[ WHITE PAPER ]

Waters UniSpray Ionization Source

S. Bajic Waters Corporation, Wilmslow, UK

INTRODUCTION The first commercial (ESI) sources for became available at the end of the 1980s. These early sources broadened the applications of (MS) to biological compounds such as peptides and proteins which would typically be infused into the source at low flow rates of 1–5 µL/min. Under these low flow conditions, atomisation was achieved via a classical Taylor cone which was formed at the end of the liquid capillary following the application of a few kilovolts between the capillary and the ion inlet cone. More recently, this type of infusion-based analysis for biological applications has been superseded by nanospray ESI capillaries that operate at extremely low flow rates (10–1000 nL/min) and offer high ionization efficiency for small sample volumes. From a commercial viewpoint, the greatest leap in the utilisation of ESI sources came from the marriage of mass spectrometry with liquid chromatography (LC-MS) which enabled analytical chemists to benefit from the enhanced specificity offered by both techniques. In order to adapt ESI sources to the high flow rates (0.1–1.0 mL/min) typically UniSpray Ionization Source. used in LC, or its modern equivalent UPLC, it was necessary to aid the atomisation process with the addition of a concentric flow of high velocity nitrogen gas at the ESI tip. However, when conducting infusion experiments with a fixed analyte concentration from flow rates of 10 nL/min to 1 mL/min, it is common to observe only a 20 times increase in analyte ion signal at the higher flow rate whilst the analyte consumption has increased by a factor of 100,000. This is known to be due to the poor ionization efficiency of high flow rate ESI which is critically dependent on a number of factors such as droplet size distribution, droplet charge per unit volume, droplet evaporation rates and additional factors such as the inlet sampling efficiency. Additionally, ionization efficiency can become particularly challenging when using highly aqueous mobile phases. In this white paper, we introduce a new ionization technique called UniSpray™ which attempts to increase ionization efficiency by interacting a high velocity spray with a high voltage, cylindrical target that is positioned in an off-axis, cross-flow arrangement. A number of physical processes will be described which are believed to be important mechanisms that lead to enhanced UniSpray sensitivity when compared to ESI, viz. high Weber number droplet impacts, the Coanda effect, vortex shedding, and counter-rotating surface microvortices.

1 [ WHITE PAPER ]

