Desorption in Mass Spectrometry

Dilshadbek Tursunbayevich Usmanov; Satoshi Ninomiya; Lee Chuin Chen; Subhrakanti Saha; Mridul Kanti Mandal; Yuji Sakai; Rio Takaishi; Ahsan Habib; Kenzo Hiraoka; Kentaro Yoshimura; Sen Takeda; Hiroshi Wada; Hiroshi Nonami

doi:10.5702/massspectrometry.S0059

Abstract

In mass spectrometry, analytes must be released in the gas phase. There are two representative methods for the gasification of the condensed samples, i.e., ablation and desorption. While ablation is based on the explosion induced by the energy accumulated in the condensed matrix, desorption is a single molecular process taking place on the surface. In this paper, desorption methods for mass spectrometry developed in our laboratory: flash heating/rapid cooling, Leidenfrost phenomenon-assisted thermal desorption (LPTD), solid/solid friction, liquid/solid friction, electrospray droplet impact (EDI) ionization/desorption, and probe electrospray ionization (PESI), will be described. All the methods are concerned with the surface and interface phenomena. The concept of how to desorb less-volatility compounds from the surface will be discussed.

RATIONALE: ABLATION VS. DESORPTION

The two technical terms, ablation and desorption, should be distinguished rigorously. As shown in Fig. 1, ablation is induced by explosion of the matrix or substrate whereas desorption is a single molecular event, i.e., detachment of a single molecule from the surface. The antonym of desorption is adsorption.

Fig. 1. Conceptual idea for (a) ablation and (b) desorption/adsorption.

In matrix assisted laser desorption/ionization (MALDI), for example, the photon energy is ultimately converted to thermal energy and ablation of the matrix takes place above the critical laser power (Fig. 1(a)). In ablation, the ablated materials are mainly composed of aggregates. Thus the ions observed in MALDI are only a small part of materials ablated. In this respect, MALDI is more likely matrix-assisted laser “ablation” ionization. In secondary ion mass spectrometry (SIMS), primary projectiles such as Ar⁺, Cs⁺, O₂⁺, Bi₃⁺, C⁺₆₀, and (Ar)_n⁺ are used. In SIMS using atomic ion projectiles, the sputtering is caused by the cascade collisions in the sample. Thus the useful yield defined as [total gas phase ions detected] divided by [total amount of sample sputtered] is rather low. The small useful yields for MALDI and SIMS (10⁻⁵–10⁻⁶) originate from the fact that photons or primary beams penetrate into the target materials leading to the sputtering of the substrate. In contrast, the useful yield for desorption-based techniques could be much higher than MALDI and SIMS if the desorbed molecules are ionized by high-efficiency ionization methods.

In this review article, “desorption methods” developed in our laboratory will be described. First, the concept of thermal desorption will be given. In the following sections, thermal and non-thermal desorption methods: flash heating/rapid cooling, Leidenfrost phenomenon assisted desorption, solid/solid friction, liquid/solid friction, electrospray droplet impact, and probe electrospray, will be described. All the methods are based on the surface or interface phenomena.

THERMAL VS. NON-THERMAL DESORPTION

The thermal desorption is the classical but most common desorption method in mass spectrometry.¹⁾ The word of “thermal” means that the system is in thermal equilibrium, i.e., all the energy modes, translational, rotational, vibrational and electronic energies in the system are in equilibrium and only under such conditions, the temperature “T” can be defined rigorously. Arrhenius equation (1) gives the dependence of the rate constant k of a chemical reaction on the temperature T (K) where A is the pre-exponential factor, ΔE the activation energy and R the gas constant.²⁾

(1)

With increase in the molecular size of analytes, the activation energy for desorption (ΔE) increases due to the intermolecular forces such as dipole–dipole interaction, dipole–induced dipole interaction, London disperse force and hydrogen bond.

Beuhler et al. made a kinetic analysis on thermal decomposition and desorption of peptides.²⁾ They found that thermal desorption becomes more favorable than thermal decomposition at higher temperature for some peptides even when the activation energy of desorption is larger than the bond dissociation energy. This phenomenon was interpreted based on the “classic” Arrhenius reaction rate theory for thermal desorption and decomposition. Figure 2 shows the conceptual idea of Arrhenius plots for thermal desorption and decomposition. Because the slope of the Arrhenius plot for thermal desorption is steeper than that for thermal decomposition, the rate constant for thermal desorption becomes larger than that for thermal decomposition in the higher temperatures above the crossing point in Fig. 2. However, if the heating rate is low, thermal decomposition will dominate over desorption during the slow heating process. Thus, the sample temperature should be raised as fast as possible to suppress the thermal decomposition. In the following sections, the methods for desorption that can suppress or even avoid the thermal decomposition will be presented.

Fig. 2. Arrhenius plots for thermal decomposition and thermal desorption. The activation energy for thermal desorption is assumed to be larger than that for thermal decomposition. The rate constant for thermal desorption becomes larger than that for thermal decomposition at temperatures higher than the temperature at the crossing point of two lines.

FLASH SURFACE HEATING/RAPID COOLING

Figure 3(a) shows the sample sitting on the heater. In this case, the bottom part of the sample is heated first. Desorbed molecules in the space between the bottom part of the sample and the heater would suffer from secondary thermal decomposition. In addition, the gas layer formed at the interface acts as a thermal insulator resulting in a slower heating rate for the upper-side sample. This is reminiscent of the fact that a wet finger does not get burned when it touches the red hot heater. Such “side effects” for the bottom heating can be largely avoided if the heater just touches the surface as shown in Fig. 3(b). In this case, the molecules on the top surface of the sample are desorbed and released to the open space without suffering from serious secondary thermal decomposition. According to this concept, we have developed the flash heating/rapid cooling desorption method as shown in Fig. 3(c).³⁾ The filament is driven along the vertical axis to the ion sampling orifice of the mass spectrometer. At the lowest position, the tip of the hot filament just touches the sample surface with the invasion depth of ≤0.1 mm and contact time of 50 ms. The high heating rate (∼10⁴°C/s) is attained by the direct and short time contact of the preheated filament (100–760°C) with the sample surface. After touching the surface, the filament is lifted upward. That is, flash heating followed by rapid cooling is attained in this system. This makes it possible to suppress the thermal decomposition for the desorbed molecules. The desorbed molecules were ionized by using a dielectric barrier discharge (DBD) ion source (ARIOS, Akishima, Japan) using helium as a discharge gas.^4,5) In this ion source, the desorbed molecules were not exposed to the plasma but were ionized mainly by H₃O⁺ and its water clusters H₃O⁺(H₂O)_n produced outside of the DBD alumina tube. That is, the present DBD ion source may be regarded as an APCI ion source.

