1. Introduction
Ultrasound, as a non-invasive method for high-precision detection and processing, has shown promising prospects in ultrasonic cleaning (Kashkoush et al. Reference Kashkoush, Busnaina, Kern and Kunesh1991; Mason Reference Mason2016; Yusof et al. Reference Yusof, Babgi, Alghamdi, Aksu, Madhavan and Ashokkumar2016), tumour treatments (Li et al. Reference Li, Zhang, Fan, Huang, Zhang and Wu2010; Zhou Reference Zhou2011; Sengupta & Balla Reference Sengupta and Balla2018), lithotripsy (Matsumoto et al. Reference Matsumoto, Allen, Yoshizawa, Ikeda and Kaneko2005; Ikeda et al. Reference Ikeda, Yoshizawa, Tosaki, Allen, Takagi, Ohta, Kitamura and Matsumoto2006; Roberts et al. Reference Roberts, Hall, Ives, Wolf, Fowlkes and Cain2006; Shpak et al. Reference Shpak, Kokhuis, Luan, Lohse, de Jong, Fowlkes, Fabiilli and Versluis2013), emulsification (Li & Fogler Reference Li and Fogler1978; Abismail et al. Reference Abismail, Canselier, Wilhelm, Delmas and Gourdon1999; Gaikwad & Pandit Reference Gaikwad and Pandit2008; Khokhlova et al. Reference Khokhlova, Canney, Khokhlova, Sapozhnikov, Crum and Bailey2011) and so on. During the propagation of ultrasonic waves, pressure fluctuations are created in the liquid, in which the pressure drop would lead to local cavitation (Noltingk & Neppiras Reference Noltingk and Neppiras1950; Nolingk & Neppiras Reference Nolingk and Neppiras1951; Neppiras Reference Neppiras1984; Atchley et al. Reference Atchley, Frizzell, Apfel, Holland, Madanshetty and Roy1988). The cavitation bubbles incept, move, grow or suddenly collapse with the vibration of the surrounding medium, generating shock waves and high-speed microjets (Reisman, Wang & Brennen Reference Reisman, Wang and Brennen1998; Brujan, Ikeda & Matsumoto Reference Brujan, Ikeda and Matsumoto2005; Supponen et al. Reference Supponen, Obreschkow, Kobel, Tinguely, Dorsaz and Farhat2017; Aganin & Mustafin Reference Aganin and Mustafin2021; Wu, Wang & Liu Reference Wu, Wang and Liu2022; Zhang et al. Reference Zhang, Ji, Zhang, Li, Wang and Zheng2022). These shock waves and microjets play an important role in ultrasonic treatment (Tomita & Shima Reference Tomita and Shima1986; Brujan et al. Reference Brujan, Ikeda and Matsumoto2005; Zhang et al. Reference Zhang, Ji, Zhang, Li, Wang and Zheng2022).
The ultrasonic frequency is a key parameter influencing its application. High-frequency ultrasound (typically higher than 15 MHz Kim Reference Kim2024 or higher than 20 MHz Cloutier et al. Reference Cloutier, Levassort, Jiang and Zhou2022) offers higher spatial resolution but has a shallower tissue penetration depth, thus being mainly applied in high-precision imaging of superficial tissues (Wilson et al. Reference Wilson, Gilany, Jamzad, Fooladgar, To, Wodlinger, Abolmaesumi and Mousavi2023). In contrast, low-frequency ultrasound (typically lower than 200 kHz (Fiorillo et al. Reference Fiorillo, Grimaldi, Paolino and Pullano2012)) has lower resolution but stronger tissue penetration ability, making it more suitable for therapeutic scenarios requiring deep tissue penetration (Cloutier et al. Reference Cloutier, Levassort, Jiang and Zhou2022). High-intensity focused ultrasound (HIFU), with typical frequency range of 1–10 MHz, is a prime exemplar of ultrasound lithotripsy, which employs high-energy pulse impacts to effectively disintegrate targets (Gail & Coussios Reference Gail and Coussios2007; Rybyanets et al. Reference Rybyanets, Nasedkin, Rybyanets, Lugovaya and Domashenkina2009; Yoshizawa et al. Reference Yoshizawa, Ikeda, Ito, Ota, Takagi and Matsumoto2009; Ikeda et al. Reference Ikeda, Yoshizawa, Koizumi, Mitsuishi and Matsumoto2016). The HIFU operating frequencies are application-dependent: solid tumour ablation (1.0–3.0 MHz) optimizes thermal coagulation (Peek & Wu Reference Peek and Wu2018; Gelet et al. Reference Gelet, Crouzet, Rouviere and Chapelon2020), while lithotripsy (1–3 MHz) balances penetration and cavitation control for stone fragmentation (Yoshizawa et al. Reference Yoshizawa, Ikeda, Ito, Ota, Takagi and Matsumoto2009; Pahk et al. Reference Pahk, Lee, Gélat, de Andrade and Saffari2021). Lower frequencies (0.3–1.0 MHz) facilitate blood–brain barrier opening via microbubble cavitation (Karakatsani et al. Reference Karakatsani, Estrada, Chen, Shoham, Deán-Ben and Razansky2024). Given the 1 MHz prevalence in clinically dominant applications (tumour ablation/lithotripsy) and its optimal trade-off between focal precision and tissue penetration (Ashar & Ranjan Reference Ashar and Ranjan2023), this frequency is selected to investigate the near-wall HIFU cavitation dynamics.
Focused ultrasound involves concentrating energy within a specified region through precise sound field design. This concentration of energy in the liquids can effectively generate cavitation clouds, which is a pivotal aspect in the HIFU treatment (Matsumoto et al. Reference Matsumoto, Allen, Yoshizawa, Ikeda and Kaneko2005). In this context, cavitation clouds refer to a densely populated region of interacting vapour cavities whose collective dynamics differs fundamentally from that of isolated bubbles (Caflisch et al. Reference Caflisch, Miksis, Papanicolaou and Ting1985; Commander & Prosperetti Reference Commander and Prosperetti1989; D’Agostino & Brennen Reference D’Agostino and Brennen1989). The phenomenon has been observed in water when ultrasound is concentrated at different frequencies (Willard Reference Willard1953; Neppiras & Coakley Reference Neppiras and Coakley1976). Based on experimental observations, Maxwell et al. (Reference Maxwell, Wang, Cain, Fowlkes, Xu, Sapozhnikov and Bailey2009, Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011) speculated that clouds initiate from a distal position within the focal volume and grow proximally, towards the transducer. Bubbles possess acoustically soft surfaces, which cause the inversion of the scattered waves respect to the incident waves. Thus, the compressive wave is converted into a tensile wave, which is responsible for the formation of cavitation clusters. Horiba et al. (Reference Horiba, Sano, Ogasawara and Takahira2017, Reference Horiba, Ogasawara and Takahira2020) experimentally and numerically investigated the dynamic characteristics of cavitation clouds in the presence of a preset bubble at the focal point. The results show that the growth of the cavitation clouds strongly depends on the maximum positive pressure at the focal point and the duration of focused ultrasound, and the cavitation cloud, caused by the backscatter of incident ultrasound, grows into a conical shape along the propagation axis of the focused ultrasound (Horiba et al. Reference Horiba, Sano, Ogasawara and Takahira2017). Once the peak value of the positive pressure of the incident wave falls below a certain extent, the cavitation cloud would no longer occur (Horiba et al. Reference Horiba, Sano, Ogasawara and Takahira2017). This indicates that there is a strong coupling relationship between the dynamical formation characteristics of a cavitation cloud and the propagation and distribution of ultrasonic waves. In numerical modelling, Hsiao, Ma & Chahine (Reference Hsiao, Ma and Chahine2017) developed a hybrid Eulerian–Lagrangian framework capturing cloud detachment and shock waves without predefined mass transfer. Ushioku & Yoshimura (Reference Ushioku and Yoshimura2025) further implemented smoothed-particle-hydrodynamics-based methods to simulate cloud oscillations. Theoretically, Margulis, Polovinkin & Yashin (Reference Margulis, Polovinkin and Yashin2024) addressed limitations in classical bubble dynamics by incorporating Bjerknes forces to predict sonoluminescence. The HIFU-induced cavitation involves complex fluid dynamics encompassing bubble formation, oscillation, coalescence and collapse within ultrasonic fields. The precise control of the cavitation clouds depends on the in-depth understanding of this unsteady interaction process. However, the coupling property of the convergence characteristic of focused ultrasonic waves and the inception condition of cavitation remain elusive, and the mechanisms underlying the growth and development of cavitation clouds are also not fully understood.
In recent years, significant progress has been made in understanding bubble dynamics (Zhang et al. Reference Zhang, Li, Xu, Pei, Li and Liu2024), phase transition mechanisms (Chen, Li & Wan Reference Chen, Li and Wan2006) and migration behaviours (Horiba, Ogasawara & Takahira Reference Horiba, Ogasawara and Takahira2020), which are fundamental to cavitation cloud evolution. Novel theoretical models now integrate liquid compressibility, phase transition, oscillation, migration and flow field interactions (Zhang et al. Reference Zhang, Li, Xu, Pei, Li and Liu2024; Xie, Zhang & Li Reference Xie, Zhang and Li2025). Studies have explored the dynamics of spherical bubbles, including metastable states (Yi et al. Reference Yi, Guo, Zuo and Liu2024), migration near walls in viscous media (Cai, Sun & Chen Reference Cai, Sun and Chen2024) and lateral migration in viscoelastic flows (Naseer et al. Reference Naseer, Izbassarov, Ahmed and Muradoglu2024). Bubble migration is particularly crucial for explaining the dynamic structuring of cavitation clouds, as evidenced by the role of backscatter-induced migration in forming distinct cloud patterns (Horiba et al. Reference Horiba, Ogasawara and Takahira2020). Furthermore, the coupling of migration and phase transition can trigger secondary cavitation through mechanisms like shockwave-induced nucleation (Chen et al. Reference Chen, Li and Wan2006). However, comprehensive algorithms capable of simultaneously resolving phase transition and migration throughout the entire lifecycle of HIFU-induced cavitation clouds are still lacking.
In practical applications, ultrasonic parameters (such as frequency, duty cycle, focus and intensity) have a decisive impact on the cavitation generation and intensity (Kim Reference Kim2024). For instance, Cloutier et al. (Reference Cloutier, Levassort, Jiang and Zhou2022) achieved precise control over the size of cavitation-induced lesions by adjusting the pulse duration, thereby enhancing the treatment accuracy. Fiorillo et al. (Reference Fiorillo, Grimaldi, Paolino and Pullano2012) found that short scanning intervals could effectively enhance or suppress cavitation activities induced by HIFU. Farny et al. (Reference Farny, Thomas, Holt and Roy2004) further demonstrated that by regulating the duty cycle, the HIFU-induced cavitation field could be manipulated, thereby influencing the morphology and size of thermal damage. Nevertheless, the mechanism through which the focal position impacts cavitation damage still awaits further investigation.
In addition, the formation of a cavitation cloud can be significantly affected by adjacent boundaries. This influence is particularly evident in various HIFU-related applications, such as in ultrasonic treatment involving tumours (Fergadi et al. Reference Fergadi, Magouliotis, Rountas, Vlychou, Athanasiou, Symeonidis, Pappa and Zacharoulis2022; Kim Reference Kim2024), ultrasonic lithotripsy targeting stones (Ikeda et al. Reference Ikeda, Yoshizawa, Tosaki, Allen, Takagi, Ohta, Kitamura and Matsumoto2006; Yoshizawa et al. Reference Yoshizawa, Ikeda, Ito, Ota, Takagi and Matsumoto2009) and ultrasonic cleaning of structural surfaces and interfaces (Tan, Tan & Tan Reference Tan, Tan and Tan2022; Ren et al. Reference Ren, Xu, Han, Ohl, Zuo and Liu2024). When HIFU is applied to a solid target, such as a lithiasis, cavitation clouds would be formed near the solid surface (Biasiori-Poulanges, Lukić & Supponen Reference Biasiori-Poulanges, Lukić and Supponen2024). The inception and collapse of these cavitation clouds can generate shock waves (Reisman et al. Reference Reisman, Wang and Brennen1998; Supponen et al. Reference Supponen, Obreschkow, Kobel, Tinguely, Dorsaz and Farhat2017) and high-speed jets (Ohl & Ikink Reference Ohl and Ikink2003; Zhang et al. Reference Zhang, Ji, Zhang, Li, Wang and Zheng2022). When a solitary bubble collapses and the distance from the bubble’s centre to the wall is less than its maximum radius, a high-speed microjet directed towards the wall is generated, thereby inflicting damage on the solid wall (Wang et al. Reference Wang, Liang, Yang, Shen, Feng and Zhang2024). Additionally, during the latter stage of the intense collapse of the cavitation cloud, the high-pressure pulses acting on the material surface generate cyclic impacts at frequencies reaching the kilohertz range (Zhao, Deng & Zhang Reference Zhao, Deng and Zhang2023). These loads can further precipitate intricate mechanical effects such as water hammer pressure (Zhao et al. Reference Zhao, Li, Xiong, Cui, Wang and Zhang2025) and circumferential stress waves (Buravova Reference Buravova1998), ultimately culminating in damage to the wall. Biasiori-Poulanges et al. (Reference Biasiori-Poulanges, Lukić and Supponen2024) found that the damaged area induced by the HIFU matched the corresponding area of cavitation on the surface. Dular et al. (Reference Dular, Bachert, Stoffel and Sirok2004) further demonstrated that the increase of the cavitation degree could lead to a significant intensification of surface damage over time. However, the results also showed that the damage rate of the surface was not positively correlated with ultrasonic amplitude (Ye, Zhu & Wei Reference Ye, Zhu and Wei2020). Maeda & Colonius (Reference Maeda and Colonius2019) examined the dynamics of cavitation clouds generated by HIFU and its destructive effect on stones. They found that the cloud strongly scattered subsequent incident waves, particularly when the size of the cloud was comparable to the wavelength of the ultrasound. The scattering effect shielded the stone from the energy acting on the stone, thereby reducing the efficiency of stone destruction. Meanwhile, unexpected cavitation areas induced by the scattering waves may cause additional damage to healthy tissues, which may lead to reduced therapeutic efficacy and even side effects (Maxwell et al. Reference Maxwell, Cain, Hall, Fowlkes and Xu2013).
Cavitation phenomena can significantly enhance HIFU efficacy, such as promoting cell ablation and drug delivery (Coussios et al. Reference Coussios, Farny, Ter Haar and Roy2007). However, uncontrolled cavitation may also cause non-target tissue damage (i.e. collateral damage) (Dibaji et al. Reference Dibaji, Al-Rjoub, Myers and Banerjee2013; Devarakonda et al. Reference Devarakonda, Myers, Lanier, Dumoulin and Banerjee2017b ). Increasing cavitation efficiency usually requires higher acoustic power or longer irradiation time, which often leads to a greater temperature rise in the surrounding treatment area and increases the risk of off-target tissue damage (Devarakonda et al. Reference Devarakonda, Myers, Giridhar, Dibaji and Banerjee2017a , Reference Devarakonda, Myers and Banerjee2018). Conversely, reducing acoustic power or shortening irradiation time to minimize collateral damage may weaken the cavitation effect and affect the final therapeutic outcome. Therefore, in specific medical applications, optimizing the balance between cavitation efficiency and collateral damage is key to successful treatment. Therefore, instead of significantly increasing ultrasonic intensity, a more accurate design of the ultrasound field according to the target characteristics is essential to enhance fragmentation efficiency and also prevent unnecessary damage to the adjacent tissues. In order to identify the regions where cavitation takes place, it is paramount to understand the coupling mechanism between cavitation clouds, solid boundaries and ultrasonic waves. However, when HIFU interacts with specific solid targets, the generation and evolution mechanism of cavitation clouds and their impact on targets are still unclear.
In order to elucidate the mechanisms responsible for the generation and development of cavitation clouds induced by HIFU and to investigate the characteristics of cavitation in the presence of solid walls, this paper aims to conduct a numerical study on the cavitation initiation and evolution process under focused ultrasonic waves considering the influence of the solid surface. The innovations of this paper are as follows:
-
(i) the whole dynamical properties of the initiation, growth and development of cavitation clouds induced by HIFU are investigated through numerical simulation;
-
(ii) a novel theoretical framework is established to explain the pressure field distribution and cavitation cloud patterns, validated against numerical data;
-
(iii) a detailed analysis is conducted to explore how the relative position of the focal point in relation to a solid wall, along with frequency variations, leads to distinct cavitation modes (embedded, adjacent, spaced).
The remainder of the paper is organized as follows. The numerical methodology are introduced in § 2. The brief descriptions and validations of numerical model are introduced in §§ 3.1 and 3.2. The propagation and convergence properties of ultrasonic waves are discussed in § 3.3. In § 3.4, a theoretical method is presented to predict the pressure distribution on the axis of symmetry. In § 4.1, the pressure distribution considering ultrasound and solid wall is discussed. In § 4.2, the inception mechanisms and distribution characteristics of both on solid wall and in flow field cavitation clouds at different relative positions between the solid wall and the focal point is investigated in detail. In § 4.3, a theoretical model of wave superposition considering incident, reflected and scattered waves is established to estimate the distribution law of the cavitation cloud on the wall at different positions. In § 4.4, the theoretical model proposed for predicting the distribution law of the cavitation cloud on the wall is extended and applied to the 2 MHz HIFU scenario. The paper concludes with a summary in § 5 and a discussion of the scope of future work. In Appendix A, the solution for the initial waveform of ultrasound in numerical investigation is presented. Appendix B delves into the dynamical evolution characteristics of cavitation clouds in the absence of a solid wall, taking into account the effects of cavitation inception. In Appendix C, the application scope of the conclusions drawn in this paper is given.
2. Methodology
2.1. Five-equation model
The HIFU transducer studied in this paper itself has an axisymmetric characteristic, and the observed cavitation cloud structure in the experiments also mainly presents an axisymmetric distribution, which provides a physical basis for using the axisymmetric model. An axisymmetric problem is considered in this work described using cylindrical coordinates
$ ( z,r )$
, which are the axial coordinate and radial distance, respectively. In the rapid growth of cavitation clouds under HIFU studied in this paper, experimental observations (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011) show that cavitation clouds develop into dense cloud states within an extremely short time. Cavitation clouds can be reasonably approximated as two-dimensional axisymmetric structures. Therefore, under the premise of effectively studying the core physical mechanisms, using the two-dimensional axisymmetric model is necessary and reasonable for the present problem. The compressible multicomponent flows are described by the five-equation model introduced by Allaire, Clerc & Kokh (Reference Allaire, Clerc and Kokh2002) considering phase transition process (Saurel, Petitpas & Abgrall Reference Saurel, Petitpas and Abgrall2008; Zein, Hantke & Warnecke Reference Zein, Hantke and Warnecke2013) to explore the cavitation inception and evolution processes in multimedia. Although the five-equation model is initially written for two-phase two-component flow, it can easily explain additional components by adding the equation of motion and volume fraction advection equation of each new medium. The model is given by (2.1)–(2.4) and consists of
$K$
(number of components) continuity equations, a momentum equation, an energy equation and
$K$
-1 volume fraction advection equations for the volume fraction. The numerical method employed in this article is governed by the following equations:
where
$k=1,2,3$
denotes the water vapour phase, water phase and air phase, respectively;
${\rho }_{k}$
,
${\alpha }_{k}$
and
${e}_{k}$
denote the internal energy, density and volume fraction of phase
$k$
, respectively;
$\boldsymbol{u}={{ ( u,v )}^{\text{T}}}$
represents the velocity vector;
$p$
denotes the pressure. The total energy
$E$
and total density
$\rho$
can be, respectively, represented as
\begin{align} E&=\sum \limits _{k=1}^{K}{{{\alpha }_{k}}{{\rho }_{k}}{{e}_{k}}+\frac {1}{2}\rho {{\boldsymbol{u}}^{2}}}, \end{align}
\begin{align} \rho &=\sum \limits _{k=1}^{K}{{{\alpha }_{k}}{{\rho }_{k}}}, \end{align}
where
$ {{\dot {S}}_{\rho ,k}}$
and
$ {{\dot {S}}_{\alpha ,k}}$
represent the source terms associated with the phase transition process;
$r$
is radial position in the axisymmetric coordinate system. The closure of the five-equation model is achieved by employing a stiffened gas equation of state (Saurel et al. Reference Saurel, Petitpas and Abgrall2008).
The Reynolds number
$Re$
and Weber number
$We$
, which, respectively, represent the importance of viscous force and capillary force, are defined as follows:
where
${\rho }_{0}$
,
$\lambda$
,
${V}_{0}$
,
$\mu$
and
$\sigma$
represent the initial density of the liquid, wavelength, initial speed of the liquid, liquid dynamic viscosity and surface tension coefficient, respectively. Considering that the Reynolds number and Weber number of the flows under consideration both reach the order of
${10}^{5}$
, inertial forces assume a dominant role. In addition, this study investigates the initiation and development of cavitation clouds induced by HIFU. The time scale associated with the growth of ultrasonically induced cavitation clouds is shorter than those over which viscosity and surface tension affect bubble dynamics. Moreover, the influence of an individual bubble remains confined to its immediate surroundings, meaning that the dynamics of a single bubble has a negligible effect on the collective behaviour of the entire cavitation cloud (Ma et al. Reference Ma, Yin, Zhang, Tian, Tian, Huang and Chen2024). Based on these considerations, the effects of viscosity and surface tension are justifiably neglected in this study.
2.2. Spatial and temporal discretization
In this study, we employ the splitting approach to decouple the hyperbolic operator and source terms present in (2.1)–(2.4). This approach enables the solution of each component independently, enhancing both accuracy and computational efficiency. For the spatial reconstructions, a fifth-order incremental stencil weighted essentially non-oscillatory scheme is utilized, as detailed by Titarev & Toro (Reference Titarev and Toro2004). This advanced technique provides robust and accurate spatial discretization, essential for capturing the intricate details of the fluid dynamics involved. To address the Riemann problem at the cell edges, a Godunov-type Harten–Lax–van Leer contact approximate Riemann solver is incorporated (Toro Reference Toro2013). In cases where phase transitions are triggered, the chemical relaxation procedure outlined by Han, Hantke & Müller (Reference Han, Hantke and Müller2017) is followed. This method treats the source terms related to phase transitions in (2.5)–(2.6), ensuring physically accurate representations of these complex processes. For time integration, a third-order total variation diminishing Runge–Kutta scheme (Gottlieb & Shu Reference Gottlieb and Shu1998) is adopted, which guarantees numerical stability and accuracy for the unsteady simulations.
2.3. Phase transition model
In the numerical method adopted in this paper, the condition for determining whether a phase change occurs is whether the local pressure in the flow field is lower than the cavitation threshold. The cavitation threshold of water is not a fixed value but fluctuates within a certain range. The actual cavitation threshold is influenced by multiple factors, making it difficult to provide an exact value (Connolly & Fox Reference Connolly and Fox1954; Angona Reference Angona1970; Hong & Son Reference Hong and Son2022; Fujikawa et al. Reference Fujikawa, Egashira, Hooman, Fukuda and Fujikawa2023; Zhong et al. Reference Zhong, Duan, Wang, Ma, Sun and Yao2023). These factors mainly include temperature, pressure, dissolved gas content, flow velocity, surface tension, impurities in water and microbubbles. In the numerical simulation study of this paper, we determine the applicable cavitation threshold parameters by comparing with experimental results. The central criterion for determining the occurrence of a phase transition is whether the local pressure within the flow field is lower than the cavitation threshold (Caupin & Herbert Reference Caupin and Herbert2006; Herbert, Balibar & Caupin Reference Herbert, Balibar and Caupin2006). We ascertain the applicable cavitation threshold parameters by comparing with experimental findings. Maxwell et al. (Reference Maxwell, Cain, Fowlkes and Xu2010) measured the negative pressure threshold for the formation of cavitation clouds to be −30 MPa through experiments in a reflector standing-wave field. It is important to note that under the action of HIFU, the cavitation threshold is not constant. A crucial phenomenon was observed during the experiments: within the first 2–3 cycles of ultrasonic exposure, no visible cavitation phenomena were detected in the flow field (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011). However, in subsequent cycles, cavitation occurred prominently (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011). This suggests that the ultrasonic irradiation during the initial cycles might have facilitated the activation, growth or generation of gas nuclei in the flow field. As a result, the content of gas nuclei increased, effectively raising the cavitation threshold under the action of subsequent pulses. This phenomenon aligns with the research on the influence of gas nucleus content (Connolly & Fox Reference Connolly and Fox1954). Taking into account the experimental observation of an elevated cavitation threshold under HIFU action, we set the cavitation threshold in the numerical model to −24 MPa. When the pressure is lower than the cavitation threshold, a phase change is considered to occur. The cavitation and phase-change process in the present solver is formulated within the relaxation-based two-phase framework originally proposed by Saurel et al. (Reference Saurel, Petitpas and Abgrall2008) and subsequently adopted in the five-equation models of Zein et al. (Reference Zein, Hantke and Warnecke2010, Reference Zein, Hantke and Warnecke2013). Here
$\upsilon$
is the relaxation parameter in the source term
$ {{\dot {S}}_{\rho ,k}}$
and
$ {{\dot {S}}_{\alpha ,k}}$
, used to quantify the ‘rate’ of the phase change process,
where
${g}_{2}$
is the Gibbs free energy of the liquid phase,
${\mu }_{1}$
is the chemical potential of the vapour phase and
$ \varrho$
is a parameter having the same dimensionality as density. Molecular dynamics simulations indicate that cavitation phase transition initiates from microscopic fluctuations within the liquid or at the interface. According to Menzl et al. (Reference Menzl, Gonzalez, Geiger, Caupin, Abascal, Valeriani and Dellago2016), cavitation in water under tension is a rare-event nucleation process, whose waiting time is governed by the stochastic thermal fluctuations needed to overcome the free-energy barrier. Once a critical bubble is formed, however, its subsequent growth occurs on subnanosecond time scales. This fast postcritical expansion is still many orders of magnitude shorter than the characteristic period of focused ultrasound (Šako et al. Reference Šako, Staniscia, Netz, Schneck and Kanduč2025). Therefore, in the present solution process, the phase transition time is negligible, and it can be assumed that
$\upsilon$
tends to infinity (Saurel et al. Reference Saurel, Petitpas and Abgrall2008; Zein et al. Reference Zein, Hantke and Warnecke2010, Reference Zein, Hantke and Warnecke2013; Pelanti & Shyue Reference Pelanti and Shyue2019). This assumption simplifies the continuous phase transition process into an algebraic problem of solving chemical potential equilibrium (
${{\mu }_{1}}={{g}_{2}}$
), and the equilibrium state after phase transition can be solved through iteration. The details of the specific calculation procedure are found in Han et al. (Reference Han, Hantke and Müller2017).
2.4. Variable definitions
The dimensionless axial position
${z}^{*}$
is described as
where
${z}_{\!{F}}$
and
$\lambda$
represent the axial position of the focal point (F) and wavelength of focused waves, respectively. The dimensionless radial position
${r}^{*}$
is described by
The dimensionless time
${t}^{*}$
is described by
where
$T$
represents the cycle of focused ultrasonic waves.
As shown in figure 1, to quantify the relative position between the solid wall and the focal point, a dimensionless parameter, denoted as
$d$
, is defined by normalizing with the wavelength,
where
${z}_{{w}}$
represents the axial position of solid wall.
Schematic of the computational domain for the high-intensity focused ultrasound (a) physical model of HIFU acting on a solid wall; (b) waveform of the initial field pressure (
$ p_w = 7.135 \, \text{MPa}$
; frequency, 1 MHz); (c) contour plot of the initial field pressure distribution.