WHAT IS UNISPRAY? The 1.6 mm-diameter, 35 mm long, cylindrical HV target is UniSpray shares some common features with high flow constructed from a cold-drawn, 316L stainless steel wire that rate ESI sources in that the liquid flow is nebulised by a is polished to a near-mirror finish with 1 µm-grade lapping high velocity, concentric nitrogen gas flow. However, in paper. The target is connected to a 0–5 kV DC power supply comparison to ESI, the high voltage is applied to a cylindrical via a 47 MΩ current-limiting resistor. Since the target has a target electrode that is positioned in close proximity to low thermal conductance to the source housing, it rapidly the grounded nebulizer such that the near-supersonic jet reaches an equilibrium temperature that is equal to the local impinges on the target surface. A schematic of the UniSpray temperature of the heater gas (typically >250 °C for a heater source is shown in Figure 1. The source is formed by directing set-point temperature of 500 °C). In order to optimise source a high velocity nebulised jet from a grounded sprayer onto sensitivity, it is critically important to adjust the point at which a cylindrical metal target that is held at a high voltage and the collimated spray impacts on the cylindrical target. As is located between the sprayer and the ion inlet orifice of shown in Figure 1, the maximum signal intensity is typically the mass . For singly charged analytes, it is obtained when the spray is asymmetrically positioned such conventional to use a positive potential of typically 1 kV for that it impacts on the upper right quadrant of the target. positive ion analysis and vice versa for negative ion analysis. Under these conditions, the gas flow becomes attached to a For multiply charged analytes, such as peptides and proteins, portion of the curved surface and results in asymmetric gas it is generally found that higher potentials of 3–4 kV are streamlines in the wake that are directed towards the ion required for signal optimisation. The pneumatically-assisted inlet orifice. This flow phenomenon is known as the Coanda nebulizer is formed from a 130 µm I.D. by 220 µm o.d. liquid effect. Under the influence of the Coanda flow field, ions and delivery capillary (stainless steel) that is surrounded by a charged droplets are directed towards the ion inlet which is 330 µm I.D. nebulizer tube (stainless steel) with a restriction surrounded by a cone gas nozzle that supports a drying gas length of 10 mm. Nitrogen gas is delivered to the nebulizer flow of nitrogen at 150 L/hr. tube at a gauge pressure of 7 bar. Under these conditions, UniSpray sources are generally found to give enhanced the nitrogen jet will be near supersonic at a distance of a few ionization efficiency when compared to high flow rate ESI millimetres from the nebulizer tip. The distance over which the which, in turn, can lead to enhanced analytical sensitivity. inner capillary protrudes from the nebulizer tube is adjustable Some general characteristics of the UniSpray source have and is typically found to optimise at very small protrusions been studied by Lubin et al1. These authors compared the (<0.2 mm) where the spray appears highly collimated. In a signal intensity obtained by ESI and UniSpray for small conventional manner, the nebulizer is also surrounded by an pharmaceutical compounds (16 analytes in positive ion annular heater that delivers hot nitrogen gas at a flow rate of mode and 7 in negative ion mode) with acidic, basic and typically 1200 L/hr. neutral mobile phase types. Figure 2 shows the signal gains (UniSpray ion intensity divided by ESI ion intensity) obtained at various mobile phase compositions and for three different flow rates. Here, each point is an average of the pooled data for each compound and each mobile phase type. From these plots it can be concluded that the signal intensity observed with the UniSpray source is, on average, higher than the ESI signal for all mobile phase compositions with relatively higher gains observed under higher aqueous conditions. The full data set exhibited a strong compound dependence where, whilst UniSpray gains in excess of x20 were observed under certain conditions, other analytes could occasionally give greater responses with an ESI source.

In view of the performance enhancements outlined above, the following sections will consider some underlying physical mechanisms that are unique to the UniSpray source and may contribute to the observed increase in ionization efficiency or ion sampling efficiency. Figure 1. A schematic of the UniSpray API source.

Waters UniSpray Ionization Source 2 [ WHITE PAPER ]

Figure 2. A comparison of the average signal intensity ratios for UniSpray and ESI versus mobile Figure 3. Typical droplet size distribution for a pneumatically phase composition for 16 positive ion analytes and 7 negative ion analytes (from reference1). assisted ESI probe at a flow rate of 0.5 mL/min of water (Malvern Instruments Spraytec system).