Fig. 3. (a) Bottom heating: the sample is placed on the heater. (b) Surface heating: the heater touches the surface of the sample. (c) Schematic diagram of the flash heating/rapid cooling system. Reproduced under permission from American Society for Mass Spectrometry.³⁾

The generated ions were entrained by helium gas toward the ion sampling orifice and detected by a time-of-flight mass spectrometer. Owing to the rapid heating/cooling operation, molecular ions and/or protonated molecules were observed as major ions without suffering from serious thermal decomposition for many less-volatility compounds. In addition, mass spectra obtained for synthetic polymers gave the information on their fundamental backbone structures.

As an example, Fig. 4(a) shows the mass spectrum for the tablet of loratadine measured at 760°C. Even at this high heater temperature, the protonated loratadine [M+H]⁺ was detected as the major ion. Similar results were obtained for other pharmaceuticals, narcotics, explosives, C₆₀ and synthetic polymers.³⁾ Figure 4(b) shows the mass spectrum for polymethyl methacrylate (PMMA), measured at 760°C. The protonated monomer units [(C₅H₈O₂)_n+H]⁺ with n=1–4 are detected as major ions. Other major ions also reflect the fundamental backbone structure of the polymer.³⁾ Minor secondary thermal decomposition observed with filament temperature even above 500°C is the useful figure of merit for the present technique, verifying the concept given by Beuhler et al.²⁾

Fig. 4. (a) Mass spectrum for loratadine. (b) Mass spectrum for polymethyl methacrylate. The filament temperature: 760°C. Reproduced under permission from American Society for Mass Spectrometry.³⁾

It should be noted that the less-volatility compounds dealt with desorbed at rather low filament temperatures (e.g., 154°C).³⁾ This suggests that there might be some tribological effect in addition to the thermal effect in the desorption process of less-volatility compounds. The friction between the filament and sample surface may result in loss of materials from the surface, i.e., tribodesorption (see the following section of solid/solid friction).

LEIDENFROST PHENOMENON-ASSISTED THERMAL DESORPTION FOR LESS-VOLATILITY COMPOUNDS BELOW 100°C

When liquid nitrogen (−196°C) is poured on the table, liquid droplets do not evaporate instantly but they levitate on the table. By the rapid evaporation of liquid nitrogen, the temperature of the droplets is kept at −196°C (boiling point of N₂) to the last moment of evaporation. Such a phenomenon is also observed on the hot frying pan in the kitchen. When water droplets are put on a heated solid surface with the temperature much higher than 100°C, the lower surface of the water droplet immediately vaporizes and the droplet levitates above its own vapor (inset 2 of Fig. 5). This is known as the Leidenfrost phenomenon.^6–8) Under this condition the liquid droplet evaporates slowly because of the poor heat conduction across the vapor layer.⁹⁾ This phenomenon explains various heat transfer mechanisms and technical applications,¹⁰⁾ but its application for the development of a highly sensitive open ion source atmospheric pressure mass spectrometric technique is unprecedented.

Fig. 5. Schematic diagram for Leidenfrost phenomenon-assisted desorption. Inset 1: Photograph of the experimental system. Inset 2: The droplet levitating on the heater, i.e., Leidenfrost phenomenon. Reproduced under permission from Elsevier.¹³⁾

Here we describe the Leidenfrost phenomenon-assisted thermal desorption (LPTD), which is applicable to less-volatility and thermally labile compounds with little or no sample pretreatment.^11–13)

As shown in the inset 1 of Fig. 5, a liquid droplet containing analyte is placed on a heated metallic sample holder situated in front of the inlet of mass spectrometer. The temperature of the sample holder was kept higher than the boiling points of solvents (e.g., 250°C for methanol and 450°C for aqueous solution). In LPTD, less-volatility compounds remain in the liquid droplets during the evaporation of the solvent till the last moment of total liquid evaporation. That is, the analytes undergo concentration during the shrinkage of the droplet. At the last moment of liquid evaporation, the less-volatility analytes desorb from the tiny droplet with the assistance of explosive solvent evaporation (see the extracted ion chronogram (EIC) in Fig. 5). This makes it possible to desorb analytes with little thermal decomposition. As the pre-concentrated analytes desorb in a very short period with enriched concentration, the sensitivity of this technique becomes several orders of magnitude higher than that of the nucleate boiling process.^11,12) The desorbed neutral molecules are ionized in a post-ionization method at open atmosphere using DBD ion-source, electrospray ionization, or corona discharge ion source before entering the mass spectrometer inlet (inset 1).

Figure 6(a) shows the extracted ion chronogram (EIC) of [M+H]⁺ for 1 ppm morphine (M) in methanol solution. In 0–0.49 min, the ion signals originating from solvent methanol predominated but those from morphine were not detected. The sharp peak of [M+H]⁺ at 0.49 min in EIC indicates that morphine desorbed at the last moment of the solvent evaporation. Figure 6(b) shows the mass spectrum at 0.49 min. [M+H]⁺ is detected as the major ion. In the inset of Fig. 6(b), 1 ppb morphine was detected by the single ion monitoring (SIM) (net amount of morphine: 20 pg).

