1. Introduction
It is a common phenomenon that droplets impact a flexible solid substrate. This phenomenon occurs not only in nature, such as a rainy droplet impact on a leaf (Gart et al. Reference Gart, Mates, Megaridis and Jung2015; Gilet & Bourouiba Reference Gilet and Bourouiba2015) or a flower (Amador et al. Reference Amador, Yamada, McCurley and Hu2013), and a rainy droplet impact on the wings of an insect (Xu et al. Reference Xu, Liu, Shang, Chen and Lian2023) and hairs or silks of animals, but also in applications such as bioprinting (van Dam & Le Clerc Reference van Dam and Le Clerc2004; Gudapati, Dey & Ozbolat Reference Gudapati, Dey and Ozbolat2016), flying robotic insect in rain (Jafernik Reference Jafernik2019) and aerosol drug delivery (Cheng et al. Reference Cheng, Irshad, Kuehl, Holmes, Sherwood and Hobbs2008; Day et al. Reference Day, Miloserdov, Wildish-Jones, Pearson and Carruthers2020). In these processes, the interactions between the droplet and the flexible substrate always cause a rich dynamics of the droplet.
A droplet impacting on a rigid solid substrate undergoes the processes of spreading, retracting and possible post-dynamic behaviours, for example, rebounding (Josserand & Thoroddsen Reference Josserand and Thoroddsen2016), splashing (Rioboo, Tropea & Marengo Reference Rioboo, Tropea and Marengo2001) and jetting (Bartolo, Josserand & Bonn Reference Bartolo, Josserand and Bonn2006; Josserand & Thoroddsen Reference Josserand and Thoroddsen2016). This droplet impact may also be accompanied by bubble entrapment and explosion (Langley, Li & Thoroddsen Reference Langley, Li and Thoroddsen2017) and lens formation (Kim, Moon & Kim Reference Kim, Moon and Kim2013). The dynamic behaviours are influenced by the contact angles of the droplet on the solid surface (Pasandideh-Fard et al. Reference Pasandideh-Fard, Qiao, Chandra and Mostaghimi1996; Ma et al. Reference Ma, Zhang, Zhang and Zhou2024; Ma, Zhang & Zhou Reference Ma, Zhang and Zhou2024), the impact Weber number (Laan et al. Reference Laan, Bruin, Bartolo, Josserand and Bonn2014; Lee et al. Reference Lee, Laan, de Bruin, Skamtzaris, Shahidzadeh, Derome, Carmeliet and Bonn2016), the impact Reynolds number (Laan et al. Reference Laan, Bruin, Bartolo, Josserand and Bonn2014), the shape of the impacted solid surface (Zhu et al. Reference Zhu, Liu, Mu, Gao, Ding and Lu2017; Abbot et al. Reference Abbot, Lannert, Kiran, Bakshi, Hussong and Roisman2024) and an external force field, such as an external electric field (Tian et al. Reference Tian, Liu, Peng, Xu, Ye, Guan, Zhou, Deng and Huang2022a , Reference Tian, Peng, Liu, Di, Zhan, Ye, Guan, Zhou, Deng and Huang2022b , Reference Tian, Wang, Zhou, Xie, Zhu, Chen, Ding and Liao2022c ) and an external magnetic field (Zhang et al. Reference Zhang, Han, Yang and Ni2016).
Unlike a rigid solid substrate, a flexible solid substrate can cause strong flow–structure interactions for a droplet wetting (de Gennes, Brochard-Wyart & Quéré Reference de Gennes, Brochard-Wyart and Quéré2004; Tamim & Bostwick Reference Tamim and Bostwick2023), droplet impact (Gart et al. Reference Gart, Mates, Megaridis and Jung2015) or droplet rebound (Weisensee et al. Reference Weisensee, Tian, Miljkovic and King2016), leading to richer physical phenomena of droplets (Xiong, Huang & Lu Reference Xiong, Huang and Lu2020; Ma & Huang Reference Ma and Huang2023; Zhou et al. Reference Zhou, Xu, Zhang, Zhang and Peng2025; Ma & Huang Reference Ma and Huang2025). Among these phenomena, splashing and spreading are two of the most studied effects, both of which can be significantly influenced by substrate flexibility.