DROPLET SIZE DISTRIBUTION: THE LIMITATION OF HIGH experimental techniques where, in particular, FLOW RATE ESI the ping-pong drift cell method4 more closely In the traditional ESI model, charged droplets are formed by positive and resembles the dynamics of LC/MS sources. negative charge separation in the high electric field region at the tip of the The diameter of a droplet, dp, after an evaporation liquid capillary. At very low flow rates (<1 µL/min), this process proceeds time, t, an evaporation rate, s, and an initial with extremely high efficiency to yield highly-charged, sub-micron droplets droplet diameter, d0, is given by that almost instantaneously give rise to gas phase ions due to Rayleigh 2 2 dp = d0 − st (i) disintegration processes2. However, at the high flow rates typically used in LC and UPLC-MS (100–800 µL/min), it is necessary to use a high velocity where s has been experimentally determined as 2 -1 nebulizer gas to atomize the liquid flow, a process which is known to 1250 and 6500 µm s for water and acetonitrile, 4 produce larger droplets with a lower charge per unit volume and hence respectively . According to equation (i), Figure lower ionization efficiency3. 4 shows a plot of initial droplet diameter versus droplet diameter after an evaporation period Whilst the use of a high velocity gas is advantageous from an atomisation of 100µs for water and acetonitrile. From these viewpoint, it has the disadvantage of reducing the residence time of data, it becomes apparent that (i) droplets above droplets between the ESI capillary and the ion inlet orifice of the mass 2 µm in diameter do not undergo significant spectrometer, which, in turn, reduces the time available for droplet evaporation in this time frame and (ii) rapid evaporation which is critical to the Rayleigh disintegration process. evaporation only occurs for diameters below In-house measurements, using Phase Doppler Anemometry (PDA) and the “knee” of the curve, which corresponds Laser Diffraction Particle Sizing (LDPS) reveal that the ESI and UniSpray to approximately 0.4 and 1.0 µm for water nebulizers typically produce initial droplet size distributions that peak at and acetonitrile, respectively. Initial droplets typical diameters of d0=10 µm in the volume domain and d0=1 µm in the diameters below the value d0(crit.) in Figure 4 number domain. Figure 3 shows typical LDPS droplet distributions obtained will completely evaporate in this time period by nebulising a flow of 0.5 mL/min of water under ambient conditions (approximately 0.35 and 0.8 µm for water and using a nebulizer as described above (no target). These distributions reveal acetonitrile, respectively). In essence, this would that although the greatest numbers of droplets are produced at micron suggest that a 100 µs residence time is totally and submicron diameters, the overwhelming volume of the spray, and inadequate for efficient evaporation of the larger hence analyte, is contained within droplets that are significantly greater droplets that dominate the volume distribution than a micron in diameter. PDA measurements also reveal that droplets in high flow rate ESI. While this laminar-flow on the spray axis can have average velocities in excess of 100 ms-1 which evaporation model may be oversimplified, it corresponds to a residence time of <100 µs for the sources used in this would suggest that only sub-micron droplets study (ignoring recirculation effects). This source residence time represents would participate in the production of gas phase the time period during which significant droplet evaporation must occur in ions by the ESI process. Since the sub-micron order to promote the Rayleigh disintegration process. Droplet evaporation population represents <1% of the total sprayed rates have been determined for a number of common solvents and

Waters UniSpray Ionization Source 3 [ WHITE PAPER ]

Figure 4. Evaporation rates for water and acetonitrile droplets for a 100 µs Figure 5. High speed photographs of the break-up of a water droplet on a hot evaporation time. stainless steel plate (from reference5). volume at high flow rates (Figure 3), this would contribute be shown from equations (ii) and (iii) that a water droplet with a to the low ionization efficiency typically observed with high diameter of 4 µm and a velocity of 100 m/s would have a Weber flow rate ESI when compared to nanospray ESI (<1 µL/min), number of 571 and would give rise to 35 visible secondary where submicron-sized droplets contribute significantly to droplets on impact with the hot target. If we consider a the volume distribution. simple, linear break-up model, this would result in secondary droplets with diameters of the order 1.2 µm. More realistically, HIGH WEBER NUMBER DROPLET IMPACTS a skewed Gaussian distribution for the secondary population The above inefficiency argument was the original inspiration would contain sub-micron droplets (equivalent to the invisible for the UniSpray ionization source that compels high velocity droplets that could not be detected by the experimental droplets to impact onto a hot metallic surface which results method used in the above work). As described in the previous in their break-up into smaller secondary droplets that can section, an increase in the sub-micron droplet population could be evaporated more efficiently at low source residence be expected to significantly increase the droplet desolvation times. A number of groups have studied the break-up of process and aid ionization efficiency. water droplets on heated stainless steel surfaces5. These In addition to shifting the droplet size distribution to smaller workers experimentally determined that the number of visible diameters, the impact process may play an additional role secondary droplets (N ) produced per impact was directly vis that may be important for ionising certain analytes that may proportional to the droplet Weber number (W ), which, in turn, e be weakly ionised by ESI due to their low surface affinity. In is directly proportional to the droplet diameter (d) and the ESI, droplets with a high surface affinity (such as detergents) square of its velocity (v): tend to preferentially ionise at the droplet surface and are ejected from the parent droplet as Rayleigh fission processes Nvis = 0.0427 We + 10.465 (ii) occur. Less surface active analytes which may previously 2 We = ρv d⁄σ (iii) have been paired with negative counterions in the core of the parent droplet may now populate the surface. However, the where is ρ the droplet density and σ is the liquid surface surface charge of the parent droplet has become reduced as tension. The Weber number can be regarded as the ratio a result of the fissions which greatly reduces the probability of of droplet kinetic energy to droplet surface energy. Thus, it further fissions and hence ionization of the less surface active stands to reason that the collision of a droplet with kinetic analytes. A prompt and violent breakup of initial droplets energy that far exceeds the droplet surface energy will result that have already partitioned due to their relative surface in instability and droplet break-up. affinity may increase the probability of secondary droplets To illustrate the break-up mechanism, Figure 5 shows a time being formed from the core of the initial droplet provided that lapse sequence of a water droplet (We=630) as it impacts on adequate charging occurs during this process. a planar, polished stainless steel surface which is held at a temperature of 260 °C. In the case of a UniSpray source, it can