3. Physical configuration and ultrasonic wave properties
3.1. Basic physical configuration
The geometric configuration of the HIFU field in this work is illustrated in figure 1. The ultrasonic generator resembles a partial spherical shape symmetric about the
$z$
-axis, with a spherical radius denoted by
${R}_{\!{F}}$
and the focal point (F) coinciding with the origin of the coordinate system. During the propagation of HIFU, nonlinear effects generate higher-order harmonics (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). These harmonics may exhibit different focusing characteristics compared with the fundamental wave, causing energy to deposit prematurely before the focal point of the fundamental wave and leading to the prefocused phenomenon (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). In this paper, F denotes the theoretical focal point of the transducer. The detailed parameters of the basic physical model are based on those in Maxwell et al. (Reference Maxwell, Wang, Cain, Fowlkes, Xu, Sapozhnikov and Bailey2009, Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), where the transducer has a focal length of
${{R}_{\!{F}}}=9\text{ cm}$
and a radial aperture of
${{R}_{{M}}}=5\text{ cm}$
. The generated ultrasonic waves have a frequency of 1 MHz. This ultrasonic generator produces converging ultrasonic waves with an initial waveform resembling a partial spherical shape. The pressure level in the focal waveform where shocks are fully developed is primarily determined by the transducer’s focusing angle characterized by
${{F}_{\textit{number}}}={{R}_{\!{F}}}\text{/}2{{R}_{{M}}}$
. The parameter
${F}_{\textit{number}}$
is related to the angular aperture
$\psi$
of the transducer through
${{F}_{\textit{number}}}={{ ( 2\sin ( \psi \text{/}2 ) )}^{-1}}$
. For transducers with different apertures but the same
${F}_{\textit{number}}$
, the shape and length of the main focal lobe on the beam axis during linear focusing are similar. Therefore, a local computational domain near the focal point is employed for numerical calculations.
Since the initial and boundary conditions of the entire fluid field are axially symmetric about the central axis (
$z$
-axis), the three-dimensional problem can be modelled using two-dimensional axisymmetric simulations. Here, a cylindrical coordinate system is used for numerical calculations, and the top-half portion of the ultrasonic field is taken as the computational domain, with the focal point located at the origin. The computational domain is defined as
$z\in [-15\text{ mm}, 17\text{ mm}]$
and
$r\in [0, 16\text{ mm}]$
. The ultrasonic wave enters the flow field from the left-hand side of the computational domain and converges towards the focal point (F), as shown in figure 1. The lower boundary is set as a symmetric boundary, while the remaining boundaries are treated as infinite boundaries. The initial liquid environment consists of homogeneous pure water at an initial temperature of 297 K and a pressure of 0.103125 MPa consistent with the experimental conditions. When calculating the initial conditions of multiphase flow within the framework of the Eulerian five-equation model, it is essential to take into account the initial volume fractions of different phases (liquid and gas) and components (water, air and water vapour). In accordance with Henry’s law (Kampbell & McInnes Reference Kampbell and McInnes2003; Saadi et al. Reference Saadi, Taleb, Yahia, Dogan, Yasin, Meraihi, Koklu, Mirjalili and Ramdane-Cherif2025), the volume fraction of the liquid phase is set to 99.999 %, and the total volume fraction of the gas phase is 0.001 %. The volume fractions of air and water vapour are determined from the thermodynamic phase equilibrium conditions.
Since only the area near the focal point is considered for numerical calculations, it is necessary to estimate the distribution structure of the ultrasonic waves emitted by the transducer after they enter the computational domain. A simplified mathematical model representing the high-amplitude pulsed sound waveform generated by the lithotripter was adopted (Church Reference Church1989; Johnsen & Colonius Reference Johnsen and Colonius2008), consisting of
where
${{p}_{0}}=1.01325\times {{10}^{5}}\text{ Pa}$
represents the environmental pressure;
${{p}_{w}}=7.135\times {{10}^{6}}\text{ Pa}$
,
$a =-0.5\times {{10}^{6}}\text{ }{{\text{s}}^{-1}}$
,
$\omega =2.618\times {{10}^{6}}\text{ }{{\text{s}}^{-1}}$
,
${{\beta }_{1}}={{\beta }_{2}}=-2\times {{10}^{7}}\text{ }{{\text{s}}^{-1}}$
,
$\varphi =2\pi \text{/}3$
and
$T=1\times {{10}^{-6}}\text{ s}$
. The initial waveform is shown in figure 1 and the solution for the initial waveform refers to Appendix A.
This paper mainly focuses on the cavitation generation mechanism near solid walls (such as stones) in applications such as HIFU lithotripsy. Given that the acoustic impedance of solids such as stones is much higher than that of water (Ali et al. Reference Ali, Raj, Kalainathan and Palanichamy2008; Le Grill et al. Reference Le Grill, Soulie, Coppel, Roblin, Lecante, Marsan, Charvillat, Bertrand, Rey and Brouillet2021), the acoustic wave is almost completely reflected at their interface, justifying the approximation of these solids as rigid walls for capturing the core dynamics of cavitation bubbles. For soft tissues with acoustic impedance similar to that of water, the transmission effect is significant and this approximation is invalid.
To systematically investigate the detailed mechanisms of the growth and development of the cavitation cloud near the solid wall under focused ultrasound step by step, three basic scenarios are considered in this work. Firstly, without a rigid wall and ignoring cavitation effects, the propagation and convergence characteristics of the ultrasonic waves themselves in the flow field is studied. Then, a small cavitation bubble near the focal point is preset as a cavitation nucleus to investigate the dynamics of the inception and growth of the cavitation cloud. Details of this case are provided in Appendix B. Finally, in the presence of a rigid wall adjacent to the focal point, the mechanism of cavitation cloud formation and its development are investigated.
Comparison between numerical simulation and experiment of focal pressure (transducer parameters, focal length
$ R_F = 9 \, \text{cm}$
; aperture diameter
$ R_M = 5 \, \text{cm}$
; frequency, 1 MHz). Experimentally measured peak values, positive pressure 85 MPa; negative pressure −19 MPa (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011).