Previous studies have shown that the substrate flexibility affects droplet splashing (Howland et al. Reference Howland, Antkowiak, Castrejón-Pita, Howison, Oliver, Style and Castrejón-Pita2016; Basso & Bostwick Reference Basso and Bostwick2020). Suppressing droplet splashing is important in both scientific and industrial fields, such as inkjet printing, where splashing could affect the printing quality (Zable Reference Zable1977; Lohse Reference Lohse2022), and fuel combustion, where splashing could cause the formation of undesirable mixtures (Gavaises, Theodorakakos & Bergeles Reference Gavaises, Theodorakakos and Bergeles1996). Splashing is a phenomenon governed by a range of physical factors, including air pressure, substrate roughness, substrate flexibility and liquid properties (Xu Reference Xu2007; Liu, Tan & Xu Reference Liu, Tan and Xu2015; Jian et al. Reference Jian, Josserand, Popinet, Ray and Zaleski2018; Yu et al. Reference Yu, Ratschow, Tao, Li, Jin, Wang and Wang2025). This study focuses on the influence of substrate flexibility. Pepper, Courbin & Stone (Reference Pepper, Courbin and Stone2008) demonstrated that reducing membrane tension can suppress splashing. Howland et al. (Reference Howland, Antkowiak, Castrejón-Pita, Howison, Oliver, Style and Castrejón-Pita2016) experimentally found that the smallest Weber number for the occurrence of droplet splashing increased for smaller Young’s modulus of the flexible substrate, and the energy required for the occurrence of splashing behaviour increased by approximately 70 % compared with a rigid substrate. Basso & Bostwick (Reference Basso and Bostwick2020) experimentally observed the increase of the threshold Weber number for the occurrence of splashing with the reduction of the substrate elasticity. Although Pegg, Purvis & Korobkin (Reference Pegg, Purvis and Korobkin2018) theoretically found that oscillation of the substrate with relatively high stiffness can promote droplet splashing, existing evidence suggests that low substrate stiffness tends to suppress droplet splashing (Howland et al. Reference Howland, Antkowiak, Castrejón-Pita, Howison, Oliver, Style and Castrejón-Pita2016; Kim, Rothstein & Shang Reference Kim, Rothstein and Shang2018). Despite these findings, there has been no systematic quantification or scaling law of the splashing threshold for droplets impacting flexible substrates.
In addition, droplet spreading has also attracted considerable attention. A flexible substrate generally reduces droplet spreading, which is undesirable in many fields such as pesticide spraying (Massinon & Lebeau Reference Massinon and Lebeau2012). Rioboo et al. (Reference Rioboo, Voue, Adao, Conti, Vaillant, Seveno and De Coninck2010) investigated the changes in contact angles of droplets on flexible surfaces and illustrated that the flexible surface suppresses the droplet spreading. Both experiments (Vasileiou et al. Reference Vasileiou, Gerber, Prautzsch, Schutzius and Poulikakos2016) and two-dimensional (2-D) numerical simulations (Xiong et al. Reference Xiong, Huang and Lu2020; Ma & Huang Reference Ma and Huang2023) showed that the spreading suppression is dominated by the flexible compliant mechanism and that the spreading suppression becomes stronger for smaller stiffness or smaller Weber number. Zhou et al. (Reference Zhou, Xu, Zhang, Zhang and Peng2025) found that the flexible substrate not only has a weakening effect on the droplet spreading but also an enhancement effect. However, the conditions of the enhancement effect are limited to low Weber numbers and low stiffnesses.
(a) Schematic of a droplet of diameter D 0, density ρ l and viscosity μ l , impacting on a flexible lotus-leaf-like disk consisting of a rigid central circular support and a flexible circular ring with the impact velocity U 0 in a downward gravity field. (b) Demonstrates the definitions of two parameters: the spreading factor β which represents the spreading diameter of a droplet on the substrate relative to the diameter of the droplet and the deflection δ at the end of the flexible substrate which represents the deflection relative to the diameter of the droplet.

The suppression of droplet splashing and the enhancement of droplet spreading are of great significance in many fields of droplet impact, such as inkjet printing, pesticide spraying and even in natural scenarios where raindrop impact contributes to the spread of diseases. Flexible substrates can be used to suppress droplet splashing (Pepper et al. Reference Pepper, Courbin and Stone2008; Howland et al. Reference Howland, Antkowiak, Castrejón-Pita, Howison, Oliver, Style and Castrejón-Pita2016), but they often also suppress the droplet spreading (Vasileiou et al. Reference Vasileiou, Gerber, Prautzsch, Schutzius and Poulikakos2016; Xiong et al. Reference Xiong, Huang and Lu2020; Ma & Huang Reference Ma and Huang2023). In addition to continuous flexible substrates, structured surfaces such as vertically oriented fibre arrays can also suppress droplet splashing by restricting droplet spreading (Rible et al. Reference Rible, Raza, Joshua, Lin, Chakpuang and Dickerson2025). However, this suppression is typically achieved at the expense of limiting droplet spreading. How to suppress droplet splashing but not reduce the droplet spreading is a significant but unresolved issue. This study aims to develop this issue. In this paper, the dynamics of droplet impact on a flexible lotus-leaf-like disk is numerically investigated. In § 2, the 2-D axisymmetric mathematical model of the droplet–solid coupling is presented by considering the droplet dynamics and the deflection of the flexible substrate. In § 3, the droplet dynamic modes during the impact process are numerically studied, and the splashing threshold and the maximum spreading factor are investigated. In § 4, the main conclusions are obtained according to the above analysis.