Fig. 6. (a) Extracted ion chronogram (EIC) of [M+H]⁺ for 1 ppm morphine (M) in methanol solution. [M+H]⁺ appeared at the last moment of the liquid evaporation. (b) Mass spectrum for 1 ppm morphine in methanol solution measured at 0.49 min. Inset: Single ion monitoring of 1 ppb morphine (net amount: 20 pg) in methanol solution. Reproduced under permission from Elsevier.¹³⁾

Figure 7(a) shows the mass spectrum of urine spiked with anabolic steroids mixture (androstadienedione, androsterone hemisuccinate, 6-dehydrocholestenone, epitestosterone and stigmastadienone, concentration of each steroids: 1 μg/mL) measured at the last moment of liquid evaporation.¹³⁾ In the figure, urea, urea dimer and creatinine peaks were predominant and the intensities of these compounds were several orders of magnitude higher compared to the targeted anabolic steroids. The enlarged section of mass range m/z: 250–420 (inset, Fig. 7(a)) shows that all of these targeted compounds were detected. Protonated molecule ion peaks [M+H]⁺ were detected for all of the steroids, except androsterone hemisuccinate. However, androsterone hemisuccinate was identified by its characteristic peak [M+H−C₄O₅H₈]⁺ at m/z 255, which was predominant in the mass spectrum of standard solution. This technique can detect as low as 2–50 ng/mL anabolic steroids from urine. The reported LODs for these steroid molecules in urine were around 300 ng/mL using reactive DESI.¹⁴⁾

Fig. 7. (a) Mass spectrum of urine spiked with anabolic steroids mixture (androstadienedione, androsterone hemisuccinate, 6-dehydrocholestenone, epitestosterone and stigmastadienone, concentration of each steroid: 1 μg/mL) measured at the last moment of liquid evaporation. (b) Mass spectrum of dichloromethane extract for a mixture of five anabolic steroids (concentration of each steroids: 1 μg/mL) in urine. The peak intensities are about 50 times higher than (a) of direct urine analysis. Reproduced under permission from Elsevier.¹³⁾

According to the World Anti-Doping Agency (WADA) guideline, the detection limit of various anabolic steroids from urine should be 2–5 ng/mL.¹⁵⁾ Although epitestosterone and androstadienedione satisfy the criteria, other three steroids showed the LOD of 10–50 ng/mL from urine. In order to overcome the disadvantages associated with the analysis of urine, we introduced a rapid dichloromethane extraction technique to suppress the matrix effect and increase the sensitivity and reproducibility of successive analyses. Dichloromethane is insoluble in water and can extract trace organic analytes from complex matrices. As dichloromethane is a volatile solvent, it shows Leidenfrost phenomenon at a lower temperature (200°C) compared to methanol (250°C). In this measurement, 500 μL steroids spiked urine was taken in a glass vial and 200 μL dichloromethane was added to that. The whole solution was vortexed for 5 s. A 50 μL droplet was pipette out from the lower dichloromethane layer and was placed on the sample holder (200°C). The total procedure takes about 1 min to analyze a single sample. Figure 7(b) shows the mass spectrum of dichloromethane extract for a mixture of five anabolic steroids (concentration of each steroids: 1 μg/mL in urine). The peak intensities indicate that about 50 times higher sensitivity was obtained compared to direct urine analysis. Using this technique the LODs for anabolic steroids from urine were ranged in between 0.1–1.0 ng/mL, which is sufficient to analyze these compounds by following WADA guideline.¹⁵⁾

Dichloromethane extraction technique also showed a better precision and recovery data compared to direct urine analysis. The % RSD values were less than 14% using this technique and recoveries were 94–112%. Basically, urine is a viscous liquid and contains a large amount of salt, which hinders the easy vapor formation. Here, using the dichloromethane extraction technique, not only the target compounds were extracted, but also they were pre-concentrated using less volume dichloromethane. The dichloromethane droplet showed good levitation property as the vapor formation is easy. Thus, the whole system showed better reproducibility compared to the direct urine analysis. The LODs obtained using dichloromethane extraction technique are comparable with the standard LC-MS or GC-MS methods for steroid analysis.^16–19)

The most fascinating aspect of this technique is that the spontaneous sample enrichment takes place inside a droplet without using any external devices. The various validation parameters studied in this work showed quite promising results for ambient MS detection of less-volatility compounds such as drugs (morphine, cocaine, atenolol, triamterene), melamine, caffeine, immunosuppressant (cyclosporine A), explosives (trinitrotoluene (TNT), 1,3,5-trinitroperhydro-1,3,5-triazine (RDX), octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX), and pentaerylthritol tetranitrate (PETN)), and steroids.^12,13) The present technique is a viable alternative of conventional GC-MS or LC-MS for rapid and high-throughput spot detection of less-volatility compounds, e.g., doping test. This scenario holds for any surface-active compounds but not for highly hydrophilic compounds such as sugars and inorganic salts that form residues after evaporation of solvent.

DESORPTION BY MECHANICAL ENERGY: SOLID/SOLID FRICTION

Tribological phenomena, observed at the interface of moving matter, are ubiquitous in nature. For example, the formation of negative ions in the air when water splashes is known as the Lenard effect. The lightning in atmosphere is caused by charging via the friction of the water/ice particles in the clouds.

Tribology is the study of interactions between surfaces in relative motion²⁰⁾ including triboemission of electrons, photons, neutral and charged particles, in addition to heat as the most degraded form of mechanical energy.^21,22) Tribological phenomena are dynamic processes taking place between the contacting surfaces. A macroscopic force is exerted on the microscopic and nanoscopic scales at the friction interface and thus it is very difficult to obtain molecular-level information by the conventional analytical techniques.²³⁾ In this respect, mass spectrometry is highly appealing because it can provide mechanistic information by the measurement of desorbed molecules and ions.

There are numerous examples in which tribological phenomena were used for desorption/ionization in mass spectrometry. Hirabayashi et al. developed sonic spray ionization (SSI) that uses the friction of a sonic gas flow to nebulize liquid flowing out of a capillary.²⁴⁾ Eberlin and coworkers further modified SSI as an easy ambient sonic spray ionization mass spectrometry (EASI-MS) that uses the sonic spray for desorption/ionization of solid samples.^25,26) Jarrold et al.^27,28) studied charge separation in liquid droplets when droplets generated by electrospray, sonic spray, and a vibrating orifice aerosol generator were introduced from atmospheric pressure to vacuum through a capillary. Charge separation was explained by the bag mechanism for droplet breakup and the electrical bilayer at the surface of liquid droplets. In deformed droplet with a bag shape, positive and negative charges are unevenly distributed because of differences in their surface active values.²⁷⁾ They also found that even negatively charged droplets were formed from positively charged electrospray droplets when they passed through a narrow capillary.²⁸⁾ Charge separation of droplets inside the capillary was applied to mass spectrometry by McEwen et al.²⁹⁾ as solvent-assisted inlet ionization (SAII). SAII produces gaseous ions with energy assistance from the pressure drop region and heat, e.g., ionization occurs in the heated inlet tube linking the atmosphere and the first vacuum region of a mass analyzer. In desorption electrospray ionization (DESI) developed by Cooks et al.,³⁰⁾ pneumatically-assisted electrospray droplets were directed onto a surface bearing an analyte. The desorbed molecules were efficiently ionized by the impact of the electrospray-charged droplets onto the surface. Dixon et al.³¹⁾ and Zhu et al.³²⁾ studied acoustic nebulization of a few μL liquid solutions of biological samples using a quartz ultrasonic transducer. Goodlett and coworkers first applied the surface acoustic wave (SAW) for nebulization of peptide solutions as a microfluidic interface for mass spectrometry.^33,34) Trimpin et al. developed an ionization method that uses only the matrix such as 3-nitrobenzonitrile or 2,5-dihydroxyacetophenon.³⁵⁾ They found that abundant analyte ions including multiple-charge proteins were formed during the sublimation of the matrix under sub-atmospheric pressure. The friction in a matrix accompanied by fracturing (e.g., crack propagation by dislocation) may form charged particles (ejecta).²³⁾

