Reconfigurable emergent patterns in active chiral fluids

Active fluids comprised of autonomous spinning units injecting energy and angular momentum at the microscopic level represent a promising platform for active materials design. The complexity of the accessible dynamic states is expected to dramatically increase in the case of chiral active units. Here, we use shape anisotropy of colloidal particles to introduce chiral rollers with activity-controlled curvatures of their trajectories and spontaneous handedness of their motion. By controlling activity through variations of the energizing electric field, we reveal emergent dynamic phases, ranging from a gas of spinners to aster-like vortices and rotating flocks, with either polar or nematic alignment of the particles. We demonstrate control and reversibility of these dynamic states by activity. Our findings provide insights into the onset of spatial and temporal coherence in a broad class of active chiral systems, both living and synthetic, and hint at design pathways for active materials based on self-organization and reconfigurability.

(vi) The phase diagram in Fig. 3 is shown up to E~3 V/\mu m. What would happen for E>3V/\mu m? Is there an experimental limitation making it impossible to explore this regime?
(vii) It could be interesting to quantify the polarization within the rotating flocks. How does the polarization change (qualitatively or quantitatively) when enhancing E?
(ix) The caption of Fig. 3 is too long I think. I suggest showing legends within some of the panels of Fig. 3 to make the figure better accessible.
Reviewer #2 (Remarks to the Author): Zhang et al report the emergence of two classes of dynamical patterns in collections of self-propelled colloidal rollers. Actuating polar colloidal dumbbells using an electrohydrodynamic instability, the authors show how to control both the translational and orbital speed of model self-propelled bodies. They provide a characterisation of their individual dynamics before addressing their collective behavior. When the polar rollers interact they can either self assemble into collection of vortices, or into a nonequilibrium phase termed rotating flocks. In rotating flocks, all rollers orbit in synchrony over macroscopic regions of space propelling along circular trajectories. The transition between the two dynamical regimes is controlled by the magnitude of the electric field responsible for the so-called Quincke rotation at the origin of self-propulsion.
The reported phenomena are truly spectacular and very carefully characterized by a number of quantitative measurements (which is unusual in the field). I have however two concerns about this work: -Firstly, the dynamics of orbiting active particles is not new. They were reported in earlier experiments from the Oiwa and Granick groups and theoretically/numerically discussed by the Chaté and Luijten groups (among others). These reference should be cited in the main text and the presented results thoroughly compared to these earlier reports.
Effective temperature concept evaluated in an active colloid mixture By Ming Han, Jing Yan, Steve Granick, and Erik Luijten PNAS July 18, 2017 114 (29)  In fairness, Zhang et al provide cleaner and more extensive experimental data, and their system makes it possible to continuously explore the phase behavior of the interacting colloids (This was also possible with in the experiments by Ha and coworkers, but not discussed as thoroughly as in the manscript of Zhang et al).
-Secondly, while the experimental characterisation of the emergent dynamics is very neat, the authors fall short of explaining the rich phenomenology they observed. As a consequence, the manuscript is plagued with rather vague statements aout the origin of the various dynamics (onebody and many-body dynamics). Reading this manuscript, I could not understand what are the mechanisms underlying the dynamical transitions observed between the Gas, Vortex and Rotating flock regimes upon increasing E. How are the symmetries and magnitude of the interactions between the particles altered y E? Is the singletrajectory curvature enough to explain the full phase behavior? Are the three regimes, three genuinely distinct dynamical phases? are the domains of the phase diagram separated by smooth crossovers? Can the observed phenomenology be accounted for by the Vicsek like models studied e.g. in Activity induced synchronization: Mutual flocking and chiral self-sorting, Phys. Rev. Research. 2019 by Demian Levis, Ignacio Pagonabarraga, and Benno Liebchen?
Given the absence of clear comparisons with the abundant earlier literature (experiments, numerics and theory), and the lack of explanations fo the rich collective dynamics observed in these beautiful experiments, I am afraid I cannot recommend the current manuscript for publication in Nature Communications.
More specific questions and comments: -What does set the location of the vortices? Does it change from one experiment to another (in the same device), or do they reflect some built in heterogeneities? What is responsible for the cohesion of these vertical structures?
-When \kappa=0, I would have expected a standard flocking phase to emerge (instead of \beta vortices). Could the authors explain why it is absent from the phase diagram? Does this reflect the polydispersity of the curvature distribution that does not peak at 0 when the <\kappa>=0?
-At the single-colloid level, could the author provide a qualitative explanation for the chirality reversal? (a quantitative explanation would e even better) - Fig. 1b. Do the authors plot the most probable or the average speed value. The suppression of a sharp bifurcation could well be a mere artifact caused by some polydispersity in the particle shape/chemistry. Plotting the most probable value of teh propulsion speed should suppress this possible artifact.
-As Fig. 1.d hardly depends on E, I feel that a 2D plot of \kappa(E) could provide a clearer illustration of the change in the roller behavior.
Reviewer #3 (Remarks to the Author): This experimental work investigates the patterns of spontaneous rotations performed by colloidal pear-shaped dielectric particles confined in a cylindrical cell and powered by a constant electric field. The experimental results show very attractive and to my knowledge novel collective behavior; the study covers an extensive range of quantities and parameters. However, the manuscript is written in an unclear manner, and after reading it quite carefully, I think there are contradictions, and several superficial or even might be wrong statements. Unfortunately, I can therefore not recommend this work for publication, at least not in its actual stage.
Following I include some more detailed comments: 1.-First, the abstract and introduction do not give clear hints of what is done in the manuscript and have in fact very little information about. In the conclusion paragraph, it is only the first few lines that summarize the work, while the second half refers to unproven and I would say unlikely projections of their results.
2.-In the first part of the manuscript the dynamics of individual particles is analyzed (around Fig. 1). The motion of the colloid is described as belonging to three modes attending to the orientation of the main axis with respect to the substrate plane. Figure 1c, and the sentence: "Transitions from \alpha to \gamma modes with the activity also inevitably lead to a chiral states reversal of the individual rollers" indicate that the trajectory of mode \beta is straight, while the other two are curved, so one would be CW and the other CCW but which one is which one ? And we can understand that fixing the applied field and the density these could not coexist then. Is this correct ?
3.-I find the discussion of the MSD and MSAD shown in Figs 1e and 1f quite confusing. 3a.--The text only states: "Initially the rollers move ballistically as msd~\tau^2 and then transition to a diffusive-like regime at longer times." As far as I see this is clear the case only for the trajectory of mode \beta, since mode \alpha seems to remain ballistic much longer and mode \gamma can become even subdiffusive. 3b.--The text further states: "At high activity levels, the curvature of the trajectories increases resulting in an appearance of characteristic oscillations in the mean square displacement". But the oscillations are only appearing in the mode \beta data, so should we understand that this mode is the only one showing rotating trajectories ? 3c.--The MSAD data seems to agree with the previous statement since mode \gamma remains ballistic, while the other data clearly show the transition from diffusive to ballistic, so this would indicate rotational motion to all modes, contradicting previous statements 4.-Later in the manuscript collective dynamics is investigated and linked to the single particle properties. Although this would be a standard solid procedure to understand the system, here is confusing. In page 4, and related to date in Fig. 2e it is stated that: "the system of pear-shaped rollers exhibits a spontaneous chirality induced phase separation from initially random distribution of chiral rollers" The initially random distribution is absolutely not proved here, and also not the transition but only an already phase separated system. Looking very carefully at all the provided movies, I would say that it seems more that the rotation is more dictated by the environment, (Visek type of behavior) and not to the individual properties as discussed here. Fig. 2f the authors refer to the subdiffusive behavior shown in Fig. 1e, but as far as I understand these two sets of data correspond to different densities, so no linked or conclusion could be made. The MSD at higher densities should then be also measured, and the collective behavior at more dilute regimes discussed.

