Dynamic surface analysis for acoustic droplet ejection

Abstract

A system for ejecting droplets from a surface of a sample using acoustic energy includes: a processor; a transducer configured to receive electrical energy and emit corresponding acoustic energy, and further configured to receive acoustic energy and generate corresponding electrical energy; a transmitter in communication with the processor and transducer, wherein the transmitter is configured to receive at least one signal from the processor and transmit corresponding electrical energy to the transducer; and a receiver in communication with the processor and transducer, wherein the receiver is configured to receive the electrical energy generated by the transducer and transmit at least one corresponding signal to the processor, wherein the transducer is configured to emit a first acoustic signal, a second acoustic signal, and a third acoustic signal, wherein the first acoustic signal is determined to cause a mound phase during which a droplet is formed at the surface of the sample, the second acoustic signal is selected to interrogate the surface of the sample during the mound phase, and the third acoustic signal is selected to further interrogate the surface of the sample during the mound phase.

Description

[0001]Attorney Docket No.22US0151-WO (66738WO01) Electronically filed on: October 24, 2023 TITLE DYNAMIC SURFACE ANALYSIS FOR ACOUSTIC DROPLET EJECTION CROSS REFERENCE TO RELATED APPLICATIONS This application claims priority to and the benefit of U.S. Prov. No.63 / 448,075, filed on February 24, 2023, the entirety of which is incorporated by reference herein. BACKGROUND Acoustic droplet ejection (ADE) is a technology that uses acoustic energy to move a liquid without any physical contact. Acoustic energy (e.g., in the form of ultrasonic pulses) is emitted from a transducer towards a volume of liquid (hereinafter, “sample”). In some examples, the beam converges on or near the upper surface of the sample, and the acoustic energy is transferred to a portion of the sample, thereby causing this portion to move upwardly away from the remainder of the sample (e.g., as a droplet). The sample may be contained in a well of a container (e.g., 96- or 384-well microplate). SUMMARY According to embodiments, a system for ejecting droplets from a surface of a sample using acoustic energy includes; a processor; a transducer configured to receive electrical energy and emit corresponding acoustic energy, and further configured to receive acoustic energy and generate corresponding electrical energy; a transmitter in communication with the processor and transducer, wherein the transmitter is configured to receive at least one signal from the processor and transmit corresponding electrical energy to the transducer; and a receiver in communication with the processor and transducer, wherein the receiver is configured to receive the electrical energy generated by the transducer and transmit at least one corresponding signal to the processor, wherein the transducer is configured to emit a first acoustic signal, a second acoustic signal, and a third acoustic signal, wherein the first acoustic signal is determined to cause a mound phase during which a droplet is formed at the surface of the sample, the second acoustic signal is selected to interrogate the surface of the sample during the mound phase, and the third acoustic signal is selected to further interrogate the surface of the sample during the mound phase. The second acoustic signal and the third acoustic signal may be part of a plurality of interrogation pings, wherein the transducer is configured to emit the interrogation pings at intervals of between approximately 50 to 100 µS. The plurality of interrogation pings may include between approximately 4 to 2000 pings. The processor may be configured to receive information corresponding to reflections of the second acoustic signal and the third acoustic signal from the surface of the sample, wherein the processor is configured to determine at least one characteristic of the mound phase based at least in part on said information. The at least one characteristic of the mound phase may include at least one of a droplet size, the droplet linear velocity, the droplet angular velocity, the droplet radial displacement, the droplet breakoff position, the surface position, the surface velocity, the surface acceleration, a mound formation rate, or the mound relaxation rate. The transducer may be further configured to emit a fourth acoustic signal, wherein the fourth acoustic signal is determined to cause a second mound phase during which a second droplet is ejected from the surface of the sample, and wherein the processor is further configured to determine the fourth acoustic signal according to the at least one characteristic of the mound phase. The first acoustic signal and the fourth acoustic signal may comprise at least one of a different power, frequency, or duration. Attorney Docket No.22US0151-WO (66738WO01) According to embodiments, a system for ejecting droplets from a surface of a sample using acoustic energy includes: a processor; a transducer configured to receive electrical energy and emit corresponding acoustic energy, and further configured to receive acoustic energy and generate corresponding electrical energy; a transmitter in communication with the processor and transducer, wherein the transmitter is configured to receive at least one signal from the processor and transmit corresponding electrical energy to the transducer; and a receiver in communication with the processor and transducer, wherein the receiver is configured to receive the electrical energy generated by the transducer and transmit at least one corresponding signal to the processor, wherein the processor is configured to cause the transducer to emit a droplet-forming signal to cause a mound phase during which a droplet is ejected from the surface of the signal, and to further emit a plurality of pings at least one of before, during, or after the mound phase, wherein the processor is further configured to receive information from the receiver corresponding to reflections of the plurality of pings, and wherein the processor is further configured to determine at least one characteristic of the mound phase according to the information received from the receiver corresponding to reflections of the plurality of pings. The processor may be further configured to cause the transducer to emit a second droplet-forming signal to cause a second mound phase, during which a second droplet is ejected from the surface of the sample, wherein the processor is further configured to determine the second droplet-forming signal based at least in part on the information received from the receiver corresponding to reflections of the plurality of pings. The second droplet-forming signal may be determined according to a machine-learning model. The droplet-forming signal and the second droplet-forming signal may be determined such that the first droplet and the second droplet have at least one of a different volume or separation velocity. The at least one characteristic of the mound phase may include at least one of a droplet size, the droplet linear velocity, the droplet angular velocity, the droplet radial displacement, the droplet breakoff position, the surface position, the surface velocity, the surface acceleration, a mound formation rate, or the mound relaxation rate. The transducer may be configured to emit the plurality of pings at intervals of between approximately 50 to 100 µS. The plurality of pings may include between approximately 4 to 2000 pings. According to embodiments, methods are provided for implementing any