1 Introduction

Desorption electrospray ionization (DESI) operational parameters have been extensively studied [1–12]. DESI optimization depends on several variables including source geometry, source construction, interrogation surface material, instrumental settings, etc. Table 1 shows a summary of operational parameters found in the literature compiled from over 25 references. The DESI parameter space is large and some parameters are correlated, so optimization is not always straightforward. Since many of the parameters are interrelated, multiple combinations of settings are capable of providing stable and strong ion currents. In addition, it is well known that signal intensities are highly variable during DESI [7, 9, 10, 13–16], although the use of an appropriate internal standard can result in inter-day variability of <10 % RSD [15]. These factors inevitably lead to the large variation in the optimization results presented in the literature.

Table 1 Descriptive Statistics Summarizing DESI Parameters Found in the Literature
Full size table

Most molecules capable of analysis by DESI are desorbed by the droplet pickup mechanism [17–20]. For droplet pickup to occur, analyte molecules are initially dissolved into a thin film of solvent formed on the interrogated surface [17]. An initial 0.1 to 0.2 min is required to reach maximum signal intensity during DESI [21, 22], which have been attributed to a delay in the formation of the liquid film. The addition of surfactant to the solvent spray reduced this initial delay, attributed to the increased wettability and solubilizing power of the solvent [22]. Once the film has formed, primary droplets colliding with the film generate two populations of secondary droplets that are ejected from the film [5, 23, 24]. Larger, slower-moving droplets leave the surface close to their specular angle while smaller, faster-moving droplets follow the dynamics of the nebulizing gas and tend to skim the sample surface [23].

After desorption, ionization is thought to occur by typical ESI mechanisms [2, 25, 26]. The spectral features of DESI and ESI share many similarities, including multiply charged ions, noncovalent complexes, and metal adducts [2]. In particular, protein spectra generated from both methods are remarkably similar, although charge state distributions are generally shifted to lower charge states in the case of DESI, suggesting that secondary droplets generated during DESI carry less charges than those generated during ESI [26]. Ion mobility studies [26] and the determination of internal energy distributions [25] have also supported an ESI-like ionization mechanism for DESI.

Analysis by DESI is comprised of two main processes, desorption of the analyte from the surface and subsequent ionization. It is likely that the optimum conditions found for DESI are a compromise between those conditions ideal for desorption and those ideal for ionization. In this study, we separately investigated the relative contributions of desorption and ionization to the DESI process. To decouple desorption from ionization, we first collected the desorbed material using surface desorption collection (SDC), a recently developed surface sampling technique [27] that closely mimics the DESI desorption process. To study ionization separately, a system that closely mimics DESI but does not require desorption or analyte pickup was required. The primary droplets produced by a DESI microsprayer are substantially different from those present after surface collision [5]. To better mimic the ionization conditions after droplet-pickup the analyte was added to the electrosprayed solvent directed at a clean surface and placed in front of the mass spectrometer, a process we call reflective electrospray ionization (RESI). A similar method was used previously for DESI optimization [28, 29].

We present results for the optimization of several different parameters and show how each parameter affects desorption or ionization processes during DESI independently.

2 Experimental

Various parameters for DESI, SDC, and RESI were optimized, including gas pressure, spray voltage, sprayer-to-inlet distance, and sprayer-to-surface distance. The standard operating conditions were inferred from Table 1 considering our sprayer design and, unless otherwise noted, are listed in Table 2.

