Introduction

Electrospray begins when the Coulomb-repulsive electrostatic force becomes greater than the surface tension of a solution. Recently, various techniques have been developed that applied the vertical or pulsed electric field to generate electrospray using dielectric materials as a liquid supporting media. Takeda et al. [1] investigated the effect of electric field on the movement of water droplets put on a super-hydrophobic film. A vertical electric field induced a water droplet on a super-hydrophobic film to jump up from the surface without splitting itself. Grimm and Beauchamp studied field-induced droplet ionization mass spectrometry [2]. A levitated neutral droplet of an electrolyte solution elongated along the axis of a strong electric field (~ 2.2 × 106 V m−1), ejecting opposing jets of positively and negatively charged progeny droplets, i.e., symmetrical Rayleigh discharge process. Schilling, Janasek, and Franzke developed inductive desorption electrospray ionization using a non-conducting piezo ceramic plate to induce an electric field to generate electrospray. Similar mass spectra were obtained as compared with conventional electrospray for reserpine and myoglobin [3]. Furthermore, they developed an electrospray interface which induced an electric field by dielectric polarization through a non-conductive barrier using a square-wave high voltage [4]. This technique made mass spectrometric measurements possible in the positive as well as in the negative mode without changing the polarity of the potential applied. Huang et al. applied inductive electrospray ionization to a desorption electrospray ionization (DESI) source [5]. This method made electrical contact to ESI emitters by simply inserting a wire conductor into the metal-coated capillary. The pulsed positive voltage was applied to a metal tube (250 μm i.d.) covering an inner silica capillary (50 μm i.d.) carrying a sample solution. The pulsed induction produced high electric fields in the DESI source that resulted in burst of charged droplets. Simultaneous production of ions of both positive and negative polarity from a single spray emitter was observed without changing the polarity of the applied potential. The precise synchronization with DESI interface was possible because the inductive pulse DC high voltage had the necessary short on/off response time of ca. 1 ms. They further developed induced nanoelectrospray ionization for matrix-tolerant and high-throughput mass spectrometry [6]. Eight nanoESI emitters loaded with different samples were mounted to a fixed pulley driven by a moving belt. The electrospray potential (2–4 kV) was applied to a single electrode that was approximately placed so that each of the spray emitter approached to within 2 mm of its turn. The applied potential was pulsed repeatedly in the positive mode at a frequency of ~ 50 Hz. The great merits of this method were its high-throughput, ultra-high sensitivity, almost simultaneous generation of ions of both polarity, and compatibility with raw serum, whole urine, and concentrated salt solution. Qiao et al. studied electrostatic-spray ionization (ESTASI) for the samples deposited in or on an insulating substrate [7]. A metallic electrode was placed close to but not in contact with the sample. Upon application of a high voltage to the electrode, an electrostatic charging of the sample occurred leading to a bipolar spray pulse. When the voltage was positive, the bipolar spray pulse consisted first of cations and then anions. This method has been applied to a wide range of geometries to emit ions from samples in a silica capillary, in a disposable pipet tip, in a polymer microchannel, and from fractions of capillary electrophoresis deposited on a polymer plate. Because a high voltage was not directly applied to the sample solution, they predicted that no electrode reaction could occur. Namely, ionization is due to the capacitive coupling effect. Further, ESTASI was applied to on-tip spyhole mass spectrometry for droplet-based microfluidics [8]. The ionization of the water-in-oil droplets generated with a volume of 3 nL was realized by the application of HV square pulses (8 kV, 10 Hz) with an electrode placed under the microchip and aligned with the spyhole and inlet of the mass spectrometer. They suggested that absence of a direct contact with the sample prevented any electrochemical reactions.

In the current work, a tip-sealed glass capillary inserted with a fine acupuncture needle (termed as glass-PESI) was tested as an electrospray emitter. After loading a sample solution on the surface of the glass probe tip, a positive square-wave HV was applied to the inner metal electrode. Glass-PESI mass spectra were measured as a function of the HV pulse width (i.e., HV duration time). The positive ions started to be detected at ~ 5 ms. Ion intensities increased gradually with the pulse width and reached a plateau in a few seconds. Glass-PESI mass spectra of cytochrome c showed a marked pulse-width dependence indicating the change of pH of the sample solution. This suggests the occurrence of electrochemical reactions at the interface between the glass surface and the sample solution. When a negative HV pulse was applied to the needle, no electrospray was generated.