of the foregoing system embodiments. BRIEF DESCRIPTION OF SEVERAL VIEWS OF THE DRAWINGS FIG.1 shows a representation of an ADE system, including a cross-sectional view a container plate including a plurality of containers (or wells) holding a respective plurality of samples, receiver plate, a transducer assembly, and a block diagram of electronic circuitry. FIG.2 shows a block diagram of a transducer assembly. FIG. 3 shows a representation of movement of the transducer assembly relative to a container plate when performing ADE on multiple samples. FIG.4 shows a top view of a container plate with a plurality of containers. FIG.5 shows a top view of a plurality of containers in a container plate and a flow illustrating a sequence for serially performing ADE on each container. FIG.6 shows a container holding a sample. FIG.7 shows a reflected signal and corresponding envelope, where the reflected signal is received at transceiver to embodiments. FIG.9A illustrates reflected signal intensities and timing from the surface of a sample during ADE. Attorney Docket No.22US0151-WO (66738WO01) FIG. 9B illustrates reflected signal intensities and timing from the surface of a sample during ADE, wherein certain ADE parameters have been adjusted according to embodiments. FIG.10A illustrates how data can be selected from collections of data, such as the collection shown in FIGS. 9A or 9B, to generate the graph shown in FIG.11A, according to embodiments. FIG.10B illustrates how data can be selected from collections of data, such as the collection shown in FIGS. 9A or 9B, to generate the graph shown in FIG.11B, according to embodiments. FIG.10C illustrates how data can be selected from collections of data, such as the collection shown in FIGS. 9A or 9B, to generate the graph shown in FIG.11C, according to embodiments. FIG.10D illustrates how data can be selected from collections of data, such as the collection shown in FIGS. 9A or 9B, to generate the graph shown in FIG.11D, according to embodiments. FIGS.11A, 11B, 11C, and 11D are graphs showing information gathered about ejected droplets, according to embodiments. FIG.12A illustrates training a machine learning model, according to embodiments. FIG.12B illustrates using a trained machine learning model in an ADE system, according to embodiments. The foregoing summary, as well as the following detailed description of certain techniques of the present application, will be better understood when read in conjunction with the appended drawings. For the purposes of illustration, certain techniques are shown in the drawings. It should be understood, however, that the claims are not limited to the arrangements and instrumentality shown in the attached drawings. Furthermore, the appearance shown in the drawings is one of many ornamental appearances that can be employed to achieve the stated functions of the system. DETAILED DESCRIPTION It may be helpful to make run-time adjustments to ejection parameters for ADE resulting in improved performance (e.g., improved control over drop placement or volume). In conventional systems, user-selected calibrations may be used to influence ADE settings such as the power and frequency for ejecting a droplet, as well as the volume(s) of the droplets being formed. Developing calibration profiles and programming an instrument may be relatively slow and expensive, requiring, for example, thousands of man hours to develop calibrations and tune individual systems throughout an instrument’s life cycle. Furthermore, such calibration profiles may be relatively static, and may lose accuracy / applicability if there are changes in the system, such as drift over time. Furthermore, adjusting ADE parameters during a given transfer may require interrupting the train of droplet ejections. This may disrupt the fluid surface behavior that has been set up in accordance with the operating ejection frequency, and may cause chaotic behavior in the system resulting in worsening performance. Pausing the drop train may also delay ADE processing, and may negatively impact workflows. Techniques described herein enable making in-train acoustic measurements that will provide information useful for adjusting ADE settings during an ejection train. This may result in improved throughput and performance. Techniques described herein enable monitoring and analysis of an ADE system in real time by gathering performance data during the droplet ejection process. Information may be gleaned that enables rapid and effective adjustment of ADE operating parameters, without undue delay. According to techniques described herein, performance data may be gathered before, during, and / or after droplet ejection. Such data may be used in accordance with machine learning algorithms, as well as conventional heuristic models. This may facilitate effective adjustment of ADE system Attorney Docket No.22US0151-WO (66738WO01) operating parameters. Such parameters may include toneburst power, ejection frequency, droplet volume, and / or toneburst configuration. FIG.1 depicts an example ADE system 100, including a cross-sectional view of a container plate 120 (e.g., a microplate) including a plurality of containers 122 (e.g., wells of a microplate) holding a respective plurality of samples 101, receiver plate 130 including a plurality of receiver wells that receive ejected liquid 102 from sample 101, and a block diagram of electronics 140. ADE system 100 further includes transducer assembly 110, a coupling liquid 160, X / Y / Z motors 150, and temperature sensors (not shown). FIG.2 further shows transducer assembly 110, including a transducer 112 and acoustic lens 113. ADE system 100 can determine characteristics of both containers 122 and samples 101, as well as cause liquid to be ejected. A sample 101 is a liquid of interest that is held within a particular container 122. Although this disclosure focuses on containers that are wells of microplates, techniques described herein can be used with other containers such as tubes, flasks, and beakers, as well as any samples contained therein, as will be recognized. In order to cause ejected liquid 102 to be ejected from sample 101, transducer 112 generates acoustic energy (e.g., ultrasonic energy), which is focused by acoustic lens 113 into beam 170. In the figures, beam 170 is shown in two dimensions, but it is understood that beam 170 is three dimensional. Furthermore, while beam 170 is shown as a perfect triangle, in practice, beam 170 can have different shapes. It may possible to vary the height of the focal point of acoustic energy beam 170 by adapting the configuration of transducer assembly 110. For example, the focal length of beam 170 may be changeable by adapting transducer assembly 110. Such a transducer assembly 110 is described in U.S. Appl. No. 16 / 369,780 (U.S. Publ.2019 / 0302063), which is herein incorporated by reference in its entirety. It may also be possible to change the height of the focal point by moving transducer assembly 110 along the z-axis (i.e., the vertical dimension between container 122 and transducer assembly 110. As depicted in FIG.1, beam 170 is focused on the upper surface of sample 101, which may be at the interface between the sample 101 and the air above the sample 101. First, beam 170 passes through coupling liquid 160, a bottom wall 123 of container 122, and then the depth of sample 101 to reach the surface 103 of sample 101. Electronic circuitry 140 includes a processor 143, a motor controller 142, transmit signal circuitry 144, receive signal circuitry 145, and temperature sensor circuitry 141. Although shown as separate components for explanatory purposes, portions of electronics 140 may be combined or integrated. Furthermore, some