺. DI-DTIMS-MS in combination with HRdm achieved a satisfactory peak-to-peak resolution (May et al., 2020). GD1a and GD1b have also been studied coupling an ultra-high resolution SLIM platform, which revealed distinct drift times with baseline separation even when using the minimal possible path (Wojcik et al., 2017). These promising results permitted the study of these species in more complex biological samples, including mouse brain extracts (Wormwood Moser et al., 2021).

Potassium adducts appear to have a lower resolution than sodium adducts in distinguishing the sn-position isomers of PC (Groessl et al., 2015). However, isomer pairs of steroid hormones such as [2M+K]⁺ adducts showed a higher resolution than sodium adducts (e.g., aldosterone and cortisone, as well as corticosterone and 11-deoxycortisol) (Rister et al., 2019).

The lithium multimeric form [2M+Li]⁺ failed to enhance resolution in DI-TWIMS-MS (Rister et al., 2019). By contrast, the [3M+Li]⁺ multimeric form of androgenic steroids (e.g., dihydrotestosterone and androsterone) could be differentiated using DI-DMS-MS (Wei et al., 2020).

Additionally, the use of lithium monomers led to significant differences in CCS values for cortisone and prednisolone in LC-DTIMS-MS with HRdm, whereas their sodium monomers were not resolved (Neal et al., 2022). Other alkali metals, such as Cs⁺ and Rb⁺, were tested in the aforementioned androgenic steroids, but the results were disappointing due to very low signal-to-noise ratios (Wei et al., 2020). Another characterisation study of glucocorticoids with the same alkali metals was also unfruitful, but in this case it was due to low resolution (Neal et al., 2022). Alternative alkaline earth metals (X²⁺) were introduced, and improvements were observed for [M+Ba+acetate]⁺ adducts (Neal et al., 2022).

Finally, it has been shown that the use of transition metals has advantages in isomer separation, as illustrated with silver, which permitted the separation of TG species as [M+Ag]⁺ adducts with exchanged fatty acyl chains in DI-DMS-MS. This method was applied to more complex biological samples of animal fats (Šala et al., 2016). Silver nitrate salt has been employed for the determination of PC 16:0/18:1 and PC 18:1/16:0 regioisomers in DI-DTIMS-MS, allowing greater differences in their K₀ and CCS values than with Na⁺ and K⁺ metal cations (Groessl et al., 2015). First row transition metals, such as Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, and Zn²⁺ have also been tested, but the resolution did not markedly improve (Neal et al., 2022).

Some examples have been reported of negatively charged molecules analysed in negative ESI mode, which helps to resolve regiosiomers; however, the lower signal obtained in most of the analyses in negative polarity limits its use. Anions can trigger conformational changes in isomers, resulting in mobility shifts between them. A good illustration of this is acetate adducts (i.e., [M+CH₃COOH-H]^-) of PC 18:0/20:4;OH differing in the location of the hydroxy group (i.e., C8, C9, C11, C12 or C15 position), which provided a greater drift separation than sodium adducts in LC-DTIMS-MS. While baseline separation could not be achieved, these molecules might be determined in complex samples using their CCS value. In this case, the hydroxyl group close to the head-group results in a faster mobility and a lower CCS value. This basis, either with acetate or with other ions, can be helpful when analysing PC regioisomers that differ in the position of their functional group (Hinz et al., 2019). As another example, steroids were successfully analysed with the use of chloride (Cl⁻) and fluoride (F⁻) anions in DI-TIMS, permitting the baseline resolution in prednisolone and cortisone pairs as chloride adducts (Cole et al., 2020).

Another interesting strategy to improve separation is the formation of non-covalent binding complexes. An example is cyclodextrin acting with BA, which creates significant mobility differences among isomers. Cyclodextrin molecules are cyclic compounds composed of 6–8 glucopyranoside monomers bound in a conical fashion with a hydrophilic outer shell and a hydrophobic inner shell. Cyclodextrin is useful because the aliphatic chain of BA can be placed inside its hydrophobic cavity. Adducts of 3-amino-3-deoxy-α-cyclodextrin (αCD) (i.e., [M+αCD+H+K]²⁺) were successfully analysed in DI-TWIMS-MS SLIM SUPER together with the use of CRIMP, permitting the separation of positional isomers of taurine- and glycine-conjugated BA with sufficient resolution (Table 1, entry 23). All approaches employed in this work proved to be essential for the determination of these molecules by IMS (Chouinard et al., 2018).

