betaFeedback

WorksPhysics and Astronomy

Direct Observation of k-Gaps in Dynamically Modulated Phononic Time Crystal

Abstract

Floquet time crystals, characterized by momentum gaps (k-gaps), have sparked intense interest across various branches of physics due to their intriguing dynamics and promising applications. Despite growing theoretical efforts, the realization and observation of phononic time crystals, especially for airborne sound, remain significant experimental challenges. In this work, we demonstrate a phononic time crystal by integrating discrete resonant meta-atoms into a one-dimensional acoustic waveguide, effectively creating a homogeneous, time-varying metamaterial. By dynamically modulating the effective compressibility, we experimentally observe exponential acoustic wave amplification, offering clear evidence of k-gap formation. Furthermore, we showcase the versatility of our platform by inducing momentum band folding and double k-gap phenomena via quasi-periodic temporal modulation. This flexible and reconfigurable approach not only enables the design of tailor-made resonant responses but also opens new avenues for realizing higher-dimensional phononic time crystals and exploring nontrivial topological dynamics in time-modulated media.

Introduction

Floquet time crystals are a new category of artificial materials distinct from conventional spatial crystals, whose material constitutive parameters are uniform in space but periodically modulated in time [1, 2, 3, 4, 5]. These periodic time interfaces induce interference between time-reflected and time-refracted waves, generating momentum band structure due to broken discrete time-translational symmetry. Unlike the energy gap (ω\omega-gap) in spatial crystals, the momentum gap (kk-gap) supports two Floquet modes: one exponential growing and the other decaying in time. Recently, numerous interesting phenomena have been theoretically predicted in photonic time crystals, including topological temporal edge states [6, 7, 8, 9], temporal Anderson localization [10], amplified emission from electrons and dipole atom [2, 11], and superluminal momentum-gap solitons [12]. To realize these effects experimentally, a variety of time-varying photonic platforms have been developed. In microwave region, dynamic transmission lines are employed to experimentally observe kk-gaps [13] and Bloch-Floquet and non-Bloch band structures are observed in an array of temporally driven resonators [14]. Additionally, time-varying metasurfaces have achieved exponential growth wave in kk-gaps by relaxing volumetric systems into surface-based photonic time crystals [15]. However, synthesizing time-varying materials at optical frequencies remains challenging due to the need for ultrafast modulation (twice the light oscillation frequency). Promising candidates include all-optically modulated transparent conductive oxides due to the high effective of refractive index change, though high pumping power requirements lead to thermal damage and limit the performance [16, 17, 18, 19, 20, 21]. Recent proposals suggest expanding kk-gaps via artificial resonators with time-varying resonating strengths [22], highlighting the potential for novel modulation schemes.

As a universal concept, Floquet time crystals have also been explored in elastic wave [23, 24, 25], water wave [26], acoustics [27] and so on. For airborne sound, realizing phononic time crystals remains considerable challenge in achieving fast, spatially uniform material modulation. Prior strategies include mechanically controlled resonators for nonreciprocal transmission [28], but these suffer from frictional loss, limiting the modulation rate and depth. Electroacoustic devices with digital feedback are used to obtain nonreciprocal mode transitions [29] by temporally switching the impedance of transducers with low modulating frequencies. Furthermore, digital virtualized meta-atoms [30], which consist of microphone and speaker pairs interconnected by an external microcontroller implementing a time-varying convolution kernel, have been proposed to investigate temporal effective medium theory [31, 32] at very high modulating frequencies and unidirectional amplification [33] at a low modulation frequency. These digital platforms provide great flexibility in tailoring resonant responses and modulation frequency, crucial for synthesizing homogeneous time-varying metamaterials.

In this report, we implement such phononic time crystals by integrating discrete resonating time-varying meta-atoms into a one-dimensional acoustic waveguide. By temporally modulating the effective compressibility, we experimentally observe the wave amplification, providing direct evidence of k-gap in phononic time crystals. Additionally, to highlight the platform’s versatility, we also present the momentum band fold and double k-gaps opening phenomenon by quasi-periodic modulating the compressibility of metamaterial. Our experiment results demonstrate that the time-varying metamaterials composed of discrete resonating meta-atoms not only provide huge flexibility in modulating tailor-made resonating response, but also offer a promising route for constructing higher dimensional phononic time crystals and extended to other wave systems for designing Floquet time crystals.