3.2. Numerical validations
A typical ultrasonic pulse comprises multiple cycles. Each cycle features a steep shock front with a high-amplitude positive pressure phase, followed by a low-amplitude negative pressure phase. The peak pressure occurs near the focal point. Figure 2 shows the pressure distribution at the focal point induced by three-cycle focused ultrasonic waves using the present numerical set-up, and the corresponding experimental result (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011) obtained using the histotripsy transducer. As shown in figure 2, the waveform exhibits significant asymmetry, in which the peak positive pressure of the pulse is several times higher than the peak negative pressure due to the combined effects of nonlinear propagation and diffraction. In general, the waveforms obtained from the present simulation are in good agreement with those measured experimentally, including the peak value and the duration of negative pressure. The peak positive and negative pressures from the numerical calculation are 85 MPa and −19 MPa, respectively, which are in good agreement with the experimental results. The experimental results show that the durations of the positive and negative pressure segments account for 30 % and 70 %, respectively, while the numerical results show 30.9 % and 69.1 %, respectively, indicating good agreement. In addition, during the propagation of the ultrasonic waves, diffraction waves are generated. The formation mechanism of diffraction waves is discussed in § 3.3. Due to the superposition of the diffraction waves and the subsequent focused waves, the amplitude of the second and third ultrasonic waves is higher than that of the first ultrasonic wave. Both the numerical and experimental results exhibit this phenomenon. The waveform discrepancies within the region indicated by the black dashed line may be partially attributed to the convergence and diffraction effects of subsequent aftershock waves in the experiment. Notably, in practical HIFU applications, multiple ultrasonic cycles (e.g. 20 cycles) are typically applied in succession. With multiple pulses, the last aftershock wave following the main wave has a relatively negligible impact on the overall cavitation cloud that ultimately forms. Therefore, these results demonstrate that the numerical method and the configuration adopted in this paper can effectively capture the propagation and focusing characteristics of focused ultrasonic waves. In the following sections, following the experimental set-up (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), pulse lengths of 20 cycles are used to investigate the characteristics of ultrasonic focusing and cavitation cloud generation.
Variation of the cavitation cloud size with the number of ultrasonic cycles. Experimental data (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), mean
$\pm$
standard deviation of the axial/radial lengths of the cavitation cloud under 5, 10, 15 and 20 cycles ultrasonic waves.

In order to verify the accuracy of the cavitation model, based on Maxwell’s experimental set-up (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), ultrasonic cavitation induced by HIFU is simulated. Figure 3 compares the simulated and experimental trends of cavitation cloud size versus the number of ultrasonic wave cycles. The axial length
${l}_{z}$
and the radial length
${l}_{r}$
represent the maximum axial and radial dimensions of the cavitation cloud. It is illustrated in figure 3 that the number of ultrasonic wave cycles has a significant impact on the axial dimension of the cavitation cloud, while the radial dimension approaches an upper bound. Furthermore, the axial dimension of the cavitation cloud exhibits a gradient increase. Although there is a slight discrepancy in the length of cavitation clouds between numerical and experimental results after 10 cycles, the dimension of cavitation clouds in the numerical results aligns well with the experiments at higher cycle numbers. In addition, the dynamic growth process of the cavitation cloud, which agree well with the experimental results (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), are presented in Appendix B. The consistency between numerical and experimental results demonstrates that the numerical model and cavitation modelling approach used in this paper can effectively simulate the cavitation problem induced by HIFU. Moreover, to ensure numerical robustness, a comprehensive grid-independence verification is documented in Appendix B.
3.3. Propagation and convergence characteristics of ultrasonic waves
In order to understand the focusing properties during the propagation of the focused ultrasound waves, in this section the characteristics of the flow field in the absence of cavitation are analysed. In figure 4, the instantaneous pressure and density schlieren at four representative time instants are presented. Figure 4(a–d) correspond to the initial instant, superposition instant, prefocused instant and the instant when the mesh structure forms, respectively.
Contours of pressure and the density schlieren in the focused ultrasound field without a solid wall (a)
$ t^* = - 8.5$
, initial waveform; (b)
$ t^* = - 6.7$
, instant of superposition; (c)
$ t^* = - 0.7$
, prefocused instant; (d)
$ t^* = 2.3$
, instant of formation of the grid structure.

Figure 4(a–c) depict the flow field of the ultrasonic waves before reaching the focal point (F). As shown in figure 4(a), each waveform comprises the focused waves propagating along the symmetry axis that converge towards F, as well as the diffraction waves emanating from the periphery of the focused waves. Figure 4(b) depicts that, within the overlap zone of these two wave types, fluctuations can be observed in both the pressure distribution and the density schlieren diagrams. As the focused waves advance towards F, both the maximum positive and negative pressure values gradually increase until they reach a peak in the immediate vicinity of F (figure 4 c). Figure 4(c) illustrates the prefocused phenomenon in focused ultrasound, where the ultrasound is focused to the left of the focal point. The prefocused phenomenon arises from the superposition of focused waves and diffraction waves (as discussed in detail later). Figure 4(d) exhibits the distribution of ultrasonic waves after the first several cycles have passed through F. After passing through the focal point (F), the focused waves undergo a transition from convergence to divergence. As shown in figure 4(d), the collision and interference of the focused waves with diffracted waves from the periphery result in a distinctive mesh structure of the wave system at the focal region. The interaction between these waves leads to a highly unsteady and non-uniform pressure distribution that can significantly influence the initial location of cavitation and the growth dynamics of the cavitation cloud.
The formation and evolution mechanism of diffraction waves and superimposed regions can be explained by the Huygens–Fresnel principle (Depasse et al. Reference Depasse, Paesler, Courjon and Vigoureux1995; Volpe, Létourneau & Zhao Reference Volpe, Létourneau and Zhao2017). Based on this principle, the propagation of ultrasonic waves can be approximately equivalent to the propagation of a series of wavelets. Every point on the initial focused ultrasonic wavefront can be regarded as a source emitting wavelets. At any subsequent moment, the envelope of these wavelets in the direction of wave propagation represents the wavefront at the present instant. As illustrated in figure 5(a), a focused ultrasonic wave, denoted as
$\text{F}{{\text{W}}_{0}}$
, converges towards F at
${t}_{0}$
. As
$\text{F}{{\text{W}}_{0}}$
propagates downstream, each point on the wavefront can be viewed as a source emitting wavelets. Figure 5(b) depicts the envelope of these wavelets that constitute the waveform
$\text{F}{{\text{W}}_{1}}$
when the ultrasonic wave
$\text{F}{{\text{W}}_{0}}$
propagates from
${t}_{0}$
to
${t}_{1}$
. Simultaneously, the wavelets at the periphery, which lack other wavelets to form an envelope with, form the diffraction wave
$\text{D}{{\text{W}}_{1}}$
at
${t}_{1}$
. Depicted in figure 5(c), the focused wave
$\text{F}{{\text{W}}_{1}}$
and diffraction wave
$\text{D}{{\text{W}}_{1}}$
continue to propagate according to the aforementioned pattern and form
$\text{F}{{\text{W}}_{2}}$
and
$\text{D}{{\text{W}}_{2}}$
at
${t}_{2}$
, respectively. A new diffraction wave
$\text{D}{{\text{W}}_{2}}$
is formed at the periphery of
$\text{F}{{\text{W}}_{2}}$
. The diffraction waves
$\text{D}{{\text{W}}_{1*}}$
and
$\text{D}{{\text{W}}_{2}}$
overlap, forming the diffraction wave pattern at time
${t}_{2}$
. Simultaneously, the diffraction wave
$\text{D}{{\text{W}}_{1*}}$
superimposes on the focused waves, creating a superimposed region. This phenomenon is also consistent with the propagation process of the focused waves and the morphology of the superposition region in figure 4.
Schematic diagram of the formation mechanism of diffraction waves (based on the Huygens–Fresnel principle). (a) At
$ t = t_{0}$
, the initial wavefront is
$\text{F}{{\text{W}}_{0}}$
; (b) at
$ t = t_{1}$
, the diffraction wave
$\text{D}{{\text{W}}_{1}}$
is formed; (c) at
$ t = t_{2}$
, the superposition region of diffraction waves.

Figure 6(a) shows the minimum pressure distribution in the focal area, and figure 6(b) shows the corresponding pressure values on the symmetry axis
$z$
, where the prefocused phenomenon is clearly represented. Due to the influence of diffraction waves in the edge region, the pressure distribution at the focal area is not symmetrical about
${{z}^{*}}=0$
, with the region closer to the wave source being narrower than the region farther away from it (figure 6
a). At the same time, as shown in figure 6(b), the peak negative pressure on the symmetry axis is at the position around
${{z}^{*}}=-1.1$
, slightly towards the wave source. Since the focusing and divergence processes of focused waves are symmetrical about
${{z}^{*}}=0$
, this asymmetry effect should be solely caused by diffraction waves. Based on the Khokhlov–Zabolotskaya–Kuznetsov (KZK) equation (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009), the prefocused position under the current condition can be estimated to be
$z_{\textit{theory}}^{*}=-{{R}_{\!{F}}}\text{/} ( \lambda G )=-1.03$
, where
$G=58.17$
represents the diffraction coefficient of the transducer. The definition of the diffraction coefficient
$G$
refers to Appendix A. This theoretically predicted value is roughly consistent with the numerical results.
Minimum pressure distribution in the focus region. (a) Two-dimensional contour plot of the focus region; (b) pressure values along the symmetry axis
$ z$
(prefocused position
$ z^{*}=-1.1$
).

