Processing math: 100%

A Quorum Sensing Active Matter in a Confined Geometry

  • Inspired by the problem of biofilm growth, we numerically investigate clustering in a two-dimensional suspension of active (Janus) particles of finite size confined in a circular cavity. Their dynamics is regulated by a non-reciprocal mechanism that causes them to switch from active to passive above a certain threshold of the perceived near-neighbor density (quorum sensing). A variety of cluster phases, i.e., glassy, solid (hexatic) and liquid, are observed, depending on the particle dynamics at the boundary, the quorum sensing range, and the level of noise.
  • Article Text

  • Bacteria are capable of adjusting their motility to form large colonies, like biofilms. While motile bacteria have the advantage to swim efficiently towards food sources, biofilm aggregates are able to resist environmental threats such as antibacterial substances. Understanding the basic physical mechanisms of biofilm growth is a topic of ongoing research by many teams worldwide. Recent studies suggest that a motility-based clustering phenomenon is involved in the formation of bacterial swarms[1] and in the transition from bacterial swarms to biofilms.[2] Moreover, it is demonstrated that synthetic active materials, such as Janus colloids, can undergo motility-induced aggregation, not only via high-density steric mechanisms,[3] or lower density mutual interactions,[4] but also by simply adjusting their velocity according to the direction[5] and the local density of their peers,[6] largely insensitive to pair interactions. These situations have been modeled theoretically using both particle-based models and field theoretical approaches.[7] In that context, it was shown that the active systems may exhibit motility-induced phase separation (MIPS), whereby self-propelled particles, with only repulsive interactions, form aggregates by reducing their swimming speed in response to a local density value greater than a given threshold (a mechanism called quorum sensing[8,9]).

    In its simplest form, MIPS has been shown to be analogous to a gas-liquid phase separation. However, recent non-equilibrium field theories have predicted intriguing behaviors, like microphase separation[10] and an active foam phase with slowly coalescing bubbles.[11] In fact, our understanding of how the microscopic details of the single-particle dynamics lead to different collective behaviors is presently far from satisfactory. Finally, it has been shown that motile E. coli bacteria spontaneously aggregate within minutes when subject to controlled convective flows produced by a microfluidic device.[12] It is still unclear, however, which physical ingredients are required for a minimal active-particles model to reproduce such a behavior.[13,14]

    Upon a closer look, it is apparent that, while the emergence of steady aggregates of motile particles is largely driven by the nature of their mutual interactions, which ultimately influence their motility, the properties of such aggregates are strongly determined by the combined action of spatial confinement and fluctuations of both the suspension fluid and the self-propulsion mechanism. In this Letter, we revisit the model of non-reciprocal particle interaction proposed by Bechinger and coworkers[6] (also see Ref. [5]), by investigating the effects of the particle dynamics against the container walls at different noise levels. Contact and far-field reciprocal (pair) interactions have been neglected; no alignment mechanism has been invoked to trigger particle aggregation: Quorum sensing under spatial confinement is the mechanism considered here. As a result, we observe a variety of cluster phases, i.e., glassy, solid (hexatic), and liquid, and determine the relevant model phase diagram.

    ModelSingle Particle Dynamics. We consider the simplest realization of a synthetic microswimmer, namely a two-dimensional (2D) Janus particle (JP).[15] An active JP of label i gets a continuous push from the suspension fluid, which in the overdamped regime amounts to a self-propulsion velocity, v0i, with constant modulus v0, and orientation θi, fluctuating with time constant τθ, under the combined action of thermal noise and the rotational fluctuations intrinsic to the specific self-propulsion mechanism. In two dimensions, its bulk dynamics obeys the Langevin equations[16]
    ˙xi=v0cosθi+ξxi(t),˙yi=v0sinθi+ξyi(t),˙θi=ξθi(t),

    (1)
    where ri = (xi, yi) are the coordinates of the particle center of mass subject to the Gaussian noises ξpi(t), with 〈ξqi(t)〉 = 0 and 〈ξqi(t)ξpi(0)〉 = 2D0δqpδ(t) for q, p = x, y, modeling the equilibrium thermal fluctuations in the suspension fluid. The orientational fluctuations of the propulsion velocity are modeled by the Gaussian noise ξθi(t) with 〈ξθi(t)〉 = 0 and 〈ξθi(t)ξθi(0)〉 = 2Dθδ(t), with Dθ = 1/τθ being the relaxation rate of the self-propulsion velocity.