Waters UniSpray Ionization Source 4 [ WHITE PAPER ]

When a droplet in a flow of gas approaches a target, the necessary conditions for a collision with the target will be determined by the Stokes number, Stk, and the turbulence of the flow where

Stk = ρvd2 ⁄18µD (iv)

Here, µ is the dynamic viscosity of the gas, D is the diameter of the target and the other terms are as defined in equation (iii). Flow turbulence is governed by the Reynolds number,

Re, which will be described in more detail in the following sections of this paper. Droplets that have Stk≤1 (termed Stokesian particles) tend to follow the gas streamlines whilst those with Stk>1 tend to deviate from the streamlines and strike the target owing to momentum dominance over the drag forces. Figure 6 shows a plot of Stk versus droplet diameter for water droplets with a velocity of 100 m/s and Figure 6. A plot of Stokes Number versus droplet diameter for 100m/s water droplets in a UniSpray source. a UniSpray target diameter of D = 1.6 mm. This shows that small droplets with diameters below dlim.= 2.5 µm would be expected to avoid collisions with the target and would follow THE COANDA EFFECT the gas streamlines. Conversely, larger droplets (d>2.5 µm), The Coanda effect is a phenomenon whereby a fluid flow which account for the majority of the spray volume in a attaches itself to a nearby surface and remains attached even UniSpray source (Figure 3), would be expected to strike as the surface bends away from the initial fluid direction. the target, thus creating smaller secondary droplets. As Figure 7(a) shows a schematic illustration of this effect previously mentioned, the impact efficiency will decrease and an actual photographic image of flow attachment to with increasing flow turbulence. Importantly, it should also the high-voltage cylindrical target in a UniSpray ionization be noted that the smaller droplets that avoid collisions source. This image shows that the gas and droplets from with the target will still benefit from other unique flow the off-axis, nebulizer jet are deflected towards the ion inlet characteristics of the UniSpray source which will be orifice as shown schematically in Figure 1. As a water droplet described in the following sections. evaporates, the gas surrounding the droplet can rapidly Although the impact of high Weber number droplets become saturated with water vapour, which ultimately is believed to be an important feature of the enhanced reduces the rate of evaporation due to re-condensation on sensitivity observed with the UniSpray source, it can be the droplet surface. In API sources, this effect is minimised by experimentally demonstrated that wide area, flat-plate supplying a flow of heated dry nitrogen gas to the nebulised stainless steel targets do not give rise to significant and spray where the nitrogen becomes “entrained” into the stable signal enhancements over ESI. In this respect, it is nebulizer flow due to the low pressure created by the high believed that the geometrical form of the UniSpray source, velocity nebulising jet. In the case of Coanda flow attachment shown schematically in Figure 1, plays an additional role to a curved surface, as occurs at the UniSpray high-voltage in the enhancement of ionization efficiency. In particular, target, it is the imbalance of the entrainment flow where flow it is believed that the curved profile of the target and the cannot penetrate from one side that ultimately leads to a off-axis, perpendicular cross-flow arrangement between deflection of the gas streamlines towards the target surface. the sprayer and the target give rise to two important gas This effect is known to create a stronger total entrainment flow phenomena that may aid the break-up and desolvation when compared to a free jet which could aid the droplet of liquid droplets in the source, viz. the Coanda effect6 and desolvation process by enhanced mixing and hence dilution surface microvortices7. of the water vapor.