2. Method
2.1. Numerical method
We consider 2-D axisymmetric numerical simulations of a droplet impacting a flexible lotus-leaf-like disk supported by a stem. This structure has been simplified into a flexible disk with a rigid central region in this study. Figure 1(a) shows a schematic of the computational domain of the flow–solid system in a downward gravity field, where a droplet of diameter D
0 with the impact velocity U
0 is above the centre of a disk with thickness h, radius R
2 = 3D
0, rigid region radius R
1 = 0.05D
0, density ρ
p
and Young’s modulus Y. The asterisk * represents a dimensional quantity in the following analysis. The droplet diameter D
0, the surface tension coefficient σ
gl
and the droplet density ρ
l
are selected as the characteristic scales to make the variables dimensionless. The reference velocity and the reference time are taken as U
ref
$=\sqrt{\sigma _{gl}/(\rho _{l}D_{0})}$
and t
ref
=
$\sqrt{\rho _{l}D_{0}^{3}/\sigma _{gl}}$
, respectively. The stiffnesses K = Yh
3/(ρ
l
U
ref
2(R
2–R
1)3) are varied with different Young’s moduli. The Weber number and Reynolds number are defined as We = ρ
l
U
0
2
D
0/σ
gl
and Re = ρ
l
U
0
D
0/μ
l
, respectively. In our simulations, the main parameters are given as: the density ratio ρ
l
/ρ
g
= 1000, and the dynamic viscosity ratio μ
l
/μ
g
= 100.
To calculate the fluid flow, the incompressible Navier–Stokes equations are as follows:
where ρ
f
is the fluid density,
v
f
is the fluid velocity vector, μ
f
is the fluid dynamic viscosity, p is the pressure,
F
su
is the surface tension force and
$\boldsymbol{g}$
is the gravitational acceleration.
The flexible structure is calculated with the shell element. The structural equation is used to describe the deformation and the motion of the solid substrate
where ρ s is the substrate density, u s is the structure displacement vector and F st is the stress force. The arbitrary Lagrangian–Eulerian method tracks the moving mesh to help finish the fluid–structure interaction.
The phase transition is defined with the phase-field method as the variation of
$\phi$
from −1 to + 1. The phase-field method describes two-phase flow with the volume fractions of different phases. To determine the fluid density and fluid dynamic viscosity in the phase-field method, the phase-field variables are introduced by
where subscripts l and g represent the liquid and the gas, respectively, and η is the volume fraction in a cell of the numerical simulation. The Cahn–Hilliard equation is used to track the diffusion of the liquid–gas interface
where Γ is the mobility,
$\varepsilon$
is the interface thickness parameter and ψ is the phase-field auxiliary variable. The mobility Γ controls the numerical stability and the interface movement accuracy, and is expressed as
where S max is the maximum element size.
The numerical simulations are conducted using COMSOL Multiphysics 5.6. The equations for fluids and solid structures are converted into the weak form, which is discretised by the finite element method. The dynamic mesh is applied to improve the precision of the mesh deformation of both the fluid and solid domains. The time discretisation is performed by the backward differentiation formula. All variables are numerically solved using the constant Newton method, with a constant damping factor 0.75. The Yeoh smoothing method is used to constrain the mesh displacement. The multifrontal massively parallel sparse direct solver is set for solving the linear systems for the problem initialisation and the transient process.
2.2. Numerical validation
The numerical model is validated against the experimental results for the maximum spreading factor β max reported by Vasileiou et al. (Reference Vasileiou, Gerber, Prautzsch, Schutzius and Poulikakos2016). The simulations are performed over different Weber numbers (We = 3, 5, 7, 11, 10, 20, 28, 37, 45 and 75) for both rigid and flexible substrates. The advancing contact angle and the receding contact angle are θ a = 161° and θ r = 158°, respectively.
Figure 2 demonstrates that the present results agree well with those of Vasileious et al. (Reference Vasileiou, Gerber, Prautzsch, Schutzius and Poulikakos2016). Figure 3 shows the grid independence test by comparing the variation of the spreading factor β for three different maximum element sizes (S max = D 0/50, D 0/100 and D 0/150). The results indicate that S max = D 0/100 provides a sufficient resolution for the present study. In addition, to save computational resources, the static contact angle is applied to our numerical model (Zhu et al. Reference Zhu, Liu, Mu, Gao, Ding and Lu2017; Zhou et al. Reference Zhou, Xu, Zhang, Zhang and Peng2025).