Results

We begin with a one-dimensional (1D), spatially homogeneous phononic time crystal along the xx-axis, as shown in Figure 1(a). Its compressibility β(x,t)\beta(x,t), normalized by the compressibility of air β0\beta_0, is modulated over time (tt) with a constant density ρ0\rho_0. For simplicity, we consider the temporal modulation with alternating phases A and B, over a modulation period TmT_m, shown as red and blue stripes representing βA\beta_A and βB\beta_B. Each phase occupies Tm/2T_m/2. For this system, the airborne acoustic wave equations are given as

Image not extracted

Figure 1. Concept of phononic time crystal and momentum bandgap. (a) Schematic picture of phononic time crystal whose compressibility switches between βA\beta_A and βB\beta_B with period TmT_m. (b) Band structure (red and blue solid lines for real and imaginary part) and momentum bandgap (orange shaded) of phononic time crystal in (a). (c) Experimental setup using a feedback system with Lorentzian resonance YA/YBY_A/Y_B. Each meta-atom consists of a speaker and microphone interconnected by a microcontroller. The feedback is a time domain convolution switching between YA/YBY_A/Y_B to realize the time-varying β(t)\beta(t). (d) Static compressibility for phase A and B(e). (e) Band structure of phononic time crystal with time-varying β\beta switching between phase A and B in (d).

∂xp(x,t)+ρ0∂tv(x,t)=0,(1)\partial_x p(x,t) + \rho_0 \partial_t v(x,t) = 0, \tag*{(1)}
∂xv+β0∂t(β(x,t)p(x,t))=0,(2)\partial_x v + \beta_0 \partial_t\left(\beta(x,t)p(x,t)\right) = 0, \tag*{(2)}

Since the medium is homogeneous and infinite along the xx direction, the wave number kk remains unchanged across time interfaces. By substituting ∂x\partial_x with ikik, Eq. (1) and (2) can be written in matrix form as

i∂tψ=ω^ψ,ω^=(0kkρ0β0)(3)i\partial_t\psi=\hat{\omega}\psi,\qquad\hat{\omega}=\begin{pmatrix}0 & k \\ \frac{k}{\rho_0\beta} & 0\end{pmatrix} \tag*{(3)}

with state vector ψ=(βp,v)T\psi=(\beta p,v)^{T}. ω^\hat{\omega} is the propagation matrix. The state vector evolves as

ψ(t)=e−iω^tψ(0).\psi(t)=e^{-i\hat{\omega}t}\psi(0).

The Floquet modes satisfies ψ(t+Tm)=e−iω^BTm/2e−iω^ATm/2ψ(t)=e−iΩTmψ(t)\psi(t+T_m)=e^{-i\hat{\omega}_B T_m/2}e^{-i\hat{\omega}_A T_m/2}\psi(t)=e^{-i\Omega T_m}\psi(t), with ω^A/ω^B\hat{\omega}_A/\hat{\omega}_B being the propagation matrix in phase A/B and Ω\Omega being the Floquet frequency. This leads to the band structure (Ω\Omega verse kk) relationship:

det⁡[e−iΩTmI2−e−iω^BTm/2e−iω^ATm/2]=0,(4)\det\left[e^{-i\Omega T_m}I_2-e^{-i\hat{\omega}_B T_m/2}e^{-i\hat{\omega}_A T_m/2}\right]=0, \tag*{(4)}

where I2I_2 is the 2 by 2 identity matrix. Solving this secular equation in Eq. (4), Figure 1 (b) presents the band structure of a non-dispersive phononic crystal with compressibility modulated between βA=1.12\beta_A=1.12 and βB=0.88\beta_B=0.88 at a modulation frequency 1/Tm=8.41/T_m=8.4 kHz. A characteristic wave number associated with the modulation frequency is defined as: k0=2π/(c0Tm)k_0=2\pi/(c_0T_m), where c0c_0 is the speed of sound in air. Then, the band structure is plotted as ΩTm\Omega T_m versus normalized wave number k/k0k/k_0, with red and blue lines denoting the real and imaginary part of Floquet frequency, respectively. The bandgap region, highlighted in yellow, near half the modulation frequency (4.2 kHz), corresponds to non-zero imaginary part of Floquet frequency, one being positive (amplifying mode) and the other being negative (decay mode). To observe wave behavior around the band gap, a medium with negligible frequency dispersion is ideal and should be modulated at twice the operational frequency of interest. However, a true non-dispersive medium does not exist. As an alternative, we operate at frequencies far from the resonance of the medium to approximate a non-dispersive response. To achieve this, we design an array of acoustic resonant meta-atoms in a 1D waveguide, as shown in Figure 1 (c). The resonance parameters of each atom can be modulated in time. Each meta-atom consists of a detector and a speaker, interconnected via a microcontroller that performs time domain convolution Y(t)Y(t) on the detected signal, alternating between YAY_A (blue) and YBY_B (cyan). This realizes a Lorentzian-type compressibility described in the frequency domain as:

β(f)≈1+c0iπflY(f),Y(f)=ifgf02−f2−2iγf.(5)\beta(f)\approx1+\frac{c_0}{i\pi f l}Y(f),\qquad Y(f)=\frac{i f g}{f_0^2-f^2-2i\gamma f}. \tag*{(5)}

Here gg, f0f_0 and γ\gamma are the resonant strength, resonant frequency and linewidth, ll is the lattice distance. By modulating the resonant strength gg between gAg_A and gBg_B, we effectively switch between two convolution kernels YAY_A and YBY_B, realizing time-dependent compressibility. The two static compressibility configurations are shown in Figure 1 (d), labeled as “Static A” and “Static B”, with the real part in red and imaginary part in blue. The compressibility values at fm/2f_m/2 are βA≈1.12\beta_A \approx1.12 and βB≈0.88\beta_B \approx0.88, marked with a red and blue star respectively. This approximates the nondispersive case depicted in Figure 1 (b). Using the dispersive characteristics of the two static phases, we calculate the band structure of the time modulated system plotted in Figure 1 (e), the details of calculated method are presented in Note1 of Supplementary Materials. Similar to Figure 1(b), the band structure shows a momentum gap around 4.2 kHz, which agree well with the non-dispersive case. This similarity confirms our ability to observe the momentum gap experimentally. Additionally, four more quasi energy bandgaps [34] are observed: two originate from the inherent resonance at 4.7 kHz, and the other two arise due to band folding induced by time modulation, though these are not the focus here.

To observe the momentum gap shown in Figure 1, we experimentally construct an array of 9 meta-atoms in a 1D acoustic waveguide. The experimental set up is shown in Figure 2 (a). As described earlier, each meta-atom comprises a detector and a speaker (circled with yellow dashed box), the detector senses the acoustic pressure filed in the waveguide and feeds the signal to the microcontroller (circled in red dashed box), which performs the convolution Y(t)Y(t). The resulting signal is then sent to the speaker through an amplifier (circled with an orange dashed box) to generate a monopolar source. This process realizes an effective compressibility β\beta, as described by Eq. (5). Initially, we evaluate the static compressibility of the two phases to verify the values of βA\beta_A and βB\beta_B in the absence of time modulation. By scanning the incident frequency, we measure the transmission tt and reflection rr coefficient and extract the effective compressibility, as shown in Figure 2 (b) with red and blue dots being the real and imaginary part. The experimental results show excellent agreement with the analytic results in Eq. (5). Notably, the imaginary part of compressibility approaches 0 within the momentum gap region.

Image not extracted

Figure 2. (a) Experimental platform of phononic time crystal with lattice distance l=0.02ml = 0.02\mathrm{m}. (b) Static compressibility βA\beta_A (in solid) in phase A and βB\beta_B (in dash) in phase B with red (blue) dots representing real (imaginary) part. The parameters are chosen as f0=4.7kHzf_0 = 4.7\mathrm{kHz}, gA=100Hzg_A = 100\mathrm{Hz}, gB=−100Hzg_B = -100\mathrm{Hz}, γ=100Hz\gamma= 100\mathrm{Hz}. (c) Experimental and simulated transmittance and reflectance for phononic time crystal with modulated compressibility between A/B phases in (b). (d) Transmission amplitude with respect to phase delay between incident wave and modulation cycle.

After establishing the static configurations, we modulate between two phases and measure the transmittance T=∣t∣2T = |t|^2 and reflectance R=∣r∣2R = |r|^2 spectra, presented in red and blue dots in Figure 2 (c). In the gap region, the transmittance is larger than 1, with larger value around half of the modulation frequency, while the reflectance remains near 0. The experiment results show great agreement with the simulation results, obtained from full-wave simulation using COSMOL Multiphysics. The central peak at half the modulation frequency fm/2f_m/2 appears anomalous corresponding to the coupling between waves with positive and negative frequencies depended on the phase difference between the incident continuous wave and the modulation cycle. To investigate this effect, we vary the phase difference between the incident signal and the modulation cycle from 0 to 2π2\pi and remeasure the transmittance, plotted in Figure 2(d). The transmittance spectrum is largely phase insensitive at all frequencies except fm/2f_m/2. Accordingly, we plot the transmittance at fm/2f_m/2 as a function of phase delay in the inset of Figure 2(d), using bright blue dashed lines for experimental results and red lines for simulation. The transmittance goes through a cycle over 2π2\pi and a minimum occurs when the incident wave is out of phase with modulation cycle. Despite the phase variation, the transmittance consistently remains greater than 1, demonstrating the robustness of the amplifying mode within the momentum band gap. Crucially, this anomalous peak only emerges for continuous wave excitation, not sound pulses, due to the requirement for persistent phase coherence across multiple modulation periods.