As described in the section of flash heating,³⁾ tribodesorption was suggested in the contact of the hot filament with the solid surface. To obtain more direct information on the desorption induced by solid/solid friction, desorption of less-volatility compounds deposited on a plastic substrate by using an ultrasonically vibrating blade was examined.³⁶⁾

Figures 8(a) and (b) show the schematic of the experimental setup.³⁶⁾ The 2 μL sample solution was deposited on the flat perfluoroalkoxy (PFA) substrate with 2 mm in diameter. After the liquid sample was dried in the air, the deposited sample on PFA was touched “very gently” with an ultrasonic cutter (20 W, 40 kHz, oscillation amplitude: 16 μm, Honda Electronics, Toyohashi, Japan). With this gentle touch, the surface of the substrate showed little frictional damage as observed by the naked eye.

Fig. 8. (a) Photograph of the solid/solid experimental system. (b) Schematic diagram of the solid/solid experiment. (c) Mass spectrum for 100 ng p-chloro-benzyl pyridinium chloride (thermometer molecule) measured with the DBD ion source turned off. (d) Mass spectrum for 100 ng p-chloro-benzyl pyridinium chloride (thermometer molecule) measured with the DBD ion source turned on. Reproduced under permission from Elsevier.³⁶⁾

A fraction of the desorbed gaseous molecules were ionized by using a He dielectric barrier discharge (DBD) ion source. The temperature of effusing gas from the DBD ion source was 49°C. The slightly higher temperature than the room temperature is due to the radiofrequency heating of the dielectric tube. The low temperature gas flow is different than direct analysis in real time (DART), which uses a much higher gas temperature for sample desorption.³⁷⁾

To obtain direct information on desorption, the nonvolatile ionic compound, p-chloro-benzyl pyridinium chloride (thermometer molecule), was examined. The result obtained for 100 ng of this compound without turning on the RF voltage of the DBD ion source is shown in Fig. 8(c). The chlorobenzyl pyridinium ion M⁺ was detected as the base peak with a much weaker signal of the fragment ion [M−pyridine]⁺ (i.e., p-chlorobenzyl cation, ClC₆H₄CH₂⁺). It is evident that the nonvolatile ionic compound desorbed mainly as an isolated monomer ion.

Figure 8(d) shows the mass spectrum obtained with the DBD ion source turned on. The strong appearance of the peak of protonated pyridine at m/z 80 clearly indicates the formation of pyridine by the fragmentation of M⁺, M⁺→[M−pyridine]⁺+pyridine. In Fig. 8(c), pyridine was not detected because no reagent ions for the ionization of neutral pyridine were present with DBD off. The strong appearance of the protonated pyridine in Fig. 8(d) with DBD on is due to the high volatility of pyridine.

For neutral analytes examined, they were detected only when the He DBD ion source was turned on. Figure 9(a) shows the mass spectrum of 2 ng cholesterol. The dehydrated peak is detected at m/z 369. Figure 9(b) presents the mass spectrum of 20 ng spinosad (insecticide). The protonated molecules at m/z 732 (spinosad A) and m/z 746 (spinosad D) in addition to the fragment ion from spinosad A and D [C₈H₁₅NO+H]⁺ at m/z 142 are detected. Figure 9(c) displays the raw urine sample spiked with 2 ng morphine. Morphine was detected in the protonated form without giving the dehydrated fragment ion [M+H−H₂O]⁺ at m/z 268. This indicated that under present experimental conditions the decomposition of morphine was negligible. In the mass spectrum for raw cow milk spiked with 2 ng melamine (Fig. 9(d)), the appearance of the protonated melamine at m/z 127 as the base peak indicated that melamine in cow milk was easily detected and this was attributed to its high proton affinity (226.2 kcal/mol).

Fig. 9. Mass spectra for neutral compounds measured by solid/solid friction. (a) 2 ng cholesterol. (b) 20 ng spinosad (insecticide). (c) 2 ng morphine spiked in urine. (d) 2 ng melamine spiked in cow milk. Reproduced under permission from Elsevier.³⁶⁾

In summary, the less-volatility compounds are desorbed with minor fragmentation in solid/solid friction. This means that the mechanical energy generated in friction between the perfluoroalkoxy substrate and the ultrasonically vibrating blade is efficiently dissipated for desorption of less-volatility compounds. In other words, dissipation of mechanical energy to thermal energy is not the major process in this method. It should be noted that when the vibrating blade was pressed to the substrate rather hard, no ion signals were detected. This suggests that the mechanical energy is mainly dissipated as shear failure.

DESORPTION BY MECHANICAL ENERGY: LIQUID/SOLID FRICTION

In the previous section, desorption of low-volatility compounds induced by solid/solid dynamic friction was mentioned.³⁶⁾ In the field of mass spectrometry, liquid/solid friction-based desorption/ionization is more familiar than solid/solid friction-based one, e.g., EASI,^25,26) DESI,³⁰⁾ and SAW.^33,34) In this section, desorption induced by liquid/solid dynamic friction using a piezoelectric microdroplet generator and an ultrasonic vibrator is described.³⁸⁾

Figure 10 shows the experimental set up. An aliquot of 1 μL sample solution of water/methanol (1/1) was deposited on the stainless steel blade of an ultrasonic cutter (frequency: 40 kHz, oscillation amplitude: ∼12 μm, SUW 30–30 CT, Suzuki, Japan). The diameter of the deposited sample spot was about 1.5 mm (∼2 mm²). The liquid droplet dried in about 10 min at room temperature.

