Nested-channel for on-demand alternation between electrospray ionization regimes

On-demand electrospray ionization from different liquid channels in the same emitter was realized using filamented capillary and gas phase charge supply. The solution sub-channel was formed when back-filling solution to the emitter tip by capillary action along the filament. Gas phase charge carriers were used to trigger electrospray ionization from the solution meniscus at the tip. The meniscus at the tip opening may be fully filled or partially empty to generate electrospray ionization in main-channel regime and sub-channel regime, respectively. For emitters with 4 μm tip opening, the two nested electrospray (nested-ESI) channels accommodated ESI flow rates ranging from 50 pL min−1 to 150 nL min−1. The platform enabled on-demand regime alternations within one sample run, in which the sub-channel regime generated smaller charged droplets. Ionization efficiencies for saccharides, glycopeptide, and proteins were enhanced in the sub-channel regime. Non-specific salt adducts were reduced and identified by regime alternation. Surprisingly, the sub-channel regime produced more uniform responses for a peptide mixture whose relative ionization efficiencies were insensitive to ESI conditions in previous picoelectrospray study. The nested channels also allowed effective washing of emitter tip for multiple sampling and analysis operations.


The nested-ESI setup and MS instruments
. A prototype of the setup. Figure S2. The position of plasma, pusher electrode, and capillary emitter.       Table S1. Types of molecular ions generated by nested-channel electrospray ionization.

Reagents, samples and types of generated ions
6. Sub-channel and the "gas-phase electrode" Figure S9. Pulsed electrospray using metal-coated emitter with sub-channel and DC power supply. Figure S10. The effect of increasing DC voltage for coated externally emitters.    Figure S14. The EIC data points during typical transitions from main-channel regime to the sub-channel regime. Figure S15. Comparing regime alternation with wire-in nanoESI and picoESI.   14. Application II: analyzing and washing online through sub-channel Figure S22. Analysis-while-washing by using the two channels of nested-ESI. Figure S23. TIC, EICs, mass spectra and tandem mass spectra obtained using analysis while washing by nested-ESI. Figure S24. Schematic of an experiment in which blank solvent was added from the front end as an alternative way to wash the emitter tip. Figure S25. TIC, EICs and mass spectra of control experiment by using capillary with filament. 4

The nested-ESI setup and MS instruments
The setup is illustrated in Figure S1 and S2. Plasma ion source, pusher electrode, and filament containing emitter tip are three core components for nested-ESI.
The plasma ions were generated by using a piezoelectric transformer (53×7.5×2.6 mm, INC model SMSTF68P10S9, Steiner & Martins), with the advantages of compact size, low power consumption and safety, and operating in atmospheric air, as shown in the inner photo of Figure S1. The piezoelectric transformer was operated by supplying an input voltage (5-25 V, Powertron Model 500A; Industrial Test Equipment Co. Inc., Port Washington, NY, USA) triggered by a sine waveform. Plasma discharge was readily generated at the tip of the output electrode under ambient conditions. The faint plasma may be observed by naked eye. A pusher electrode (44×44 mm) was placed behind the capillary emitter to create an auxiliary electric field, which pushed positive or negative plasma ions to the capillary emitter.
Typical alignment parameters for plasma ion source, pusher electrode and capillary emitter were listed in Figure S2.  An LTQ-XL linear ion trap and an LTQ velos Orbitrap mass spectrometer (Thermo Scientific, San Jose, CA) were used in the experiments. Unless mentioned otherwise, the LTQ-XL was operated with following parameters: inlet temperature 125 ºC, inlet capillary voltage 9 V, tube lens voltage 100 V. LTQ velos Orbitrap was operated under the conditions: inlet temperature 125 ºC, S-lens RF level 55%. All mass spectra were recorded as peak profiles with 1 microscan and 10 ms maximum injection time. Woburn, MA). The emitter tips were checked by bright-field microscopy (Olympus IX73), as well as measured by a field emission scanning electron microscopy (TESCAN LYRA3) after sputter coating a 20 nm layer of Pd/Au. A micro butane torch was used to seal the proximal end of emitters when needed. 6 Figure S3 shows the input for the piezoelectric (PT) discharge plasma (15 V at 67.3 kHz). The actual resonance frequency of the PT slightly alters from crystal to crystal.