Table 2 Standard Operating Conditions for SDC, RESI, and DESI Used in the Present Study
Full size table

2.1 Instrumentation

A linear ion trap mass spectrometer (LTQ; Thermo Scientific, Waltham, MA, USA) was combined with a 3-dimensional translational stage (Purdue University, West Lafayette, IN, USA) for DESI and RESI analysis and SDC collection. For the generation of a pneumatically-assisted solvent spray, a typical spray ionization source was constructed in-house [30]. Briefly, the source was comprised of a 1/16-in. Swagelok T-piece, a fused silica solvent line (o.d. 190 μm, i.d. 75 μm), and a coaxial stainless steel nebulizing gas capillary (o.d. 1/16”, i.d. 250 μm, ca 1.5 cm long). The spray potential was applied to the liquid junction on the stainless-steel syringe needle used to deliver the spray solvent. The tip of the extended mass spectrometer inlet capillary was bent at an angle of 10° to facilitate efficient droplet/ion transfer during DESI and RESI [10]. Analysis of the samples collected by SDC was performed with a spectrofluorometer (FL920; Edinburgh Instruments, Reading, England) by measuring the fluorescence of rhodamine 6 G at 547 nm at an excitation wavelength of 500 nm.

A schematic of the various analytical techniques used in the current study is presented in Figure 1.

Figure 1
figure 1

A schematic of spray desorption collection (SDC) and reflective electrospray ionization (RESI)

Full size image

2.2 Surfaces and Samples

Two surfaces were required for this study, the sample surface used in all experiments and the collection surface used during SDC. Glass was chosen as the sample surface for DESI, SDC, and RESI experiments as discussed below. For the collection surface used during SDC, silica gel TLC plates were found to give quantitative results after extraction of the analyte followed by fluorescence spectroscopy.

Rhodamine 6 G was used as the analyte for each study. It is a red colored dye that is readily found in some permanent markers and is easily visible to the naked eye. It has a high fluorescence quantum yield [31] and has been used before in many DESI studies [6, 7, 14, 21, 27, 32, 33].

2.3 DESI

For each data point, five straight lines of rhodamine 6 G dye were drawn on a clean glass surface using a red permanent marker. The sample was scanned through the charged solvent spray of pure MeOH at a speed of 150 μm/s using the translational stage. The lines were drawn 1 cm apart to ensure that there was no carryover during the scans. For each line that was scanned, a peak was observed in the extracted ion chronogram for rhodamine 6 G (m/z 443.33). The area of each peak was calculated and used to generate the average response along with the standard deviation for the data point.

2.4 SDC

Similar to DESI, a red permanent marker was used to draw a single line of rhodamine 6 G along the straight edge of a clean glass surface. A clean silica TLC plate was positioned immediately in front of and blocking the standard atmospheric pressure inlet capillary. The sample edge was placed on the translational stage so that a 1 mm gap existed between the sample surface and the collection surface. The sample was scanned horizontally at 150 μm/s using the translational stage and a spray of pure MeOH. For each data point, material desorbed from the sample surface was collected in a single spot on the TLC plate for 133 s (2 cm of translation). Collection was typically observed to occur in a thin line just above the horizon of the sample surface [27]. The visible lines were ca 1 mm wide and 1 cm long. After changing the value of the parameter of interest the collection surface was moved so that sample collection occurred at a new spot approximately 5 mm from the previous spot. After sample collection was complete, 3 mm × 20 mm squares containing each collected spots were cut from the silica gel TLC plates and the silica was carefully scraped from the plastic substrate into labeled centrifuge tubes. One mL of methanol (MeOH) was added to each tube and the tubes were agitated using a vortex mixer (Genie 2, Diagger, Vernon Hills, IL) for 30 min to ensure complete extraction of rhodamine 6 G. This step was followed by centrifugation of each tube to remove any suspended particles. The final solutions were analyzed by fluorescence spectroscopy. The average fluorescence measurements of each data set were then normalized for comparison between the data for DESI, SDC, and RESI.

2.5 RESI

A stock solution of rhodamine 6 G in pure MeOH was serially diluted to a final concentration of 11.1 nM to be used as the working solution during RESI experiments. The solution was nebulized by the microspray source during the analysis and directed at the sample surface. The surface was scanned horizontally at 150 μm/s while the parameter of interest was changed at 30-s intervals. The average intensity of the rhodamine 6 G peak was determined for each parameter value by averaging over the 30 s in the chronogram corresponding to that setting. This procedure was repeated three times for each parameter that was optimized. To allow comparison between trends observed for the mass spectral and fluorescence data, the average intensities of the DESI, SDC and RESI data sets were normalized to their highest values.