Fig. 10. Experimental setup for the liquid/solid friction experiment. Inset 1: microdroplets with ∼30 μm o.d. ejected from the piezo electric microdroplet generator. Inset 2: Schematic diagram for the liquid/solid friction experiment.

Microdroplets of water/methanol (1/1) were generated by a piezoelectric microdroplet generator (Microjet, IJHC-10, Shiojiri, Japan) manufactured for ink-jet printers. The liquid microdroplets ejected from the nozzle are shown in the inset 1. The droplets of ∼30 μm in diameter (volume: ∼10 pL) were directed perpendicularly to the blade on which the samples were deposited as shown in the inset 2 (red circle). The frequency of the droplet generator was set at 100 Hz (net flow rate: ∼60 nL/min). The distance between the nozzle of the microjet generator and the blade was set at 1.5 mm. With this travel distance the speed of the microdroplet was ∼10 m/s. The blade bearing an analyte was moved manually across the microdroplet beam with a speed of ∼0.5 mm/s by using an x–y–z manipulator. With the power of ultrasonic generator less than 20 W, the injected droplets on the blade accumulated on the blade and a liquid pool was formed. However, with the power of ≥20 W, the blade did not get wet by the injected droplets. This indicates the occurrence of “cavitation” between the droplet and the vibrating blade. Desorbed molecules were ionized by using a He DBD ion source (inset 2 in Fig. 10).

Figure 11 shows the results obtained for neutral compounds. In Fig. 11(a) for dried apple juice, ions originating from mono- and di-saccharides are detected. In Fig. 11(b) for 10 ng opium, all the major components contained in opium, i.e., morphine, codeine, thebaine, paraverine, protopine, and noscapine were detected as protonated forms. In Fig. 11(c), the blade after cutting the plastic hose was served for the measurement. The protonated plasticizer, bis(2-ethylhexyl)phthalate, was detected as the major ion. This suggests that the present technique would be useful for the analysis of real-world samples by just cutting them by the blade.

Fig. 11. Mass spectra for neutral samples measured by the liquid/solid friction experiment. (a) Apple juice. (b) 10 ng opium. (c) The blade after cutting the water spray hose. Inset 1: Occurrence of cavitation when the water droplet hits the ultrasonically vibrating blade. Inset 2: Asymmetric collapse of the bubble formed on the surface of the ultrasonically vibrating substrate.

As shown in the inset 1 of Fig. 11, the cavitation occurs at the solid/liquid interface. This is the most important point of the liquid/solid friction experiment. When the bubble is formed close to or on the surface of the boundary, the asymmetric collapse of the bubble takes place as shown in the inset 2 of Fig. 11. This happens because the upper side pressure becomes much higher than the opposite side of the bubble. Note that gas is compressible but liquid is incompressible. The asymmetric collapse causes one side of the bubble to accelerate inward more rapidly than the opposite side resulting in the development of a high-speed liquid microjet which penetrates the bubble toward the interface. Accordingly, the acoustic wave is formed at the interface.³⁹⁾ The observed efficient desorption of less-volatility compounds at the interface may be mainly caused by the acoustic wave excitation.

Figure 12 shows the mass spectra for nonvolatile ionic compounds; ionic liquid, thermometer molecule, and rhodamine B. In this measurement, the DBD ion source was turned off. Figures 12(a) and (b) show the positive- and negative-mode mass spectra, respectively, for ionic liquid (1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide). The molecular cation (C⁺) and anion (A⁻) are detected as major ions. That is, ionic liquid was desorbed mainly as monomer ions by cavitation. Similar results were obtained for thermometer molecule (p-chloro-benzyl pyridinium chloride) and rhodamine B as shown in Figs. 12 (c) and (d), respectively. Strong appearance of single molecular ions indicates that desorption but not ablation is the major process under present experimental conditions.

Fig. 12. (a) and (b): positive- and negative-mode mass spectra, respectively, for ionic liquid (1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide). (c) Positive-mode mass spectrum for thermometer molecule, p-chloro-benzyl pyridinium chloride. (d) Positive-mode mass spectrum for rhodamine B.

The basic principle common to all the techniques described above for flash heating, Leidenfrost phenomenon-assisted thermal desorption, solid/solid friction, and liquid/solid friction is that the thermal or mechanical energy is imparted primarily on the sample surface. This results in efficient desorption of less-volatility and thermally labile molecules with little fragmentation. These methods were found to be widely applicable to compounds with molecular masses of less than ∼500 u. In the next section, seemingly an “unrealistic” method that can desorb/ionize proteins with masses of ≈20,000 u and even inorganic materials such as metals and semiconductors is described.

ELECTROSPRAY DROPLET IMPACT

In the inset 2 of Fig. 10, the liquid droplet hits the solid surface with the speed of ∼10 m/s. Desorption of analytes did not occur when the ultrasonic vibrator was tuned off. This means that the collision of a droplet with a speed 10 m/s is not enough to desorb the sample. If the speed of the droplets is increased to the sound velocity and above, the supersonic collision will take place at the colliding interface. The supersonic collision of a liquid droplet with the solid would be followed by the efficient energy dissipation via shock wave propagation. According to this concept, we developed an ionization/desorption technique using supersonic massive water cluster ions as the projectile.

Figure 13 shows the schematic diagram of the electrospray droplet impact (EDI)/secondary ion mass spectrometer.^40–42) The charged liquid droplets generated by electrospraying 1 M acetic acid or 0.01 M trifluoroacetic acid aqueous solution at atmospheric pressure are introduced into the first vacuum chamber through an orifice with 400 μm diameter. The voltages applied to the stainless steel capillary (i.d. 0.1 mm, o.d. 0.2 mm) is +3 kV. The charged droplets sampled through the orifice are transported into a first quadrupole ion guide for collimation, and accelerated by 0.5–10 kV after exiting the ion guide. The electrospray droplets (i.e., the multiply-charged massive clusters) are allowed to impact the solid sample prepared on a stainless steel substrate with the angle of 60° to the surface normal. The total current of the electrospray charged droplets irradiated on the target was measured to be ∼1 nA using a Faraday cup installed at the target position.