components shown may include multiple different subcomponents not specifically shown. For example, processor 143 may include multiple processors, either located together in a single chip or distributed in different locations. Processor 143 causes or controls transmit signal circuitry 144 to generate an analog electrical signal (electronic transmission signal, such as a radio frequency signal), which is communicated to transducer 112. Transducer 112 then vibrates in response to the analog signal (amplitude and frequency), such that a corresponding acoustic signal is emitted. Transducer assembly 110 may also receive acoustic signals (e.g., acoustic signals reflected from container 120 and / or samples 101 in response to the emitted acoustic signal) and vibrate responsively. This vibration may generate an analog electrical signal (electronic reception signal), which is then communicated to receive signal circuitry 145. Processor 143 may receive information corresponding to the received acoustic signals from the receive signal circuitry 145 in the form of an electronic reception signal. The information in the electronic reception signal will be analyzed by processor 143. Processor 143 can also communicate with motor controller 142 to control the location of transducer assembly to and in Attorney Docket No.22US0151-WO (66738WO01) order control the relative movement between transducer assembly 110, container plate 120, and / or receiver plate 130. Processor 143 can control one or motors 150 to position transducer assembly 110 underneath a given container 122 in container plate 120 (e.g., transducer assembly 110 is centered with respect to a center of given container 122), and then to move transducer assembly 110 underneath another given container 122 in container plate 120. In some embodiments, the ADE system 100 may include temperature sensor(s) (not shown) that can be located in coupling liquid 160, in a region between container plate 120 and receiver plate 130, or other locations. Temperature sensor circuitry 141 receives signals (e.g., electrical or wireless) from temperature sensor(s), and communicates with processor 143 such that temperature(s) (e.g., of coupling liquid 160, containers 122, samples 101, air temperature) can be measured. In some embodiments, transducer assembly 110 can have a cylindrical shape. In some examples, instead of using a single transducer 112 to both transmit and receive acoustic signals, transducer assembly 110 may include separate transmitter and receiver transducers, for example, as disclosed in U.S. Patent No.10,787,670, which is herein incorporated by reference in its entirety. According to one technique, receiving transducer can substantially surround the transmitting transducer and acoustic lens. FIG. 3 shows a representation of movement of transducer assembly 110 relative to container plate 120 when performing ADE on multiple samples 101. Transducer assembly 110 is moved from container-to-container 122 along the x-axis. Transducer assembly 110 can also move along the y-axis to additional containers 122 (not shown) as further described with respect to FIG.5. For each container 122, transducer assembly 110 may be centered underneath container 122. Transducer assembly 110 may move vertically along the z-axis to emit and receive acoustic signals at different z- positions beneath container 122. Transducer assembly 110 can be positioned along the z-axis to focus beam 170 on the surface 103 of sample 101 to cause ejected liquid 102 to be ejected (ADE). As further described below, transducer assembly 110 can be positioned along the z-axis to focus beam 170 at a predetermined height with respect to container plate 120 to destroy bubbles without performing ADE, for example, destroying bubbles that are located at the interface between container 122 and sample 101. FIG.4 shows a top view of container plate 120 having a plurality of containers 122. Container plate 120 shown is a 384-well microplate (e.g., a polypropylene microplate, designated 384-PP). FIG. 5 a top view of a plurality of container wells 122 and an example pattern (a serpentine pattern) for performing ADE and / or other ultrasonic techniques (such as bubble destruction) on each container well 122 and sample 101 therein, as described with respect to FIG.3. In this example, motors 150 move the transducer assembly 110 along the x- and y-axes to position it under various container wells 122. Any other suitable pattern may be used (e.g., a raster pattern). It may also be possible to move container plate 120 or a combination of container plate 120 and transducer assembly 110 to achieve similar effects. FIG.6 shows container 122 (or well) holding sample 101, which has a surface 103 (i.e., upper surface or free surface). Bottom wall 123 of container 122 defines a top-of-the-bottom interface (TB) 124 and a bottom-of-the-bottom interface (BB) 125. FIG.7 shows a reflected signal and corresponding envelope, where the reflected signal is received at transducer assembly 110, in response to an emitted signal emitted by the transducer assembly 110. BB indicates a reflection from BB 125. TB indicates a reflection from TB 124. SR indicates a reflection from surface 103 of sample 101. The overall time at which the reflections appear are referred to as time of flight, or ToF. For each reflection BB, TB, SR in FIG.7, there is a different, respective ToF. FIGS.8A, 8B, 8C, and 8D illustrate a sequence of transmissions and reflections of acoustic signals, according to embodiments. These figures are depicted as a sequence. At FIG. 8A, a mound formation phase is initiated by Attorney Docket No.22US0151-WO (66738WO01) transmitting a first acoustic signal from the transducer towards the surface 103 of a sample 101. The first acoustic signal (and other acoustic signals depicted) may be transmitted by a transducer assembly (not shown in FIGS.8A-8D), such as transducer assembly 110. The emitted acoustic energy beam may be substantially focused at the surface 103. In FIG. 8A, the surface 103 may be substantially at equilibrium prior to transmission of the first acoustic signal. The first acoustic signal may be a droplet-forming signal, and may sometimes be referred to, but need not necessarily be, a toneburst. A droplet-forming signal may include multiple frequencies and may be of relatively long duration. Examples of such frequencies and durations are disclosed in U.S. Pat. Nos.10,112,212 and 10,325,768, which are incorporated by reference herein in their entireties . A droplet-forming signal may be sufficient to cause a droplet to be ejected from the surface 103, but droplet ejection is not required. A droplet-forming signal may be sufficient to cause a droplet to form, but the droplet may not necessarily separate from the surface 103. The surface 103 need not be at equilibrium, but it is shown in FIG. 8A as such for illustrative purposes. The energy of the first acoustic signal may be such that a mound begins to form in the surface 103, as shown, for example, in FIG.8B. If the energy of the first acoustic signal is sufficient (or the amount of energy transferred to the surface 103 is sufficient), a droplet may be ejected. The formation of the mound may be the beginning of a mound phase, which may last until the surface 103 returns to equilibrium. A droplet may or may not be ejected during a mound phase. At FIG.8B, a second acoustic signal is transmitted towards the surface 103, and a portion of the energy of the second acoustic signal is reflected back towards the transducer assembly (not shown). The second acoustic signal may be of lesser energy than the first acoustic signal for the second acoustic signal). Examples of such energies are disclosed in