Experimental

Mass Spectrometer

Measurements were made using an orthogonal-acceleration time-of-flight mass spectrometer (oa-TOFMS) (JEOL, Akishima, Tokyo, Japan). The temperature of the ion sampling orifice was kept at 100 °C. The ions generated by glass-PESI were sampled through a 0.4-mm-diameter ion sampling orifice into the vacuum. The mass spectra were acquired using an ADC/continuous averager ion detection system. Mass Center V1.1.5 software (JEOL) was used for the data processing and the signal integration. The default desolvation chamber was removed and the glass-PESI probe system was installed.

Experimental System

In probe electrospray ionization (PESI) developed in our laboratory [9, 10], a small amount of sample solution is loaded on a metal needle by touching the liquid surface with the needle tip, and electrospray is generated by the application of a HV to the needle. The experimental procedure in glass-PESI was exactly the same except that the glass-covered metal needle was used. The experimental system is shown in Figure 1. The tip of the borosilicate glass capillary (PM Micropipettes for PMM Ultra-thin, PINU06-20FT, Primetech, Tokyo, Japan) manufactured for artificial fertilization was heated and sealed by a small flame (inset, bottom left). The absence of a pinhole in the glass tip was confirmed by dipping the glass needle in the red ink. If there was a pinhole, red ink was sucked into the capillary by the capillary effect. After sealing, an acupuncture needle (J type No.02, body diameter 0.12 mm, tip diameter 700 nm, Seirin, Shizuoka, Japan) was inserted in the glass capillary to the terminus of the capillary. The needle tip was in direct contact with the inner surface of the glass capillary. The needle was fixed by epoxy resin at the opposite terminus of the glass capillary. The thickness of the glass wall at around the needle tip is ~ 20 μm.

Figure 1
figure 1

Schematic diagram of the glass-PESI system. Inset (left): the sealed glass capillary inserted with an acupuncture needle. The glass probe is dipped into the sample solution with a dipping depth of ~ 1.5 mm. After reaching highest position, a HV (4.5 kV) was applied to the acupuncture needle with the delay time of 150 ms. Inset (right): the pulse shape measured using an oscilloscope; pulse width: 400 μs; pulse height: 2.1 kV. Rise and fall time: 3 μs

The glass-PESI probe was positioned vertically in front of the inlet of the mass spectrometer. The optimum position of the needle tip was 3 mm to the side and 2 mm above the apex of the ion-sampling cone of the mass spectrometer. The needle was driven down along the vertical axis by a linear motor-actuated system (SCN-5-010-050-S03, Dyadic Systems Co., Ltd.) with a speed of 400 mm s−1 and was dipped into the sample solution in the reservoir to a depth of ~ 1.5 mm for 50 ms [11]. The level of the liquid meniscus in the reservoir was adjusted manually by using an x-y-z manipulator (see Figure S1). The amount of liquid sample (aqueous or H2O/CH3OH(1/1, v/v)) captured on the glass capillary tip with the dipping depth of 1.5 mm was measured to be ~ 390 ng using a microbalance.

The timing of the application of a HV and the needle motion was synchronized using a homemade control system (Figure 1). After the needle reached the highest position, a HV was applied to the metal needle with the delay time of 150 ms. The HV pulse width (i.e., HV duration time) was changed from μs to seconds. The inset in Figure 1 (bottom, right) shows the waveform of the 400 μs HV pulse (2.1 kV) applied to the needle measured using an oscilloscope. The rise and fall time of the voltage was about 3 μs.

Video of the Taylor cone of electrospray generated at the tip of the glass capillary was measured by a CCD camera (Keyence, High-speed camera unit, VW-600C lens; Keyence, Osaka, Japan). The temporal profile of electrospray current was measured using an oscilloscope (DSO-X 2014, Agilent Technologies).