Fig. 13. Schematic diagram for the electrospray droplet impact/secondary ion mass spectrometer (EDI/SIMS). Reproduced under permission from Springer.⁴²⁾

The first RF quadrupole ion guide shown in Fig. 13 is adjusted for sampling charged droplets in the m/z range 1×10⁴–5×10⁴. From the principles of mass spectrometry analysis, only m/z values can be evaluated for the charged particles, but not independent values of m or z. However, in the experimental system in Fig. 13, using electrospray as a source of charged projectiles, the m and z values may be roughly estimated. It is generally accepted that the charge states of the primary electrospray droplets and their offspring droplets are close to the Rayleigh limit.⁴²⁾ The relation between the charge number z and radius of the droplet r is given by the Rayleigh limit equation:

(2)

where γ is the surface tension, ε₀ is the vacuum permittivity, and e is the electronic charge. From the surface tension of water (γ=72×10⁻³ J/m²), ε₀=8.9×10⁻¹² C²/J⋅m, and the density of the droplet (assumed to be equal to that of water, 1 g/cm³), the masses of the droplets and the number of charges are calculated to be in the range of 6.2×10⁵–1.6×10⁷ u, and 62–311, respectively. The radii of the droplets are estimated to be in the range 5–20 nm. As an example, the kinetic energy and velocity of a typical projectile [(H₂O)_90,000+100H]¹⁰⁰⁺ accelerated by 10 kV are 1×10⁶ eV and 12 km/s, respectively. The diameter of the droplets beam can be adjusted in the range of 2–0.05 mm using the ion beam focusing lenses.

A hypothetical model describing the ionization/desorption induced by EDI⁴³⁾ is shown in Fig. 14. Water droplets with a diameter of ∼10 nm impact the sample surface with a velocity of ∼12 km/s, which is higher than the sound velocity in various solids. For example, the sound velocity in polystyrene, ice, and Si is ∼2.4, ∼3.2, and ∼8.4 km/s, respectively. Thus, the supersonic or even hypersonic collision takes place at the moment of impact. Immediately following the collision, an enormous pressure would be exerted at the colliding interface because of the high momentum of the water droplet impact (Fig. 14(b)). The force F exerted on the surface is expressed as

(3)

where dp is the momentum change of the projectile (i.e., p=m×v) during the collision, and dt is the interaction time of the droplet with the surface. The water droplet is composed of a tight hydrogen bond network. Thus, the supersonic collision may be followed by shock wave propagation through the hydrogen bond network in the water droplet as shown in Fig. 14(b). If one assumes that the shock wave velocity propagating in the water droplet is roughly the same as the projectile velocity, the transit time of the shock wave through a water droplet with 10 nm diameter may be ∼1 ps (i.e., 10 nm/(12×10³ m/s)≈10⁻¹² s). Because the shock wave has an open end at the opposite side of the droplet, the droplet would be disintegrated as finer microclusters within ∼1 ps (Fig. 14(c)). That is, the dissipation process of the kinetic energy of the projectile should be largely nonthermal.

Fig. 14. Hypothetical model for ionization/desorption taking place in electrospray droplet impact (EDI). Reproduced under permission from American Institute of Physics.⁴³⁾

One of the characteristic features of EDI is its capability to form abundant positive and negative ions for many organic compounds, e.g., amino acids,⁴⁴⁾ peptides,⁴⁵⁾ proteins,⁴⁵⁾ biological tissues,⁴⁶⁾ pigments,⁴⁷⁾ C₆₀,⁴⁴⁾ and inorganic compounds (e.g., silicon, silicon oxide, gold, indium, alkali halides, etc.).^48–50) For example, Fig. 15 shows the mass spectra for 10 fmol gramicidin S, 10 pmol bradykinin, and the natural SiO₂ layer formed on Si substrate.⁵¹⁾ For peptides, the protonated species [M+H]⁺ are detected as major ions. For SiO₂, proton-bound adduct ions containing SiO₂ are detected in both positive and negative ion mass spectra. The efficient formation of both positive and negative ions may be ascribed to the disproportionation proton transfer reaction of water molecules, i.e., the occurrence of electrolytic dissociation reaction of water molecules at the collision interface.^44,47,48)

(4)

For the supersonic collision system, the endoergic Reaction (4) may be one of the primary routes for the collisional energy dissipation. The H₃O⁺, hydronium ion, is known to be a very strong acid, and is likely to transfer its proton to the analyte M to form the protonated molecule [M+H]⁺ at the collision interface.

(5)

The occurrence of Reaction (5) explains the strong appearance of [M+H]⁺ for molecules that have larger proton affinities than H₂O (691 kJ/mol).

Fig. 15. (a) Positive-mode EDI mass spectrum for 10 fmol gramicidin S, (b) Positive-mode EDI mass spectrum for 10 pmol bradykinin, (c) Positive-mode EDI mass spectrum for SiO₂, and (d) Negative mode EDI mass spectrum for SiO₂. Reproduced under permission from Springer.⁵¹⁾

The OH⁻ in Reaction (4) is also known to be a very strong base and it deprotonates from M to form [M−H]⁻.

(6)

In a sense, the very strong acid H₃O⁺ and also the very strong base OH⁻ are inherently incorporated in the supersonic collisional events in EDI using the projectile of water droplets.

In EDI mass spectra, some compounds give strong radical cations and anions, (e.g., for C₆₀,⁴⁴⁾ coronene, and organic pigments⁴⁷⁾) in addition to [M+H]⁺ and [M−H]⁻ ions. For example, mass spectra for C₆₀, C⁺₆₀ and C⁻₆₀ ions are detected in EDI with approximately equal abundances.⁴⁴⁾ Some of these ions may be formed by the disproportionation electron transfer Reaction (7) caused by the supersonic collision in EDI.⁴⁷⁾

(7)

The high ionization efficiency of EDI indicates that at the colliding interface the kinetic energy of the droplet is efficiently converted into internal energies of the species. When an enormous pressure is exerted on the colliding interface at the moment of supersonic collision, the wave functions of the species at the colliding interface overlap and highly excited quasielectronic states will be formed originating from the exchange repulsion (note that electrons are fermions).⁵²⁾ As the particles recede from each other, transitions to the electronic excited states can occur at potential energy curve crossings.⁵²⁾ Thus, the formation of quasielectronic states opens new channels for ionization and dissociation of the species taking part in the collision, and this may be one of the major dissipation channels for the kinetic energies brought about by the impinging water droplets. The shock wave in EDI is a planar (coherent) compressional wave. The following rarefaction of the compressional wave may lead to the efficient desorption of ionized analytes on the surface, i.e., ionization/desorption.