In summary, it appears that positive metal adducts provide a higher isomer resolution than negative adducts formed with negatively charged molecules or ions, regardless of their ionisation efficiency. The selection of ions to incorporate into an IMS-MS workflow for optimal separation depends strongly on the ion mobility spectrometer, the type of isomerism addressed and the analytes of interest. Currently, there is no predictive model or trend that can be reliably applied for ion selection for adduction, but the examples given in the literature may serve as a preliminary guide.

Nitrogen (N₂) is the most common gas used in IMS-MS analyses (Matz et al., 2002) and the majority of the values in CCS databases have been measured with N₂. Nevertheless, other gases of different polarities have been tested in IMS instruments (see Table 2).

There are several examples in IMS-based lipidomics where the composition and polarity of the buffer gas have been changed. According to the data in Table 2 , He and Ar, although less polar than N₂, have been implemented for some time. Helium was found to improve the separation of standards mixtures in DI-DMS-MS. An example is GL isomers with different fatty acid positions, which were resolved under He-rich gases (i.e., 70:30 He/N₂) (Shvartsburg and Smith, 2008). However, the use of less polar gases may require other combinatory strategies such as adduct formation. For example, potassium adducts of prednisolone and cortisone in LC-DTIMS-MS in an Ar stream showed a high resolution (Neal et al., 2022).

Nevertheless, the trend seems to be the use of more polar gases. A comparative study between N₂, He, Ar, and CO₂ in DI-DTIMS-MS for endogenous steroid hormones concluded that more polar drift gases (i.e., CO₂) yielded a marked improvement in mobility separation, especially for testosterone-related metabolites (Chouinard et al., 2017a). Similar results were obtained in a study comparing identical buffer gases in the same ion mobility instrumentation for cortisone and prednisolone isomers. However, in these studies other molecules showed better resolution under other gas conditions (Neal et al., 2022). Meanwhile, mixtures of gases were also tested, such as CO₂ with breathable air (0.11% O₂ and 79.89% N₂). Steroids in DI-DTIMS achieved a better separation than with the use of these gases. A specific drift gas mixture worked better for some regioisomer pairs due to a better separation. For corticosterone and 21-deoxycortisol, a 55/45 CO₂/mixture showed 85% separation) (Kaszycki et al., 2019).

As is the case with adduct formation, the separation conditions using buffer gases (i.e., used singly or as mixtures) differ depending on the isomerism tested and the pair or group of isomers examined, requiring specific and optimal configurations for each.

Pressure or temperature modifications are usually combined with other strategies to boost changes in mobility (e.g., combination with buffer gas additives or adduct formation). Increased pressure provides an increased number of ion-molecule collisions, thus improving the R _p . An example is TG analysis using DI-DMS-MS, illustrated by the improved resolution between sn-regioisomers when combining silver-ion adduction, chemical modifiers and higher pressures of N₂ (up to 41 psi) (Šala et al., 2016). Also, the same instrumentation using 35 psi of N₂ provided a higher resolution than with lower pressures for PC regioisomers as silver adducts (Maccarone et al., 2014). Another example is the analysis of oxysterols in LC-TWIMS-MS, where up to 3.5 mbar was applied (Kylli et al., 2017). In the case of DTIMS, pressure is not easily manipulable, but commercial low- and high-pressure platforms are available. High-pressure platforms provide a higher R _p (Dodds and Baker, 2019). Up to 1,400 mbar pressure employed in high-pressure DTIMS-MS led to better separation of structural isomers of PC and gangliosides (Groessl et al., 2015; Kaszycki et al., 2019).

The temperature also influences the mobility of ions, showing an indirect relationship with resolution; that is, the lower the temperature the better the resolution achieved (Dodds and Baker, 2019). However, IMS experiments should be performed at relatively high temperatures to reduce uncertainties in the measurement of reduced mobilities (Fernandez-Maestre and Daza, 2021). Therefore, a compromise in temperature ranges must be reached to improve resolution. We were unable to find any recent lipidomic studies to highlight the temperature changes that could lead to substantial isomer resolution.