    The simplifications introduced in Eq. (1) are not limited to the reduced dimensionality of the system. All noise sources have been treated independently, although strictly speaking, spatial and orientational fluctuations are statistically correlated to some degree. Moreover, we ignore hydrodynamic effects which may favor the clustering of active particles at high packing fractions. However, we make sure that the parameters used in our simulations are experimentally accessible, apparently on expressing times in seconds and lengths in microns. The stochastic differential Eq. (1) is numerically integrated by means of a standard Euler–Maruyama scheme.[17] To ensure numerical stability, the numerical integrations have been performed using an appropriately short time step, 10−3.

    Boundary Conditions. In this study, the JPs are confined to a restricted area, say, a circular cavity of radius R [Fig. 1(a)]. One can think of motile bacteria spreading on a Petri dish. Equation (1) still applies away from the walls; we only need to set a prescription to treat the particle collisions with the boundaries. Following Refs. [18,19], we assume that, upon hitting it, a JP is captured by the wall and immediately re-injected into the cavity, at an angle ϕ with respect to the radius (i.e., the perpendicular) through the collision point [Fig. 1(a)]. A finite (short) trapping time does not affect the conclusions of the present work. A commonly accepted distribution for such a boundary scattering angle is[18]
    p(ϕ)=2exp(λcosϕ)/[πI0(λ)],

    (2)
    where I0 is the modified Bessel function of the first kind, and the parameter λ depends on the temperature and the physio-chemical properties of the particle and the cavity wall. Notice that in the limit λ → ∞ we recover the reflecting boundary conditions adopted in Ref. [6], namely, p(ϕ) = δ(ϕ). We consider other cavity geometries as well; an example is discussed at the bottom of the forthcoming section.
    Fig. Fig. 1.  Schematics of our model. (a) Illustration of the non-reciprocal interaction mechanism for N active Janus particles with visual half-angle α and horizon dc. The distribution, p(ϕ), of the boundary scattering angle, Eq. (2), is plotted for λ = 100: the distance of a point on the chart line from the scattering point on the boundary is proportional to the probability that the particle gets scattered in the direction of that point. (b) Quorum sensing protocol: for values of the sensing function above the threshold P0(α), Eq. (4), the particle turns from active to passive. (c) Example of passive cluster formation for λ = 100, α = (7/8)π, dc = 16, Dθ = 0.001, v0 = 0.5, R = 45, r0 = 1, N = 304, and D0 = 0.01 (snapshot taken at t = 2 × 104). Active/passive particles are represented by red/blue circles. (d) Passive (blue), active (red) and total (black) radial particle distributions for the parameters in (c). These distributions have been averaged over time (10 snapshots taken every 1000 time units starting at t = 104) and initial conditions (200 realizations).
    Non-Reciprocal Interaction (Sensing). When N is unchanged, independent active particles of Eq. (1) are confined into the cavity, interactions among them cannot be neglected. In our simulations we consider only two kinds of interactions: (i) hard-core repulsion, whereby the particles are modeled as hard discs of radius r0. Further reciprocal interactions have been discarded; (ii) neighbor perception, a mechanism governing the motility of each particle depending on the spatial distribution of its neighbors. In biological systems this process is mediated by some form of inter-particle communication (mostly chemical in bacteria colonies[8,9]). On the other hand, the motility of artificial microswimmers grows less efficient with increasing density.[15] Without entering the details of the specific perception mechanisms, we can define the sensing function of particle i as follows:[6,20]
    Pi(α)=jVαi12πrij,

    (3)
    where rij is the distance between particles i and j, and Vαi denotes the visual cone of particle i, centered around the direction of its self-propulsion velocity, v0i, with finite horizon, rijdc. This means that each particle senses the presence of other particles only within a restricted visual cone and a finite distance dc. For a uniform active suspension of density ρ0 = N/πR2, the sensing function of a particle placed at the center of the cavity reads[5]
    P0(α)=(α/π)ρ0R.

    (4)
    We assume now that the particle motility is governed by the following simple quorum sensing protocol [Fig. 1(b)],
    |v0i|={v0,Pi(α)P0(α),0,Pi(α)>P0(α).

    (5)
    Clearly, this form of particle interaction is non-reciprocal, since j may be perceived by i and, therefore, influence its dynamics, without being affected by the presence of i. The dynamical implications of the non-reciprocal interactions in biological matter are discussed at length by Bechinger and coworkers in Refs. [5,6]. For an earlier and more elaborated quorum sensing model of synthetic active matter, readers can refer to Ref. [21]. What matters here is that for appropriate choices of the horizon range dc and the visual angle α, clustering may occur, as illustrated in Fig. 1(c).