Maximum spreading factor β
max
for different values of We in the present simulations and the experimental results from Vasileiou et al. (Reference Vasileiou, Gerber, Prautzsch, Schutzius and Poulikakos2016). The key parameters of the fluids are given as: droplet radius D
0 = 2.91 mm, droplet density ρ
l
= 998 kg m−3, droplet dynamic viscosity μ
l
= 0.001 Pa · s, air density ρ
g
= 1.205 kg m−3, air dynamic viscosity μ
g
= 1.825 × 10−5
$\rm{Pa} \boldsymbol{\cdot} \rm{s}$
and the surface tension σ
gl
= 0.072 N m−1. The key parameters of the flexible substrate are given as: the thickness h
s
= 10 μm, the length L
s
= 30 mm, the natural frequency f
s
= 36 Hz and the effective mass m
s
= k
b
/(2πf
s
)2 = 9.5 mg, where k
b
is the bending stiffness that is directly used for the numerical calculations.

Time variations of the spreading diameter with three different maximum element sizes for We = 50 and K = 0.05.

3. Results and discussion
The results for a representative case of hydrophobic substrate (θ = 130°) are analysed over the range of We ∈ (1, 500) and K ∈ (10–2, 102) with a constant Reynolds number Re = 1000. Section 3.1 demonstrates the droplet behaviours on impacting this lotus-leaf-like substrate. Section 3.2 investigates the splashing threshold on flexible substrates. Section 3.3 explores the variation of the maximum spreading factor with different We and K and discusses the prediction of the maximum spreading factor.
3.1. Droplet modes and phase diagram
In this section, we focus on the flow dynamics of impacting droplets. The morphological changes of the droplets impacting this lotus-leaf-like substrate will be explored. The droplet impacting process can be divided into two stages – spreading and retraction processes, whose characteristics will serve as the basis for classifying our results. Five typical modes of droplet morphologies are identified (see figure 4), i.e. no splashing and simple rebound (NSR), no splashing and rebound with satellite droplet (NRSD), splashing and simple rebound (SSR), splashing and rebound with satellite droplet (SRSD) and annular retraction with detaching from the surface (ARDS).
Snapshots of the morphology of droplet impacting on the substrate for (a) We = 10 and K = 0.01, (b) We = 200 and K = 0.05, (c) We = 500 and K = 10, (d) We = 400 and K = 0.1 and (e) We = 450 and K = 0.01 representing the NSR, NRSD, SSR, SRSD and ARDS modes, respectively. The black arrows are the velocity vector of the droplet. The green arrows are the subsequent movement direction of the substrate.

The NSR mode and the NRSD mode involve spreading without splashing, while the SSR mode and the SRSD mode are associated with splashing. The primary distinction between them lies in whether tiny secondary droplets are generated at the edge of the impacting droplet during the spreading process. Besides, the ARSD mode can exist either with or without splashing. As shown in figure 4(a), spreading without splashing and simple rebound occur in the NSR mode. As shown in figure 4(b), the retraction stage generates a liquid column (jet flow) after spreading without splashing and this column breaks up into satellite droplets after being elongated, which is the NRSD mode. From figure 4(c), the prompt splashing and the subsequent simple rebound occur in the SSR mode. From figure 4(d), the prompt splashing and the subsequent rebound with the generation of the satellite droplets are the characteristics of the SRSD mode. From figure 4(e), a droplet in the shape of an annular liquid ring detaches from the substrate during the retraction stage, which is called the ARDS mode. It should be noted that annular liquid rings may exist in all other modes, but detachment only occurs in the ARDS mode. In the ARDS mode, the formation of the annular liquid ring is mainly attributed to the deformation of the substrate. The periodic deformation of the substrate is from the elastic waves generated by the impact of the droplet. The deformation trend near the rigid central region is opposite to that of the edge part (see figure 4 e).
Figure 5 demonstrates the five typical modes of impacting droplets in a (We, K) parameter space. In the low We region, the NSR and NRSD modes dominate. In the high We region, where inertial forces dominate, the SSR and SRSD modes exhibit rebounding in the droplet entirety and rebounding with satellite droplets, respectively. Besides, the formation region for satellite droplets is centred in the moderate We region. The ARSD mode appears exclusively in the low K and high We region, where the significant deformation of the substrate and enhanced inertial forces facilitate droplet detachment. Upon droplet impact, the generation of elastic waves induces the oscillation of the substrate. When the droplet retracts and moves upward, the substrate tends to be concave downward, as shown in figure 4(e). The different movement tendencies lead to the detachment of the droplet from the substrate. It then completes the contraction of the annular liquid in the air (see figure 4 e).
Distribution of typical modes in a parameter space (K, We). Triangles and circles represent deposition and splashing, respectively. Green triangles, red triangles, blue circles, black circles and the circles (or triangles) with green borders denote the NSR, NRSD, SSR, SRSD and ARDS modes, respectively. The blue dashed line is the splashing threshold We t,ri for the rigid substrate (Mandre & Brenner Reference Mandre and Brenner2012). The splashing threshold for large enough K is found to be close to that (We t,ri = 253, see the Appendix A) for the rigid substrate case (Mandre & Brenner Reference Mandre and Brenner2012).