In the previous section, we have experimentally demonstrated that the momentum band gap (k-gap) emerges in phononic time crystal when time modulating two static phases. In this platform, an advanced strategy involves using Lorentzian (loss) and anti-Lorentzian (gain) responses as the two phases in time modulation to enables a greater modulating depth, which results in stronger transmission and reflection within the k-gap. Furthermore, by balancing gain and loss through time domain, the effective acoustic impedance of the phononic time crystal can be engineered to approach that of air, thereby eliminating reflections at the spatial boundaries between air and the phononic time crystal. The detailed illustration is given in Note 3 of the Supplementary materials. In Figure 3(a), we present the experimental transmittance as a function of modulating depth Δa\Delta a and frequency ff. We observe that the transmittance increases from 1.7 to 13.4 as the modulating depth Δa=2Δgc0f02l\Delta a = \frac{2\Delta g c_0}{f_0^2 l} is raised from 4.8×10−24.8 \times10^{-2} to 10.79×10−210.79 \times10^{-2}, Δg\Delta g is the resonating strength difference of each meta-atom between phases A and B. Notably, two transmission dips located at 3.7 kHz and 4.7 kHz are preserved, corresponding to the fixed resonating frequency and its lower first-order harmonic. Similarly, the reflectivity peak becomes more pronounced with increasing modulating depth in the k-gap.

Image not extracted

Figure 3. Transmittance (a) and reflectance (b) spectra with increased modulation depth in experiment. (c) Schematic picture for radiating source in phononic time crystal. (d) Purcell factor as a function of incident frequency.

It is well known that the amplified emission within the k-gap is a unique property of Floquet time crystals [2, 11], whose eigenmodes grow temporally over time, independent of initial phase. To demonstrate the amplified emission in phononic time crystal, we place an acoustic source at the center of the phononic time crystal structure, as shown as Figure 3(c). To quantify the emission enhancement, we first deactivate all meta-atoms, and measure the outgoing sound pressure at both ends of the 1D empty waveguide, denoted as initial state: pip_i. Next, we activate all meta-atoms to generate a k-gap around 4.2 kHz and record the amplified sound pressure, defining the final state: pfp_f. To character the amplified capacity of the phononic time crystal, we define the Purcell factor as PF=∣pfpi∣PF = \left|\frac{p_f}{p_i}\right|, which is plotted in Figure 3(d) as a function of incident frequency. We find that the PF can exceed 10210^2 in magnitude for a modulating depth of Δa=10.79×10−2\Delta a = 10.79 \times10^{-2} under finite six modulating periods. In fact, the order of amplified emission can be further enhanced by increasing either modulating depth or modulating cycles, leading to laser-like emission behavior.

In spatial phononic crystals, it is well known that expanding the 1D monoatomic crystal to the diatomic crystal causes the ω\omega-band structure to fold into the first Brillouin zone and open new bandgaps [35, 36]. In the same spirit of route, we construct phononic time crystal with quasi-periodic time modulation (stack-ABAC), as illustrated in Figure 4(a). The static phases A, B, C correspond to nondispersive compressibilities βA=1.3\beta_\mathrm{A}=1.3, βB=0.9\beta_\mathrm{B}=0.9, βC=0.5\beta_\mathrm{C}=0.5, respectively. Each phase last for Tm/2T_m/2. To avoid the Fabry-Pérot resonance around 2.5 kHz in the meta-atom array, we select a modulation frequency of fm=6.6f_m=6.6 kHz as modulating frequency for this section. Thus, here the wave number correspond to the time modulation as k0=2πfmc0=120.8 rad/mk_0=\frac{2\pi f_m}{c_0}=120.8\ rad/m.