    Results. The number of tunable parameters of our model is quite large. In our simulations we keep the particle radius r0 and self-propulsion speed v0 fixed, amounted to setting space and time units. The particle number N and the cavity radius R play no key role as long as the suspension packing fraction ϕ0 = N(r0/R)2 is kept sufficiently small (typically ϕ0 < 0.2), to avoid steric clustering.[3] We remind that the active-passive transition threshold P0(α) of Eq. (4) scales like N/R. All remaining parameters D0, Dθ = 1/τθ, α, dc, and λ are varied to shed light on the underlying collective dynamics.

    Our main findings are summarized in Fig. 2 for the optimal case of full visual perception α = π, persistence length lθ = v0τθ much larger than the cavity diameter, and small translational noise D0 (whereby the time for a particle to diffuse a distance of the order of its diameter is much larger than to self-propel the same distance). The resulting 2D parameter space (λ, dc) is traversed by a continuous separatrix curve, dc vs λ, whereby below (above) it all JPs retain (lose) their active nature. Since the packing fraction of the simulated active suspension is too small to trigger steric clustering, no active aggregates are detected (region I). For small λ values the suspension maintains its initial uniform distribution, whereas at large λ, transient, short-lived active clusters form and dissolve (see Fig. 1 of the Supplemental Material[22])

    Fig. Fig. 2.  Phase diagram in the space parameter (λ, dc). Four distinct regions are distinguishable, namely, I (below the red curve): all particles remain active; II: a liquid condensate of passive particle coexists with a gas of active particles; III: the passive condensate solidifies into a steady-state hexatic structure; and IV: periodic formation of hexatic passive clusters (also see Fig. 3). Snapshots of the relevant suspension patterns taken at time t = 2 × 104 are shown as insets for different (λ, dc); red and blue circles denote respectively active and passive particles.
    Fig. Fig. 3.  Cluster stability. Top panel: the fraction of passive particles, Np(t)/N, vs t for different dc (see legend). The scattering angle distribution here is p(ϕ) = δ (ϕ), which corresponds to the limit λ → ∞ of Eq. (2). Inset: frequency spectrum of the subtracted function [Np(t) – Np(∞)]/N in the stationary regime for t ∈ (2 × 103, 2 × 104). Bottom panels: snapshots showing the aggregation and evaporation of a hexatic cluster for dc = 10. In panels, active/passive particles are denoted by red/blue circles and the remaining simulation parameters are α = π, Dθ = 0.001, v0 = 0.5, R = 45, r0 = 1, N = 304, and D0 = 0.01.

    Above the separatrix curve, the active-passive transitions induced by quorum sensing, Eq. (4), sustain the formation of large clusters, with core consisting of passive particles. For large dc values, region II, one typically observes a large passive condensate surrounded by a low-density gas of active particles [also see the active and passive radial distributions of Fig. 1(d)], which bears a certain similarity with the situation analyzed in Ref. [23]. The passive constituents of the condensate keep fluctuating in thermal noise, as expected in a liquid phase. At large λ, when the wall scatters the colliding particles mostly toward the center of the cavity, the clustering mechanism exhibits additional distinct features. Lowering dc, we can detect two more regions, III and IV, characterized by very dense clusters made of a passive core surrounded by an active layer; in both of them, the particles are closely packed into hexatic structures.[24] The particles in the cluster active layer show larger motility than in the cluster core, but substantially lower than the surrounding active gas particles, As a major difference, in region III the clusters are stationary in time, whereas in region IV the clusters appear and disappear over time. The time oscillating clustering process of region IV is further analyzed in Fig. 3. In the top panel there we plot the fraction of passive particles, Np/N, versus time for λ → ∞. One notices immediately that for dc values corresponding to the regions I–III, this ratio approaches a steady state value after a transient of the order of the ballistic cavity crossing time, R/v0. We also remark for all curves Np(t → ∞)/N < 1, no matter what is dc, which suggests that a gas of active particles is always at work. Vice versa, as anticipated above, for system configurations in the region IV, Np/N appears to execute persistent irregular time oscillations. A spectral analysis of the time-dependent ratio Np(t)/N confirms that: (i) Np(t) fluctuates around a stationary asymptotic value Np(∞) < 1; (ii) the spectral density of the subtracted ratio, [Np(t) – Np(∞)]/N (an example is shown in the inset of Fig. 3), peaks around a finite frequency of the order of Dθ. We further observe that, with increasing Dθ, region IV in Fig. 2 shrinks and finally disappears.