Chemicals

Water was purified and deionized using a Simplicity UV system (Millipore, Bedford, MA). High-performance liquid chromatography grade methanol, NaCl (UGR), acetic acid (UGR), and ammonium acetate (UGR) were purchased from Kanto Chemicals (Tokyo, Japan). Cytochrome c and gramicidin S were purchased from Sigma-Aldrich. Yogurt (fermented milk, Meiji Probio, 1073R-1, sodium: 48 mg, proteins: 3.6 g, calcium: 129 mg, lipids: 0.67 g, sugars: 13.3 g, carbohydrates: 13.9 g in 112 mL) was purchased from a local supermarket. UltraMark1621 for the mass calibration (10−5 M sodium dodecyl sulfate, 10−5 sodium taurocholate, 0.001% UltraMark1621 in acetonitrile/methanol/acetic acid (50/25/1)) was purchased from ThermoFischer Scientific.

Results and Discussion

Electrospray Generated from the Tip of the Glass Probe

The purpose of the present work is to examine whether the intermittent inductive electrospray is generated when a strong electric field was applied to the liquid film captured on the glass-PESI probe. To our surprise, continuous cone-jet mode electrospray was observed for a few seconds after the application of a HV (4.5 kV) to the metal electrode until all the liquid solution was depleted from the glass probe. Figure S2 shows the video of electrospray for 10−5 M cytochrome c and 1% acetic acid in H2O/CH3OH (1/1) generated from the tip of the glass probe tip under illumination taken by a CCD camera. The reflected light from the red spot marked at the apex of the Taylor cone was monitored as shown in the inset. The unit of the time (y axis) is μs. The time interval of the reflected light was ~ 200 μs, i.e., the frequency of cone-jet mode electrospray was about 5 kHz. It is worth noting that electrospray was generated only at the tip but not from the side wall of the glass probe. It was apparent that the electric field was highest at the tip of the glass probe although the metal needle was recessed from the probe tip by ~ 300 μm. This is reasonable because the electrolyte solution behaved as an electric conductor and the highest electric field was generated at the apex. The electrospray lasted for about 3–5 s for the sample amount of ~ 400 nL captured on the glass probe tip. That is, the flow rate of the sample solution was about 5 μL min−1 under present experimental conditions. This high flow rate may be due to the rather large tip diameter (~ 100 μm) of the glass probe.

Contrary to the occurrence of normal electrospray in the positive mode, no electrospray was generated when a high negative HV was applied to the metal needle.

Pulse Width Dependence on Mass Spectra

In this experiment, the sample solution was not in direct contact with the metal electrode. Under such experimental conditions, two mechanisms may be conceivable for the charging of the liquid sample: (i) intermittent inductive charging caused by the application of pulsed electric field and (ii) supply of excess charges by the occurrence of electrochemical reactions. To clarify the working mechanism for the sample charging, glass-PESI mass spectra were measured as a function of a HV pulse width from μs to 5 s. Figure 2 shows mass spectra for 10−5 M cytochrome c and 1% acetic acid (AcOH) in H2O/CH3OH (1/1) measured with the pulse widths of 5 ms, 0.2 s, 0.5 s, and 5 s. No ions were detected with the pulse width shorter than a few ms. The major ion [M + 9H]9+ at m/z 1360 started to be detected with the pulse width of ~ 5 ms and its intensity increased with increase in the pulse width. As shown in Figure 2, the relative intensities of ions with higher charge states [M + nH]n+ (n > 9) became stronger than those with the shorter pulse widths with the pulse width of 5 s.

Figure 2
figure 2

Mass spectra of 10−5 M cytochrome c and 1% acetic acid in H2O/CH3OH (1/1) with the HV duration time of (a) 5 ms, (b) 0.2 s, (c) 0.5 s, and (d) 5 s. Mass spectrum is the time-averaged one for each duration time

Figure 3 shows the intensities of major ions measured as a function of the HV pulse width for 10−5 M cytochrome c with the addition of 0, 0.1, and 1% acetic acid (AcOH) and 10 mM ammonium acetate (AcONH4) in H2O/CH3OH (1/1). Ion intensities for all the samples showed gradual increase until the sample solutions were all depleted at ~ 5 s.