A critical feature in figure 5 is the splashing threshold, delineated by a splashing–non-splashing boundary that shifts downward with increasing K, indicating the splashing threshold We decreases. This splashing threshold for flexible substrates is larger than that for rigid substrates (the blue dashed line) from Mandre & Brenner (Reference Mandre and Brenner2012) (see the Appendix A), suggesting a higher required inertial force for splashing onset in this system. This lotus-leaf-like substrate can prevent the occurrence of droplet splashing.
3.2. Splashing threshold
The occurrence of splashing is influenced not only by the droplet’s inertia and surface tension but also by the physical properties of the substrate. On flexible substrates, the interaction between the liquid and the solid substrate becomes more complex. As shown in § 3.1, flexible substrates tend to suppress splashing. There have been few detailed studies on the splashing threshold under flexible substrate conditions. Compared with rigid substrates, flexible substrates introduce additional motion of the substrates, making it difficult to apply the splashing threshold of rigid substrates to predict the present splashing threshold. A unified criterion for both substrate types is important for understanding droplet–solid interactions during impact.
To investigate this, we first analyse the initial spreading stage, which is essential for understanding the onset of splashing (Mongruel et al. Reference Mongruel, Daru, Feuillebois and Tabakova2009; Mani, Mandre & Brenner Reference Mani, Mandre and Brenner2010; Riboux & Gordillo Reference Riboux and Gordillo2014). Figure 6 demonstrates the spreading factor β versus the growth time
$\varDelta$
t =
$\varDelta$
t
*/T
ref
at the initial spreading stage. From figure 6, a larger K results in a larger β during the initial spreading process, indicating a faster growth of the liquid film. A faster rate can help the liquid film eject from the restraint of the surface tension, promoting the droplet splashing (Riboux & Gordillo Reference Riboux and Gordillo2014). This result matches the results in figure 5 well, where a higher K corresponds to a lower Weber number threshold for splash.
Spreading factor β versus the growth time
$\varDelta$
t for different K at We = 300.

Snapshots of the droplet impacting on the substrate with We = 400 and K = 0.01. Here, α is defined as the angle between the tangent of the substrate near the centre contact area and the horizontal direction.

The interaction between the droplet and the flexible substrate also deserves attention. The part that first comes into contact with the droplet undergoes deformation initially, after which the deformation propagates from the centre to the periphery (see figure 7). To quantitatively describe the local deformation at the contact region, the angle α is defined as the angle between the tangent of the substrate near the centre contact area and the horizontal direction. At early times, the contact area is small and the deformation is strong, leading to a large α. As the droplet spreads, the substrate lifts upward and α decreases (see figure 7 b–e). When the droplet starts to retract, α increases again (see figure 7 f). This evolution shows that the substrate deformation exhibits a quasi-periodic behaviour that is coupled with the droplet spreading and retraction dynamics. Furthermore, different from previous studies on flexible substrates (Ma & Huang Reference Ma and Huang2023), the maximum spreading diameter and the maximum substrate deflection do not occur at the same time in the present system. Instead, the maximum substrate deflection is delayed, which can be attributed to the wave-like propagation of substrate deformation. Notably, the deformation of the substrate in the contact area changes rapidly, accompanied by substantial changes in the substrate geometry. Although substrate stiffness affects the impact results, the rapid change of substrate shape before splashing makes it difficult to analyse the process using only instantaneous structural features. These observations emphasise the necessity of considering the comprehensive, time-dependent interaction when analysing splashing in the present study.
To quantify the stiffness effect, we perform an energy-based analysis. During the droplet impact process, energy conversion primarily occurs among the kinetic energy E k , the surface energy E s u and the substrate’s elastic potential energy E e . Simulations for K = 0.05, 0.5 and 50 at We = 50 (figure 8 a–c) show that the maximum of E e decreases as K increases. Here, E e is stored during the initial stage of spreading and released afterwards, reducing the energy available for sheet ejection. As shown in figure 9, for cases where splashing occurs, the substrate typically reaches its maximum energy absorption before splashing. As stiffness increases, the magnitude of the time difference shows an overall reduction and approaches zero. Higher Weber numbers also reduce this time difference. It demonstrates that softer substrates absorb more energy. The negative time difference indicates that the maximum energy absorbed by the substrate provides an upper bound for the energy transferred from the droplet because the maximum energy absorption finishes before splashing. Therefore, it is necessary to quantify the energy absorption ability of the flexible substrate.
Energy variation and the spreading factor β plotted against the dimensionless time t with We = 50 for (a) K = 0.05, (b) K = 0.5 and (c) K = 50. (d) The corresponding morphologies of droplets and substrates when the maximum spreading factor is reached (red points in panels a–c) with We = 50 for K = 0.05, K = 0.5 and K = 50, respectively. The grey areas represent the initial spreading stage before reaching the maximum of E e in panels (a–c).