Waters UniSpray Ionization Source 5 [ WHITE PAPER ]

Figure 7(b) illustrates the effect of “droplet beam steering” that occurs due to the Coanda effect in UniSpray sources. As the nebulizer jet impact point is adjusted from a central to an off-axis position on the upper right hand quadrant of the Ø1.6 mm target, the ion signal intensity rapidly increases until an optimum is reached. This characteristic is found to be compound dependent but is generally a subtle, as opposed to a critical, tuning parameter that allows an acceptable compromise tuning position to be found for multi-polarity analyte mixtures. Conversely, Figure 7(b) shows that impacting on the upper left hand quadrant of the target results in a severe loss of ion signal as the deflected wake is directed away from the ion inlet cone and ion sampling efficiency is adversely affected. It should be noted that the term beam steering is used as opposed to focussing since the gas streamlines will not cross to form a focal point. It is also anticipated that this flow geometry will give rise to a momentum separation effect where smaller droplets will tend to follow the Coanda flow towards the MS ion inlet whilst the larger non-Stoksian droplets will continue in a forward direction and become separated from the Coanda flow.

Effects that occur in the “wake” of the deflected gas flow could also be beneficial in terms of droplet break up and enhanced droplet desolvation as charged droplets are directed towards the ion inlet cone. One such effect is periodic vortex shedding that occurs behind a cylindrical object where the Figure 8. A typical illustration of van Karmen vortex flow frequency of shedding increases as the velocity of the gas flow increases. behind a cylinder for moderate Reynolds numbers. Figure 8 shows an example of classic vortex shedding that occurs behind a cylinder as the gas flow moves from the top to the bottom of the image. parallel plane in front of, or behind, the plane

The transition between laminar (uniform) flow and turbulent flow in these of Figure 8. However, as Re progressively systems is determined by the Reynolds number (Re) where increases above a value of 2300, the flow becomes turbulent, or three-dimensional. It can R = ρvD⁄µ (v) e be shown that the near-supersonic gas flow at where ρ and µ are the density and viscosity of the gas. At low Re (<10), the the UniSpray target will have a highly turbulent gas will flow uniformly around the cylinder which results in a uniform wake flow wake with a Re of typically 30,000. Under that contains no vortices. As Re is increased to a value of, for example 400, these conditions, it is proposed that primary the flow wake will include a number of vortices as shown in Figure 8. This or secondary droplets will be subjected to flow regime is not considered to be turbulent since the “vortex street” is enhanced desolvation due to increased shear relatively two-dimensional, such that the same vortex pattern exists in a forces, increased entrainment (mixing) and an increased residence time in the turbulent flow.

To understand the importance of turbulence in aiding droplet break-up, we should reconsider the expression for the Weber number (equation (iii) above). In pneumatic nebulisers, liquid droplets are created with low initial velocities,

vl, and are surrounded by a high speed gas

jet with a velocity, vg. It is the difference in these two velocities that gives rise to initial droplet shearing where v2 in equation (iii) is 2 now substituted by (vg−vl) . In a pneumatically- assisted ESI probe, where the gas flow is significantly less turbulent than the gas flow at Figure 7. (a) A schematic and photographic evidence of Coanda gas flow in a UniSpray source; a UniSpray target, the initial droplets reach a (b) The dependence of UniSpray ion signal versus the position of the spray impact point.

Waters UniSpray Ionization Source 6 [ WHITE PAPER ]

terminal velocity at a distance of approximately 5 mm from the tip. At this point, vg−vl becomes zero and the shearing process stops. In comparison, the UniSpray source creates chaotic, turbulent flow over a greater proportion of the droplet in-source trajectory which reduces the probability that vg−vl=0, particularly for the larger, non-Stokesian droplets. In the case of the UniSpray source geometry, it can be shown that the frequency of shedding (fs) is given by

fs = v Sr ⁄D (vi) where v is the freestream gas velocity, D is the cylinder diameter and Sr is the dimensionless Strouhal number. Sr is Figure 9. A schematic representation of the stagnation zone that is formed related to the Reynolds number (Re) by as gas flows over a cylinder.