Image not extracted

Figure 4. (a) Conceptual picture of quasi-phononic time crystal with ABAC sequence modulation. (b) Static compressibilities for different configurations to approximate nondispersive βA=1.3\beta_A = 1.3, βB=0.9\beta_B = 0.9, βC=0.5\beta_C = 0.5 around gap region. (c) Band structure of quasi-phononic time crystal in (a) with time-varying nondispersive compressibilities βA=1.3\beta_A = 1.3, βB=0.9\beta_B = 0.9 and βC=0.5\beta_C = 0.5. (d) Band structure of dispersive quasi-phononic time crystal with static phases in (b). Experimental and simulated results of transmittance and reflectance spectra around the first (e) and second k-gap (f) region.

According to Eq. (4), the momentum band structure can be obtained as shown in Figure 4(c) for nondispersive phononic time crystal. Remarkably, two distinct kk-gaps emerge at k0/2k_0/2 (light pink region) and k0/4k_0/4 (light cyan region), where the real part (red line) vanishes and the imaginary part (blue cycle) becomes non-zero. To experimentally observe the double kk-gaps in quasi-periodical modulation, we design the system with two resonances in Y(f)Y(f) from Eq. (5), enabling an approximation of the desired nondispersive compressibility in both gap regions. The resulting dispersive compressibility is given by:

β(f)=1+a1f12f12−f2−2iγ1f+a2f22f22−f2−2iγ2f,(6)\beta(f)=1+\frac{a_1f_1^2}{f_1^2-f^2-2\mathrm{i}\gamma_1f}+\frac{a_2f_2^2}{f_2^2-f^2-2\mathrm{i}\gamma_2f}, \tag*{(6)}

where the resonating frequencies are f1=1.8f_1=1.8 kHz, f2=3.6f_2=3.6 kHz with decay rates γ1=50\gamma_1=50Hz, γ2=100\gamma_2=100Hz. Different resonating strengths can obtain different desired compressibility values. To realize βA\beta_\mathrm{A} at fm/4f_m/4 and fm/2f_m/2, we set a1=0.04a_1=0.04, a2=0.055a_2=0.055, as shown by the red line in Figure 4(b). The detailed values are highlighted by the stars in the insets. The yellow Line, representing static case B corresponds to a1=−0.013a_1=-0.013, a2=−0.0186a_2=-0.0186, while static case C require a1=−0.067a_1=-0.067, a2=−0.092a_2=-0.092. Figure 4(d) presents the momentum band structure of the dispersive phononic time crystal under quasi-periodic modulation. Comparing the band structures shown in Figure 4(c) and Figure 4(d), we observe that the dispersive phononic time crystal firmly preserves both kk-gaps in k0/4k_0/4 and k0/2k_0/2 with only a slight frequency shift. Due to the high flexibility of our time-varying metamaterials in tailoring resonating responses, we experimentally implement quasi-periodic modulation for dispersive media with the compressibility shown in Figure 4(b). Figure 4(e) shows the transmissivity and reflectivity in the first kk-gap around k0/4k_0/4. Red lines and symbols represent simulated and experimental transmissivity, while blue lines and symbols denote the reflectivity results. The experiment results confirm the opening of the kk-gap at k0/4k_{0}/4, as indicated by transmission spectrum greater than 1. A similar high transmission result is also observed at the second kk-gap near k0/2k_{0}/2, as shown in Figure 4(f). The stronger transmission in the second kk-gap arises because the finite spatial length L=0.18L = 0.18 m of phononic time crystal covers a broader range of wave numbers at higher operational frequencies, enhancing wave interaction. In brief, our versatile time-varying metamaterials successfully demonstrate the double kk-gaps formation through quasi-periodic time modulation.

Conclusion

In this work, we have experimentally demonstrated the emergence and control of momentum band gaps (kk-gaps) in phononic time crystals for airborne sound through time-periodic and quasi-periodic modulation of compressibility. Using acoustic meta-atoms with time-varying Lorentzian responses, we observed significant wave amplification within the kk-gap. Building on this, we introduced quasi-periodic time modulation using three compressibility phases to realize double kk-gaps at fractional modulation frequencies k0/4k_{0}/4 and k0/2k_{0}/2, supported by transmission spectral measurements. The demonstrated ability to engineer multiple, tunable kk-gaps and realize laser-analog wave amplification opens new pathways for developing high-gain acoustic amplifiers, filters, and coherent sources. Our findings highlight the remarkable flexibility of time-varying acoustic metamaterials for tailoring phononic time crystals and can be extended to other Floquet metamaterials such as optics, electromagnetics, and elastic waves, promising innovative applications in wave-based technologies.