U.S. Pat. Nos.10,112,212 and 10,325,768. The second acoustic signal may be relatively shorter in duration than the first acoustic signal. The second acoustic signal may have different frequenc(ies) than the first acoustic signal. For example, the second acoustic signal may not substantially influence mound formation and behavior. Instead, the second acoustic signal may be a “ping,” and may serve to gather information of one or more characteristics of the sample during the mound phase. The second acoustic signal may be reflected by different features in the sample, and the intensities of these reflections (e.g., amplitude of measured reflections) may be measured and recorded along with the associated time of flight information. For example, the ADE system (e.g., system 100) may determine the time of flight between the surface 103 and the transducer (e.g., transducer 112) and the intensity of the reflection at the surface 103. The time of flight indicates distance from the transducer 112. For example, there may be a relatively strong reflection (a peak in the reflected signal) from the surface 103. It may then be possible to determine a height of the surface 103 at a given location at a given point in time during the mound phase. In addition to time of flight and intensity information, reflections from a ping may also include frequency and / or related phase information. Such information may be extracted, for example, by performing a Fast Fourier Transform (FFT) on the time domain surface reflection received by the transducer. This FFT returns a power spectrum across a frequency range of roughly 5-20MHz, for example, that can be used to extract additional information about the fluid surface shape and kinetics In addition to reflections from the surface 103, the reflection of a ping may indicate reflection(s) from other features in a sample. As will be further discussed, the ADE system may be able to determine characteristics about the surface and droplet ejection based on information determined from the reflection of one or more pings. Such characteristic(s) may include one or more of droplet size, droplet linear velocity, droplet angular velocity, droplet radial displacement, droplet breakoff position, droplet surface position, droplet surface velocity, the surface acceleration, a mound formation rate, or the mound relaxation rate. At FIG. 8C, a third acoustic signal is transmitted towards the surface 103, and a portion of the energy of the third acoustic signal is reflected back towards the transducer assembly (not shown). The third acoustic signal may be of lesser energy than the first acoustic signal, and may be of similar energy as the second acoustic signal. For example, the Attorney Docket No.22US0151-WO (66738WO01) third acoustic signal may not substantially influence mound formation and behavior. Instead, the third acoustic signal may be similar to the second acoustic signal and may be a ping, and may serve primarily to gather information of one or more characteristics of the surface 103 (and / or droplet formation) during the mound phase. As with the second acoustic signal, the ADE system may determine the time of flight between the surface 103 and the transducer, the reflected signal intensity, frequency information, and / or phase information related to the frequency information. The third acoustic signal may optionally differ from the second acoustic signal. For example, the third acoustic signal may be varied or adapted from the second acoustic signal, depending on the information received in the reflection of the second acoustic signal. As shown in FIG.8C, the mound has continued to grow in surface 103, and a droplet has begun to form. The reflection of the third acoustic signal may be processed, and information may be determined about the changing characteristic(s) of the mound. Additional relatively low-energy acoustic signals (e.g., pings) may be emitted, similar to the second acoustic signal and the third acoustic signal. Each of these signals may be emitted at, optionally, a periodic interval (e.g., between 50-100 µS, such as 60 µS). For example, in an example where the mound phase is 0.96 mS, it may be possible to transmit sixteen of pings at 60 µS intervals. As an example, the duration of an entire reflection from a ping may be between approximately 30-40 µS. A subsequent ping may be emitted after the transducer receives the entire reflection. It may be possible to transmit, for example, up to 2000 pings or possibly more, depending on the duration of the mound phase and the rate at which pings are transmitted. The second, third, and possible subsequent pings may be used to gather data about the surface 103 during the mound phase. As shown in FIG.8D, the mound phase has ended and surface 103 is at equilibrium again. The surface 103 need not be at equilibrium, but it is shown as such for illustrative purposes. A fourth acoustic signal (e.g., droplet-forming signal or toneburst) is transmitted towards the surface 103. The fourth acoustic signal may be similar to the first acoustic signal, in that it has sufficient energy to cause mound formation (and possibly droplet ejection). However, the fourth acoustic signal may be different from the first acoustic signal based on information determined by the ADE system from the reflection of the second acoustic signal and the reflection of the third acoustic signal (and possibly additional lower- energy acoustic signals). For example, characteristics of the fourth acoustic signal may be determined at least in part by the characteristic(s) of the surface 103 or other aspects of the sample determined in conjunction with the low-energy pings. For example, the first acoustic signal and the fourth acoustic signal may have at least one of different power, frequenc(ies), associated phase(s), and / or duration than previous droplet-forming signals or tonebursts. As a result of the different signals, it may be possible to eject droplets having different characteristics at different times during a sequence, such as droplets with differing droplet volume and / or separation velocity. FIG. 9A illustrates measurements made during ADE in response lower-energy acoustic signals (e.g., pings), such as the second acoustic signal discussed in context of FIG.8B, and the third acoustic signal discussed in context of FIG. 8C. Each row of pixels, starting from the top and advancing downwards indicates information determined from low-energy acoustic pings transmitted before, during, and after the mound phase. Each row of pixels represents information gathered from a single ping, where the pings are successively transmitted at 60 µS intervals. Therefore, the Y-axis represents time as pings are successively transmitted in sequence downwardly along the Y-axis. The X-axis indicates the time-of-flight for the pings. At one particular row where there is a spike in time-of-flight, it can be seen that a droplet is forming and / or the droplet is being ejected and the signal extends to the far right of the graph. The brighter pixels (Z-axis) indicate more reflected energy received from a given ping at a given time during the time of flight at the ADE system. The yellow pixels indicate higher reflected energy, and the blue pixels indicate lower reflected energy (for example, the deepest blue indicates the least reflected energy). As shown, the surface of the mound (indicated by the last significant reflection from a given ping) rises up until the point of droplet formation and / or ejection, and then begins to sink back down such that the surface returns to equilibrium. As discussed above, the reflected signal may include more information, such as frequency and / or associated phase information (for one