Enhanced axial migration of a deformable capsule in pulsatile channel flows

High-throughput measurements of single-cell behaviour under confined channel flow is of fundamental importance and technical requirement in bioengineering applications such as cellular-level diagnoses for blood diseases. Although several attempts have addressed this issue and gained insights into (soft) particle dynamics in microchannels (Ciftlik et al., 2013; Fregin et al., 2019; Ito et al., 2017), cell manipulation including label-free cell alignment, sorting, and separation still face major challenges. Along with the aforementioned experimental studies, recent numerical simulations revealed the mechanical background regarding the lateral movement of particles, e.g. in (Alghalibi et al., 2019; Takeishi et al., 2021, 2022). The lateral movement of deformable spherical particles in almost inertialess conditions was originally reported in Karnis et al. (1963), and these results have been the fundamental basis to describe the phenomena observed in microfluidics (Kim et al., 2019) but also in in vivo microcirculations (Secomb, 2017). In particular, it was found that a deformable spherical particle tends to move towards the channel axis and settles there. Hereafter we will call this phenomenon as “axial migration”.

It is known that the presence of axial migration or non-axial migration depends on particle shape and initial orientation angles. An RBC modelled as a biconcave capsule does not always exhibit axial migration especially in the tank-treading slipper shape, obtained with high $Ca$ and high $\lambda$ (Guckenberger et al., 2018; Takeishi et al., 2021). Furthermore, RBCs have bistable flow mode, so-called rolling and tumbling motions, which depend on the initial cell orientations (Takeishi et al., 2022). Thus, the original spherical shape is one of requirements for the axial migration in (almost) inertialess conditions. In a recent work, the framework of the axial migration of a droplet has been extended by Santra and Chakraborty (2021) by including the effect of an electric field, and finding that as the strength of the electric field increases, droplets can reach the centreline at a faster rate with reduced axial oscillations. Furthermore, a deformation-dependent propulsion of soft particles, including biological cells, were confirmed experimentally by Krauss et al. (2022) and numerically by Schmidt et al. (2022).

Despite these efforts, the effect of a pulsatile flow on the axial migration of capsules has not been described and understood yet. The objective of this study is thus to clarify whether frequency-dependent axial migration of the spherical capsule occurs in confined channel flows. More precisely, can the time necessary for the axial migration be controlled by the channel pulsations? Is there an optimal pulsation frequency to do that? As we will describe in the following, our investigation of the capsule dynamic show that it indeed exists an optimal frequency to speed-up the capsule axial migration by up to $80\%$ in the range of parameters investigated here.

First, we investigate the trajectory of the capsule centroids for different frequencies $f^{\ast}=f/\dot{\gamma}_{\mathrm{m}}$. The time history of the radial position of the capsule centroid $r$ is shown in Fig. 2(a), together with the capsule shape at the initial ($\dot{\gamma}_{\mathrm{m}}t$ = 0) and final states ($\dot{\gamma}_{\mathrm{m}}t$ = 50). The capsule, initially spherical, migrates towards the channel centerline while deforming, finally reaching its equilibrium position at the centerline, where it achieves an axial-symmetric shape. While the trajectory obtained with the highest frequency investigated ($f^{\ast}$ = 5) well collapses on that obtained with a steady flow, see Appendix §A , when $f^{\ast}$ is small enough, the trajectory paths depend on the pulsation frequency, with the appearance of oscillations and with different axial migration speed.

The time history of the capsule deformation is shown in Fig. 2(b), quantified by the Taylor parameter $D_{12}=|a_{1}-a_{2}|/(a_{1}+a_{2})$, where $a_{1}$ and $a_{2}$ are the lengths of the semi-major and semi-minor axes of the capsule. Note that, we compute $D_{12}$ from the eigenvalues of the inertia tensor of an equivalent ellipsoid approximating the deformed capsule (Ramanujan and Pozrikidis, 1998). The capsule deformation is maximized just after the flow onset when the capsule is subject to the high shear near the wall. As time passes, $D_{12}$ decreases and settles to a value which is around one order of magnitude smaller than the maximum (i.e., $O(D_{12})=10^{-2}$) when reaching the channel axis.

The migration speed is also affected by the amplitude of the oscillation $\nabla p^{\mathrm{amp}}$, as shown in Fig. 3, where the side views of the capsule during its axial migration for different $\nabla p^{\mathrm{amp}}$ (= $\nabla p_{0}$ and $4\nabla p_{0}$) are shown in Figs. 3(a) and 3(b), respectively. The snapshots clearly show the capsule deformation and position as a consequence of the change of the background flow directions and oscillatory amplitudes. As $\nabla p^{\mathrm{amp}}$ increases, the capsule appears to migrate faster toward the channel centerline (Fig. 3c).

To properly quantify the changes in axial migration, we define the migration time $T^{\ast}$ as the time needed by the capsule centroid to reach the centerline (within a distance of $\sim$6% of its radius to account for the oscillations in the capsule trajectory). The ratio of the elapsed time $T^{\ast}$ and that in a steady flow is reported in Fig. 4(a) as a function of $f^{\ast}$, for various pulsation amplitudes. The results clearly suggest that there exist a specific frequency to minimize the migration time. A very minor increase of the optimal frequency with the pulsation amplitude can be observed in the data. While the optimal frequency is almost independent of the pulsation amplitude, the migration time can be strongly reduced by its increase. Indeed, while the elapsed time is reduced by $18\%$ at the lowest amplitude investigated ($\nabla p^{\mathrm{amp}}=\nabla p_{0}/4$), it is reduced by $80\%$ at the highest one ($\nabla p^{\mathrm{amp}}=4\nabla p_{0}$). Interestingly, the optimal frequency that minimizes the migration time ($O(f^{\ast})=10^{-2}$) is one order of magnitude smaller than the one which maximizes $D_{12}$ (Fig. 8 in Appendix §B), thus, suggesting that the axial migration time is unrelated to the maximum capsule deformation which happens in the initial stage of the capsule motion.