5.-Discussing
6.-The radius of the individual trajectories seem to be importantly affected by the intensity of the applied field and/or by the density, as shown in Figs 2c and 2f. I find this very interesting, but it has not been even discussed at the single partied regime level, and I think should be investigated for all range of E and \phi values here presented. If there is a reason not to link these values it should be explained, otherwise a more in deep analysis should be presented. For example the frequencies are claimed to increase linearly with activity, but how linearly, quadratically, do the authors present an argument of why is this behavior expected ?
8.-The persistence length Lp is here defined "as the distance particles travel where the velocity temporal correlation function Ct decays to 1/e". I assume this has to be equivalent, or at least related to the more intuitive and more commonly used definition of the decay of the orientation correlation along the trajectory as most, is this the case ? All movies and the trajectories shown in Figs. 3c, 3f show that the distribution of such Lp's should be broad, since it can be very different depending on the particle relative position to the vortex center or boundary area. This is not discussed and I think it will have a big influence in the relevance of these quantities. 11.-I also assume now that the beta vortex is able to choose both CW and CCW, and therefore is this type (with its very limited set of possible parameters) the only one able to show the discussed coexisting vortexes of opposite rotations. In the text now, it seems that this behavior is more the rule than the exception.
12.-On the other hand, I am not sure if this contradicts the discussion at the beginning of the manuscript were the rotation direction was stated to be determined the applied field (see Fig. 1d). The rotations we supposed to have opposite direction for \alpha and \gamma, while non-existing for \beta. It could be that the explanation for this is that the effect emerges with increasing density, but something should be said and probably characterized about it, or simply that the single particle motion need more clarification.
13.-The authors also claim around Fig. 2 that the discussed patterns are quite stable on time. However, there is not much information about for how much time is this checked, or if this happens for all types of patterns, or only for the two shown in Fig. 2. 14.-The spinner phase is only shown in the most diluted case. I would find interesting to see how this is for higher densities.
15.-If I understand properly the PDF's of \psi_1 an\psi_2 shown in the manuscript and in more detail in the supplementary note correspond just to a very particular configuration, this is one realization at one particular time. This would mean that these peaks can change with time and would disappear if an averaged PDF would be shown. Would the rest of the distribution then remain stable in that case, and I would wonder about the relevance of the peaks is a general discussion of the system as the one presented in this work. More important would be to discuss the polar order shows two maximums in the vortex phase, but not the nematic order.
16.-Finally I am interested on the importance of the system size. Would it possible to make experiments with cell of different sizes ? Both smaller or larger would provide important information. In case this is not possible, a discussion of the possible outcome, and the consequent universality of the obtained phase diagrams seems relevant here. -The frequently refereed purple color is a bit darker blue in my prints.
- Caption Fig 3a: I guess that the colors do not refer to the symbols as stated in the caption, but to the background color.
-The scale bars would be more clear referring to the same length in all plots.
-Why there is is no data for \phi=0.001 in Fig, 3e ? -Symbols are not always defined before being used.