3 Results and Discussion

Several authors have reported comparative studies between multiple surfaces that have contributed to our current understanding of how surface properties affect the analytical performance of DESI [2–4, 6, 15, 17, 20, 34, 35]. These surfaces include PMMA, PTFE, glass, paper, stainless steel, porous silicon, and TLC plates. “Rough” surfaces tend to show better stability and reproducibility than smooth ones, probably by preventing the washing and coffee-ring effects observed with the latter[2, 6, 15]. The coffee-ring effect [6], sometimes referred to as the Marangoni effect [36], refers to the heterogeneous sample deposition often associated with samples prepared by allowing aliquots to air-dry on a surface [37]. Capillary action within the drying droplets results in preferential analyte deposition along the periphery of the droplet. The DESI surface should also be nonconductive as charge losses can occur to the surface [2, 34, which is exacerbated by solvent spreading [34]. Thus, hydrophobic surfaces perform better with the polar solvents typically used during DESI. The ideal interrogation surface appears to rely heavily upon the nature of the analyte [2, 15, 34]. We investigated PTFE, glass, and plastic mesh for instrumental response, signal stability and reproducibility for rhodamine 6 G [14], an analyte typically used for DESI optimization. While PTFE showed the highest instrumental response, glass displayed the most stable signal for the analyte. Hence, glass was chosen as the sample surface for DESI, SDC and RESI experiments.

3.1 Applied Potential

Although a high potential is not required for obtaining a DESI signal [2, 11, 13, 32, 34], it has been demonstrated that instrumental response increases with increasing spray voltage up to 8 kV [2, 5, 7, 11]. Increasing the voltage produces smaller droplets with larger velocities [5], leading to more efficient ionization. However, oxidative species in the electrospray also increase with increasing voltage, leading to the presence of oxidized products present in the spectra [38–40] and increased spectral complexity. Figure 2 shows optimization results for the applied potential between 0 to 8 kV, the instrumental limit, for DESI, SDC, and RESI.

Figure 2
figure 2

Results from the optimization of the applied potential for SDC, RESI, and DESI

Full size image

The SDC curve in Figure 2 remains close to its optimum value across the range of voltages tested, indicating that desorption does not significantly depend on the high voltage applied during DESI. This result is consistent with the droplet-pickup model, since desorption is thought to occur due to analyte solubility in the spray solvent followed by collisional ejection from the film, processes that we do not expect to depend significantly on the applied potential. These results are also in agreement with previous studies where it was shown that an electrostatic contribution is not necessary to replicate the droplet impact characteristics observed during DESI [5] using computational fluid dynamics [23] and that a high voltage is not necessary for efficient desorption during desorption sonic spray ionization [13].

Unlike the SDC curve, the RESI and DESI curves both increase with increasing applied potential. This suggests that the applied potential dependence of DESI is based mostly on ionization or ion transfer processes and not desorption.

3.2 Gas Pressure

In DESI, instrumental response usually increases with increasing gas pressure until an optimum value is achieved, after which instrumental response plateaus or decreases [2, 5, 7, 9]. A similar result is presented in Figure 3 where gas pressure was optimized between 100 psi to 180 psi at the regulator, corresponding to nebulizing gas volumetric flow rates at the emitter of 0.96 L/min at 100 psi up to 1.72 L/min at 180 psi. The desorption and ionization contributions to the dependency of DESI on the nebulizing gas flow rate were studied by SDC and RESI respectively.