In view of the above, changes in the composition, pressure and temperature of the drift gas can be performed on most of the available IMS-MS instruments (Dodds and Baker, 2021). However, the pressure or temperature ranges vary for each instrument. Furthermore, it is likely that achieving the best separation of different lipid classes or pairs of structural lipid isomers will require specific conditions. For all these reasons, it is advisable to optimise the methods in order to obtain a broader isomer coverage in the analyses.

Isomeric resolution by IMS-MS has been reported with alternative fragmentation methodologies, such as electron impact excitation of ions from organics (EIEIO). This has been performed in DMS analysers equipped with an ExD cell (a branched radio-frequency electron-ion reaction device) (Baba et al., 2016), and has provided structural information of lipid class, acyl length, and sn- and db-position of SM (Table 1, entries 18–20). The technique could also remove isobaric interferences in the combined IMS analysis of animal tissue extracts. Double-bond location was based on the presence of -2H mass shifts in the products and a characteristic “V” shape in the EIEIO fragmentation spectra (Baba et al., 2016). Accordingly, more extensive information can be obtained with the use of a single spectrum provided by this technique. The same approach was recently implemented to increase the lipid coverage when analysing GL, GP, and SP together, which allowed for the structural characterisation of over 300 regioisomer lipids in complex animal extracts (Baba et al., 2018).

Ultraviolet photodissociation (UVPD, at 193 nm) of unsaturated lipids enables a high-energy photoactivation process, resulting in the cleavage of C-C bonds adjacent to a C=C bond. This process yields diagnostic ions with a distinctive mass difference of 24 Da (Fang et al., 2020). An application of this method, coupling AP-DTIMS with a UVPD-enabled mass spectrometer and multiplexing, could unequivocally determine PC species differing in their db-locations by their UVPD spectra - for example, PC 18:1(6Z)/18:1(6Z), PC 18:1(9Z)/18:1(9Z) and PC 18:0/18:2(9Z,12Z) (Table 1, entries 8–9) (Sanders et al., 2022).

Ozone-Induced Dissociation (OzID) involves a gas-phase reaction between ozone vapour and mass-selected unsaturated lipid ions. This yields diagnostic fragment ions (differing in 16 Da), pinpointing the db-location in the precursor ion (Xia and Wan, 2021). Several groups have explored the impact of OzID coupled to IMS-MS, as it has wide compatibility with many IMS instruments. For example, it can be coupled directly before IMS in DTIMS (i.e., the ion trap) (Poad et al., 2018b); directly to the IMS cell in TWIMS (Poad et al., 2017; Vu et al., 2017; Claes et al., 2021); or immediately after in DMS analysers (Maccarone et al., 2014; Steiner et al., 2016; Poad et al., 2018a; Berthias et al., 2021), thus modifying the mass spectrometers.

Both GL and (mostly) GP have been broadly characterised at their db-position using OzID fragmentation spectra. Some examples included lipid commercial standards of TG and PC (Table 1, entries 5 and 12). Different cations have been tested, and analyses with Ag⁺ showed the best IMS resolution and best OzID performance (Berthias et al., 2021). Moreover, isomerism was resolved for sphingosines (SPH) (Table 1, entry 22) (Poad et al., 2018a).

The backbone substitution of acyl chains and cis/trans isomerism have also been studied using OzID, owing to the different reaction rates between isomers. Double bonds present in sn-2 acyl chains and db with trans geometry in PC species react faster to OzID. For instance, sn-regioisomers PC 16:0/18:1(9Z) and PC 18:1(9Z)/16:0 differed in intensity in their OzID spectra and similar results were observed for cis/trans isomers PC 18:1(9E)/18:1(9E) and PC 18:1(9Z)/18:1(9Z) (Table 1, entries 10–11) (Vu et al., 2017).

Interestingly, methodologies combining OzID with CID−also known as COzID−provide more information in terms of sn- and db-positions and db-geometries (Table 1, entries 13–17). Some examples include PC 16:0/18:1(7Z) versus PC 16:0/18:1(9Z) (Poad et al., 2017), which were differentiated due to unequal spectra; and PE 18:1(9E)/18:1(9E) versus PE 18:1(9Z)/18:1(9Z) (Poad et al., 2018b), which showed the same spectra but different peak intensities. Another possibility is its combination with cation adduction for better IMS resolution among isomers (e.g., silver (Maccarone et al., 2014) and sodium (Claes et al., 2021) adducts). Furthermore, LC-DMS-CID/OzID was found to enable the complete characterisation of complex molecules such as an atypical 1-deoxysphingosine (SPH) after dimethyl disulfide (DMDS) derivatisation (Steiner et al., 2016) and could locate the db-position (Liao and Huang, 2022) (Table 1, entry 21).