The time difference between the maximum value of E e and the splashing.

As shown in figure 5, flexible substrates with smaller stiffness exhibit lower splashing thresholds. This is because flexible substrates absorb part of the impact energy, reducing the energy available for droplet splashing. The energy balance of the droplet motion from the moment of initial impact until just before splashing can be written as
where E
k0 and E
su0 denote the initial kinetic and the initial surface energy, E
k1 and E
su1 denote the kinetic and the surface energy just before splashing, E
e
is the energy of the substrate and E
diss
is the viscous dissipation. In addition, since splashing occurs in the region with the high We (We
$\geqslant$
200), the initial kinetic energy is much higher than the surface energy of the droplet. The splashing process is governed by the effective impact energy before splashing E
eff
= E
k1 + E
su1 + E
diss
, which is affected by the amount of energy absorbed by the substrate E
e
.
Ma & Huang (Reference Ma and Huang2023) introduced an additional inertial length scale associated with the substrate deformation and proposed an acceleration scale a ∼ U 0 2/(D 0 + δ max ), where δ max is the maximum substrate deflection during impact. Here, (D 0 + δ max ) is the acceleration displacement, which represents the displacement that occurs when the velocity U 0 decreases to 0 (Clanet et al. Reference Clanet, Béguin, Richard and Quéré2004; Ma & Huang Reference Ma and Huang2023). In the prediction of the maximum spreading diameter, δ max represents the additional acceleration displacement caused by the flexible substrate (Ma & Huang Reference Ma and Huang2023). Although their analysis focused on the maximum spreading diameter, the related physical mechanism, which is the reduction of effective impact energy due to the substrate deformation, is not specific to spreading and is expected to influence splashing. This is because splashing is likewise governed by the inertia energy during the early impact stage.
According to the previous studies (Clanet et al. Reference Clanet, Béguin, Richard and Quéré2004; Ma & Huang Reference Ma and Huang2023), the droplet acceleration at the splashing time should be obtained by determining the additional acceleration displacement of the splashing time, which is represented by the flexible substrate deformation. However, different from the previous studies (Ma & Huang Reference Ma and Huang2023; Zhou et al. Reference Zhou, Xu, Zhang, Zhang and Peng2025), the deformation in the present study propagates gradually from the centre to the periphery. The substrate deformation undergoes significant evolution before splashing occurs, particularly for low-stiffness substrates (see figure 7). The substrate deformation first increases and then decreases before splashing (see figure 7). Thus, the substrate configuration at the instant of splashing does not characterise the influence of the flexible substrate on the splashing process, which is different from the spreading prediction of the previous studies. It means that the substrate deformation at the splashing moment cannot sufficiently quantify the acceleration displacement.
As shown in figure 9, the maximum value of E e occurs shortly before, or nearly with, the onset of splashing. It indicates that the substrate has completed the primary energy absorption from the droplet before splashing. During the early spreading stage, this energy transformation leads to a large deformation angle (see figure 7 a, b). As the contact area expands, the substrate region in contact with the droplet is lifted upward, resulting in a small angular deflection (see figure 7 c–e). Although the interactions between droplet and substrate persist after the maximum E e is reached, the primary reduction of the droplet’s inertial energy has already finished.
These observations suggest that the flexibility’s influence on splashing is more appropriately described by a quantity that accounts for the deformation ability rather than by the instantaneous measurement. As illustrated in figure 7, the substrate deformation propagates in a wave manner from the centre to the periphery. The temporal evolution of the substrate deflection at the edge provides a reliable indicator of the maximum inertial energy that can be absorbed by substrates of different stiffnesses. Although the maximum deflection δ max is reached after splashing, it can still effectively represent the measurement of the substrate’s ability to absorb the droplet’s initial energy and the ability to prolong the acceleration displacement. Thus, the proportion of energy absorbed by the flexible substrate can be estimated (Ma & Huang Reference Ma and Huang2023) as follows:
where δ max is the maximum deflection of the substrate (see figure 1). Thus, the effective impact energy is E eff = 1−E e = E k0/(1 + δ max ). Figure 10 verifies and supports the prediction of the effective impact energy at the splashing time. The collapse of the data across different substrate stiffnesses confirms that the effective impact energy is primarily governed by the substrate deformation ability, rather than by instantaneous deformation for the present study.
The effective impact energy versus (a) the initial kinetic energy E k0 and (b) E k0/(1 + δ max ).

In general, the Weber number is commonly used to characterise the impact energy of a droplet. Compared with the rigid condition, there is an influence from the substrate energy E e . On this basis, we extend the energy distribution from (3.2) to construct the relation between the threshold Weber number We t,fl for the flexible condition and the threshold Weber number We t,ri for the rigid condition, We t,fl /(1 + δ max ) = We t,ri , where We t,ri is constant, as shown in figure 5. Therefore, the threshold Weber number for the flexible condition is
This relationship reflects that the greater deformation of the flexible substrate reduces the effective impact energy of the droplet, suppressing the onset of splashing. The new splashing threshold We t,fl is generated as shown in figure 11. The splashing threshold We t,fl for flexible substrates matches the boundary between the splashing mode and the non-splashing mode. In addition, the value of We t,fl converges to the value of We t,ri with the increase of substrate stiffness.