Sr = 0.198(1−19.7/Re) (vii) This latter point is a further feature that is unique to the cross- For a near-supersonic nitrogen gas flow (v=300 ms-1) and a flow geometry of the UniSpray source and will be considered Ø1.6 mm cylinder, it can be shown that the vortex shedding in more detail in this section. frequency in the wake of the UniSpray target may be as high as 37 kHz (ultrasonic). As an experimental observation, For a cylinder in cross-flow, a uniform gas flow will become it is found that the point of signal optimisation shown in inherently unstable (three dimensional) in the stagnation Figure 7(b) is accompanied by both a visual increase in region where the flow becomes attached to the curved “wake disturbance” and an audible increase in sound from surface. These instabilities take the form of a linear series the UniSpray source. On this latter point, it is not clear of counter-rotating vortices whose axes of rotation are whether significant energy is mechanically imparted into aligned with the streamlines of the gas flow. A single vortex the cantilevered target, but experiments in this laboratory pair is shown schematically in Figure 10(a). The disturbance with ultrasonic nebulisation via a similar 40 kHz agitation wavelength, λ, between adjacent counter-rotating pairs is of a target (no gas) reveal enhanced ionization of non-polar known to be directly proportional to the cylinder diameter analytes that are difficult to ionise by ESI; relatively large (D) and inversely proportional to the square root of the local sensitivity enhancements over ESI for certain non-polar Reynolds number (Re) so that compounds is also a characteristic of UniSpray under λ = κD⁄√Re (viii) certain conditions.

where κ is a constant and Re is defined in equation (v) above. CROSS-FLOW SURFACE MICROVORTICES Kestin and Wood7 have used wind tunnel tests with oil coated As described in the section above, a gas flow will become cylinders to experimentally verify the relationship between λ attached to the surface of a cylinder at the stagnation point and Re and their results are reproduced in Figure 10(b). Using and will detach from the surface at a downstream point the Tu=4% (tubulence intensity) plot and assuming a nebulizer known as the separation point. This is shown schematically velocity of Mach-1 and D=1.6 mm, Figure 10(b) would predict in Figure 9 where the region bounded by the stagnation a disturbance wavelength (λ) of approximately 37 µm for the and separation points is known as the stagnation zone. surface microvortices on a UniSpray target electrode. Firstly, For the near supersonic nebulizer conditions encountered this disturbance dimension is significant in that it is of the with UniSpray, the stagnation zone can be shown to have a same order as the size of the larger droplets and thus could radial thickness of the order 10 µm over which a flow velocity be expected to efficiently impart energy into the droplets to gradient exists from zero at the surface to the freestream aid break-up. This efficiency dependence on the size of the velocity. Primary or secondary droplets that enter this perturbation is analogous to, for example, microwave heating reduced velocity region would be expected to experience where the microwave frequency is chosen to match the natural enhanced desolvation due to (i) an increased residence time, rotational frequencies of water molecules. Secondly, the highly (ii) greater thermal transfer from the hot target surface and mobile smaller droplets that enter the vortex region would be (iii) increased agitation/entrainment with the hot nitrogen flow highly agitated compared to those in the freestream and would due to surface vortex effects. benefit from enhanced mixing and enhanced heat transfer.

Waters UniSpray Ionization Source 7 [ WHITE PAPER ]

Therefore, in a similar manner to the vortex shedding region described in the previous section, this surface boundary layer region represents an additional desolvation enhancement zone that subjects droplets to an increase in residence time, shearing and entrainment.