3.4. Theoretical superposition model of ultrasonic waves
As deduced from § 3.3, the propagation and convergence characteristics of focused ultrasound exhibit remarkable complexity. On the one hand, it is characterized by strong unsteadiness. On the other hand, the focusing behaviour significantly modifies the intensity and distribution of the ultrasonic waves, accompanied by a certain degree of nonlinearity. In light of these factors, it is imperative to establish a theoretical model capable of quantitatively characterizing the properties of focused ultrasound. This model will serve as a valuable tool for facilitating subsequent in-depth quantitative analyses of the flow field characteristics in the context of cavitation. The current models used to predict the pressure distribution of focused acoustic fields include the Gaussian beam model (Lock Reference Lock1995; Andrews et al. Reference Andrews, Al-Habash, Hopen and Phillips2001; Vaveliuk, Ruiz & Lencina Reference Vaveliuk, Ruiz and Lencina2007), the angular spectrum method (Maynard, Williams & Lee Reference Maynard, Williams and Lee1985; Williams Reference Williams1999), the complex source point method (Felsen & Marcuvitz Reference Felsen and Marcuvitz1994), among others. However, the Gaussian beam model exhibits significant errors in non-paraxial regions (Vaveliuk et al. Reference Vaveliuk, Ruiz and Lencina2007); the angular spectrum method involves computationally intensive numerical integration (Williams Reference Williams1999); and the complex source point method suffers from far-field divergence issues (Felsen & Marcuvitz Reference Felsen and Marcuvitz1994). Therefore, we propose to develop a simplified model for the rapid prediction of pressure distributions in focused acoustic fields. As shown in figure 6, the pressure distribution of the ultrasonic field is jointly determined by focused waves and diffraction waves. Based on Huygens’ principle (Depasse et al. Reference Depasse, Paesler, Courjon and Vigoureux1995), a simple assumption is made that during propagation, the pressure wave can be regarded as the superposition of focused waves and diffraction waves. Therefore, by separately constructing the propagation models for focused waves and diffraction waves and then superposing them, we can obtain the final distribution of the ultrasonic pressure field. The schematic diagram of the wave superposition model is shown in figure 7(a). Here
$\text{F}{{\text{W}}_{0}}$
represents the converging wave at the initial moment. Here
$\text{FW}$
and
$\text{DW}$
represent the focused wave and the diffraction wave, respectively, after propagation for a certain period of time. The propagation of the focused waves can be approximated as one-dimensional spherical waves (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). The propagation model for focused waves in two dimensions can be analogously described as
Validation of the wave superposition model. (a) Schematic illustration of the model (FW, focused wave; DW, diffracted wave); (b) comparison between the numerical and theoretical pressure distributions along the symmetry axis.

where
${K}_{\!p}$
represents the nonlinear gain coefficient;
$A$
represents the linear amplification coefficient;
${p}_{0}$
represents the initial pressure amplitude;
$\xi$
represents the distance to the focal point;
${\theta }_{\textit{FW}}$
represents the angle with the
$z$
-axis. The definitions of
${K}_{\!p}$
and
$A$
refer to Appendix A, and
${g}_{\textit{FW}}$
represents the effective range of the focused wave,
\begin{equation} {g}_{\textit{FW}}\left ( {{\theta }_{\textit{FW}}} \right )=\begin{cases} 1, & 0\leqslant {{\theta }_{\textit{FW}}}\leqslant \psi \text{/}2, \\ 1, & \pi -\psi \text{/}2\leqslant {{\theta }_{\textit{FW}}}\leqslant \pi , \\ 0, & \text{others}. \end{cases} \end{equation}
Although this model takes into account the diffraction effect during the propagation of ultrasonic waves, it can only predict the pressure distribution along the
$z$
-axis and cannot fully reflect the prefocused effect. Therefore, in this section, diffraction waves are added to correct the model, to achieve the prediction of the pressure field in the focusing area and to be used for the analysis of the cavitation area in § 4.3. Similar to focused waves, the propagation model for diffraction waves can be expressed as
where
$\xi =\left | {{\xi }_{0}}-{{\xi }_{1}} \right |$
;
${\xi }_{1}$
represents the distance from the diffraction wavefront to the diffraction wave starting point;
${g}_{\textit{DW}}$
represents the effective range of the diffraction wave,
\begin{equation} {g}_{\textit{DW}}\left ( {{\theta }_{\textit{DW}}} \right )=\begin{cases} -\dfrac {{{\theta }_{\textit{DW}}}{{\left ( \dfrac {\psi }{2}-{{\theta }_{\textit{DW}}} \right )}^{2}}}{{{\left ( \dfrac {\psi }{2} \right )}^{3}}}, & \text{0}\lt {{\theta }_{\textit{DW}}}\leqslant \dfrac {\psi }{2}\text{ }\&\text{ } 0\leqslant {{\theta }_{\textit{FW}}}\leqslant \dfrac {\psi }{2}, \\[15pt] \dfrac {\left ( \dfrac {\psi }{2}+{{\theta }_{\textit{DW}}} \right )^{2}}{{{\left ( \dfrac {\psi }{2} \right )}^{2}}}\text{,} & -\dfrac {\psi }{2}\leqslant {{\theta }_{\textit{DW}}}\lt \text{0}\text{ }\&\text{ }\dfrac {\psi }{2}\lt {{\theta }_{\textit{FW}}}\lt \pi -\dfrac {\psi }{2}, \\[12pt] 0, & \text{others}, \end{cases} \end{equation}
where
${\theta }_{\textit{DW}}$
represents the angle with the line (OA). The
${\theta }_{\textit{DW}}$
of the point located below the line (OA) is positive. In the left-hand focusing area (
$\text{0}\leqslant {{\theta }_{\textit{FW}}}\leqslant \psi \text{/}2$
), the diffraction waves are mainly used to correct the prefocused phenomenon. Comparison shows that this model can effectively predict the pressure distribution in the prefocused area. In the diffraction area (
$\psi \text{/}2\lt {{\theta }_{\textit{FW}}}\lt \pi -\psi \text{/}2$
), in order to ensure the continuity of the pressure field,
${K}_{\!p}$
is set to unity at
${{\theta }_{\textit{DW}}}=0$
and decreases as
${\theta }_{\textit{DW}}$
increases. In the right-hand focusing area (
$\pi -\psi \text{/}2\leqslant {{\theta }_{\textit{FW}}}\leqslant \pi$
), considering that the intensity of the diffraction waves in this area is weak, the distribution of the pressure field can still be approximately predicted without correction, and therefore
${K}_{\!p}$
is set to zero. The relationship between
${K}_{\!p}$
and
$\xi$
is obtained from the KZK equation (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). The ultrasonic pressure field can be represented as the superposition of focused waves and diffraction waves:
To verify the accuracy of this simplified model, figure 7 compares the minimum pressure distribution on the
$z$
-axis between the computational results and the theoretical model. It is shown that the theoretical predictions exhibit a congruent overall trend with the computational results. Both clearly manifest the asymmetry, and their peak values demonstrate a remarkable degree of agreement. Additionally, the fluctuations caused by the diffracted waves have been accurately captured. Consequently, it is concluded that the wave superposition model is capable of effectively forecasting the pressure distribution within the flow field.
In summary, this section discusses the characteristics of the flow field without considering the cavitation effect, and establishes a simplified model for the pressure distribution of the flow field. Appendix B further elaborates on the dynamic evolution characteristics of cavitation clouds within the focused ultrasound field, taking into account the impacts of cavitation inception.
4. Dynamic characteristics of ultrasound and cavitation clouds in the presence of solid wall
In practical applications (Ikeda et al. Reference Ikeda, Yoshizawa, Koizumi, Mitsuishi and Matsumoto2016; Biasiori-Poulanges et al. Reference Biasiori-Poulanges, Lukić and Supponen2024), ultrasonic is often applied near some specific boundaries. In the presence of a solid wall, both the convergence characteristics of ultrasonic waves and the formation of cavitation clouds might change (Okita Reference Okita2018). Therefore, based on the above basic understanding of the properties of ultrasonic focusing (in § 3.3) and cavitation evolution (in Appendix B), the mechanism of the inception and growth of the cavitation cloud in the presence of solid wall is mainly discussed in this section.
4.1. Propagation and convergence characteristics of ultrasonic waves
Distribution of minimum pressure under (a i)
$d=-1$
, (a ii)
$d=0$
and (a iii)
$d=1$
;
$r$
–
$t$
–
$p$
map of the solid wall pressure under (b i)
$d=-1$
, (b ii)
$d=0$
and (b iii)
$d=1$
; schematic diagram of the incident wave reflection at the solid wall under (c i)
$d\lt 0$
and (c ii)
$d\gt 0$
.

For the ultrasonic field with a solid wall, ultrasonic waves are reflected due to the presence of a solid wall. Therefore, the incident wave and the reflected wave are superimposed unsteadily, and when combined with the multidimensional effect of focused ultrasonic waves, the superimposed ultrasonic field exhibits a unique distribution characteristic. Furthermore, the superimposed ultrasonic field varies further with different relative positions between the solid wall and the focal point. First, three different relative positions are considered (
$d=-1,0,1$
) to study the ultrasonic propagation and convergence characteristics in the presence of a wall, where the ultrasonic waveform is the same as that in § 3.3 without considering the cavitation effect. Figure 8 illustrates the distributions of ( figure 8
a) the minimum pressure of the flow field, (figure 8
b) the
$r-t-p$
map at the solid wall and (figure 8
c) the schematic diagram of wave superposition at the wall.
Different from the results of the case in absence of a solid wall (figure 6
a), figure 8(a i–a iii) indicate that the superposition of the reflected and incident waves creates a banded pattern in the area where the cavitation threshold is reached. For three different
$d$
, in the region where
${{z}^{*}}\gt -2$
, the minimum pressures show similarities, all appearing as ellipsoids spaced at intervals. However, in the region where
${{z}^{*}}\lt -2$
, there are significant differences in the minimum pressures. Under the cases of
$d=0$
and
$d=1$
, the minimum pressures exhibit obvious convergence characteristics, which are caused by the superposition of diffraction waves and focused waves. Since the diffraction waves converge at the position of
${{z}^{*}}=-1.1$
, when the solid wall is located at
${{z}^{*}}=-1$
, the diffraction waves are reflected by the solid wall and propagate in the opposite direction with a similar waveform, resulting in minimum pressure distributions that lack the convergence characteristics at
${{z}^{*}}\lt -2$
region.
Figure 8(b) shows the wall pressure distribution over time under different
$d$
, where the distribution experiences an alternation of positive and negative pressure. The area surrounded by the black dashed line corresponds to the negative pressure peak, and the area surrounded by the black solid line corresponds to the positive pressure peak. There are certain differences in the structures of the areas of positive pressure peak under the three operating conditions. Figure 8(c i) shows the schematic diagram of the superposition of the incident wave and the reflected wave at the wall under the working condition of
$d=-1$
. Since the solid wall is located on the left-hand side of the focal point, the regions with larger radial positions are closer to the wave source than the regions with smaller radial positions. Therefore, the incident wave first reaches the regions with larger radial positions on the solid wall and forms a reflected wave. The ultrasonic wave and the reflected wave are superimposed to form a superposition region. As the incident wave propagates, it reaches the regions on the solid wall with smaller radial positions, causing the superposition region to move along the solid wall towards the regions with smaller radial positions, as presented in figure 8(bi).
The schematic diagram of the superposition of the incident wave and the reflected wave at the wall under the working condition of
$d=1$
is shown in figure 8(c ii). Since the solid wall is located on the right-hand side of the focal point, the convergent wave turns into a divergent wave after passing through the focal point. Therefore, the incident wave first reaches the region with smaller radial positions on the solid wall, causing the superposition area of the incident wave and the reflected wave to emerge first in the region with smaller radial positions. As the incident wave propagates, it reaches the regions on the solid wall with larger radial positions, which causes the superposition region to move along the solid wall towards the regions with larger radial positions, as presented in figure 8(b iii).
The condition of
$d=0$
represents an intermediate state between
$d=-1$
and
$d=1$
, and a turning point occurs in the distribution of its positive pressure peaks. In the region where
${{r}^{*}}\leqslant 0.25$
, the distribution of the positive pressure peak region is similar to that of the case when
$d=-1$
. In the region where
${{r}^{*}}\gt 0.25$
, the distribution of the positive pressure peak region is similar to that of the case when
$d=1$
. In addition, for the three operating conditions, influenced by the positive pressure distribution, the inclination angles of the negative pressure distribution regions show certain differences, but their distribution ranges in time and space are similar.
In summary, due to the solid wall altering the propagation direction of ultrasonic waves, the position of the solid wall has a significant impact on the distribution of the pressure field in the near wall region, and thus greatly influences the cavitation effect in this region. Therefore, in the next section, the formation and development mechanism of the cavitation cloud in the near wall region when the solid wall is at different positions is discussed.
4.2. Initiation and development of cavitation clouds
Due to changes in the focused characteristics of ultrasonic waves as discussed in § 4.1, the inception and growth mechanism of cavitation clouds are also altered in the presence of a solid wall. In this section, to investigate the inception and growth mechanism of cavitation clouds and to compare it with that in Appendix B, the case of cavitation evolution under 20 consecutive cycles of focused ultrasonic waves is studied, considering the cavitation effect and the presence of a solid wall at
${z}_{{w}}$
. For the region far away from the solid wall, the cavitation threshold is set to be −24 MPa (Horiba, Ogasawara & Takahira Reference Horiba, Ogasawara and Takahira2018). In the near-wall region, the cavitation threshold might be influenced by the wetting properties of the surface (Caupin & Herbert Reference Caupin and Herbert2006; Wu, Liu & Wang Reference Wu, Liu and Wang2021),
where
$\xi$
represents the surface contact angle;
${E}_{b}$
denotes the energy barrier during cavitation;
$\sigma$
is the surface tension coefficient and
${p}'$
is the pressure at which the vapour is at the same chemical potential as the liquid at the local pressure
$p$
. The static contact angle of stones is affected by various factors such as stone composition and surface characteristics. Taking calcium oxalate stones as an example, calcium oxalate crystals themselves exhibit a certain degree of hydrophilicity, as they can form within biological organisms and interact with water in the body (Al-Atar et al. Reference Al-Atar, Bokov, Marshall, Teichman, Gates, Ye and Branda2010). However, if the stone surface is coated with lipids or proteins (such as in gallstones), forming a hydrophobic layer, the contact angle usually increases significantly (Monroe et al. Reference Monroe, Barry, DeStefano, Aydogan, Jiao, Robinson-Brown, Webber, Crumlin, Han and Shell2020). In this section, it is assumed that the surface contact angle at the wall is approximately
${75}^{\circ }$
which corresponds to the cavitation threshold of −20 MPa.
Cavitation cloud distributions at different values of
$ d$
(after the action of 20 ultrasonic waves). Here (a)
$ d = - 1.5$
, embedded mode (Mode E); (b)
$ d=-0.25$
, Mode E; (c)
$ d = 0$
, adjacent mode (Mode A); (d)
$ d = 0.25$
, Mode A; (e)
$ d = 1.0$
, spaced mode (Mode S); (f)
$ d = 1.25$
, Mode S.

To explore the variation of the cavitation cloud distribution with the position of the solid wall, numerical calculations with the value of
$d$
ranging from −1.5 to 1.25 are conducted. The distribution of the final cavitation cloud for different
$d$
is illustrated in figure 9. The results show that the cavitation clouds with different
$d$
values exhibit similar distributions, which can be divided into three or four subregions of cavitation clouds. The cavitation cloud generally exhibits a ‘tree-like’ structure. Cavitation Cloud I (marked as CC I in the figure) is located on the solid wall and has a conical structure, which can be regarded as the ‘tree root’. Cavitation Cloud II (marked as CC II in the figure) is located at larger radial positions on the solid wall. Cavitation Cloud III (marked as CC III in the figure) is located in the flow field and is relatively sparsely distributed, representing the ‘tree crown’. For larger values of
$d$
, Cavitation Cloud IV (marked as CC IV in the figure), which is formed between Cavitation Cloud I and Cavitation Cloud III, is distributed near the
$z$
-axis and serves as the ‘tree trunk’.
According to the relative positions of Cavitation Cloud I and Cavitation Cloud III, the observed results under different
$d$
can be classified into three different modes. When
$d\lt 0$
, corresponding to embedded mode (Mode E), there is considerable overlap between Cavitation Clouds I and III. When
$d=0$
, corresponding to adjacent mode (Mode A), Cavitation Clouds I and III are adjacent to each other. When
$d\gt 0$
, corresponding to spaced mode (Mode S), Cavitation Clouds I and III are separated by Cavitation Cloud IV. Then, the inception and growth mechanisms of cavitation clouds in different modes are discussed, respectively. Considering that the cavitation cloud on the solid wall plays a more significant role in ultrasonic lithotripsy and ultrasonic cleaning, with a focus on the cavitation effect on the solid wall, our primary attention is centred on the dynamic characteristics of Cavitation Clouds I and II.
Mode A can be regarded as a transitional mode between Mode S and Mode E and has certain similarities with both of them. Hence, the formation and development mechanisms of the subcavitation clouds for Mode A are analysed first.
Figure 10 depicts the formation and development of Cavitation Cloud I at different instants for Mode A when
$d=0$
. Meanwhile, figure 11 shows the corresponding evolution processes of Cavitation Cloud II and III. Here,
${{t}^{*}}=0$
represents the instant when the first ultrasonic wave reaches the focal point. The upper and lower sides of each panel show the numerical schlieren image and the corresponding schematic diagram of the wave system structure, respectively. In the schematic diagram, the solid red line represents the incident wave (W); the green line represents the wave reflected from the solid wall (RW); the dashed blue line represents the scattered wave (SW); the blue shadow represents the existing cavitation cloud at the instant; the yellow shadow represents the newly formed cavitation cloud, which is also circled with the black dotted–dashed line, and the corresponding waves involved in the superposition is indicated in square brackets beside it. The subscript number represents the sequence number of the incident wave corresponding to each wave. Besides, the small subfigure located in the lower left-hand corner (as shown in figure 10
b for example) further shows the superposition mechanism of the waves, where the red line represents the incident wave, the green line represents the reflected wave and the dashed black line represents the overlapped wave.
The formation and development of Cavitation Cloud I for Mode A when
$d=0$
at (a)
${{t}^{*}}=-0.2$
, (b)
${{t}^{*}}=0.6$
, (c)
${{t}^{*}}=1.0$
, (d)
${{t}^{*}}=1.3$
, (e)
${{t}^{*}}=1.5$
and (f)
${{t}^{*}}=1.8$
.