Figure 3
figure 3

Ion intensities of cytochrome c as a function of the HV duration time (ms) for 0, 0.1 and 1% acetic acid, and 10 mM ammonium acetate. The number in the bracket stands for n for the multiply charged ions of [M + nH]n+ (M: cytochrome c). The most abundant ions were [M + 9H]9+ for 0.1 and 1% acetic acid solution, whereas [M + 7H]7+ for 0% acetic acid and 10 mM ammonium acetate solutions. Ion intensities correspond to the time-averaged ones for the samples

The increase in charge states with the pulse width in Figure 2 suggests the decrease of pH of the solution with time. Figure 4 shows the snap shots of mass spectra measured with the HV pulse width of 5 s. The onset of the total ion chromatogram (TIC) in the inset corresponded to the application of a HV. Electrospray lasted for about 4 s and the ion signal decreased to a background level 4 s after the HV application. In TIC, two peaks appeared. Figure 4a, b shows the respective mass spectra measured at P1 and P2 in TIC. The increase in the charge state with P1 → P2 clearly indicated that the denaturation (i.e., unfolding) of cytochrome c proceeded with time. This was likely to be due to the decrease of pH of the solution. These results suggest the occurrence of electrochemical reactions that supply excess charges to the sample solution.

Figure 4
figure 4

Snap-shots of the mass spectra for 10−5 M cytochrome c and 1% acetic acid in H2O/CH3OH (1/1) at P1 and P2 in TIC shown in the inset. HV: 4.5 kV

Figure 5a, b shows the respective mass spectra measured at P1 and P2 in TIC shown in the inset for the sample of 10−5 M cytochrome c and 100 mM NaCl in H2O/CH3OH (1/1). It should be noted that NaCl cluster ions were detected first at P1 that were taken over by the much more surface-active multiply protonated cytochrome c [M + H]n+ (n = 8–11) at P2. This trend was opposite to the results obtained in our previous PESI experiment [12]. In our previous work [12], NaCl clusters were detected at the last stage of electrospray after more surface active analytes than NaCl were depleted. Thus, the appearance of NaCl cluster ions at the earliest stage of spray event in Figure 5 cannot be due to the normal electrospray that is governed by the surface-active values of analytes. Therefore, the occurrence of polarization-induced electrospray must be invoked to explain the detection of the non-surface-active NaCl cluster ions at the initial stage of the spray event.

Figure 5
figure 5

Snap-shots of the mass spectra for 10−5 M cytochrome c, 1% acetic acid and 100 mM NaCl in H2O/CH3OH (1/1) at P1 (a) and P2 (b) in TIC shown in the inset. HV: 4.5 kV. (c) Mass spectrum measured with the glass capillary operated with the repetition rate of 1 Hz and the HV duration time of 0.8 s

The most probable electrochemical oxidation reactions in the present positive-mode electrospray may be reactions (1) and (2).

$$ 2{\mathrm{H}}_2\mathrm{O}\to 4{\mathrm{H}}^{+}+{\mathrm{O}}_2+4{\mathrm{e}}^{-} $$
(1)
$$ 2{\mathrm{Cl}}^{-}\to {\mathrm{Cl}}_2+2{\mathrm{e}}^{-} $$
(2)

The standard electrode potentials for oxidation reactions (1) and (2) are 1.25 and 1.36 V vs. 2H+/H2. That is, reaction (1) is more favorable than reaction (2) to occur. This may explain the reason why the ions [M + nH]n+ took over NaCl cluster ions with time in Figure 5. Figure 5c shows the mass spectrum measured with the repetitive operation of the needle with 1 Hz. The HV duration time was 0.8 s for each cycle. The multiply charged ions [M + nH]n+ were observed as the major ions with much weaker Na14Cl13+ at m/z 782. It seems likely that accumulated protons in the sample solution on the glass probe were preferentially detected as [M + nH]n+ resulting in the suppression of NaCl cluster ions.

Figure S3 shows the mass spectra measured at P1, P2, and P3 in the inset for the sample of 10−5 M gramicidin S and 100 mM NaCl in H2O/CH3OH (1/1). In contrast to the case of cytochrome c, [M + H]+ and [M + Na]+ were detected as the major ions with negligible NaCl cluster ions. The ion intensities of [M + H]+ and [M + Na]+ are more than one order of magnitude greater than those of [M + 9H]9+ for cytochrome c in Figure 4. It is evident that gramicidin S is more surface active than cytochrome c. In general, [M + 2H]2+ at m/z 571 was observed as strong as [M + H]+ at m/z 1142 in normal ESI [12]. In this respect, the absence of [M + 2H]2+ in Figure S3 is unique. The generation of excess protons in the present system must be much slower than the conventional electrospray in which the solution is in direct contact with the metal electrode.