    The numerical results of Figs. 2 and 3 demonstrate the role of the boundary in the cluster formation. At large λ, the distribution of the scattering angle, p(ϕ), is mainly peaked around ϕ = 0; the boundary exerts a lensing effect on the active particles by re-directing them toward the center of the cavity. This clearly enhances the probability that the sensing function, Pi(α), of the particles there overcome the threshold value, P0(α), of Eq. (4), thus triggering the clustering process. This is the key “herding” function of the gas of active particles continuously bouncing between the cavity and the cluster border. Accordingly, we note that in region II the clusters tend to be denser along the border. In the opposite limit of wide scattering angle distribution, i.e., for small λ, the active suspension is no longer focused toward the cavity center and clustering is suppressed. The separatrix curve in Fig. 2 clearly shows that for λ → 0 clustering requires that dcR, as implicit in the quorum sensing protocol adopted with Eqs. (3) and (4).

    Discussion and Conclusions. The phase diagram of Fig. 2 is obtained for a convenient choice of the tunable parameters D0, Dθ, and α. We now briefly discuss the role of these parameters in the clustering process.

    (i) Role of Spatial Noise, D0. The additive noises ξxi(t) and ξyi(t) in Eq. (1) keep the particles diffusing even after they undergo the active-to-passive transition. This has a twofold consequence. On the one hand, it hampers the cluster formation by delaying it in time and pushing the separatrix curve of Fig. 2 to higher dc values [compare Figs. 4(a) and 4(b)]. On the other hand, for D0 = 0 the passive particles come immediately at rest after having reset their self-propulsion velocity to zero. This leads to the buildup of frozen clusters with an amorphous glassy structure [Fig. 4(c)]. Vice versa, adding a little amount of noise allows clusters to rearrange themselves in the denser hexatic structures of region III [Fig. 4(d)].

    Fig. Fig. 4.  Role of additive noise. Left panels: D0 = 0 and snapshot time t = 105; right panels: D0 = 0.001 and t = 2 × 104; top panels: dc = 24 and Dθ = 0.01; bottom panels: dc = 16 and Dθ = 0.001. The remaining simulation parameters are α = π, v0 = 0.5, R = 45, r0 = 1, and N = 304. The ϕ distribution and the particle color code are the same as in Fig. 3.

    (ii) Role of Rotational Noise, Dθ. In Fig. 2 we assume that the particle persistence length lθ = v0/Dθ is much larger than the cavity diameter. That choice is convenient in that it enhances the role of the boundary dynamics in the clustering process. Indeed, under this condition, active JPs may hit the cavity walls repeatedly before grouping at the center, where eventually undergo the active-to-passive transition. To clarify the role of the persistence time τθ = 1/Dθ, we simulate the time evolution of the same suspension for increasing values of Dθ and observe that cluster formation gets, indeed, progressively suppressed (see the Supplemental Material[22] for details). This comes as no surprise, since upon increasing Dθ, the persistence length lθ decreases and the active particles’ dynamics resembles more and more standard Brownian motion with strength v20/2Dθ.

    (iii) Role of the Visual Angle, α. We now consider the cases of α < π, contrarily to Fig. 2. This means that the neighbor perception of particle i is restricted to a visual cone directed along its instantaneous self-propulsion velocity vector, v0i.[5] This enhances the non-reciprocal nature of the particle interactions. As a consequence, the active-passive transitions at the periphery of the forming clusters become more frequent. Indeed, an incoming particle perceives a comparatively much larger neighbor density than a particle moving outward. We remind here that all particles, active and passive alike, keep rotating randomly [third equation in (1)] with correlation time τθ. This mechanism tends to destabilize the forming clusters, so that one expects that shrinking the visual cone eventually suppresses clustering. Our simulations confirm this guess, even though the asymptotic value of the ratio Np(t)/N exhibits a non-monotonic α dependence with a maximum for α/π ≳ (3/4) (Fig. 5), compared the snapshots for dc = 16 and α = (7/8)π in Fig. 1(c) and α = π in Fig. 2. We attribute its behavior to the combined effect of the above mechanism and the α dependence of the sensing threshold P0(α) (see the Supplemental Material[22] for details).