The formation and development of Cavitation Cloud II and III for Mode A when
$d=0$
at (a)
${{t}^{*}}=3.4$
, (b)
${{t}^{*}}=7.3$
, (c)
${{t}^{*}}=10.3$
, (d)
${{t}^{*}}=11.7$
, (e)
${{t}^{*}}=14.3$
and (f)
${{t}^{*}}=24.1$
.

Firstly, in the formation and development of Cavitation Cloud I, as demonstrated in figure 10, the cavitation clouds are generated at both wall and near-wall regions. As the incident wave propagates towards the solid wall, it is reflected from the wall. Since the acoustic resistance of the solid wall is much greater than that of the fluid, the reflection from the solid wall does not change the properties of the wave itself. Therefore, the compression wave (positive pressure segment of the wave) is still a compression wave after reflection, and the rarefaction wave (negative pressure segment of the wave) is also still a rarefaction wave after reflection. In the process of propagation and reflection, there must be a superposition effect of different phases of the reflected wave and the subsequent propagated wave.
The superposition of the reflected wave and the incident wave in the negative pressure segment leads to the negative pressure peak on the wall (as shown in figure 10
b), which may induce the occurrence of cavitation on the wall once the local pressure value is lower than the local cavitation threshold. As shown in figure 10(b), due to the superposition of
$\text{R}{{\text{W}}_{1}}$
and
${\text{W}}_{1}$
, cavitation is triggered on the wall (the yellow-shaded area). Given the relatively low intensity of the first wave as shown in figure 2, the cavitation effect close to the solid wall is relatively week. As the waves evolve continually, when the negative pressure section of
$\text{R}{{\text{W}}_{1}}$
superimposed with the negative pressure section of the second incident wave
${\text{W}}_{2}$
, a new cavitation zone near the solid wall is triggered as illustrated in figure 10(c). As the intensity of the second ultrasonic wave is stronger than that of the first, the cavitation area in figure 10(c) is larger than that in figure 10(b).
Additionally, the positive pressure section of
${\text{W}}_{2}$
converges and propagates to the wall, where it is scattered by the cavitation cloud previously generated on the wall and the nature of the wave changes upon scattering (the positive pressure phase becomes negative.) This scattered wave (
$\text{S}{{\text{W}}_{2}}$
) overlaps with the remaining following negative pressure stage of
${\text{W}}_{2}$
, thus a new small cavitation area is induced between the previous two cavitation areas, as shown in figure 10(d).
While one part of
$\text{S}{{\text{W}}_{2}}$
propagates back into the flow field, another part propagates along the solid wall, and it overlaps with the negative pressure superposition section of
${\text{W}}_{2}$
and
$\text{R}{{\text{W}}_{2}}$
, which is similar with the subpanel in figure 10(b). The superposition of the negative pressure section of these three waves causes the pressure on the solid wall to drop below the cavitation threshold, thereby inducing a new cavitation area on the solid wall, as depicted in figure 10(e).
In the flow field, by focusing on the interaction between the waves and the cavitation clouds in the flow field, it can be seen that the following incident wave
${\text{W}}_{3}$
scatters on the leftmost interface of the cavitation clouds, giving rise to scattered waves
$\text{S}{{\text{W}}_{3}}$
. The overlap of
${\text{W}}_{3}$
and
$\text{S}{{\text{W}}_{3}}$
creates a lower pressure zone near the left-hand front surface of the cavitation clouds, which promotes the formation of a new cavitation area, as shown in figure 10(f). This development mechanism is similar to that presented in figure 23(b). At this point, Cavitation Cloud I is basically formed. The morphology and development process of the cavitation clouds under the first few ultrasonic excitations are consistent with the findings observed in previous studies (Osuga et al. Reference Osuga, Yasuda, Jimbo, Yoshizawa and Umemura2016). The growth of Cavitation Cloud I relies chiefly on the incident, reflected and scattered waves.
Then, turning to the other two cavitation clouds, Cavitation Cloud II on the solid wall and Cavitation Cloud III in the flow field. As these two cavitation clouds develop simultaneously, the evolution process of the two cavitation clouds is discussed and analysed in the following paragraphs along with the evolution of the flow field at different instants, as depicted in figure 11. After the formation of Cavitation Cloud I, ultrasonic waves are scattered on the left-hand surface of the cavitation cloud, giving rise to scattered waves, which is similar to the mechanism presented in figures 10(f) and 23(b). Under the influence of these scattered waves and the incident waves, the cavitation cloud commences to grow in the reverse direction, as presented in figure 11(a). The fifth incident wave and its scattered waves on the surface of the cavitation cloud continuously induce the leftward growth of the cavitation cloud, which is a prototype of Cavitation Cloud III. It is worth mentioning that the structures of the scattered waves and the incident waves are associated with the focusing characteristic. Figure 6 demonstrates the prefocused phenomenon of focusing ultrasound, where the ultrasonic waves converge at
${{z}^{*}}=-1.1$
. Owing to the convergence, the intensities of the incident and scattered waves are relatively high at around
${{z}^{*}}=-1.1$
. As the cavitation cloud grows in the reverse direction, the scattering fronts also shift to the left-hand side and gradually away from
${z}^{*}$
. As a result, the intensities of the incident and scattered waves decline. Based on this mechanism, the radial lengths of the cavitation clouds induced by the fifth and subsequent ultrasonic waves show a downward trend.
Once the superposition of a single scattered wave with the incident wave fails to reduce the pressure in the left-hand region to the cavitation threshold, the cavitation cloud instead expands towards a more distant radial position, as illustrated in figures 11(c) and 11(e). This radial growing mechanism of the cavitation clouds is quite similar to that of the third stage related to the case in Appendix B. Subsequently, scattered waves from cavitation clouds at different radial positions continue to prompt the reverse growth of the cavitation cloud, as illustrated in figure 11(f). Under this mechanism, a ‘crown-like’ cavitation cloud comes into being. Cavitation Cloud III is mainly distributed in the prefocused region and chiefly reliant on the incident and scattered waves.
While Cavitation Cloud III is growing, Cavitation Cloud II is also expanding radially along the wall, as shown in figures 11(b) and 11( d ). The growth mechanism of Cavitation Cloud II is more complex and requires the combined action of multiple ultrasonic waves.
As depicted in figure 11(b), the ultrasonic wave
${\text{W}}_{8}$
is scattered at various positions on the lateral surface of the Cavitation Cloud I and III. Here, figures 11(a), 11(b) and 11(c) correspond to the cavitation cloud in prefocused region (
${{z}^{*}}=-1.1$
), the middle of Cavitation Cloud I, and the cavitation cloud that has formed on the solid wall, respectively. The scattered waves
$\text{S}{{\text{W}}_{8}}$
from figures 11(a), 11(b) and 11(c) travel towards the solid wall and their reflected waves
$\text{RS}{{\text{W}}_{8}}$
by the solid wall (the formation mechanism of the reflected wave
$\text{RS}{{\text{W}}_{8}}$
is shown in figure 8
c, and
$\text{RS}{{\text{W}}_{8}}$
is not marked in this figure). Besides, when the incident wave
${\text{W}}_{8}$
is reflected on the solid wall, a reflected wave
$\text{R}{{\text{W}}_{8}}$
is formed. The negative pressure sections of
${\text{W}}_{8}$
and
$\text{R}{{\text{W}}_{8}}$
are superimposed, and so are those of
$\text{S}{{\text{W}}_{8}}$
and
$\text{RS}{{\text{W}}_{8}}$
. Since the moving speeds of the superposition regions on the solid wall vary (which is related to the phases of the waves), when these superposition regions overlap, the pressure on the solid wall further decreases, thus inducing the cavitation effect. The growth mechanism of the cavitation cloud shown in figure 11(d) is similar to that in figure 11(b). The difference lies in that the cavitation cloud is formed at a larger radial position. Therefore, the phase difference between the incident wave and the reflected wave is greater, and one wavelength is spanned between them. The cavitation effect is generated by the superposition of the superposition region of
${\text{W}}_{13}$
and
$\text{R}{{\text{W}}_{13}}$
and the superposition region of
$\text{R}{{\text{W}}_{12}}$
and
$\text{RS}{{\text{W}}_{12}}$
.
Initiation and development mechanism of cavitation clouds for Mode E when
$d=-1.25$
at (a)
${{t}^{*}}=0.6$
, (b)
${{t}^{*}}=2.1$
, (c)
${{t}^{*}}=2.6$
, (d)
${{t}^{*}}=9.2$
, (e)
${{t}^{*}}=15.9$
and (f)
${{t}^{*}}=24.1$
.

The reason why the growth mechanism of Cavitation Cloud II is so complex lies in the intensity of the ultrasonic waves. As mentioned in § 4.1, the negative pressure superposition of the reflected wave and the incident wave leads to a decrease in the pressure on the solid wall. On this basis, the further superposition of the scattered wave and this superposition region causes the pressure to drop further and induces the cavitation effect, as shown in figure 10(e). However, on the solid wall at a larger radial position, the intensities of both the reflected wave and the incident wave decrease. Moreover, the amplitude of the scattered wave also decreases during propagation. Therefore, the superposition of a single scattered wave with the reflected wave and the incident wave is insufficient to induce the cavitation effect on the solid wall at a larger radial position, and multiple scattered waves are required.
In summary, the cavitation clouds in Mode A can be partitioned into three distinct regions. Cavitation cloud I serves as the ‘tree root.’ In the early stage, it is comprehensively affected by the incident wave and the wall-reflected wave. In the later stage, it is jointly affected by the incident wave, the reflected wave, and the scattered wave. Cavitation Cloud II grows along the solid wall and is jointly affected by multiple superposition regions of waves and reflected waves. Cavitation Cloud III, as the ‘tree crown’, is mainly distributed within the focusing region and is affected by both incident waves and scattered waves.
The growth mechanisms of cavitation clouds in Mode E (
$d\lt 0$
) are discussed next. Similar to that in Mode A, the cavitation cloud in Mode E can be divided into three parts, as presented in figure 9. The difference is that there is an overlap between Cavitation Cloud I and Cavitation Cloud III. In this section, the analysis is mainly focused on the similarities and differences between Mode A and Mode E.
For instance, figure 12 presents the numerical results of focused ultrasonic waves over 20 cycles at different instants when
$d=-1.25$
. The upper and lower sides of each panel also show the numerical schlieren image and the corresponding schematic diagram of the wave system structure, respectively. Figure 12(a) shows the distribution of Cavitation Cloud I. Under the combined influence of the incident and reflected waves, Cavitation Cloud I initiates its formation. Subsequently, it grows from the solid wall into the fluid field, a process that closely resembles what occurs under the Mode A condition.
During the subsequent growth process of the cavitation cloud, owing to the closeness of Cavitation Cloud I to the convergence zone, the generated scattered waves are more intense, resulting in earlier cavitation effects near the solid wall and the formation of Cavitation Cloud II, as shown in figure 12(b). The growth mechanism of Cavitation Cloud II is the same as that in Mode A, being jointly affected by
$\text{S}{{\text{W}}_{\text{4 }}}\text{and} \text{ RS}{{\text{W}}_{\text{4}}}$
negative pressure superposition region and
${{\text{W}}_{\text{4 }}}\text{and}\text{ S}{{\text{W}}_{\text{4}}}$
negative pressure superposition region. Subsequently, the cavitation cloud proceeds to an alternating growth in the reverse and radial directions, as shown in figure 12(c–d). Cavitation Clouds I and III have a substantial overlap. Cavitation Cloud I functions in a manner similar to Cavitation Cloud III in figure 11(e) and can be regarded as an early stage of Cavitation Cloud III for Mode A. Hence, Cavitation Cloud III enters a phase of growth towards larger radial positions earlier, as depicted in figure 12(c). Subsequently, similar to the case for Mode A, the scattered waves from different radial positions combine with the incident wave, triggering the reverse growth of the cavitation cloud, as depicted in figure 12(d). Due to the earlier radial growth of Cavitation Cloud III, it rapidly encloses Cavitation Cloud I, reducing the number of ultrasonic waves reaching Cavitation Cloud I and prefocused region. Therefore, the growth of the cavitation cloud in the near-wall regions is mainly influenced by the scattered waves from Cavitation Clouds II (point b) and III (point c), as well as the incident and reflected waves, as shown in figure 12(e). Figure 12(e) shows the formation position of the scattered wave
$\text{S}{{\text{W}}_{17}}$
. Compared with that in Mode A, its radial position is larger. When the scattered wave formed here propagates to the solid wall, its attenuation is weaker. Therefore, it can play a role similar to that of the cavitation clouds in prefocused region and Cavitation Cloud I for Mode A. Finally, as shown in figure 12
(f), a more widely distributed Cavitation Cloud II is formed compared with the case for Mode A.
For Mode S, as shown in figure 9(e–f), the corresponding cavitation clouds can be divided into four parts. That is, Cavitation Cloud IV is formed between Cavitation Cloud I and Cavitation Cloud III. However, a slight change in the value of
$d$
can cause a significant change in the distribution of the Cavitation Cloud II. In general, there are two different forms. In this section, two cases with two different
$d$
related to each form are analysed, respectively.
Firstly, for the case of
$d=1$
, the evolution of the cavitation cloud over 20 consecutive cycles is illustrated in figure 13. Similar to Mode A and Mode E, Cavitation Cloud I develops from the solid wall into the fluid field under the combined influence of the incident and reflected waves as presented in figure 13(a). The incident wave
${\text{W}}_{3}$
scatters on the left-hand side of the previously generated cavitation cloud to form the scattered wave
$\text{S}{{\text{W}}_{3}}$
, where the properties of the scattered wave change (the positive pressure changes to negative pressure). When the scattered wave
$\text{S}{{\text{W}}_{3}}$
propagates to the left-hand side, its negative pressure section overlaps with the negative pressure sections of the incident waves
${\text{W}}_{3}$
and
${\text{W}}_{4}$
at positions
${{z}^{*}}=-0.25$
and
${{z}^{*}}=-0.75$
, respectively, reducing the pressure and inducing the formation of the cavitation cloud, as shown in figure 13(a). As the scattering position is far from the focal region, the peak pressure of the incident wave
${\text{W}}_{3}$
at scattering position is relatively low, as shown in figure 6. The intensity of the formed scattered wave
$\text{S}{{\text{W}}_{3}}$
is also weak. As a result, the cavitation effect can only be induced when the negative pressure peak of the incident wave
$\text{S}{{\text{W}}_{3}}$
overlaps with that of the reflected wave
${\text{W}}_{3}$
and
${\text{W}}_{4}$
. In this case, the cavitation clouds form a linear structure.
Subsequently, the incident wave
${\text{W}}_{5}$
is scattered on the surface of these two linear cavitation clouds, generating the scattered wave
$\text{S}{{\text{W}}_{5}}$
. Here
$\text{S}{{\text{W}}_{5}}$
coincides with the incident waves
${\text{W}}_{5}$
and
${\text{W}}_{6}$
, inducing spherical cavitation clouds at
${{z}^{*}}=-1$
and
${{z}^{*}}=-0.5$
, as shown in figure 13(b). Different from the previous instant, the scattering position is closer to the focal region, and the generated scattered wave
$\text{S}{{\text{W}}_{5}}$
is stronger. Therefore, two more concentrated spherical cavitation clouds are induced. Due to their more concentrated distribution, the scattered waves from the surface of the spherical cavitation cloud possess a higher intensity. In contrast, the linear cavitation cloud is dispersed, causing the incident waves to scatter at different positions and resulting in relatively weaker scattered waves after superposition. The cavitation cloud formed at the position of
${{z}^{*}}=-1$
(point a) is crucial for the subsequent growth of Cavitation Cloud II along the solid wall, which is mentioned in figure 11. The scattered waves formed by ultrasonic waves at
${{z}^{*}}=-1$
propagate towards the solid wall, where they combine with the incident waves and scattered waves from the Cavitation Cloud I (point b), thus inducing newly cavitation regions on the solid wall, as depicted in figure 13(c). The growth mechanism of Cavitation Cloud II along the solid wall is the same as that in Mode A, involving the interaction of incident wave
${\text{W}}_{9}$
, reflected wave
$\text{R}{{\text{W}}_{9}}$
, scattered wave
$\text{S}{{\text{W}}_{9}}$
and reflected scattered wave
$\text{RS}{{\text{W}}_{9}}$
. Thereafter, the scattered waves from different positions within Cavitation Cloud III prompt the continued left-hand growth of the cavitation cloud. The cavitation cloud formed after the action of 20 ultrasonic waves is shown in figure 13(d). An additional Cavitation Cloud IV forms between Cavitation Clouds I and III, mainly concentrated near the
$z$
-axis.
Initiation and development mechanism of cavitation clouds for Mode S when
$d=1$
at (a)
${{t}^{*}}=3.1$
, (b)
${{t}^{*}}=4.0$
, (c)
${{t}^{*}}=9.8$
and (d)
${{t}^{*}}=24.1$
.