The changes in the migration time are clearly reflected in the migration speed ${\mathcal{V}}^{\ast}={\mathcal{V}}/V_{\mathrm{max}}^{\infty}$, reported in Fig. 4(b), which shows that when the migration time is minimum, the axial migration speed reaches almost its maximum. Here, the migration speed ${\mathcal{V}}$ is defined as the ratio of the elapsed time $T$ and the traveled distance ${\mathcal{L}}$ (i.e., ${\mathcal{V}}={\mathcal{L}}/T$), defined as ${\mathcal{L}}=\int_{0}^{\mathcal{L}}|d{\boldsymbol{r}}|=\int_{0}^{\mathcal{L}}d{\boldsymbol{r}}\cdot\hat{{\boldsymbol{t}}}=\int_{0}^{T}{\boldsymbol{v}}dt\cdot\hat{{\boldsymbol{t}}}$, where $\hat{{\boldsymbol{t}}}={\boldsymbol{r}}/|d{\boldsymbol{r}}|$ is the unit tangential vector along the trajectory of the capsule centroid and ${\boldsymbol{v}}$ is the the capsule centroid velocity.

The distance traveled by the capsule before completing the axial migration is reported in Fig. 4(c) for the sake of completeness, showing that the optimal frequency to minimize the migration time, roughly corresponds to the minimization of the the traveled distance too. Note that, the distance traveled during the migration $L^{\ast}$ depends not only on $f^{\ast}$ but also on $Ca$ (see Fig. 9 Appendix §C).

In summary, so far we have shown that, for a fixed $Ca$ and $\lambda$, there is an optimal frequency for the channel pulsation, able to minimize the capsule migration time by maximizing the migration speed and minimizing the traveled distance. To complete our investigation, the effects of $Ca$ and $\lambda$ on the migration time $T^{\ast}$ are shown in Fig. 5. In particular, the results in Fig. 5(a) shows that the migration time depends on $Ca$, thus suggesting that the optimal frequency $f^{\ast}$ is also a function of $Ca$. On the other hand, as shown in Fig. 5(b), the migration time remains almost independent of the viscosity ratio for $\lambda\lesssim 5$.

We also investigate the effect of the radial channel size on the migration time. Figure 6 shows the ratio of $T^{\ast}$ to that in a steady flow $T^{\ast}_{\mathrm{steady}}$ for two different channel size ratios $D/d_{0}$ = 2.5, 3.75, and 5, corresponding to $D(=2R)=20$ $\mu$m, $30$ $\mu$m, and $40$ $\mu$m for $d_{0}(=2a_{0})=8$ $\mu$m, as a function of the pulsation frequency $f^{\ast}$. For all cases, the initial position $r_{0}$ is set to be the same above (i.e., $r_{0}/R$ = 0.55). The results show that, independently of the channel size, the qualitative picture discussed above remains unchanged. While the amount of the speed-up of axial migration achieved with a pulsation remains almost unaltered (around $50\%$ for this case), the value of the optimal frequency changes with $D$, (the peak frequency reduces when $D$ is increased).

In conclusion, we have proved that the axial migration speed of an elastic capsule in a pipe flow can be substantially accelerated by making the driving pressure gradient oscillating in time. We found that, the axial migration speed increases with the amplitude of the oscillation, while the most effective frequency revealed to be independent of the oscillatory amplitude. Also, we showed that the optimal frequency depends on $Ca$, but is basically independent of the viscosity ratio $\lambda$, overall proving that the changes in the axial migration are mostly due to the membrane elasticity.

The behaviour of capsules under pulsatile channel flows has been investigated in in some previous works (Lafzi et al., 2020; Maestre et al., 2019). However, our study provides the first conclusive evidence of the acceleration of the axial migration of a capsule by pulsatile flow. Although it may be expected that the capsule is trapped in a state of resonance at the optimal frequency $f^{\ast}$ to minimize the migration time (Fig. 4a), there is currently no clear theoretical framework on the resonance frequency of capsule in confined channel flows. Indeed, in our case the capsule configuration and its centroid are changing simultaneously, making the problem more complicated than what investigated in previous theoretical and numerical studies which assumed small deformations (i.e., weakly nonlinear problem) of drops (Chan and Leal, 1979; Magnaudet et al., 2003) and bubbles (Sugiyama and Takemura, 2010).

Given that the migration speed can be controlled by oscillatory frequency as well as background flow strength (or amplitude), the results obtained here can be utilised for label-free cell alignment/sorting/separation techniques, e.g., for circulating tumor cells in cancer patients or precious hematopoietic cells such as colony-forming cells. Our numerical results obtained physiologically relevant RBC properties in size $a_{0}$ and membrane elasticity $G_{s}$ form a fundamental basis for further studies on cellular flow mechanics in confined environments.

N.T. was supported by JSPS KAKENHI Grant Number JP20H04504. M.E.R. was supported by the Okinawa Institute of Science and Technology Graduate University (OIST) with subsidy funding from the Cabinet Office, Government of Japan. he presented study was partially funded by Daicel Corporation. N.T. thanks Dr. Naoto Yokoyama for helpful discussion. Finally, the collaborative research was supported by the SHINKA grant provided by OIST.