Figure 3
figure 3

Results from the optimization of the nebulizing gas pressure for SDC, RESI, and DESI

Full size image

As indicated by the results shown for SDC in Figure 3, desorption efficiency initially improves up to 140 psi before decreasing again at higher pressures, following the same trend that is observed for DESI and optimizing at the same pressure. It was previously shown that with increasing gas pressure droplet velocities increase and a certain threshold velocity is required for the efficient production of secondary droplets after surface collision [5, 23, 24]. At gas flow rates lower and higher than the optimum value, sample re-deposition [10] was also observed, which occurs when material desorbed from the spray, deposits on the surface past the point of impact. This is likely caused by decreasing impacting droplet velocity for low gas flow rates and a larger force on the secondary droplets by the gas for high flow rates. It is further possible to explain the observed ion signal dependency on gas pressures at the hand of increasing size of the impact zone or wetted spot. With higher gas pressures, when the solvent spray becomes more focused [12] and evaporation increases, the spot size is somewhat reduced. At very high gas pressures it is possible that the system becomes too dry, the wetted spot on the surface disappears and desorption efficiency is reduced. By probing the charge distribution on a nonconductive surface without the presence of an inlet providing suction, Cooks et al. demonstrated that a higher gas pressure focuses the charges arriving at the surface [12], a result that agrees with the smaller desorption spot sizes observed visually with higher gas pressures [5].

The dependence of ionization efficiency on nebulizing gas pressure settings was studied by RESI. At the lowest gas flow rate (≤100 psi), a high signal intensity is observed because there is not enough driving force exerted by the nebulizing gas to effectively direct droplets produced by the microsprayer toward the sample surface. These droplets and ions fly directly toward the atmospheric pressure inlet without impacting the surface, which leads to the initially high ion signal with RESI. On the other hand, in DESI, when no analyte ions are present in the spray solvent, no (or low) signal intensities are observed in this pressure regime. Ignoring this initial anomalously high response with RESI, under conditions where the primary droplets do collide with the surface, the signal continuously increases with increasing gas pressure. If the spray is sufficiently wet for the formation of the liquid film, secondary droplet production efficiency increases with increasing gas pressure. It was previously shown that droplet diameters decrease with increasing gas pressure settings, confirming that, as is the case for ESI, smaller droplet sizes lead to more efficient ionization with RESI and thus with DESI [5].

While both desorption and ionization are affected by changes in gas flow rates, the close correlation between the SDC and DESI trend lines indicates that desorption effects dominate during optimization of the gas pressure. The mild increase in ionization efficiency observed at higher gas pressure settings does not compensate for the loss in desorption efficiency caused by sample re-deposition and the reduced size of the wetted spot at high gas pressure settings and the DESI signal intensity decreases, even as ionization efficiency improves.

3.3 Sprayer-to-Impact Zone Distance

Increasing the distance between the sprayer and the impact zone was shown to decrease the velocity of droplets impacting the surface as well as the velocity of droplets leaving the surface [5]. In the present study, the distance between the impact zone and mass inlet or collection surface was kept constant while the distance between the sprayer and the impact zone was varied from 1 to 8 mm by moving the sprayer diagonally away from the surface. Figure 4 shows the results for the optimization of the sprayer-to-impact zone distance for DESI, SDC, and RESI.

Figure 4
figure 4

Results from the optimization of the sprayer-to-impact zone distance for SDC, RESI, and DESI

Full size image

The dependence of the signal intensity observed for DESI on the distance between the sprayer and the impact zone displays an interesting morphology. Instrumental response initially increased with increasing distance up to 3 mm, followed by a ridge at 4 mm, after which signal intensity decreases. This shape of the DESI dependence on sprayer-to-impact zone distance curve can be explained by the combined contributions of desorption and ionization efficiencies.

Similar to the DESI curve, the SDC curve increases to a maximum at 3 mm sprayer-to-impact zone distance, after which the curve continuously decreases. At short distances, desorption efficiency may suffer as the nebulizing gas and droplet velocities are high, which results in sheeting of the solvent along the surface instead of secondary droplet formation. As the droplet velocities decrease with increasing distance, desorption efficiency increases until an optimum value is achieved. After this value, desorption efficiency decreases along with droplet size and velocity. At this point, droplets may contain too little kinetic energy to effectively remove material from the surface and some removed material may begin re-depositing past the impact zone instead of making it to the inlet. As shown by these results, DESI response is initially low due to low desorption efficiency. As desorption efficiency increases, so does DESI response, although DESI response increases more steeply due to the concurrent increase in ionization efficiency as indicated by the RESI results.