Initiation and development mechanism of cavitation clouds for Mode S when
$d=1.25$
at (a)
${{t}^{*}}=3.3$
, (b)
${{t}^{*}}=3.8$
, (c)
${{t}^{*}}=8.6$
and (d)
${{t}^{*}}=24.1$
.

Next, the other form of Mode S is analysed. For the case of
$d=1.25$
, the evolution of the cavitation cloud over 20 consecutive cycles is illustrated in figure 14. The Cavitation Cloud I develops from the solid wall into the fluid field under the combined influence of the incident and reflected waves as presented in figure 14(a). Then the incident wave
${\text{W}}_{3}$
scatters on the left-hand side of the previously generated cavitation cloud to form the scattered wave
$\text{S}{{\text{W}}_{3}}$
, which is similar with the case under
$d=1$
. As the scattered wave
$\text{S}{{\text{W}}_{3}}$
propagates leftward, the distance between the convergence zone and the conical cavitation clouds increases by
$0.25\lambda$
compared with the
$d=1$
case. The regions of overlapped negative pressure are shifted to
${{z}^{*}}=0$
,
${{z}^{*}}=-0.5$
and
${{z}^{*}}=-1$
, as presented in figure 14(a). Likewise, a spherical cavitation cloud is formed at
${{z}^{*}}=-0.75$
, as shown in figure 14(b). The cavitation clouds at
${{z}^{*}}=-1$
are linear cavitation clouds (point a), and their scattered wave intensity is relatively low. Therefore, newly generated cavitation clouds only form in the region with a small radial position on the solid wall, as depicted in figure 14(c). The superposition of incident wave
${\text{W}}_{8}$
, reflected wave
$\text{R}{{\text{W}}_{8}}$
, scattered wave
$\text{S}{{\text{W}}_{8}}$
and reflected scattered wave
$\text{RS}{{\text{W}}_{8}}$
is unable to induce cavitation effects in the region with a relatively larger radial position. After the Cavitation Cloud III is formed, it shields part of the incident wave, further inhibiting the growth of Cavitation Cloud II along the solid wall. The cavitation cloud formed after 20 cycles of ultrasonic waves is shown in figure 14(d). The distribution of cavitation cloud II is considerably smaller than that in the
$d=1$
case.
In summary, the growth of Cavitation Cloud II in Mode S is closely related to the structure of the cavitation cloud in the prefocused region (
${{z}^{*}}=-1$
). When the cavitation cloud in the prefocused region is a spherical one, Cavitation Cloud II has a wider distribution. Conversely, the growth of Cavitation Cloud II is inhibited. The structure of the cavitation cloud in the prefocused region is related to the phase difference between the incident wave and the scattered wave. When the position of the solid wall changes from
$d=1$
to
$d=1.25$
, the superposition region of the negative pressure peak of the incident wave and that of the scattered wave also changes by a quarter wavelength synchronously. As a result, the cavitation cloud in the prefocused region changes from a spherical one to a linear one, leading to a significant change in the distribution of Cavitation Cloud II.
(a) The maximum radial dimension of the cavitation cloud and (b) cavitation cloud coverage area varying with
$d$
after the action of 20 cycles of a 1 MHz ultrasonic wave.

The distribution of Cavitation Cloud II is a crucial metric for evaluating the cavitation effect on the solid wall. The non-dimensional area
$S$
of the solid wall covered by the cavitation cloud can be formulated as
\begin{equation} S=\frac {\int _{0}^{{{R}_{\textit{max}}}}{\chi \boldsymbol{\cdot }r{\rm d}r}}{\int _{0}^{{{R}_{\textit{max}}}}{r{\rm d}r}},\text{ }\chi =\begin{cases} 1, & {\alpha }_{{b}}\geqslant 0.1, \\ 0, & {\alpha }_{{b}}\lt 0.1, \end{cases} \end{equation}
where
${R}_{\textit{max}}$
stands for the maximum radial dimension of the cavitation cloud in the near-wall region among all cases with
$d$
ranging from −1.5 to 1.25;
$\chi$
indicates the parameter to identify whether local on the wall cavitation occurs. Figure 15 depicts the trends of the non-dimensional maximum radial dimension
$r_{\textit{max}}^{*}=({1}/{\lambda })\max r,\text{ where }{{\alpha }_{{b}}} ( r )\geqslant 0.1$
of the cavitation cloud and the non-dimensional area
$S$
with the change of
$d$
, respectively.
The results reveal that both the maximum radial extent and the cavitation cloud coverage area reach their peak values at
$d=0$
(Mode A). As shown in figure 11, the formation and growth of Cavitation Cloud II in this case are ascribed to the combined impact of incident waves, scattered waves from
$z^{*}=-1$
, scattered waves from the middle and root of the conical Cavitation Cloud I, and their respective waves reflected from the solid wall.
In Mode E, Cavitation Cloud III enwraps Cavitation Cloud I and obstructs a portion of the ultrasound that propagates towards it. The scattered waves from
${{z}^{*}}=-1$
and Cavitation Cloud I are either non-existent or feeble. The growth of Cavitation Cloud II is primarily influenced by the scattered waves from the surface of Cavitation Cloud III and the incident waves (and their reflected waves on the solid wall). The intensity of the scattered waves depends on the scattering position. When
$d=-0.25$
, the scattering position lies within the ultrasonic diffraction zone. As
$d$
decreases, the scattering position approaches the convergence zone. When the scattering position is closer to the ultrasonic convergence zone, the incident diffraction waves are more potent, resulting in stronger scattered waves and an increasing tendency in
${r}_{\textit{max}}$
. However, as
$d$
decreases, the intensity of the incident focused wave declines, restricting the area where cavitation occurs within the dominant region of the focused wave. Hence, a reduction in
$d$
leads to a decrease in
$S$
. Nevertheless, there are two exceptions: the cases of
$d=-0.5$
and
$d=-1.25$
. When
$d=-0.5$
, the middle section of the Cavitation Cloud I is positioned near
${{z}^{*}}=-1$
, which is the prefocused area. The intense scattered waves from this middle section cause
${r}_{\textit{max}}$
and
$S$
to reach peak values. In addition, as previously stated, Cavitation Cloud III is mainly distributed in the convergence zone, i.e.
${{z}^{*}}\lt -1$
. When
$d=-1.25$
, Cavitation Cloud III is also located in the near-wall regions, contributing to the peak value of
$S$
.
Evolution of the cavitation cloud area on the wall (per ultrasonic cycle) with the number of ultrasonic cycles for a 1 MHz HIFU.

In Mode S, due to the relatively large distance between the solid wall and the convergence zone, the influence of the scattered waves from Cavitation Cloud III is weaker. The formation and development of Cavitation Cloud II are affected by the scattered waves from the cavitation cloud at
${{z}^{*}}=-1$
, the scattered waves from Cavitation Cloud I and incident waves (and their reflected waves on the solid wall). Since the structure and intensity of the scattered waves from Cavitation Cloud I and incident waves are similar under different
$d$
values, the growth of Cavitation Cloud II in Mode S depends on the scattered waves from the cavitation cloud at
${{z}^{*}}=-1$
. As mentioned earlier, the cavitation cloud at
${{z}^{*}}=-1$
has two kinds of structures: spherical and linear. The spherical cavitation cloud generates stronger scattered waves, while the linear one generates weaker scattered waves. The structure of the cavitation cloud at
${{z}^{*}}=-1$
changes every
$0.25\lambda$
as the solid wall position increases. Therefore,
${r}_{\textit{max}}$
and
$S$
vary periodically with
$d$
, with a period of
$0.5\lambda$
.
In practical application scenarios, the number of HIFU cycles is a crucial parameter that significantly influences the treatment efficacy. It needs precise control and optimization in accordance with specific application contexts and objectives. The number of cycles determines the duration of energy transfer and the cumulative effect, thereby affecting the damage extent of the targeted object. Figure 16 shows the evolution of the cavitation cloud area on the wall (induced per ultrasonic cycle) as a function of the number of ultrasonic cycles (or time) for a 1 MHz HIFU. For Mode E, during the initial 10 cycles, the cavitation cloud induced by the ultrasonic waves exhibits a remarkable growth rate, which is higher than that of Mode A and Mode S. This finding is consistent with the numerical results in figure 15. The underlying reason is that the scattered waves from Cavitation Cloud I trigger the formation of Cavitation Cloud III at an earlier stage. The cases of
$d = 0$
and
$d = 1$
share similarities. In the first 10 cycles, the growth of the cavitation cloud shows a slowdown trend. However, during the subsequent 10 cycles, a substantial increase in the cavitation cloud is observed. In the first 10 cycles, the ultrasonic waves induce the formation of Cavitation Clouds I and IV. In the following 10 cycles, the ultrasonic waves scatter on the surface of the preformed cavitation clouds, and the scattered waves prompt the radial growth of Cavitation Cloud II. In the case of
$d = 1.25$
, owing to the relatively weak intensity of the scattered waves originating from the prefocused region, the growth rate of the cavitation cloud is comparatively lower than that in the other cases.
4.3. Wave superposition model for the distribution of cavitation clouds in the near-wall region
Based on the aforementioned mechanism of how scattered waves affect the development of cavitation clouds in near wall regions, it is feasible to construct a theoretical model describing the variation of Cavitation Cloud II distribution with the relative position between the solid wall and the focal point. In order to facilitate the establishment of the model, the following assumptions are made.
-
(i) As depicted in § 4.2, the growth of Cavitation Cloud II is primarily influenced by the incident wave, scattered waves from the prefocused area (
${{z}^{*}}=-1$
), scattered waves from Cavitation Cloud I, scattered waves from the previously generated Cavitation Cloud II and their respective reflected waves on the solid wall. Although scattered waves are also generated at other positions, their intensities are very weak compared with these major waves, and thus contribute little to the development of cavitation. Therefore, it is reasonable to neglect the impact of other scattered waves and only consider these major waves when establishing the model. -
(ii) The cavitation clouds are densely distributed, similar to point sources, and can be regarded as single bubbles. For the propagation process of the wave, a one-dimensional spherically symmetric propagation model is considered here to describe its propagation behaviour (Zhou & Zhong Reference Zhou and Zhong2006),
(4.3)where
\begin{equation} p({{R}_{i}})=p({{R}_{0}})\frac {{{R}_{i}}}{R}, \end{equation}
${R}_{0}$
represents the initial radius of the scattered waves. We adopt the equivalent radius
${{R}_{0}}=\sqrt [{}]{ {{{S}_{0}}}/{4\pi }}$
of the cavitation cloud at the cavitation site as the initial radius of the scattered waves. Here
${S}_{0}$
represents the cross-sectional area of the cavitation cloud at the scattering position in two dimensions.
-
(iii) The solid wall is assumed to be a rigid wall. The reflection process does not result in wave energy loss.
-
(iv) The sound wave maintains a spherical wave during propagation and it ignores local variations in sound speed and pressure field non-uniformity caused by nonlinear effects (such as shock wave formation).
In addition, the nonlinear interaction between bubbles and cavitation clouds can also affect the pressure distribution in the flow field (Ma et al. Reference Ma, Yin, Zhang, Tian, Tian, Huang and Chen2024). For example, multiple reflections of ultrasonic waves at the cavitation cloud interface may lead to increased complexity in the local pressure distribution. However, the model adopted in this paper focuses on predicting the achievable peak negative pressure in the flow field. Considering that the acoustic intensity of the ultrasonic wave after multiple reflections is usually lower than that of the incident wave and the first reflected wave (the acoustic intensity decreases significantly with the increase in the number of reflections), it can be reasonably inferred that their direct impact on the flow field peak pressure is relatively small; thus, the model simplification is justified.
As indicated by the numerical calculation results, Cavitation Cloud I adheres directly to the solid wall and exhibits a conical distribution pattern. It plays a crucial role in the formation of the entire tree-like cavitation cloud. The scattered waves on its surface induce the development of Cavitation Cloud II (radial growth) and Cavitation Cloud III (reverse growth within the flow field). Consequently, our focus is on the formation mechanism of Cavitation Cloud I. Therefore, we first aim to predict its distribution using the wave superposition model. Cavitation Cloud I is induced by the superposition of the first reflected wave and the negative-pressure phase of subsequent incident waves. Under the assumption of axisymmetric conditions, the incident wave propagates along the
$+z$
direction, while the reflected wave travels along the
$-z$
direction. The space–time pressure field is described as follows:
where ‘in’ represents the incident wave, ‘r’ represents the reflected wave. The time domain is
$t\in [ 0,3T ]$
, corresponding to three consecutive periods. The first reflected wave is generated by the reflection of the first incident wave on the solid wall. Given that the wall is rigid (with an acoustic impedance significantly higher than that of the fluid), the pressure reflection coefficient is + 1. Hence, the reflected waveform is identical to the incident waveform, expressed as
Consequently, the pressure can be expressed as
Substituting the wave superposition model,
the pressure can be expressed as
The region where the pressure falls below the cavitation threshold represents the initial formation area of Cavitation Cloud I. A comparison between the theoretical and numerical results for
$d = 0$
is presented in figure 17. The close agreement in distribution between the two validates that the wave superposition model employed in this study can accurately predict the distribution of the pressure field under the action of ultrasonic waves and the formation of Cavitation Cloud I.
Comparison between theoretical and numerical solutions for the initial formation region of Cavitation Cloud I when
$d = 0$
.