Positive and Negative Mode of Operation

So far, the experimental results obtained by the positive mode of operation have been described. In the inductive electrospray, positive as well as negative ions were detected [3,4,5,6,7]. It is suggested that the almost simultaneous generation of positive and negative ions is due to the induced voltage in positive followed by negative voltage in solution by the application of a positive HV pulse [5]. In the present experiment, however, no negative ions were detected when a positive HV pulse with pulse width from μs to seconds was applied to the needle. Moreover, no electrospray plume was generated under any experimental conditions when a negative HV pulse was applied to the needle. The liquid on the glass probe was standing still with the application of a negative HV. For example, UltraMark1621 gave no negative ions in the negative mode while it did strong enough ion signals to detect in the positive mode (not shown). When a negative HV was increased up to about − 5 kV, an arc discharge was formed through the glass tip which led the metal needle to melt down. This indicated that the accumulation of negative excess charges to the sample solution in the negative mode was not enough to generate either inductive or normal electrospray under present experimental conditions.

Occurrence of Sequential Electrospray

In probe electrospray ionization (PESI), sequential and exhaustive electrospray for the analytes with different surface active values was observed [12,13,14,15]. The sequential electrospray is a unique nature of PESI that is based on the single shot electrospray [13]. In PESI, excess charges are continuously supplied to the liquid sample in contact with the needle tip by the occurrence of electrochemical reactions until all the solution is depleted. In charging of solution, excess charges are preferentially enriched in the thin liquid surface skin of the Taylor cone leading to the high electric field generated at the liquid meniscus. The thickness of the liquid surface skin enriched by the excess charges (i.e., electric double layer) is roughly approximated as the Debye length λD in eq. (3), where C0 is the molar concentration (M) of the electrolyte [10, 16].

$$ {\lambda}_{\mathrm{D}}\left(\mathrm{nm}\right)=0.305\times {C_0}^{-1/2} $$
(3)

As an example, λD for 100 mM NaCl is calculated to be ~ 1 nm.

In Figure S3 measured in the present glass-PESI experiment, [M + H]+ and [M + Na]+ for gramicidin S were detected as major ions with much weaker NaCl cluster ion at m/z 782. This indicated that the thin surface layer of the Taylor cone enriched by surface active [M + H]+ and [M + Na]+ was preferentially stripped off and the major component of NaCl remained in the bulk solution.

Figure 6 shows the glass-PESI mass spectra for yogurt measured with a HV duration of 5 s. Sample of yogurt was served for the measurement without dilution. As for this sample, electrospray lasted for about 3.5 s. Figure 6a, b shows the mass spectra measured at P1 and P2, respectively, in TIC (inset). At P1, adduct ions of Na+, K+, and Ca2+ with lactose were detected as major ions. It is apparent that lactose which is the major component of saccharides in yogurt forms strong bonds with these metallic ions. At P2, a series of ions in the range of m/z 600–920 were newly detected, i.e., the occurrence of sequential electrospray. Despite of our efforts, these ions could not be identified. We conjecture that these compounds may be the additives in yogurt to protect lactobacillus in the highly acidic stomach.

Figure 6
figure 6

Snap-shots of the mass spectra for yogurt at P1 and P2 in TIC shown in the inset. HV: 4.5 kV

Hayati et al. observed the axisymmetric circulating meridional motion of the tracer particles (lycopodium powder) in the conical base of the Taylor cone [17]. The velocity of the liquid at the surface layer was the largest, whereas there was a backflow in the center. This liquid circulation (i.e., vortex) is driven by the surface shear stress induced by the tangential electric field [18]. The axis of the doughnut-shaped toroidal vortex coincides with the centerline of the cone. If the solution is composed of analytes with different surface active values, more surface active analytes will be enriched on the surface of the Taylor cone and they are electrosprayed at the initial stage of electrospray. Owing to the continuous supply of excess charges to the solution, less-surface active ions temporarily reserved in the vortex are enriched on the surface of the Taylor cone and are electrosprayed after the more-surface active ions are depleted from the Taylor cone (see Figure 6).