Distribution of typical modes in a parameter space (K, We). The blue dashed line is the splashing threshold We t,ri for the rigid substrate. The green dash-dotted line is the splashing threshold We t,fl calculated by (3.3) for the flexible substrate.

3.3. Spreading process
The maximum spreading factor
$\beta _{\textit{max}}=\beta _{\textit{max}}^{*}/D_{0}$
is a primary measure of spreading ability. In this section, the effects of the Weber number We and the disk’s stiffness K on the maximum spreading factor are analysed. The prediction of the maximum spreading factor β
max
is also discussed.
Figure 12 demonstrates the variations of the maximum spreading factor with K for different We. There is almost no influence of stiffness on the maximum spreading factor, as shown in figure 12. Under high We conditions, β max will slightly fluctuate as K increases. However, β max changes little with variations in K for the low We conditions. Compared with the increase in the stiffness K, the increase in We has a more significant impact on β max , which will increase as We increases.
Variations of the maximum spreading factor β max with K for different We.

Figure 13 shows the variation of the spreading time
$\varDelta$
t
βmax
required for the droplet to reach its maximum spreading factor as the value of We or K increases. As shown in figure 13(a), the spreading time
$\varDelta$
t
βmax
exhibits a general decreasing tendency for rising We. As shown in figure 13(b), the tendency of the spreading time
$\varDelta$
t
βmax
for varying K remains essentially consistent across the entire K range, with no obvious trending shift observed. On the whole, the variation in
$\varDelta$
t
βmax
is more strongly influenced by We and less by K. However, in regions where We is relatively large, a situation arises where the spreading time corresponding to extremely low stiffness K is slightly higher than in other cases. This particular change corresponds to the ARDS mode from the parameter space, in which mode the deformation of the substrate is relatively substantial.
Variation of the spreading time
$\varDelta$
t
βmax
with (a) We for different K and (b) K for different We.

Different from the stiffness influence on the splashing in § 3.2, the maximum spreading factor β max remains unaffected by substrate flexibility. Initially, the small contact area between the droplet and the substrate produces a concentrated force on the substrate, causing significant local deformation (see figure 7). As spreading proceeds, the contact area increases continuously, causing the force on the substrate to gradually become uniform. Combined with the constraint effect of the rigid central region, the substrate shape in contact with the droplet gradually approaches that observed under high K conditions before reaching the maximum spreading factor (as shown in figures 7 and 8 d). This transformation, from an initially suppressed state (as shown in figure 6) to a raised central region with a sloped periphery, promotes spreading in the later stages and offsets the initial stage suppression. In addition, the proportion of E e of the substrate is relatively small compared with E su and E k . All these factors ultimately result in nearly identical β max values across stiffnesses.
In general, the theoretical prediction of β
max
is challenging because the deformation of the substrate has a significant effect on it (Ma & Huang Reference Ma and Huang2023; Zhou et al. Reference Zhou, Xu, Zhang, Zhang and Peng2025). Ma & Huang (Reference Ma and Huang2023) and Zhou et al. (Reference Zhou, Xu, Zhang, Zhang and Peng2025) considered the substrate deflection, which represents the effect of the stiffness, to modify the traditional prediction model. However, the variation of stiffness K has a little effect on the maximum spreading factor under the conditions of this study (figures 12 and 13). For the droplet impacting on the rigid substrate conditions, Laan et al. (Reference Laan, Bruin, Bartolo, Josserand and Bonn2014) introduced the impact parameter P = WeRe
−2/5 and derived
$\beta _{\textit{max}}\,{\textit{Re}}^{-1/5}=P^{1/2}/(A+P^{1/2})$
with the Padé approximation (where A is the fitting parameter). Lee et al. (Reference Lee, Laan, de Bruin, Skamtzaris, Shahidzadeh, Derome, Carmeliet and Bonn2016) have considered the effect of wettability based on the previous prediction. The prediction model for the rigid substrate with the influence of the maximum spreading factor at nearly zero impact velocity β
v→0 from Lee et al. (Reference Lee, Laan, de Bruin, Skamtzaris, Shahidzadeh, Derome, Carmeliet and Bonn2016) is used to describe our results. There is a constant Reynolds number Re = 1000 for the present study. Thus, the impact parameter P could be simplified as a function of only one parameter, We. The applicable prediction model of the maximum spreading factor β
max
for Re = 1000 is derived as
where the fitting parameter is B = 7.6 and β ν→0 is related to the contact angle (Lee et al. Reference Lee, Laan, de Bruin, Skamtzaris, Shahidzadeh, Derome, Carmeliet and Bonn2016) as follows:
\begin{equation} \beta _{v\rightarrow 0}=\left\{\begin{array}{l} \left(\dfrac{4\sin ^{3}\theta _{v\rightarrow 0}}{2-3\cos \theta _{v\rightarrow 0}+\cos ^{3}\theta _{v\rightarrow 0}}\right)^{1/3}\quad \mathrm{if}\,\theta _{v\rightarrow 0}\lt 90^{\circ},\\[5pt] \left(\dfrac{1}{\left(2+\cos \theta _{v\rightarrow 0}\right)\sin ^{3}\left(\theta _{v\rightarrow 0}/2\right)}\right)^{1/3}\quad \mathrm{if}\,\theta _{v\rightarrow 0}\gt 90^{\circ}. \end{array}\right. \end{equation}
From figure 14, the rescaled variable
$(\beta _{\textit{max}}^{2}-\beta _{\boldsymbol{v}\rightarrow 0}^{2})^{1/2}{\textit{Re}}^{-1/5}$
is a function of We. It shows all rescaled results with different stiffnesses collapse to a single curve with (3.4), which proves the lesser effect from stiffness K.