In EDI, no ablation (i.e., sputtering) but atomic- and molecular-level etching takes place. The etching rates with EDI were determined experimentally for several materials, for example, ∼2 nm/min for bradykinin,^44,53) 3–5 nm/min for polystyrene,^54,55) 2 nm/min for polyethylene terephthalate,^54,55) 0.2 nm/min for SiO₂,³⁷⁾ 0.12 nm/min for TiO₂,⁵⁶⁾ 0.7 nm/min for Ta₂O₅,⁵⁶⁾ and 0.3 nm/min for CuO/Cu.⁵⁷⁾ Roughly, the etching rates for organic materials (synthetic polymers and peptides) are about one order of magnitude higher than those for inorganic materials (metal, metal oxides, and SiO₂).

The characteristic features of EDI may be summarized as follows. (1) 10⁶ eV huge water cluster ions represented as [(H₂O)₉₀₀₀₀+100H]¹⁰⁰⁺ are used for the projectile. (2) By using such large-size cluster ions, the impact energy is localized at the colliding interface. (3) Because the velocity of ∼12 km/s is higher than the sound velocities in solids, supersonic collision takes place. (4) Supersonic collision at the colliding interface results in the efficient ionization/desorption for only the top-surface molecules without causing the damage underneath the sample. The unique nature of EDI is ascribed to the fact that the excitation by supersonic collision is instantly followed by the energy dissipation as the shock wave propagation in the media in the time scale shorter than ns, i.e., the processes is almost entirely non-thermal. The rapid excitation coupled with rapid energy dissipation is the key to achieve very high-efficiency ionization/desorption with minor fragmentation of the analytes and little chemical modification of the sample surface left behind.

Finally, mild ion desorption from the liquid caused by Coulomb repulsion, i.e., electrospray, will be described in the next section.

PROBE ELECTROSPRAY IONIZATION

Electrospray is based on the field desorption of ions or charged droplets from liquids that is free from thermal process. In 1917, Zeleny observed the electrospray of liquid glycerol from a capillary.⁵⁸⁾ The electrified liquid meniscus, which took the shape of a cone, was theoretically analyzed by Taylor.⁵⁹⁾ Dole et al. described that analyte ions might be liberated from the electrospray droplet,⁶⁰⁾ and the first electrospray ionization mass spectrometry (ESI-MS) was demonstrated by Fenn et al.^61,62)

ESI-MS has now become one of the indispensable tools for the analysis of biomolecules. In order to cope with small volume of sample solution, miniaturized ESI ion sources with low flow rates have been developed using fine glass capillaries, i.e., nanoESI.^63–66) Wilm and Mann described theoretically and demonstrated experimentally that a narrower spraying capillary with much reduced flow rate could generate smaller initial droplets, and hence better ion desorption from the liquid.^65,66)

While conventional ESI and nanoESI use metal or glass capillaries, several designs of electrospray ion sources that used non-capillary probes had been put forward. Shiea et al. electrosprayed sample solutions deposited on a copper wire ring,⁶⁷⁾ optical fibers coiled with copper or platinum wires,⁶⁸⁾ a glass rod,⁶⁹⁾ and nanostructured tungsten oxide⁷⁰⁾ and mass spectra of proteins similar to those of conventional ESI were obtained.

In 2007, probe electrospray ionization (PESI) using a sharp metal needle as an electrospray emitter was developed in our laboratory.⁷¹⁾ Because PESI is free from clogging, robust, and tolerant to salts⁷²⁾ and buffers,⁷³⁾ it can be applicable to the direct analysis of various biological samples including plants,^74–76) biological samples,^77,78) cancer tissues,^79–81) and metal complexes.^82,83)

The novel feature of PESI compared to the capillary-based electrospray is sequential and exhaustive desorption of ions depending on the analytes’ surface activity values.⁸⁴⁾ The schematic diagram of the PESI system is drawn in Fig. 16(a). In PESI, the needle probe was moved up and down along the vertical axis using a linear motor-actuated system. High voltage of 2–3 kV was applied to the needle when it was at its highest position for electrospray generation, and was held at ground potential when it was in motion or at the position to touch the sample. The bottom position of the needle tip was adjusted to just touch the surface of the biological or liquid sample that was mounted on the xyz manipulation stage. The amount of liquid adhered to the tip depends on the size of the needle, viscosity and surface tension of the sample. When an acupuncture needle with 0.12 mm o.d. and 700 nm tip diameter is used, the sample volume is about tens pL with the invasion depth of ≤1 mm. A single sample loading on the tip of the needle with the volume of ~ten pL is enough for getting strong ion signal abundance.

Fig. 16. (a) Schematic diagram for the probe electrospray ionization (PESI) system. (b) Sequential and exhaustive electrospray taking place in PESI. Excess charges are continuously supplied to the liquid by electrochemical reactions taking place at the interface of the metal and the liquid. (c) Capillary-based electrospray. No supply of excess charges after the droplets detach from the capillary. Reproduced under permission from IM Publication LLP.⁸⁵⁾

Figure 16(b) shows the conceptual idea why sequential and exhaustive electrospray take place in PESI.⁸⁵⁾ Here, the sample solution is supposed to be composed of three components, the most surface-active (red), the next most surface active (blue), and the least surface-active or even non-surface active component (purple). In PESI at the first stage of electrospray, the most surface-active ions are enriched on the surface of the Taylor cone and red ions are preferentially electrosprayed. After depletion of red ions, blue ions are replaced as the enriched ions on the surface of the Taylor cone, resulting in the sequential appearance of blue ions. After depletion of the more surface-active red and blue ions, the least or even non-surface active purple ions (e.g., [(NaCl)_n+Na]⁺) can be electrosprayed at the last stage of electrospray. This kind of sequential and exhaustive electrospray of analyte ions is realized by the “discontinuous” sampling by PESI. By the isolation of liquid sample at the needle tip, excess charges are continuously supplied to the liquid by electrochemical reactions taking place at the liquid/metal interface. In contrast, by capillary-based ESI as shown in Fig. 16(c), the excess charges cannot be supplied to the charged droplets any more after they are detached from the metal capillary. Because the less surface-active ions are apt to remain in the primary droplets, the suppression effect cannot be avoided for the capillary-based ESI.