Schematic diagram of the wave superposition model when
$d = 0$
(incident wave
$p_{\textit{in}}$
, scattered waves
$p_{{s}}^{{a}}$
,
$p_{{s}}^{{b}}$
,
$p_{{s}}^{{c}}$
and their reflected waves).

The maximum distribution range of Cavitation Cloud II on the solid wall can then be predicted based on the distribution of Cavitation Cloud I. Based on the above assumptions, we can derive the pressure distribution on the wall which is directly related to the formation and growth of Cavitation Cloud II. Based on the first assumption, the pressure on the solid wall can be represented as
where the subscript ‘s’ represents the scattered wave; the superscripts ‘a’, ‘b’ and ‘c’ refer to the points marked in figure 18 and, respectively, represent the prefocused region, the Cavitation Cloud I and the previously generated Cavitation Cloud II. Based on the third assumption, the intensity of reflected waves can be represented as
Substituting (4.10) into (4.9), we have
Based on the second assumption, the three scattered waves can be considered as one-dimensional spherical waves originating from the scattering locations. Neglecting transmission, due to the difference in acoustic impedance, the amplitude of the scattered waves undergoes a phase inversion relative to the incident waves,
\begin{equation} \begin{aligned} p_{{s}}^{{a}}({{z}_{{w}}},r,t)&=-{{p}_{\textit{in}}}\left({{z}_{{a}}},{{r}_{{a}}},t-\frac {\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{a}}} \right )}^{2}}+{{\left ( r-{{r}_{{a}}} \right )}^{2}}}-{{R}_{{a0}}}}{c}\right)\\&\quad\times\frac {{{R}_{{a0}}}}{\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{a}}} \right )}^{2}}+{{\left ( r-{{r}_{{a}}} \right )}^{2}}}}, \\ p_{{s}}^{{b}}({{z}_{{w}}},r,t)&=-{{p}_{\textit{in}}}\left({{z}_{{b}}},{{r}_{{b}}},t-\frac {\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{b}}} \right )}^{2}}+{{\left ( r-{{r}_{{b}}} \right )}^{2}}}-{{R}_{{b0}}}}{c}\right)\\&\quad\times\frac {{{R}_{{b0}}}}{\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{b}}} \right )}^{2}}+{{\left ( r-{{r}_{{b}}} \right )}^{2}}}}, \\ p_{{s}}^{{c}}({{z}_{{w}}},r,t)&=-{{p}_{\textit{in}}}\left({{z}_{{c}}},{{r}_{{c}}},t-\frac {\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{c}}} \right )}^{2}}+{{\left ( r-{{r}_{{c}}} \right )}^{2}}}-{{R}_{{c0}}}}{c}\right)\\&\quad\times\frac {{{R}_{c0}}}{\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{c}}} \right )}^{2}}+{{\left ( r-{{r}_{{c}}} \right )}^{2}}}}, \end{aligned} \end{equation}
where
$ ( {{z}_{{a}}},{{r}_{{a}}} )$
,
$ ( {{z}_{{b}}},{{r}_{{b}}} )$
and
$ ( {{z}_{{c}}},{{r}_{{c}}} )$
, respectively, represent the coordinates of the three regions where scattering occurs;
${R}_{{a0}}$
,
${R}_{{b0}}$
and
${R}_{{c0}}$
denote the initial radius of the three scattered waves which are equivalent to the equivalent radius of the cavitation cloud where scattering occurs; the pressure of the incident waves
${p}_{\textit{in}}$
is defined using (3.6).
For Modes A and S, scattering in the prefocused region occurs on the
$z$
-axis. For Mode E, however, due to Cavitation Cloud III covering Cavitation Cloud I, the scattering location is at a certain distance from the
$z$
-axis. Besides, for Modes A and S, the relative positions of the scattering locations in the Cavitation Cloud I remain constant with respect to the solid wall. For Mode E, due to Cavitation Cloud III covering Cavitation Cloud I, its radial position is farther away from the
$z$
-axis. Based on the numerical results and wave superposition model, the scattering positions are obtained:
${{z}_{{a}}}=\min ( -\lambda ,{{z}_{{w}}}-\lambda )$
;
${{z}_{{b}}}={{z}_{{w}}}-0.6\lambda$
;
${{z}_{{b}}}={{z}_{{w}}}-0.01\lambda$
;
${{r}_{{a}}}=0$
;
${{r}_{{b}}}=0.48\lambda$
;
${{r}_{{c}}}=0.9\lambda$
for
$d\lt 0$
and
$0.75\lambda$
for
$d\geqslant 0$
.
Ignoring the scattering effects of newly formed cavitation clouds, the occurrence of cavitation can be determined by assessing the peak negative pressure. When the pressure reaches the cavitation threshold, the region is considered to contain cavitation clouds. Figure 19 presents a comparison of the maximum radial extent of the cavitation clouds in the near-wall regions between the numerical and theoretical results obtained from this model. Despite some discrepancies in quantitative values in individual cases, the wave superposition model effectively captures the variation trends of near-wall cavitation clouds for both Mode E and Mode S as a function of
$d$
. This shows that this theoretical wave superposition model, based on the mechanism analysis in § 4.2, can effectively describe the main phenomena of the flow field and give an accurate prediction of the wall cavitation distribution.
Comparison of the maximum radial dimension of the cavitation clouds in near wall regions between numerical and wave superposition model results.

4.4. Mechanisms under different frequencies and model expansion
The ultrasonic frequency is a pivotal parameter that significantly influences its applications. In this section, the theoretical model proposed for predicting the maximum radial distribution of the wall cavitation cloud is extended and applied to the 2 MHz HIFU scenario. Moreover, the model is adaptively optimized and validated in light of the special physical mechanisms observed at this frequency. Given that the initiation of cavitation predominantly hinges on the intensity of the negative pressure phase of the acoustic field, while maintaining the negative pressure peak consistent with the 1 MHz operating condition, the following initial ultrasonic waveform is adopted:
where
${{p}_{0}}=1.01325\times {{10}^{5}}\text{ Pa}$
represents the environmental pressure;
${{p}_{w}}=5.6367\times {{10}^{6}}\text{ Pa}$
,
$a =-1.0\times {{10}^{6}}\text{ }{{\text{s}}^{-1}}$
,
$\omega =5.232\times {{10}^{6}}\text{ }{{\text{s}}^{-1}}$
,
${{\beta }_{1}}={{\beta }_{2}}=-4\times {{10}^{7}}\text{ }{{\text{s}}^{-1}}$
,
$\varphi =2\pi \text{/}3$
and
$T=1\times {{10}^{-6}}\text{ s}$
. The positive and negative pressure peaks are 60 MPa and −19 MPa, respectively, which are in accordance with the calculation results based on the KZK equation.
Under the action of 2 MHz HIFU, when
$d = 0$
and
$d = 0.25$
, numerical simulations unveil a complex cavitation cloud evolution mechanism distinct from that of the 1 MHz case, as depicted in figure 20. Specifically, the scattered waves generated by the initial cavitation cloud
$a_1$
can induce the formation of a new cavitation cloud
$a_2$
. Consequently, Cavitation Cloud I transforms from a conical shape in the 1 MHz case to a cylindrical one, as shown in figure 20(a). The presence of the
$a_2$
cloud exerts a notable shielding effect on the incident wave, impeding the propagation of a portion of the acoustic energy towards the target cavitation cloud
$b$
. More critically, in this scenario, the acoustic wave experiences a multiscattering process: the incident wave is first scattered by the cavitation cloud
$a_2$
, and then this scattered wave propagates to the cavitation cloud
$b$
where it undergoes a second scattering. Eventually, the resulting scattered wave, carrying dual scattering information, propagates in the direction of the wall.
To precisely capture the impact of this special mechanism on the distribution of wall cavitation clouds, this study makes two crucial optimizations to the original theoretical model.
-
(i) For the
$d = 0$
and
$d = 0.25$
conditions, the scattering source
$a_1$
cloud is replaced by the
$a_2$
cloud, which has a shielding effect on the
$a_1$
cloud, as illustrated in figure 20(b). The position and intensity of the
$a_2$
cloud emerge as the core factors influencing the subsequent scattering process. -
(ii) For the
$d = 0$
and
$d = 0.25$
conditions, a portion of the incident waves undergoes two consecutive scatterings at
$a_2$
and
$b$
, leading to a double phase flip of the scattered waves. This physical process substantially modifies the characteristics of the scattered acoustic field that ultimately reaches the wall region. Based on the analysis of this double scattering path, the expression for the intensity of the secondary scattered wave propagating towards the wall under this specific mechanism is deduced as(4.14)
\begin{align} p_{{s}}^{{b}}({{z}_{{w}}},r,t)&={{p}_{\textit{in}}}({{z}_{{a2}}},{{r}_{{a2}}},t-{{t}_{{a2,b}}}-{{t}_{{b,w}}})\frac {{{R}_{{a2}}}}{\sqrt {{{\left ( {{z}_{{b}}}-{{z}_{{a2}}} \right )}^{2}}+{{\left ( {{r}_{{b}}}-{{r}_{{a2}}} \right )}^{2}}}}\nonumber\\&\quad\boldsymbol{\cdot }\frac {{{R}_{{b0}}}}{\sqrt {{{\left ( {{z}_{{w}}}-{{z}_{{b}}} \right )}^{2}}+{{\left ( r-{{r}_{{b}}} \right )}^{2}}}}, \end{align}
where
${{t}_{{a2,b}}}={\sqrt {{{ \!( {{z}_{{b}}}\!-\!{{z}_{{a2}}} )}^{2}}\!+\!{{ ( {{r}_{{b}}}\!-\!{{r}_{{a2}}} )}^{2}}}\!-\!{{R}_{{a2}}}}/{c}$
and
${{t}_{{b,w}}}={\sqrt {{{ \!( {{z}_{{w}}}\!-\!{{z}_{{b}}} )}^{2}}\!+\!{{ ( r\!-\!{{r}_{{b}}} )}^{2}}}\!-\!{{R}_{{b0}}}}/{c}$
, respectively, denote the time required for the ultrasonic wave to propagate from
$a_2$
to
$b$
and from
$b$
to the wall. Substituting (4.14) into (4.11) enables the acquisition of the pressure distribution on the wall under the action of 2 MHz HIFU. The reason that this mechanism is only present for
$ d = 0$
and
$ d = 0.25$
is as follows: for other values of
$ d$
, the cavitation cloud in the prefocused region is either too far from the cloud at location
$ b$
or is surrounded by Cavitation Cloud III. Consequently, this mechanism does not come into play when
$ d$
falls within other intervals.
The novel evolution mechanism of the cavitation cloud under the action of 2 MHz HIFU. (a) The distribution of the cavitation cloud after the application of 20 ultrasonic cycles when
$ d = 0$
. (b) A schematic illustration of the distribution of Cavitation Cloud I and the scattered waves originating from Cavitation Cloud I when
$ d = 0$
.

Based on the above optimized model that incorporates the analysis of the special mechanism, we systematically predict the trend of the maximum radial distribution of the wall cavitation cloud as a function of
$d$
in the 2 MHz HIFU acoustic field. To rigorously validate the predictive ability and generality of the model, the theoretical predictions of the model are compared with the numerical results (figure 21). The comparison indicates discrepancies in the predictions for
$ d \leqslant -0.75$
. Under these conditions, the distribution of cavitation clouds resembles that shown in figure 12. This discrepancy arises because the model predicts the pressure field based solely on scattered waves from a portion of the cavitation clouds. Nevertheless, overall, the prediction curves of the optimized model exhibit good agreement with the numerical calculation results across the entire range of focus positions investigated in this study. This compellingly demonstrates that the theoretical model framework proposed in this study has remarkable generalization capabilities. The model can transcend the initial 1 MHz frequency constraint of the original modelling and successfully adapt to the higher frequency (2 MHz) HIFU acoustic field. It can accurately predict the maximum radial distribution characteristics of the wall cavitation cloud at different focus positions, particularly in special conditions featuring complex multiscattering and shielding effects (such as the formation of the
$a_2$
cloud). This validation fully proves the effectiveness of the model’s core physical concept and the broad applicability of its scope.
Notably, when the characteristics of ultrasonic waves change significantly, the growth mechanism of cavitation clouds is affected. Thus, the conclusions derived in this paper are applicable to certain ultrasonic operating conditions. In Appendix C, an analysis of the applicable scope of the conclusions presented herein is provided.
Comparison of the maximum radial dimension of the cavitation clouds in near wall regions under 2 MHz HIFU between numerical and wave superposition model results.