The plasma charge supply
So, the effects of driving frequency for the discharge status and input current need to be characterized, as shown in Figure S4. The input current at the maximum was 64 mA, corresponding to a brightest corona discharge at the tail end of the output electrode ( Figure S4a). So, 67.3 kHz was finally chosen as the optimized condition. Under this condition, the power consumption was ~960 mW.

Charge transport to the meniscus
Several experiments were designed to investigate how charges were transported from the plasma source to the ESI meniscus.
Firstly, capillary emitter with a fire-sealed proximal end was tested. As shown in Figure S6, electrospray from the sealed capillary emitter had no significant difference with that from the emitter with open proximal end. This suggests that the channels inside the capillary emitter are not needed for charge/ion transportation.
Secondly, two experiments further designed to verify the transportation from the surface and/or the space out of capillary, as shown in Figure S7. Cu tape wrapped around the middle section of the capillary was not able to shut down the electrospray by grounding. However, a Cu film (100×100 mm) placed perpendicularly to the capillary was able to effectively shut down the electrospray. These results suggest that the exterior space is critical for the charge transport. Fig. S7b also confirms that the 9 inner channels of the emitters are not responsible for charge transport.
Thirdly, grounding the emitter tip was found to effectively shut down the ESI ( Figure S8), suggesting charges build up at the tip is critical for the ESI. Based all these results, it is concluded that the charges were transported via the exterior space surrounding capillary to the tip of the emitter, which leads to electrospray.

Rapid alternation of ESI regimes
Video S1 shows the rapid alternation of ESI regimes by turning the pusher voltage on and off. The on and off was controlled by setting voltage to 0 or 1.5 kV using the user software of the mass spectrometer. When the voltage was set to 0, the spray was stopped, and the solution started to accumulate in the main channel (7-14 s). Right after turning the pusher voltage back to 1.5 kV, a bright plume of ESI droplets was generated, until the solution in the main-channel was consumed (14-17 s). After that the ESI entered sub-channel mode, which is characterized by an "empty" emitter tip, fainter ESI plume, and higher relative ion intensity for saccharide (0-7 s, 17-19 s).

Reagents and Samples and types of generated ions
Methanol (HPLC grade), water (HPLC grade), ammonium acetate, sodium chloride, Samples without special note were prepared to the target concentrations by serial dilution using methanol and water (v:v, 1:1).

Sub-channel and the "gas-phase electrode"
The gas-phase electrode (plasma ions + pusher electrode) was more effective than conventional voltage supplies in charging small meniscus. Here, applying DC voltage to metal-coated emitter tip was tried. As an unsuccessful attempt in triggering ESI from the sub-channel, it serves as a comparison to illustrate the effective charging mechanism of the gas-phase electrode.

Flow Rate Measurements
In this work, the flow rates of the ESI were determined using one of the following two methods.
Measurement method #1 is based on gravimetric analysis of the capillary emitter before and after spraying for a period. Given the spray time, the weight lost, and the density of the solution, the flow rate can be determined. The weight measurements were carried out using a Mettler Toledo MX5 microbalance (Mettler-Toledo, Columbus, OH; repeatability reported by manufacturer is ±0.8-0.9 µg). In our work, the total weight of Lengths in the video were calculated using a known object, 2.14 mm/228 pixels. The volume was calculated by measuring the height (h) and radius (r = kh for a fixed cone shape) of the cone shaped solution. This calculated volume by =

Stability of nested-ESI
The robustness of nested-ESI was exemplified in multiple experiment that spray more than 3 hours. An initial flow rate was measured by regime alternation and then the setup was left untouched. In one experiment, nested-ESI was set up at 0.4 nL min -1 and then sprayed continuously for almost 5 hours, shown in Figure S13. The initial increase of TIC and EICs after 5 minutes was likely due to the fluctuation of flow rate of the system. Despite signal fluctuations, the TIC and EICs indicate that there was no separation effect inside the channels caused by the external plasma and electric fields.
Note that at 185 min, the flow rate decreased because a particle accidentally entered the emitter. Interestingly, this did not lead to a complete clogging of the emitter. Instead, it was slowly washed down the sub-channel and then ejected by ESI. After that, the flow and ion signals quickly recovered. (v:v, 1:1). At 185 min, the flow rate decreased because a particle accidentally entered the emitter. Interestingly, this did not lead to a complete clogging of the emitter.