    Fig. Fig. 5.  Role of the visual angle α. Asymptotic value of Np(t)/N vs α for different choices of the tunable parameters (λ, dc). The remaining simulation parameters are Dθ = 0.001, v0 = 0.5, R = 45, r0 = 1, N = 304, and D0 = 0.05.

    We conclude this report briefly by discussing a special case, where the cavity wall is replaced by periodic boundaries. We consider a square 2D simulation box of size L: particle dynamics and quorum sensing protocols are the same as those for the circular cavity; as a difference, a particle i crossing a box side is re-injected into the box through the opposite side with same self-propulsion vector v0i. In this regard, periodic boundaries are reminiscent of the scattering wall of the initial model for λ → 0, in that the self-propulsion direction of the re-injected particle tends to be uniformly distributed. Similarly to Fig. 2, one then expects that clustering only occurs for d0L/2, as a consequence of the very definition of the active-passive transition threshold, P0(α). Direct numerical simulations (not shown) confirm this expectation.

    Acknowledgment: This work was supported by the National Natural Science Foundation of China (Grant Nos. 12375037 and 11935010).
  • [1]
    Schwarzendahl F J, Mazza M G 2022 Europhys. Lett. 140 47001 doi: 10.1209/0295-5075/aca11c

    CrossRef Google Scholar

    [2]
    Grobas I, Polin M, Asally M 2021 eLife 10 e62632 doi: 10.7554/eLife.62632

    CrossRef Google Scholar

    [3]
    Fily Y, Marchetti M C 2012 Phys. Rev. Lett. 108 235702 doi: 10.1103/PhysRevLett.108.235702

    CrossRef Google Scholar

    [4]
    Ginot F, Theurkauff I, Detcheverry F, Ybert C, Cottin-Bizonne C 2018 Nat. Commun. 9 696 doi: 10.1038/s41467-017-02625-7

    CrossRef Google Scholar

    [5]
    Lavergne F A, Wendehenne H, Bäuerle T, Bechinger C 2019 Science 364 6435 doi: 10.1126/science.aau5347

    CrossRef Google Scholar

    [6]
    Bäuerle T, Fischer A, Speck T, Bechinger C 2018 Nat. Commun. 9 3232 doi: 10.1038/s41467-018-05675-7

    CrossRef Google Scholar

    [7]
    Cates M E, Tailleur J 2015 Annu. Rev. Condens. Matter Phys. 6 219 doi: 10.1146/annurev-conmatphys-031214-014710

    CrossRef Google Scholar

    [8]
    Miller M B, Bassler B L 2001 Annu. Rev. Microbiol. 55 165 doi: 10.1146/annurev.micro.55.1.165

    CrossRef Google Scholar

    [9]
    Parsek N R, Greenberg E P 2005 Trends Microbiol. 13 27 doi: 10.1016/j.tim.2004.11.007

    CrossRef Google Scholar

    [10]
    Tjhung E, Nardini C, Cates M E 2018 Phys. Rev. X 8 031080 doi: 10.1103/PhysRevX.8.031080

    CrossRef Google Scholar

    [11]
    Shi X, Fausti G, Chaté H, Nardini C, Solon A 2020 Phys. Rev. Lett. 125 168001 doi: 10.1103/PhysRevLett.125.168001

    CrossRef Google Scholar

    [12]
    Yadzi S, Ardekani A M 2012 Biomicrofluidics 6 044114 doi: 10.1063/1.4771407

    CrossRef Google Scholar

    [13]
    Li Y Y, Zhou Y X, Marchesoni F, Ghosh P K 2022 Soft Matter 18 4778 doi: 10.1039/D2SM00500J

    CrossRef Google Scholar

    [14]
    Ghosh P K, Zhou Y, Li Y, Marchesoni F, Nori F 2023 ChemPhysChem 24 e202200471 doi: 10.1002/cphc.202200471

    CrossRef Google Scholar

    [15]
    Walther A, Müller A H E 2013 Chem. Rev. 113 5194 doi: 10.1021/cr300089t

    CrossRef Google Scholar

    [16]
    Ghosh P K, Misko V R, Marchesoni F, Nori F 2013 Phys. Rev. Lett. 110 268301 doi: 10.1103/PhysRevLett.110.268301

    CrossRef Google Scholar

    [17]
    Kloeden P E, Platen E 1992 Numerical Solution of Stochastic Differential Equations Berlin Springer