Figure 11 shows a scanning electron microscope (SEM) image of a 1.6 mm diameter, stainless steel UniSpray target which was used for the analysis of analytes contained in protein-precipitated human plasma. The granular, circular “halo” is due to the deposition of involatile components of the plasma and is not relevant to the present discussion. The SEM image was taken in the same direction as the impinging droplet stream and nebulizer gas jet. The cross (+) in Figure 11 is an approximation of the point of impact, and hence stagnation point, of the incoming nebulizer jet and was located from an optical micrograph of the same target. A close examination of the circled region of the image reveals a linear series of striation marks which are aligned with the direction of the flow streamlines. These striation marks, which are not visible at optical wavelengths, are evidence of the existence of counter-rotating surface vortices as described above. From this image, we can estimate a disturbance wavelength (λ) of 23 µm which assumes that three striation marks represent the outer extent and centre of one counter- rotating vortex pair. Thus, there would appear to be some correlation Figure 11. An electron micrograph of an UniSpray target between the observed experimental data and the theory of surface vorticity showing evidence for the existence of a linear series of for cylinders in cross-flow. surface microvortices at the flow separation point.

Figure 10. (a) A schematic of a counter-rotating vortex pair on the surface of a cylinder with a cross flow of gas; (b) The relationship between disturbance wavelength (λ) and the inverse of the square root of the Reynolds number for a cylinder in cross flow (from reference7).

Waters UniSpray Ionization Source 8 [ WHITE PAPER ]

Figure 12. The effect of a large surface defect on the sensitivity of a Figure 13. A schematic representation of the formation of surface liquid UniSpray source. filaments and secondary droplets at the flow separation point on a UniSpray target.

THE ROLE OF THE TARGET SURFACE target. It can be shown that by rotating the position of the In this paper, we have sought to provide an explanation of groove with respect to the stagnation region, significant how the unique geometry of the UniSpray source gives rise sensitivity decreases are observed when the groove overlaps to a number of hydrodynamic and aerodynamic phenomena the stagnation zone. Figure 12 shows the effect of target that may aid in the nebulisation and desolvation of primary groove position on the relative signal intensity for a UniSpray- droplets from the liquid capillary. Whilst these hypotheses MS analysis of busiprone and reserpine which were infused may be supported by the observation of a greater increase into the source at a concentration of 0.125 pg/µL and a flow in sensitivity over ESI for high flow rates and highly aqueous rate of 0.8 mL/min. Referring to Figure 12, the highest signal mobile phases, where nebulisation and desolvation are intensity is observed when the groove is positioned well away particularly difficult, it is clear that a full understanding of from the stagnation zone (upper right hand quadrant). The these effects is far from complete. The processes described lowest sensitivity is observed when the groove completely thus far are highly interdependent and highly complex from overlaps the upper right hand quadrant, where presumably, a modelling perspective. In particular, the exact role of the the stagnation region is overwhelmed by turbulence such that metallic surface in the flow stagnation region is not a clear definition between a stagnation zone and freestream fully understood. flow no longer exists. The two additional reference points for busiprone and reserpine were obtained from a different It is known that any significant damage (gouging) to the 1.6 mm-diameter target which contained no groove. Whilst target surface in the stagnation region will severely affect this experiment does not distinguish between the relative the UniSpray source performance. Referring to Figure 11, if importance of the dominant processes described in this we assume that the cross represents the location of the flow paper, it further reinforces the hypothesis that the curved stagnation point and the end of the striation marks represent surface is central to the enhanced signal and ionization the flow separation point, we can determine from a simple efficiency observed with the UniSpray source. geometric projection that the UniSpray target stagnation zone subtends a radial angle of approximately 46 degrees. Throughout this paper, we have considered the dominant For a 1.6 mm diameter target, this equates to a stagnation ionization mechanism to be based on an electrospray-type zone that has a 0.65 mm circumferential length. Since the process where gas phase ions are created by evaporating effects of Coanda steering, surface microvorticity and vortex charged liquid droplets. Whilst some droplet charging may shedding are all associated with the formation of a stagnation originate at the point of nebulisation, it is likely that charging zone, one would assume that any gross interference will occur at the point of impact on the high voltage target via with this region would have detrimental effects on the a process that resembles (i) electrospray charge separation, 8 performance of the UniSpray source. To test this hypothesis, (ii) spray electrification or (iii) statistical charging such as 9 10 an experiment was conducted where a surface groove, with observed in sonic spray or thermospray ionization . It is an equivalent width to the stagnation length (0.65 mm), was plausible that an explanation of all experimental data obtained cut longitudinally into the 1.6 mm diameter stainless steel over a wide range of analyte classes, mobile phases and flow