5. Conclusion
This paper presents a numerical investigation into the problem of ultrasound-induced cavitation evolution for both with and without the existence of the solid wall. It is demonstrated that the numerical method used in this paper can accurately capture the propagation and convergence of ultrasonic waves, as well as the initiation and development process of cavitation clouds. The main contributions of this paper are summarized as follows.
First, the convergence behaviour of convergent ultrasound is quantitatively analysed, and the basic characteristics of convergent ultrasound and the generation of diffraction waves are analysed based on Huygens–Fresnel principle. The theoretical model is given for the convergent ultrasonic propagation to explain its prefocused phenomenon.
Second, in the presence of the solid wall, the propagation and convergence characteristics of ultrasonic waves and the inception and growth mechanism of cavitation clouds are investigated. Different related position of the solid wall
$d$
ranging from −1.5 to 1.25 is considered to explore the variation law of the cavitation cloud distribution. It is found that the cavitation cloud distributions can be divided into three or four subcavitation clouds and generally present a ‘tree-like’ structure. According to the relative positions of Cavitation Cloud I and Cavitation Cloud III, three modes are classified: embedded mode (Mode E), adjacent mode (Mode A) and spaced mode (Mode S). The formation and development mechanism of cavitation cloud in each region under each model is discussed in detail. In particular, for Mode S (
$d\gt 0$
), the structure of the cavitation cloud in the prefocused region is related to the phase difference between the incident wave and the scattered wave. The cavitation cloud structure at prefocused region changes every quarter wavelength increase in the solid wall position; therefore, the distribution of Cavitation Cloud II demonstrates a periodic variation with a period of half wavelength.
Third, based on the understandings of mechanism of cavitation evolution, a theoretical wave superposition model is constructed, describing the distribution of Cavitation Cloud II along the solid wall as it varies with the relative position from the focal point to the solid wall. In this model, the influences of the main wave patterns, including the incident waves, scattered waves from the prefocused area (
${{z}^{*}}=-1$
), scattered waves from Cavitation Cloud I, scattered waves from previously generated clouds on the solid wall and their reflected waves on the solid wall, are all considered. Then, the wall pressure as well as the maximum radial dimension of the cavitation clouds on the solid wall can be theoretically obtained. The theoretical results obtained from the wave superposition model show good agreement with the numerical results and can effectively capture the variation trends of near-wall cavitation clouds for all modes as a function of
$d$
. Finally, the universality of the core mechanism and model has been verified through a 2 MHz HIFU case study. The growth of the cavitation cloud under 2 MHz HIFU remains driven by the superposition of the negative pressures of the incident and scattered waves. Comparative analyses of numerical and model results across different frequencies also exhibit remarkable consistency.
This understanding of the ultrasonic cavitation mechanism may support precise control in future HIFU applications. Although this paper centres on ultrasonic cavitation, the insights gained regarding the evolution of complex wave systems across media, the resulting transient mechanisms, the associated simulation methods and the developed superposition model of wave systems are all transferable to a wide range of fluid-related scenarios and applications, such as underwater explosion dynamics (Bogatko & Kolton Reference Bogatko and Kolton1974; Shi et al. Reference Shi, Huang, Wang, Wang and Li2015; Yan, Zhao & Niu Reference Yan, Zhao and Niu2017), biomedical ultrasound (Sankin, Yuan & Zhong Reference Sankin, Yuan and Zhong2010) and cavitation erosion research (Wu et al. Reference Wu, Liu and Wang2021). This framework might also provide a foundational theoretical and modelling basis for further investigation and design in related engineering and scientific contexts.
Funding
This work was supported by the National Natural Science Foundation of China (nos. 12472249; 12325206), Young Elite Scientists Sponsorship Program by CAST (no. 2022QNRC001), the Fundamental Research Funds for the Central Universities (XSQD-6120220065) and the BIT Research and Innovation Promoting Project (no. 2024YCXY017).
Declaration of interests
The authors report no conflict of interest.
Appendix A. The solution for the initial waveform
The waveform of the initial field used in numerical calculations is formed after the ultrasonic wave emitted by the transducer propagates a certain distance. During the propagation process, the diffraction coefficient
$G$
represents the linear amplification coefficient relative to the pressure on the transducer surface (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009; Rosnitskiy, Yuldashev & Khokhlova Reference Rosnitskiy, Yuldashev and Khokhlova2015; Khokhlova et al. Reference Khokhlova, Yuldashev, Rosnitskiy, Maxwell, Kreider, Bailey and Sapozhnikov2016). The definition of
$G$
is
where
${f}_{0}$
and
${c}_{0}$
represent the frequency of ultrasound and wave speed, respectively;
$R$
represents the distance to the focal point. The propagation of the converging wave can be approximated as a one-dimensional spherical wave (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). For a converging spherical wave, the linear solution of the pressure amplitude on its
$z$
-axis can be expressed as
where
${p}_{0}$
represents the pressure amplitude at the source. Substituting the position of the focal point
$z=0$
,
$A ( 0 )=G$
means that, for a spherical wave with an initial amplitude of
${p}_{0}$
, its amplitude becomes
$G{{p}_{0}}$
when it linearly propagates to the focal point. Assume that the position of the wavefront in the computational domain is at
${z}_{0}$
, substituting
$z={{z}_{0}}=-8.67\lambda$
into the (A2) allows us to obtain the amplitude of the initial field for numerical calculation
$p ( {{z}_{0}} )$
.
Furthermore, in nonlinear acoustic beams, due to the combined effects of nonlinearity and diffraction, the focal waveform becomes asymmetrical; the positive pressure peak significantly exceeds the negative pressure peak (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009). This nonlinear effect is influenced by the nonlinear coefficient (Rosnitskiy et al. Reference Rosnitskiy, Yuldashev and Khokhlova2015),
where
$\varepsilon$
represents the nonlinearity of the medium. In this case, the nonlinear gain coefficients for positive pressure and negative pressure at the focal point are, respectively,
${{K}_{\!p+}}=3.7$
and
${{K}_{\!p-}}=0.67$
. In theory, the positive and negative pressures at the focal point are
${{K}_{\!p+}}G{{p}_{0+}}$
and
${{K}_{\!p-}}G{{p}_{0-}}$
, respectively. In Maxwell’s experiment (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), the measured positive and negative pressure peaks at the focal point were
${{p}_{\!{F+}}}=85\text{ MPa}$
and
${{p}_{\!{F-}}}=-19\text{ MPa}$
, respectively. The basic parameters of the ultrasonic waves in this study are set to match those used in the experiment. Thus, the pressure at the wave source is
For numerical simulation, at the initial instance, the nonlinearity of the ultrasonic wave is not fully pronounced, and
$K_{\!p+}^{\prime }\approx 1.05,K_{\!p-}^{\prime }\approx 0.95$
at
$z={{z}_{0}}$
(Rosnitskiy et al. Reference Rosnitskiy, Yuldashev and Khokhlova2015). Therefore, the pressure amplitude of the initial field in numerical calculation can be expressed as
\begin{equation} \begin{aligned} & {{p}_{\textit{max}}}=\frac {{{p}_{\!{F+}}}K_{\!p+}^{\prime }A\left ( {{z}_{0}} \right )}{{{K}_{\!p+}}A\left ( -{{R}_{\!{F}}} \right )}, \\ & {{p}_{\textit{min}}}=\frac {{{p}_{\!{F-}}}K_{\!p-}^{\prime }A\left ( {{z}_{0}} \right )}{{{K}_{\!p-}}A\left ( -{{R}_{\!{F}}} \right )}. \end{aligned} \end{equation}
Based on
${p}_{\textit{max}}$
and
${p}_{\textit{min}}$
,
${p}_{w}$
and
$\alpha$
can be calculated. Besides, the process of positive pressure convergence is accompanied by the development of shock waves. Based on Godunov’s shock wave model (Godunov et al. Reference Godunov, Klyuchinskii, Fortova and Shepelev2018), the length of the wavefront remains constant, which determines the
$\omega$
,
$\varphi$
and
$\beta$
.
Development of cavitation clouds together with contours of pressure and density schlieren diagrams in the focused ultrasound field at (a)
${{t}^{*}}=1.2$
, (b)
${{t}^{*}}=2.3$
, (c)
${{t}^{*}}=7.0$
, (d)
${{t}^{*}}=10.9$
, (e)
${{t}^{*}}=11.6$
and (f)
${{t}^{*}}=24.1$
.

Appendix B. Initiation and development of cavitation clouds in the absence of a solid wall
On the basis of understanding the propagation and convergence characteristics of focused ultrasonic field in the §§ 3.3 and 3.4, this appendix further discusses the dynamical evolution properties of cavitation clouds in the absence of a solid wall considering the effect of the cavitation inception. Referring to the experiment of Maxwell (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), 20 consecutive cycles of focused ultrasonic waves are emitted into the flow field to observe the growth and evolution process of cavitation region. A small cavitation bubble near the focal point is preset as a cavitation nucleus. We observed in the experiment (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011) that isolated bubbles exist in the flow field after the first few ultrasonic waves. When a high positive pressure phase of the ultrasound wave acts on the bubble, the bubble, as an acoustically soft interface, backscatters a negative pressure wave. The negative pressure phases of the scattered and incident waves constructively interfere, producing stronger local negative pressure. This superimposed negative pressure exceeds the cavitation threshold, inducing new bubble nucleation and forming the initial cavitation cloud. In order to be consistent with the experimental results, we preset a vapour bubble, positioned at one wavelength to the right of the focal point, with a radius of 0.1 mm to achieve a comparative simulation of the formation process of cavitation clouds under this working condition. The threshold of cavitation inception is consistent with that in § 4.2.
The numerical results at different instants of the whole evolution process are presented in figure 22. The upper and lower halves of each panel show the numerical schlieren image and the pressure distribution, respectively. As shown in figure 22, the development of the cavitation clouds undergoes three typical stages, namely (i) cavitation initiation, (ii) reverse growth and (iii) expansively growth. In order to clearly show the physical mechanism of the growing process of the cavitation cloud, figure 23 further illustrates the corresponding schematic diagram. The solid red line represents the incident wave, while the dashed blue line depicts the scattered wave originating from the front of the cavitation cloud, the blue shadow represents the cavitation region, and the yellow shadow corresponds to the newly formed cavitation region. For simplicity, diffraction waves are not marked in the figure 23.
The schematic diagram of initiation and development mechanism of cavitation clouds.

The first stage, as shown in figures 22(a,b) and 23(a), involves the first several incident waves focusing and cavitation initiation process. In order to capture the formation of the cavitation cloud, a small bubble with a radius of 0.1 mm is prepositioned at a position one wavelength to the right-hand side of the focal point, simulating a cavitation nucleus and initial cavitation bubble. In fact, in the previous experimental results (Maxwell et al. Reference Maxwell, Wang, Cain, Fowlkes, Sapozhnikov, Bailey and Xu2011), it can be seen that this small bubble is universal. As the first ultrasonic wave reaches the surface of the preset bubble, scattered waves are generated and deflected in propagating direction compared with the incident wave, as depicted in figure 22(a). The back-propagating scattered wave overlaps with the following incident waves. Due to the different acoustic impedance between bubble and the surrounding liquid, the properties of the scattered waves change, then the negative pressure segment of the scattered wave corresponds to the positive pressure segment of the primary incident wave. When the negative pressure segment of scattered wave overlaps the negative pressure segment of incident wave, a local pressure drop is induced. If the local pressure reaches the cavitation threshold, the cavitation would be generated, as depicted in figures 22(b) and 23(a). This is how the initial cavitation cloud is initiated.
Then, the following incident waves continues to propagate and enters the second stage, as shown in figures 22(b) and 22(c), involves a reverse growth mechanism depicted in figure 23(b). As incident waves reach the left-most surface of the generated cavitation cloud, they would turn into the back propagating scattered waves. Similarly with the first stage, the scattered wave continues to propagate backward and superimposes again with the subsequent incident wave. As the cavitation threshold is reached again due to the superposition of the negative pressure segments of scattered wave and subsequent incident wave. Besides, the subsequent incident waves are scattered at the surface of the cavitation cloud and newly scattered waves are generated. The newly scattered waves are also deflected in propagating direction compared with the incident wave and overlaps the incident waves, leading to the cavitation cloud growing in reverse, as depicted in figures 22(c) and 23(b). In other words, the interaction between the incident and scattered ultrasonic waves creates the local pressure drop near the left-hand side of the cavitation clouds, which promotes the formation and growth of cavitation clouds in the reverse direction of incident ultrasound waves propagation. However, the reverse growth of cavitation clouds is limited. When the superimposed negative pressure of the scattered wave and the incident wave remains above the cavitation threshold, the backward growth of the cavitation clouds would be ceased. Besides, this superposition induced the cavitation effect explains the stepwise rise of the axial dimension of the cavitation cloud, as shown in both the numerical and experimental results in figure 3.
The third stage, as shown in figure 22(d–f), involves an expansive growth (growth in both reverse and radial direction) mechanism depicted in figure 23(c–d). When the axial dimension of the cavitation cloud reaches a certain length, the position where the incident wave scatters is relatively far from the focus, resulting in a decrease in the intensity of the scattered wave. Thus, the cavitation cloud appears first at the
$z$
-axis, as illustrated in figures 22(d) and 23(c). Then, ultrasonic waves scatter at both the newly formed cavitation clouds at
$z$
-axis and the previously generated cavitation clouds. Under the combined effect of the incident wave and two scattered waves, the cavitation cloud begins radial growth, as shown in figures 22(e) and 23(d). At this stage, the cavitation cloud exhibits dynamic characteristics of alternating reverse growth and radial growth. Additionally, subject to the combined effects of the incident wave and the scattered waves, the volume of the bubbles increases, resulting in an increase in the radial dimension of the cavitation cloud.
Figure 24 compares the axial pressure profiles and cavitation-cloud morphology obtained on the three grid resolutions. The pressure distributions at
$t^{*} = 7.0$
(figure 24
a) show nearly identical wave structures, indicating good convergence of the acoustic field. The corresponding cavitation-cloud shapes (figure 24
b–d) also display consistent overall morphology and evolution across the grids, with finer meshes revealing only local details. These results demonstrate that the key physical processes – wave focusing, cavitation onset, and early cloud growth – are already grid-independent. Accordingly, the
$4800\times 2400$
grid is selected for the simulations discussed in the paper.
Grid-independence verification based on axial pressure profiles and cavitation cloud structures at
$t^{*}=7.0$
. (a) Axial pressure distributions for three grids. (b–d) Cavitation cloud structures for
$3200\times 1600$
,
$4000\times 2000$
and
$4800\times 2400$
grids.

Appendix C. Applicable scope of cavitation cloud growth mechanism and wave superposition model
The mechanism of HIFU-induced cavitation cloud formation near solid walls obtained in this paper is closely related to the characteristics of ultrasound. When the ultrasonic intensity is high enough, the negative pressure of the incident wave alone can induce cavitation in the flow field without the need for the superposition of reflected waves from the solid wall and incident waves. In this condition, the initial formation and development mechanism of the cavitation cloud changes. Therefore, the application scope of the conclusions drawn in this paper can be given. Under HIFU, the peak negative pressure at the focus can be expressed as (Bessonova et al. Reference Bessonova, Khokhlova, Bailey, Canney and Crum2009; Rosnitskiy et al. Reference Rosnitskiy, Yuldashev and Khokhlova2015)
where, the linear amplification coefficient
$G$
can be expressed as
where,
${f}_{0}$
represents the ultrasonic frequency;
${R}_{\!{F}}$
represents the distance to the focus;
$\psi$
represents the angular aperture of the transducer;
${c}_{0}$
represents the speed of sound (approximately 1500 m s−1 in water);
${{F}_{\textit{number}}}={{R}_{\!{F}}}\text{/2}{{R}_{{M}}}$
represents the F-number. The relationship between the nonlinear gain coefficient
${K}_{\!P-}$
and the frequency is
where
$N$
is the nonlinear coefficient, which can be expressed as
Substituting
$G$
and
$N$
into
${K}_{\!P-}$
gives
Based on the above discussion, the peak negative pressure can be expressed in terms of the ultrasonic frequency
${f}_{0}$
. The research in this paper is conducted under the condition that the peak negative pressure does not reach the cavitation threshold. Thus, the applicable range of the mechanism derived from this study can be presented: when
the dynamic mechanism of cavitation cloud growth in the near-wall region induced by HIFU, as analysed in § 4.3, and the wave superposition model established based on this mechanism in § 4.4 are valid.


























































