    Google Scholar

    [18]
    Kärger J, Ruthven D M 1992 Diffusion in Zeolites and Other Microporous Solids New York Wiley

    Google Scholar

    [19]
    Codina J, Mahault B, Chaté H, Dobnikar J, Pagonabarraga I, Shi X 2022 Phys. Rev. Lett. 128 218001 doi: 10.1103/PhysRevLett.128.218001

    CrossRef Google Scholar

    [20]
    Chen Q, Patelli A, Chaté H, Ma Y, Shi X 2017 Phys. Rev. E 96 020601R doi: 10.1103/PhysRevE.96.020601

    CrossRef Google Scholar

    [21]
    Barberis L, Peruani F 2016 Phys. Rev. Lett. 117 248001 doi: 10.1103/PhysRevLett.117.248001

    CrossRef Google Scholar

    [22]
    see the Supplemental Material.

    Google Scholar

    [23]
    Kümmel F, Shabestari P, Lozano C, Volpe G, Bechinger C 2015 Soft Matter 11 6187 doi: 10.1039/C5SM00827A

    CrossRef Google Scholar

    [24]
    Redner G S, Hagan M F, Baskaran A 2013 Phys. Rev. Lett. 110 055701 doi: 10.1103/PhysRevLett.110.055701

    CrossRef Google Scholar

  • Related Articles

    [1]Zezhong Li, Lankun Han, Yili Sun, Shanshan Zhang, Zhenyuan Zeng, Shiliang Li. Antiferromagnetic Order and Possible Quantum Spin Liquid in Kagome Antiferromagnet LuCu3(OH)6Br2[Brx(OH)1-x] [J]. Chin. Phys. Lett., 2025, 42(2): 027504. doi: 10.1088/0256-307X/42/2/027504
    [2]LI Tao, XIANG Ying, LIU Yi-Kun, WANG Jian, YANG Shun-Lin. Transient Reorientation of a Doped Liquid Crystal System under a Short Laser Pulse [J]. Chin. Phys. Lett., 2009, 26(8): 086108. doi: 10.1088/0256-307X/26/8/086108
    [3]WU Zhi-Min, WANG Xin-Qiang, WANG Fang-Wei. Magnetic Properties and Spin State Transition of Gallium Doped Perovskite Cobaltite Oxide [J]. Chin. Phys. Lett., 2007, 24(11): 3249-3252.
    [4]SONG Jing, LIU Yong-Gang, MA Ji, XUAN Li. Low Driving Voltage and Analysis of Azobenzene Polymer Doped Liquid Crystal Grating [J]. Chin. Phys. Lett., 2006, 23(12): 3285-3287.
    [5]WANG Cheng-Zhi, LI Chun-Xian, GUO Guang-Can. Ground-State Entanglement and Mixture in an XXZ Spin Chain [J]. Chin. Phys. Lett., 2005, 22(8): 1829-1832.
    [6]CAO Gui-Xin, ZHANG Jin-Cang, SHA Yan-Na, YAO Kai, CAO Shi-Xun, JING Chao, SHEN Xue-Chu. Reentrant Spin Glass Behaviour and CE-Type Antiferromagnetic Phase in Half Doping (La,Pr)1,2Ca1/2 MnO3 Manganites [J]. Chin. Phys. Lett., 2005, 22(3): 682-685.
    [7]KOU Su-Peng. Fermion Bound States Around Skyrmions in Doped Antiferromagnets [J]. Chin. Phys. Lett., 2003, 20(8): 1353-1355.
    [8]MENG Chuan-Min, SHI Shang-Chun, DONG Shi, SUN Yue, JIAO Rong-Zhen, YANG Xiang-Dong. Equation of State of Dense Liquid Nitrogen in the Regionof the Dissociative Phase Transition [J]. Chin. Phys. Lett., 2002, 19(2): 252-254.
    [9]AN Jin, GONG Chang-De, LIN Hai-Qing. Competition Between Two Ordering Processes in Two-Dimensional Doped Antiferromagnets [J]. Chin. Phys. Lett., 2001, 18(3): 419-421.
    [10]LU Wei, LIU Pulin, SHEN Xuechu, M. von Ortenberg, J. Tuchendler, J . P. Renard. Haldane Gap Related Energy States in Spin s = 1 Linear ChainHeisenberg Antiferromagnet [J]. Chin. Phys. Lett., 1994, 11(3): 177-180.

Catalog

    Figures(5)

    Article views (41) PDF downloads (290) Cited by()

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return