Waters UniSpray Ionization Source 9 [ WHITE PAPER ]

rates cannot be attributed to a single ionization mechanism. Referring to References Figure 2, it is known that the general behaviour of the UniSpray source 1. A. Lubin, S. Bajic, D. Cabooter, P. Augustijns, and F. undergoes a transition at flow rates between 100 and 200 µL/min, which Cuyckens, J. Am. Soc. Mass Spectrom., 28, 286-293 (2017). results in distinct differences in the gain versus mobile phase composition 2. P. Kebarle, J. Mass Spectrom., 35, 804-817 (2000). plots for the low flow rate data. In view of this observation, an alternative 3. R. Juraschek, T. Dulcks, and M. Karas, J. Am. Soc. model could be considered where the actual generation of charged droplets Mass Spectrom., 10, 300-308 (1999). occurs due to wetting of the target surface and the subsequent effects of 4. R.L. Grimm and J.L. Beauchamp, Anal. Chem., 74, the high velocity gas flow on the wetted zones. A schematic of this process 6291-6297 (2002). is shown in Figure 13 where the size of the wetted region is massively 5. S.W. Akhtar, G.G. Nasr, and A.J. Yule, Atomization exaggerated. Wetting is known to occur on the underside of the cylindrical and Sprays, 17, 659-681 (2007). target and is particularly prevalent at high flow rates and with highly 6. A. Dumitrache, F. Frunzulica, and T.C. Ionescu, Mathematical Modelling and Numerical aqueous mobile phases. The surface liquid could exist just below the Investigations on the Coanda Effect, Chapter 5 of flow separation point shown in Figures 9 and 13 where the local surface Nonlinearity, Bifurcation and Chaos - Theory and gas flow is stagnant but the flow velocity is high at a small distance (>10 µm) Applications, ISBN 978-953-51-0816-0 (2012). from the surface. This gas flow could create secondary droplets from the 7. J. Kestin and R.T. Wood, J. Fluid Mech., 44, 461-479 (1970). resulting liquid filaments shown in Figure 13. In fact, it is conceivable 8. L.B. Loeb, Static Electrification, ISBN 978-3- that the surface liquid at this junction could be “squeezed” between the 642-88243-2 (1958). counter-rotating microvortices (Figure 10(a)) to form a linear series of 9. A. Hirabayashi, M. Sakairi, and H. Koizumi, Anal. secondary droplet emitters which may be responsible for the striation Chem., 66, 4557-4559 (1994). marks shown in Figure 11. Although this model differs from one based on 10. V. Katta, A.L. Rockwood, and M.L. Vestal, Int. J. Mass Spectrom. Ion Processes, 103, 129-148 (1991). high We-number droplet impacts, it does however depend on the same aerodynamic principles outlined in the previous sections.

Although not covered in any detail here, it is also believed that the nebulizer/target gap current and its effect on the charging of both the target surface and gas phase molecules is another important parameter that will influence the performance of a UniSpray source. In fact, experiments with an ambient UniSpray source in this laboratory have shown that gas phase ion/molecule reactions are highly likely to occur under these operating conditions. Furthermore, such reactions can be expected to benefit from the enhanced desolvation processes that occur with the UniSpray geometry, as highlighted in the previous sections.

Waters Corporation 34 Maple Street Milford, MA 01757 U.S.A. Waters and The Science of What’s Possible are registered trademarks of Waters Corporation. UniSpray is a trademark of Waters Corporation. All other trademarks are the property of their respective owners. T: 1 508 478 2000 F: 1 508 872 1990 ©2017 Waters Corporation. Produced in the U.S.A. June 2017 720006009EN KP-PDF www.waters.com