or more frequencies). This information is Attorney Docket No.22US0151-WO (66738WO01) not shown in FIG.9A or 9B, but it is understood that such information can be determined from the reflected signal and stored and further processed. In FIG. 9A there is a discontinuity towards the top of the graph, because the timeline along the Y-axis is discontinuous. This discontinuity corresponds to when the system emits the power signal (e.g., first acoustic signal, described in FIG.8A) to cause the mound to form. For this interval, it may not be possible to transmit a ping, because the transducer may be being used to transmit the higher energy signal. FIG. 9B is similar to FIG. 9A, but the measurement conditions are different. FIGS. 9A and 9B show two examples of how pings can be used to measure characteristic(s) of a sample, and particularly the surface 103. The gathered data from the pings can be used to develop models, such as machine-learning models, that can be used to improve instrument performance. FIG.10A illustrates one example of how data captured by the ADE system from reflections of the pings may be assessed, heuristically. FIG. 10A shows a collection of data, such as what is shown in FIGS. 9A or 9B, where the reflections of two pings (ping 6 and ping 9 along the Y-axis) are highlighted. In each reflection, a point is designated – point A for ping 6 reflection and point B for ping 9 reflection. The designation of points A, B is arbitrary in this case, and is only for illustrative purposes. The points could be designated for other ping reflections or other times of flight in a given ping reflection. The points could be designated before, during, or after mound formation. The graph in the lower left corner of FIG.10A corresponds to the ping 6 reflection, with amplitude on the Y-axis (corresponding to the Z-axis (color, as depicted in U.S. Prov. No.63 / 448,075, filed on February 24, 2023, incorporated herein) of the graph above) and the time of flight on the X-axis. Point A is again depicted. The graph in the lower right corner of FIG. 10A corresponds to the ping 9 reflection, with amplitude on the Y-axis (corresponding to the Z-axis (color, as depicted in U.S. Prov. No.63 / 448,075, filed on February 24, 2023, incorporated herein) of the graph above) and the time of flight on the X-axis. The ratio of A and B may be determined not for just one mound phase, but for thousands or millions or any plausible number of mound phases. For example, one or more parameters of the droplet-forming signal may be varied (e.g., power, duration, frequenc(ies), and / or related phase information) in a given droplet-forming signal, and the ratio of A / B may be determined. Further, features of the sample itself may be varied, such as sample composition and / or surface height. Again, the designation of A and B may be arbitrary. FIG. 10B is similar to FIG. 10A, except now points C and D are designated. The ratio of C and D may be determined not for just one mound phase, but for thousands or millions or any plausible number of mound phases. For example, one or more parameters of the droplet-forming signal may be varied (e.g., power, duration, frequenc(ies), and / or related phase information) in a given droplet-forming signal, and the ratio of C / D may be determined. Further, features of the sample itself may be varied, such as sample composition and / or surface height. Again, the designation of C and D may be arbitrary. FIG. 10C is similar to FIG. 10A, except now points E and F are designated. In this instance, points E and F indicate data responsive to the same ping, rather than different pings as shown in FIGS.10A and 10B. The ratio of E and F may be determined not for just one mound phase, but for thousands or millions or any plausible number of mound phases. For example, one or more parameters of the droplet-forming signal may be varied (e.g., power, duration, frequenc(ies), and / or related phase information) in a given droplet-forming signal, and the ratio of E / F may be determined. Further, features of the sample itself may be varied, such as sample composition and / or surface height. Again, the designation of E and F may be arbitrary. Attorney Docket No.22US0151-WO (66738WO01) FIG. 10D is similar to FIG. 10A, except now points G and H are designated. The ratio of G and H may be determined not for just one mound phase, but for thousands or millions or any plausible number of mound phases. For example, one or more parameters of the droplet-forming signal may be varied (e.g., power, duration, frequenc(ies), and / or related phase information) in a given droplet-forming signal, and the ratio of G / H may be determined. Further, features of the sample itself may be varied, such as sample composition and / or surface height. Again, the designation of G and H may be arbitrary. For some or each of these mound phases, the surface of the sample and / or characteristics of any ejected droplet may be measured with one or more different systems. For example, a phase-Doppler interferometer may measure characteristics of an ejected droplet, including horizontal velocity on an X-axis substantially parallel to the surface of the sample, horizontal velocity on a Y-axis substantially parallel to the surface of the sample, vertical velocity on a Z-axis substantially perpendicular to the surface of the sample, the placement of a droplet (e.g., the trajectory of a droplet, including the radial displacement of the droplet), and / or the angular velocity of the droplet. Such a phase-Doppler interferometer may employ three positioned lasers and corresponding photodetector arrays. As another example, the measurements from reflected pings at the ADE system may measure additional characteristics, such as the vertical (Z- axis) droplet breakoff position, the velocity of the surface, the acceleration of the surface, the mound formation rate, and / or the mound relaxation rate. As another example, the measurements from a camera (e.g., visible wavelength camera) may measure characteristics of an ejected droplet and / or surface, such as the shape of the mound formed by the acoustic pressure wave, the shape and trajectory of the droplet formed, the presence of secondary satellite drops being formed, and the overall fluid surface behavior harmony the toneburst ejection frequency. FIGS.11A, 11B, 11C, and 11D show graphs illustrating correlations between regions of mound phase images similar to FIGS. 9A and 9B. Each data point corresponds to a given mound phase evaluation. The data processing associated with generating FIGS. 11A-11D (and analogous compilations of data) may be performed by a processor, such as processor 143 in the ADE system 100. Or data processing may be performed by other processors in combination with or separately from processor 143. For example, as discussed below, data may be gathered from other systems, such as phase-Doppler interferometer system(s) and camera(s). Those other systems may generate data that is stored and processed separately from the ADE system 100. Such separate processing may also receive data generated by the ADE system 100, including acoustic transmission and reception data. Together, this data from different systems may be used to train a machine learning model using one or more algorithms. Once the machine learning model for an ADE system has been developed through training, the final model may be stored on the ADE system 100. The model may be accessed or exploited by processor 143. For example, processor 143 may input information associated with transmission and reception of a first acoustic signal. Based on that information, the machine learning model may suggest different transmission parameters for subsequent