The RESI curve continuously increases with increasing sprayer-to-impact zone distance. Increasing the distance between the sprayer and the impact zone results in smaller droplets impacting the surface, which has been shown to increase the efficiency by which ions are transferred from the surface to the mass spectrometer inlet [41]. This transfer efficiency appears to increase over the entire range of sprayer-to-impact zone distances that was investigated in the current study. As desorption efficiency decreases after 3 mm, DESI response continues to remain high due to the increasing ionization and ion transfer efficiencies. After 4 mm, however, the decreasing desorption efficiency overcomes the benefit of increasing ionization efficiency as there is little material desorbed and therefore nothing to ionize.

3.4 Impact Zone-to-Inlet Distance

For a fixed impact-zone to inlet distance, progeny droplets initially remain close to the surface after formation at the impact zone. Their velocity decreases with increasing distance from the impact zone due to aerodynamic drag and eventually increase in height due to dispersion [5]. We also observed collection to occur in a narrow line level with the horizon of the sample surface during SDC [27], and Kertesz et al. have shown that the ‘edge sampling’ geometry gives the best response for the same reason [33]. Figure 5 shows the results for the optimization of the impact zone-to-inlet distance for DESI, SDC, and RESI.

Figure 5
figure 5

Results from the optimization of the impact zone-to-inlet distance for SDC, RESI, and DESI

Full size image

While studying desorption efficiency using SDC, we could not optimize for distances shorter than 2 mm since it was necessary to leave a small space between the sample surface and the collector surface. The SDC response was not expected to change substantially with an increase in distance between impact zone and inlet since this region reflects what happens to droplets and ions after desorption. From the SDC curve, it can be seen that desorption falls less than 20 % from its optimal value at the largest distance. The observed decrease is likely due to dispersion of gas and desorbed material after leaving the sample surface, leading to a slight increase in the size of the collection zone. In DESI, this dispersion may also reduce the collection efficiency of the produced progeny droplets and ions by the atmospheric pressure interface.

The RESI curve was observed to increase with increasing distance until a maximum was achieved between 3 and 4 mm, after which the curve decreases. The smaller, faster moving progeny droplets formed during droplet collision with the surface [23] tend to remain close to the surface [5]. At very short distances, these droplets might not achieve the height necessary to avoid hitting the bottom rim of the inlet capillary or re-depositing onto the sample surface. The dispersion of these small, fast-moving droplets increases with increasing distance from the impact zone [5], so as the impact zone-to-inlet distance increases, the sampling of progeny droplets and ions by the inlet also increases as they gain height from the surface. In addition longer flight times will provide more time for evaporation and ion production from the progeny droplets. After the optimum distance, the decrease in the RESI curve is likely due to a decreasing sampling efficiency by the atmospheric pressure interface as the dispersion of the plume continues to increase, decreasing the fraction of the reflected plume that is sampled by the inlet.

The DESI curve follows the same trend as the RESI curve, although the DESI curve is lower initially. For both techniques, an increasing portion of the charges bypass the surface and head directly toward the inlet due to the increased electric field at very short distances [16]. For DESI, this results in decreased desorption efficiency and low signal while for RESI the response is higher due to the presence of analyte ions already present in the spray solution. The DESI dependence on the impact zone-to-inlet distance appears to depend mostly on the ion transport and dispersion processes occurring post-desorption.

4 Conclusion

We have found that the optimum conditions for desorption do not always coincide with those for ionization during the concerted DESI process. Often, the optimum conditions for DESI are a compromise between those conditions that lead to efficient desorption and ionization respectively. In fact, the response trends for desorption and ionization were different within each parameter investigated, which demonstrates the inherent complexity of the optimization process for DESI.

In the present study, a single analyte and surface combination was used to optimize several parameters. Different analytes may optimize differently based on their individual desorption and ionization requirements. However, the general optimization trends shown here for each parameter agree with published results.