Equivalent Circuit

In this work, a gradual accumulation of excess charges (protons) in the sample solution was observed, i.e., the occurrence of electrochemical reaction. Apparently, the thin-wall (~ 20 μm) glass capillary did not behave as a perfect insulator but it did as a high-resistance material. The equivalent circuit of the present system is shown in Figure 7. The capacitor (capacitance: C) is composed of the internal metal needle electrode, the glass capillary, liquid sample captured on the glass probe, and the air space between the glass probe and the counter electrode (mass spectrometric inlet system). When a HV pulse (V = 4.5 kV) was applied to the metal needle that was separated from the inlet by 3 mm, the liquid on the glass probe was exposed to the electric field of ~ 1.5 × 106 V m−1 (4.5 × 103 V divided by 3 mm). At the moment of HV application to the metal electrode, polarization-induced electrospray took place (see Figure 5). By the occurrence of inductive electrospray induced the sharp rise of the voltage, positively charged droplets are emitted from the neutral droplets towards the counter electrode. Due to the loss of positive charges, the liquid sample will be negatively charged after the inductive electrospray. In order to observe the temporal behavior of the inductive electrospray that is followed by normal electrospray, the ion current to the counter electrode was measured for 10−5 M gramicidin S and 1% acetic acid in water/methanol (1/1) using an oscilloscope. The distance between the glass probe and the counter metal electrode was 3 mm. When a HV pulse of 4.5 kV with the duration of 10 s was applied to the metal electrode, the positive displacement current at the start and the negative one at the end (negative current) of the pulse were observed. However, the dc component of electrospray was found to be below the background level between two sharp displacement currents. In the same experimental setup, PESI experiment was made by replacing the glass-PESI needle with a 0.2 mm o.d. titanium needle wire. The electrospray current of ~ 50 nA lasted for a second after the application of HV. These results indicated that the electrospray current in glass-PESI was more than two orders of magnitude lower than normal PESI. Apparently, high-resistance glass capillary limited the electrospray current. Despite the low current, the mass spectrometer is sensitive enough to detect ions generated by glass PESI.

Figure 7
figure 7

Equivalent electrical circuit for the glass-PESI probe system. R stands for the resistance of the glass capillary

Gradual increase of the ion signal intensities (Figure 3) indicated that the excess charges induced on the metal needle at the moment of the high voltage application were slowly transferred to the liquid sample on the glass probe by the electric current through the high-resistance glass. The charging of the glass surface in contact with the liquid solution generates the electric double layer at the interface between the glass surface and the liquid solution. The electric double layer as generated induces the electrochemical reactions for the sample solution. According to Gauss’ law, the excess charges generated by the electrochemical reactions are enriched on the surface of the liquid meniscus because the electrolyte solution is regarded as an electric conductor. When the electric field on the liquid meniscus reached the Rayleigh limit, charged droplets start to be emitted from the tip of the Taylor cone (e.g., 5 ms in Figure 2). According to Figures 2 and 3, the time to reach the steady-state electrospray is estimated to be ~ 1 s. At this moment, the rate of charging by the electrochemical reactions and that of discharging (loss of excess charges by electrospray) becomes nearly equal. This time (~ 1 s) is about two orders of magnitude longer than the normal PESI [14]. This is due to the high resistance of the glass capillary in glass-PESI.

Conclusions

A tip-sealed fine glass capillary inserted with an acupuncture needle (glass-PESI) was used as the electrospray emitter. The mass spectra were measured as a function of the pulse width of a positive HV (+ 4.5 kV) applied to the metal needle. The positive ions started to be detected with the pulse width of ~ 5 ms for 10−5 M cytochrome c in H2O/CH3OH. The ion abundance increased with the pulse width up to a few seconds. The occurrence of polarization-induced electrospray to generate positive ions was likely right after the application of high electric field to the liquid sample. Despite the abundant positive ion signals, no negative ions were detected.

In addition to the polarization-induced electrospray, normal electrospray generated by the occurrence of electrochemical reactions taking place at the interface between the glass probe and the liquid sample solution was observed. The electrospray current observed for glass-PESI was more than two orders of magnitude lower than that for PESI. In contrast to the positive-mode operation, no electrospray was generated when a negative HV was applied to the metal needle.

As in the case of PESI, the occurrence of sequential electrospray was observed in glass-PESI. It was suggested that the sequential electrospray was caused by the vortex flow in the Taylor cone.