droplet-forming signal(s). The graphs of FIGS. 11A-11D show correlations of the droplet ejection dynamics. For FIG.11A, the X-axis shows the ratio of A / B, as defined in FIG.10A. For FIG.11B, the X-axis shows the ratio of C / D, as defined in FIG.10B. For FIG.11C, the X-axis shows the ratio of D / E, as defined in FIG.10C. For FIG. 11D, the X-axis shows the ratio of G / H, as defined in FIG.10A. For each graph, the color of a point indicates a Z-axis, which is the average power emitted from the RF-pulser for the duration of the droplet-forming signal. The blue color indicates a lower power while a red color indicates a higher power. Colors are shown in U.S. Prov. No. 63 / 448,075, filed on February 24, 2023, and incorporated herein More particularly, the range of power is indicated by blue, green, orange, and red, as power increases. In FIG.11A, the horizontal drop velocity along an X-axis above and substantially parallel to the sample (i.e., not the X- axis of FIG. 11A) is plotted against the A / B ratio. The horizontal drop velocity along the X-axis for each droplet was measured by a phase-Doppler interferometer, and is shown in units of meters / second. In FIG.11B, the horizontal drop velocity along a Y-axis above and substantially parallel to the sample (i.e., not the Y-axis of FIG.11B) is plotted against Attorney Docket No.22US0151-WO (66738WO01) the C / D ratio. The horizontal drop velocity along the Y-axis for each droplet was measured by the phase-Doppler interferometer, and is shown in units of meters / second. In FIG.11C, the vertical drop velocity along a Z-axis above and substantially perpendicular to the sample (i.e., not the Z-axis of FIG.11C) is plotted against the E / F ratio. The vertical drop velocity along the Z-axis for each droplet was measured by the phase-Doppler interferometer, and is shown in units of meters / second. In FIG.11D, the drop placement error magnitude is plotted against the G / H ratio. The ratios in FIGS. 11A-D can include, but are not limited to the ping and time-of-flight constraints shown in Figure 10. The drop placement error magnitude for each droplet was measured by the phase-Doppler interferometer, and is shown in units of meters / second. The drop placement error magnitude may be related to drop trajectory, and may be determined, for example, by the velocities discussed above in context of FIGS.11A, 11B, and 11C. As shown, there may be relatively strong correlations (e.g., R-squared > 0.8) that provide an indication on how a given droplet will travel and whether it will be a successful ejection event (in terms of drop placement) based on conditions and the droplet-forming signal. Instead of determining relationships between a limited number of points and measured characteristics, machine learning may be used to model complex relationships between many (e.g., thousands) of different specific data components of an acoustic droplet profile, such as the collected data discussed above in context of FIGS.9A and 9B and data collected from other systems including phase-Doppler interferometer systems and / or cameras. Such modeling may provide relatively accurate predictions on, for example, drop trajectories. A trajectory may be determined by assessing a vertical velocity vector (perpendicular to the transducer face) and two horizontal orthogonal velocity vectors to determine axial displacement magnitude and direction. As ADE proceeds, for example, during a drop train, data from the pings can be gathered and processed to adapt aspect(s) of the droplet-forming signal (e.g., power, frequency, or duration) to improve performance. For example, from assessing data from the pings during ejection of a train of droplets, it may be determined that the droplet(s) exhibit an undesirably large radial displacement. The ADE system may adjust parameter(s) of the droplet-forming signal to reduce the degree of radial displacement of the droplet(s). As another example, from assessing data from the pings during ejection of a train of droplets, it may be determined that the droplet(s) exhibit an undesirably slow vertical velocity. The ADE system may adjust parameters of the droplet-forming signal to increase the vertical velocity of the droplet(s). FIG.12A illustrates a system for training a machine learning model, in accordance with embodiments disclosed herein. A machine learning model 1240 that is being trained may receive data from an ADE system 1210 (e.g., similar or identical to ADE system 100), a phase-Doppler interferometer 1220, and / or a camera 1230. The ADE system 1210 is operated in accordance with principles discussed herein, and data is generated. This data may include droplet-forming signals data (e.g., the parameters for one or more droplet-forming signals), ping transmission data (e.g., the parameters for one or more pings), and / or ping reflection data. Some or all of this data may be provided to the machine learning model 1240. The machine learning model 1240 may be trained in a separate system from the ADE system 1210, phase- Doppler interferometer 1220, and / or camera 1230. The phase-Doppler interferometer 1220 may collect information corresponding to droplet(s) that are caused to be ejected during operation of the ADE system. The phase-Doppler interferometer 1230 may provide the ejected droplet data to the machine learning model 1240. The camera 1220 may collect information (e.g., visible-wavelength information) from the surface(s) of the sample(s) during operation of the ADE system 1210. The surface data may be provided to the machine learning model 1240. Training of the machine learning model 1240 may be implemented in a separate database and / or processor(s) than those in the data-generation systems. Machine learning model 1240 may implement one or more convolutional neural network(s) (CNNs). The CNNs may effectively scan through the acoustic data and may extract some or all the features of data generated by ADE system 1210. CNNs are known to be used in image identification models, in which case they may extract lines, curves, shapes, or colors from an image in order to build a feature set that defines the object Attorney Docket No.22US0151-WO (66738WO01) it is trying to classify. Machine learning model 1240 may implement CNNs that are presented with acoustic data to extract features, in a similar manner to image features. Examples of such features include the contours of the acoustic signal density profiles, the curvature of the leading or trailing edges of the non-zero signal regions, or the verticality of the pre- and post-ejection region. The machine learning model 1240 may further include a dense neural network (DNN), which may receive the features extracted by the CNNs. The DNN may tune weights and biases associated with the extracted features through an iterative process to reduce error. During model training, a feature set extracted from the acoustic signal profile may correspond to several attributes measured from any combination of systems 1210, 1220 and 1230. For example, if training a model to predict droplet vertical velocity as measured in system 1220, thousands of images may be presented to the model during training and be ‘tagged’ with a vertical velocity value. Each time an image is fed into the model, the CNN may break it down into its features and then tune the weights and biases of the DNN to reduce overall model error, for example, represented by a Mean Squared Error (MSE). Each subsequent image and vertical velocity tag fed into the model may result in a similar process without destroying the ability to predict the previous event. This may be achieved through loss-reduction algorithms that allow for a model to settle into an optimal state, or minimize MSE, where