Application I: desalting and decreasing non-specifically adducting by alternating main-channel and sub-channel
Salt adducts is one type of non-specific adducts that complicates data interpretation.
The presence of salt also causes ionization suppression. In ESI, a protein in NaClcontaining solution may form adducts ions 3, 4 : [M+(z-n+m)H+nNa+mCl] z+ in which M represents the neutral analyte, z is the charge state, n and m are integers.
Electrospray using smaller emitters tips were found to reduce salt adduct ion signal and non-specific adduct. 5,6 Here we demonstrate regime alternation as an effective method to reduce and identify non-specific adduct. As shown in the figure below, the TIC shows the different intensities between main-channel regime (i) and sub-channel regime (ii).
In main-channel regime, salt clusters ions (m/z 315-667 with 58 gaps) were generated and cyt c was suppressed. A non-specific adduct with HCl (1772) was produced with significant intensity. The weak HCl adduct was confirmed by the fact it would be lost with minimum energy input during ion isolation by the ion trap. In the sub-channel mode, most of the salt cluster ions disappeared. Cyt c ions became the dominant peak with a 10-times higher absolute intensity. All the produced ions correspond to the folded conformation. The HCl non-specific adduct was not formed.
Despite the complete removal of salt cluster ions for this 5 mM NaCl solution, desalting by sub-channel regime did not work well for more concentrated (100 mM) salty sample solutions. Calculation suggests that, for a 5 mM solution, a droplet containing less than 2 NaCl molecules need to be smaller than 10 nm in diameter.
Removing salt cluster from more concentrated solution would require even smaller initial droplets, which probably exceed the limit of the current setup. This limited desalting capability also confirms that there was no obvious electrophoretic effects 5,7,8 during the regime alternation.

Application II: analyzing and washing online through sub-channel
The operation process was illustrated in Figure S22 below. More sample solutions may be analyzed by repeating step 1-3. The solvent in the sub-channel typically last for 4-6 analyses, before blank solvent can be added by step 0. Figure S22. Analysis-while-washing by using the two channels of nested-ESI.

25
10 µM caffeine aqueous solution (m/z=195) was chosen as another example, as shown in Figure S23. After solvent filled in the subchannel of capillary, the separate operation 1 and 2 were to pause the MS recording to dip the tip of capillary into sample solution about 9-12 s, and then continued detecting. the process of analysis with washing took about less than 0.1 min, the backfilled solvent can last 2.4-3.0 min during each operation. In one operation, the TIC and EICs shows the analyte's signals (m/z=195) were shown firstly, and then gradually decreased because of rinsing; the solvent was continued to flush the tip (as seen the EIC of solvent in the mass range of 233-235). That the analytes cannot be observed in full scan and tandem mass (the relative intensity decreased to at least 10%) is used to prove cleanliness. Mass signals of caffeine and solvent illustrate the analysis is valid and washing is clean without residue. Figure S23. (a) TIC, EICs, (b-e) mass spectra and tandem mass spectra obtained using analysis while washing by nested-ESI. Sample solution is an aqueous 10 µM caffeine.
As an alternative washing method, aspirating blank solvent from the front end of the emitter tip was tested ( Figure S24). 10 µM angiotensin II in aqueous solution was used in this control experiment. As shown in Figure S25, full scans and EIC of angiotensin II in operation 1 indicate the solution at the tip was gradually consumed without any supply from the sub-channel. Then the capillary was removed and dipped into blank solvent for 10 s. The following analysis showed angiotensin II contamination.
Even after three such washing operations and spraying for more than 10 minutes, 26 angiotensin II contamination was still evident. In comparison, the previous washing method which generally takes 0.1-0.5 minutes, was significantly more efficient. Figure S24. Schematic of an experiment in which blank solvent was added from the front end as an alternative way to wash the emitter tip. Figure S25. Comparison experiment in which the emitter tip was washed by dipping the emitter tip into bulk solution. The operations were shown in Figure S24. Sample was a 10 µM angiotensin II solution. Carryover contamination was not removed.