Hydrogen/deuterium exchange (HDX) method is based on a chemical reaction in which a covalently bonded hydrogen atom is replaced by a deuterium atom from the solvent. Its combination with DTIMS and CID fragmentation revealed that each ion exhibits a unique deuterium uptake profile (Maleki et al., 2018). In the aforementioned study, the technique did not allow the separation of isomeric GP species (e.g., PC 14:1(9Z)/14:1(9Z) and PC 14:1(9E)/14:1(9E)), but it showed a strong potential for isomer resolution upon a better understanding the behaviour of HDX (Maleki et al., 2018).

There are many other advanced tandem mass spectrometry techniques, such as those listed in Figure 1C, however, they have yet to be implemented in IMS-MS analysis. Nevertheless, many hold promise for in-depth lipid characterisation, as they have been proven successful in conventional LC-MS methodologies (Liang et al., 2007; Jones et al., 2015; Feng et al., 2019).

The use of adduct ions is a proven alternative for the separation of stereoisomers by altering gas-phase ion structures. New approaches in metal adduct formation with sodium in IMS-MS technologies can resolve R and S enantiomers of sphingolipids. For example, Cer 18:0;2OH[S] and Cer 18:0;2OH[R] could be discriminated using LC-DTIMS upon sodium ion binding in a standard mixture. Sodium coordinated in such a manner that the orientation of the hydroxyl group in the R enantiomer repelled the ceramide chains, acquiring an open conformation. Contrastingly, a closer conformation was adopted in the S enantiomer where the orientation was located in the opposite direction (Kyle et al., 2016).

Sodium adduct formation has been more extensively used for steroid epimers. For instance, in the case of 25-hydroxy-vitamin D3 and its epimer in carbon 3, using DTIMS (Chouinard et al., 2017b; Oranzi et al., 2018), non-conjugated muricholic acids (α-, β-, ω- and γ-MCA) and taurine conjugated muricholic acids (Tα-, Tβ-, Tω- and Tγ-MCA) using TWIMS (Hadavi et al., 2022), and androgenic steroid hormones (e.g., testosterone and epitestosterone), using TWIMS as well (Rister et al., 2019).

Due to the lack of resolution, in some cases adduction with other alkali metal has proven to be efficient, such as lithiated multimers in FAIMS for androsterone and its trans epimer (Wei et al., 2020). As previously stated for structural isomers, each pair of epimers requires different adducts to be resolved, and their selection must be carried out empirically. In the case of steroid hormones, extensive work has been done comparing sodium with other metal cations (potassium and lithium). The best resolution of dimers was achieved when using lithium for androsterone and epiandrosterone, potassium for α- and β-estradiol, and sodium for testosterone and epitestosterone (Rister et al., 2019).

Separation with other metal groups is both promising and challenging because of the greater number of possible adducts that can be formed and used to distinguish isomers. Alternative cation adducts have been tested for androsterone and trans-androsterone epimers in DTIMS, including alkaline earth metals (Mg²⁺, Ca²⁺, Sr²⁺, Rb²⁺) and first-row transition metals (Sc³⁺, Cr³⁺, Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, and Zn²⁺). A small improvement in separation was observed for alkaline earth metals, but interestingly, first-row transition metal adducts led to enhanced resolution, notably with copper and zinc adducts (Chouinard et al., 2017a).

The previously mentioned αCD, which forms complex [M+αCD+H+K]²⁺ adducts, has also been used in epimer separation as illustrated for the BA tauroursodeoxycholic acid (TUDCA) and taurochenodeoxycholic acid (TCDCA), as well as for glycoursodeoxycholic acid (GUDCA) and glycochenodeoxycholic acid (GCDCA) in DI-TWIMS SLIM SUPER platform with the use of CRIMP software (Table 3, entry 1) (Chouinard et al., 2018).