it can confidently predict events across a sample population without favoring a certain event type. This may be referred to as model generalization. FIG. 12B illustrates a system for implementing a machine learning model 180 in an ADE system 100, in accordance with embodiments disclosed herein. Once a model has been sufficiently trained, the trained model 180 may be stored in ADE system 100. A trained model 180 may be implemented in software and / or hardware. A trained model 180 may be implemented in software executed by processor 143 and / or hardware incorporated by processor 143. To be clear, even though FIG.12B shows the trained model 180 as being outside of processor 143, this is for illustrative purposes only. Processor 143 may cause interactions with transducer 112, according to principles discussed herein. Processor 143 may transmit information that causes transducer 112 to emit particular droplet-forming signals and pings into sample(s). Processor 143 may receive information reflected off of the pings by sample(s), where this information is received at the transducer 112 from the sample(s). Processor 143 may provide data corresponding to the droplet-forming signals and pings (e.g., power, duration, and / or frequenc(ies)) to the trained model 180. Processor 143 may also process ping reflection information received from transducer 112 and provide some or all of this processed data to the trained model 180. The trained model 180 may recognize pattern(s) from the inputted data, and in response, may provide new droplet-forming signal data (e.g., power, duration, and / or frequenc(ies)) to processor 143. Processor 143 may then generate droplet-forming signal information from the droplet-forming signal data received from the trained model 180, and provide this droplet-forming signal information such that it causes transducer 112 to emit a new droplet-forming signal. By utilizing a trained model 180, it may be possible to detect when a droplet velocity and size have changed, even without systems such as a phase-Doppler interferometer 1220 or camera 1230. For example the reflection of a ping signal may have a width or amplitude profile that may have changed at a particular time following droplet ejection, indicating that the drop is slightly smaller and is moving with less velocity towards our target. In response to this information, it may be possible to adjust the power of the droplet-forming signal to shift the drop back into stable state and continue with the transfer of droplets. This may be performed in real-time without having to interrupt a drop train. This may increase the speed of ADE when processing a plate while improving performance. As another example, the stability of the surface may be evaluated and accounted for with the trained model 180. During droplet ejection, the Attorney Docket No.22US0151-WO (66738WO01) droplet-forming signal may not be synchronized with movement of the surface of the sample. In such a case, a relatively high degree of drop placement error may occur. With the trained model 180, it may be possible to adjust the rate or timing at which droplet-forming signals are emitted by transducer 112 to better synchronize ADE and surface movement, such as observed oscillations. FIG.11A is a graph showing vertical drop velocity and the amplitude ratio of the acoustic signals. The color of a given point is the power applied to create the droplet, with blue being low power and red being high power. The colors are shown in U.S. Prov. No. FIG. 11A illustrates a relatively strong correlation between acoustic signals and vertical velocity of a drop. FIG.11B illustrates a similar chart, where the Y-axis is the velocity vector magnitude. The data processing associated with FIGS. 10A, 10B, 11A, and 11B may be performed by the ADE system 100 in, for example, processor 143. Many of the embodiments described herein, as will be understood, may be implemented on or in conjunction with a computer storage product with a non-transitory computer-readable medium (also can be referred to as a non- transitory processor-readable medium) having instructions or computer code thereon for performing various computer- implemented operations. These embodiments may include the ones that involve processor 143 (or relevant portions of such embodiments). The medium can include one or more distinct media. The code may be executed on one or more processors, such as processor 143 (which itself may include multiple processors). The computer-readable medium (or processor-readable medium) is non-transitory in the sense that it does not include transitory propagating signals per se (e.g., a propagating electromagnetic wave carrying information on a transmission medium such as space or a cable). The media and computer code (also can be referred to as code) may be those designed and constructed for the specific purpose or purposes. Examples of non-transitory computer-readable media include, but are not limited to: magnetic storage media such as hard disks, floppy disks, and magnetic tape; optical storage media such as Compact Disc / Digital Video Discs (CD / DVDs), Compact Disc-Read Only Memories (CD-ROMs), and holographic devices; magneto-optical storage media such as optical disks; carrier wave signal processing modules; and hardware devices that are specially configured to store and execute program code, such as Application-Specific Integrated Circuits (ASICs), Programmable Logic Devices (PLDs), Read-Only Memory (ROM) and Random-Access Memory (RAM) devices. Other embodiments described herein relate to a computer program product, which can include, for example, the instructions and / or computer code discussed herein. Computer-readable media may include a machine learning model 1240 that is being or will be trained. Computer- readable media may include a machine learning model 180 that has been trained. Some embodiments and / or methods described herein can be performed by software (executed on hardware), hardware (e.g., processor 143), or a combination thereof. Hardware modules may include, for example, a general-purpose processor, a field programmable gate array (FPGA), and / or an application specific integrated circuit (ASIC). Software modules (executed on hardware) can be expressed in a variety of software languages (e.g., computer code), including C, C++, Java™, Ruby, Visual Basic™, and / or other object-oriented, procedural, or other programming language and development tools. Examples of computer code include, but are not limited to, micro-code or micro-instructions, machine instructions, such as produced by a compiler, code used to produce a web service, and files containing higher-level instructions that are executed by a computer using an interpreter. For example, embodiments may be implemented using imperative programming languages (e.g., C, Fortran, etc.), functional programming languages (Haskell, Erlang, etc.), logical programming languages (e.g., Prolog), object-oriented programming languages (e.g., Java, C++, etc.), interpreted languages (JavaScript, typescript, Perl) or other suitable programming languages and / or development tools. Additional examples of computer code include, but are not limited to, control signals, encrypted code, and compressed code. It will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the novel techniques disclosed in this application. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the novel techniques without Attorney Docket No.22US0151-WO (66738WO01) departing from its scope. Therefore, it is intended that the novel techniques not be limited to the particular techniques disclosed, but that they will include all techniques falling within the scope of the appended claims.