PESI is widely applicable to liquid or wet samples. As an example, Fig. 17 shows the results obtained for a tissue of human breast cancer.⁸⁴⁾ The PESI mass spectra were measured after 10 μL of mixed solvent of H₂O/MeOH/HOAc (50/50/1) was dropped on a tumor tissue. In this measurement, 0.5 mm o.d. Ti wire was used for the needle probe. Figure 17(a) shows the total ion chromatogram (TIC) that has three peaks. That is, electrospray events are composed of three steps just as shown in Fig. 16(b). At the first step (T1), heme and α and β chains of hemoglobin were detected as the major ions with weaker phosphatidylcholins (PC) and triacylglycerides (TAG) as shown in (b). The sharp decay of the first peak indicates that hemoglobin is almost totally exhausted and electrospray current decreased. During low current regime, the excess charges are being accumulated to the liquid. After enough charging, the second peak (T2) appeared that is composed of PC, TAG and hemoglobin fragments as shown in (c). After the depletion of PC, TAG, and hemoglobin in T2, the second low-current regime appears in 9–12 s. In this time interval, excess charges are continuously accumulated in the liquid. After the enough accumulation of excess charges, the liquid is ready for third electrospray, T3. The third peak mainly composed of sodiated TAG as shown in (d). Because TAG is a neutral molecule, it is rather difficult to detect it by capillary-based electrospray. In contrast, TAG was detected with highest intensity in Fig. 17. The detection of sodiated TAG suggests that Na⁺ ions are enriched in the droplet in the second low-current regime in 9–12 s in (a). The sodium source in cancer tissue is likely to be NaCl contained in the original sample. In the positive-mode electrospray operation, Cl⁻ will be oxidized to Cl₂ on the metal needle electrode, 2Cl⁻→Cl₂ and Na⁺ ions will lose their negative counterpart ions of Cl⁻ in the droplet. After reaching the Rayleigh limit with excess Na⁺ ions, the Taylor cone starts to emit Na⁺ ions that are accompanied with neutral TAG molecules [TAG+Na]⁺. That is, the presence of NaCl in the sample promotes but not suppresses the detection of TAG. In this respect, PESI is highly suitable for the cancer diagnostics because TAG is apt to be accumulated in the tumor.

Fig. 17. Mass spectra for a human breast cancer tissue. (a) Total ion chromatogram (TIC). (b) Mass spectrum measured at T1. (c) Mass spectrum measured at T2. Inset: Expanded mass spectrum at m/z 750–950. (d) Mass spectrum measured at T3. Inset: Expanded mass spectrum at m/z 840–940. Asterisks (*) and open circles (○) stand for peaks from phosphatidylcholine (PC) and sodiated triacylglycerides (TAG), respectively. Reproduced under permission from American Society for Mass Spectrometry.⁸⁴⁾

Because PESI uses a sharp needle, it is highly suitable for imaging mass spectrometry and single cell analysis. Chen et al. applied PESI to the ambient imaging of the mouse brain.⁸⁶⁾ The mapping of phosphatidylcholines and galactosylceramides was made with spatial resolution of about 60 μm. Further, PESI was applied to the single cell analysis by several groups. Yu et al. inserted the surface-activated sharp stainless steel probe into plant cells of rhodendron petal, soybean sprout, and geranium leaf.⁸⁷⁾ Metabolites loaded on the tip surface were extracted by the auxiliary solvent generated by a piezo-electric microdroplet generator and electrosprayed after applying a high voltage. Gong et al. detected metabolites at cellular and subcellular levels by PESI-MS.⁸⁸⁾ Tungsten probes with a tip diameter of ∼1 μm were inserted into live A. cepa (onion) cells. By spraying the solvent vapor to the needle tip, metabolites were electrosprayed from the tip of the probe. Recently, Chen et al. developed a method for single-cell analysis and lipid profiling by combining drop-on-demand inkjet cell printing and PESI-MS.⁸⁹⁾ Through inkjet sampling of a cell suspension, droplets including single cells with a volume of ∼500 pL were generated. They were precisely dripped onto a tungsten-made electrospray ionization needle, and immediately spayed under a high-voltage electric field. As mentioned, PESI has a wide scope for applications due to its novel features.

SUMMARY AND CONCLUSION

While aggregates are formed in ablation, isolated molecules are released from the surface in desorption. In this article, various techniques for desorption of less-volatility compounds are described.

In flash heating/rapid cooling, by the touch and go contact of the hot filament with the sample surface, less-volatility compounds are desorbed with minimal thermal decomposition. Gas layer at the interface acts as the thermal insulator resulting in the suppression of secondary thermal fragmentation.

In Leidenfrost phenomenon-assisted thermal desorption (LPTD), less-volatility compounds enriched on the surface of the tiny solvent droplets are desorbed by the assist of final explosive evaporation of solvent. Thermal degradation is minimal because the evaporation takes place below the boiling points of the solvents.

In solid/solid friction, a slight contact between the ultrasonically vibrating blade and perfluoroalkoxy substrate leads to desorption of less-volatility analytes deposited on the substrate. The mechanical energy may be mainly dissipated as the acoustic wave in the solids. Desorption process is found to be non-thermal.

In liquid/solid friction, cavitation occurs between the ultrasonically vibrating blade and the microdroplets. Desorption may be ascribed to the cavitation, i.e., collapse of the bubble.

In electrospray droplet impact (EDI), very efficient ionization/desorption takes place by the supersonic collision between the water droplets and the sample. Shock wave propagation through the hydrogen bonds in water droplet plays a role for the rapid dissipation of supersonic collisional energy. In EDI, atomic- and molecular-level etching with little surface chemical modification is attained. This technique is unique because it can be applicable not only to the organic but also to inorganic materials. Thus this technique is highly promising for the nano-scale three dimensional imaging. However, this technique needs the vacuum system. For the next-generation surface analysis, ambient atomic- and molecular-level desorption/ionization methods are highly demanded.

In probe electrospray ionization, sequential and exhaustive ionization of analytes takes place depending on the surface-active values of the analyte ions in solution. Suppression effect that is inevitable in capillary-based electrospray is largely moderated in PESI.

As a whole, the most important concept for soft desorption of less-volatility compounds is based on the rapid energy dissipation, i.e., large phase space for the energy disposal. This can be realized by the phenomena in which surface or interface is concerned.

Acknowledgments

The financial supports for this work from the Japan Science and Technology Agency and Grant-in-Aid for Scientific Research (S) are gratefully acknowledged.

REFERENCES

Corresponding author