Value of
$(\beta _{\textit{max}}^{2}-\beta _{\boldsymbol{v}\rightarrow 0}^{2})^{1/2}{\textit{Re}}^{-1/5}$
as a function of We for different stiffnesses K. The solid line is plotted by (3.4).

4. Conclusions
We consider the impact of droplets on a hydrophobic lotus-like substrate (a rigid central region and a flexible periphery) and explore the effects of the Weber number We and the disk’s stiffness K on the spreading behaviour during droplet impact. In this study, the droplet impact regimes are classified into five modes: the NSR, NRSD, SSR, SRSD and ARDS modes. These five modes can be separated into two categories based on whether secondary droplets are generated during the spreading process: no splashing (the NSR and NRSD modes), splashing (the SSR and SRSD modes) and the ARDS mode exists in both categories. In this mode, the droplet exhibits premature bouncing behaviour, i.e. it detaches from the substrate during the retraction of the liquid ring and completes retraction in the air.
The results demonstrate that the flexible periphery effectively suppresses splashing by absorbing energy through adaptive deformation. Our analysis shows that the substrate’s instantaneous state at splashing does not adequately reflect the substrate’s effect on the droplet. The maximum energy absorbed by the substrate is already attained before splashing occurs. We give the splashing threshold boundary by considering the maximum deflection at the edge. The stiffness influences the initial stage of the spreading process, while it has a minimal effect on the maximum spreading factor. This contrast arises from the evolution of substrate deformation. The structure deformation of the substrate will offset the spreading factor that is suppressed in the initial stage. Besides, the periodic deformation of the flexible region is attributed to elastic waves generated by the droplet impact, which propagate through the substrate and shape its dynamic response. Energy analysis reveals that elastic potential energy plays a role during early deformation but remains negligible at the maximum spreading. Most kinetic energy is converted into surface energy, while elastic energy is quickly dissipated. In addition, a delay between the maximum spreading and the maximum deflection is observed, which is different from the previous studies.
These findings enhance our understanding of droplet–substrate interactions under the conditions of such specially structured substrates, emphasising the coupling relationship between dynamic deformation and fluid flow. The study provides a substrate structure that can suppress splashing without inhibiting maximum spreading, i.e. a rigid structure can be added to the central part of the flexible substrate. Future work can build on this foundation by investigating the effects of varying the ratio or transition modes between rigid and flexible regions on the droplet spreading dynamics.
Funding
This research was supported in part by the National Natural Science Foundation of China (No. 12372263).
Declaration of interests
The authors report no conflicts of interest.
Appendix A. Calculation of the splashing threshold under the rigid substrate condition
To identify the threshold Weber number We
t,ri
for the onset of splashing under the rigid substrate condition, a critical impact velocity is calculated based on the theory proposed by Mandre & Brenner (Reference Mandre and Brenner2012). The splashing threshold impact velocity
$U_{t,ri}$
can be estimated by balancing inertia and surface tension as (Mandre & Brenner Reference Mandre and Brenner2012)
\begin{equation} U_{t,ri}=C\left(\frac{2{\sigma_{gl}} ^{3}}{\mu _{g}^{2}\rho _{l}D_{0}}\right)^{1/4}\mathrm{,}\quad0.1\lt C\lt 0.3 ,\end{equation}
where C is a variable factor. For the present study, the threshold velocity belongs to the range U t,ri ∈ (3.01, 9.47) m s−1, as determined by equation (A1). The minimum value is taken as the velocity threshold combined with the present results. The corresponding splashing threshold Weber number is We t,ri = 253.