Claims

Attorney Docket No.22US0151-WO (66738WO01) CLAIMS 1. A system for ejecting droplets from a surface of a sample using acoustic energy, the system comprising: a processor; a transducer configured to receive electrical energy and emit corresponding acoustic energy, and further configured to receive acoustic energy and generate corresponding electrical energy; a transmitter in communication with the processor and transducer, wherein the transmitter is configured to receive at least one signal from the processor and transmit corresponding electrical energy to the transducer; and a receiver in communication with the processor and transducer, wherein the receiver is configured to receive the electrical energy generated by the transducer and transmit at least one corresponding signal to the processor, wherein the transducer is configured to emit a first acoustic signal, a second acoustic signal, and a third acoustic signal, wherein the first acoustic signal is determined to cause a mound phase during which a droplet is formed at the surface of the sample, the second acoustic signal is selected to interrogate the surface of the sample during the mound phase, and the third acoustic signal is selected to further interrogate the surface of the sample during the mound phase.

2. The system of claim 1, wherein the second acoustic signal and the third acoustic signal are part of a plurality of interrogation pings, wherein the transducer is configured to emit the interrogation pings at intervals of between approximately 50 to 100 µS.

3. The system of any one of claims 1 or 2, wherein the plurality of interrogation pings comprises between approximately 4 to 2000 pings.

4. The system of any one of claims 1-3, wherein the processor is configured to receive information corresponding to reflections of the second acoustic signal and the third acoustic signal from the surface of the sample, wherein the processor is configured to determine at least one characteristic of the mound phase based at least in part on said information.

5. The system of any one of claims 1-4, wherein the at least one characteristic of the mound phase comprises at least one of a droplet size, the droplet linear velocity, the droplet angular velocity, the droplet radial displacement, the droplet breakoff position, the surface position, the surface velocity, the surface acceleration, a mound formation rate, or the mound relaxation rate.

6. The system of any one of claims 1-5, wherein the transducer is further configured to emit a fourth acoustic signal, wherein the fourth acoustic signal is determined to cause a second mound phase during which a second droplet is ejected from the surface of the sample, and wherein the processor is further configured to determine the fourth acoustic signal according to the at least one characteristic of the mound phase.Attorney Docket No.22US0151-WO (66738WO01) 7. The system of any one of claims 1-6, wherein the first acoustic signal and the fourth acoustic signal comprise at least one of a different power, frequency, or duration.

8. A system for ejecting droplets from a surface of a sample using acoustic energy, the system comprising: a processor; a transducer configured to receive electrical energy and emit corresponding acoustic energy, and further configured to receive acoustic energy and generate corresponding electrical energy; a transmitter in communication with the processor and transducer, wherein the transmitter is configured to receive at least one signal from the processor and transmit corresponding electrical energy to the transducer; and a receiver in communication with the processor and transducer, wherein the receiver is configured to receive the electrical energy generated by the transducer and transmit at least one corresponding signal to the processor, wherein the processor is configured to cause the transducer to emit a droplet-forming signal to cause a mound phase during which a droplet is ejected from the surface of the signal, and to further emit a plurality of pings at least one of before, during, or after the mound phase, wherein the processor is further configured to receive information from the receiver corresponding to reflections of the plurality of pings, and wherein the processor is further configured to determine at least one characteristic of the mound phase according to the information received from the receiver corresponding to reflections of the plurality of pings.

9. The system of claim 8, wherein the processor is further configured to cause the transducer to emit a second droplet- forming signal to cause a second mound phase, during which a second droplet is ejected from the surface of the sample, wherein the processor is further configured to determine the second droplet-forming signal based at least in part on the information received from the receiver corresponding to reflections of the plurality of pings.

10. The system of any one of claims 8-9, wherein the second droplet-forming signal is determined according to a machine- learning model.

11. The system of any one of claims 8-9, wherein the droplet-forming signal and the second droplet-forming signal are determined such that the first droplet and the second droplet have at least one of a different volume or separation velocity.

12. The system of claim 8, wherein the at least one characteristic of the mound phase comprises at least one of a droplet size, the droplet linear velocity, the droplet angular velocity, the droplet radial displacement, the droplet breakoff position, the surface position, the surface velocity, the surface acceleration, a mound formation rate, or the mound relaxation rate.Attorney Docket No.22US0151-WO (66738WO01) 13. The system of claim 8, wherein the transducer is configured to emit the plurality of pings at intervals of between approximately 50 to 100 µS.

14. The system of claim 8, wherein the plurality of pings comprises between approximately 4 to 2000 pings.

15. A method for implementing any one of claims 1-14.

Images