This work was supported by Scientific research projects of colleges and universities in Anhui Province (Grants No. 2022AH040114); National Key Research and Development Program of China (2022YFA1404303, 2020YFA0211300); National Natural Science Foundation of China (12234010); J L acknowledge support from the EPSRC via the META4D Programme Grant (No. EP/Y015673/1).

Direct Observation of k-Gaps in Dynamically Modulated Phononic Time Crystal

Ziling Liu; Xinghong Zhu; Zhi-Guo Zhang; Wei-Min Zhang; Xue Chen; Yong-Qiang Yang; Ru-Wen Peng; Mu Wang; Jensen Li; Hong-Wei Wu

Band structure for time-varying dispersive media with a Lorentzian-type resonance

When considering the dispersion of the medium, we can write the acoustic wave equation as:

∂xp(x,t)+ρ0∂tv(x,t)=0,(7)\partial_x p(x,t) + \rho_0\partial_t v(x,t) = 0, \tag*{(7)}
∂xv(x,t)+β0∂t(p(x,t)+M(x,t))=0,(8)\partial_x v(x,t) + \beta_0\partial_t \left(p(x,t) + M(x,t)\right) = 0, \tag*{(8)}

with M(x,t)M(x,t) being the monopolar polarization and governed by a Lorentzian-type model

∂t2M(x,t)+2γ∂tM(x,t)+ω02M(x,t)=a(t)ω02p(x,t),(9)\partial_t^2 M(x,t) + 2\gamma\partial_t M(x,t) + \omega_0^2 M(x,t) = a(t)\omega_0^2 p(x,t), \tag*{(9)}

Combining S1-S3 and replacing ∂x\partial_x with ikik, we can construct a 4 by 4 eigenvalue problem as

i∂tψ=ω^ψ,ω^=(0kβ00ikρ0000000−i−ia(t)ω020iω02−2iγ)(10)i\partial_t\psi=\widehat{\omega}\psi,\qquad\widehat{\omega}= \begin{pmatrix} 0 & \frac{k}{\beta_0} & 0 & i \\ \frac{k}{\rho_0} & 0 & 0 & 0 \\ 0 & 0 & 0 & -i \\ -ia(t)\omega_0^2 & 0 & i\omega_0^2 & -2i\gamma \end{pmatrix} \tag*{(10)}

where the state vector is ψ=(p,v,M,−∂tM)T\psi=(p,v,M,-\partial_t M)^T. The propagation matrix ω^\widehat{\omega} switches between the two phases ω^A\widehat{\omega}_A and ω^B\widehat{\omega}_B, corresponding to the resonant strengths aAa_A and aBa_B in the two phases. Then, using the Floquet mode theory, we can solve the band structure plotted in Figure 1 (e) in the main text by solving the secular equation

det⁡[e−iΩTmI4−e−iω^BTmBe−iω^ATmA]=0,(11)\det\left[e^{-i\Omega T_m}I_4-e^{-i\widehat{\omega}_B T_m B}e^{-i\widehat{\omega}_A T_m A}\right]=0, \tag*{(11)}

where I4I_4 is the 4 by 4 identity matrix.

Band structure for quasi-periodic dispersive media with two Lorentzian-type resonances

Following the same spirit of route, when we have two monopolar resonances in each phase, we only need to add another response function into the wave equation by rewriting S2 and S3 as

∂xv(x,t)+β0∂t(p(x,t)+Mi(x,t))=0,(12)\partial_xv(x,t)+\beta_0\partial_t\left(p(x,t)+M_i(x,t)\right)=0, \tag*{(12)}
∂t2Mi(x,t)+2γi∂tMi(x,t)+ω0i2M(x,t)=ai(t)ω0i2p(x,t),(13)\partial_t^2M_i(x,t)+2\gamma_i\partial_tM_i(x,t)+\omega_{0i}^2M(x,t)=a_i(t)\omega_{0i}^2p(x,t), \tag*{(13)}

with i=1,2i=1,2, representing the first and second resonance mode. Combined with S1, we can construct a 6 by 6 eigenvalue problem as

i∂tψ=ω^ψ,ω^=(0kβ00i0ikρ000000000−i00−ia1(t)ω0120iω012−2iγ10000000−i−ia2(t)ω022000iω022−2iγ2)(14)i\partial_t\psi=\widehat{\omega}\psi,\qquad\widehat{\omega}= \begin{pmatrix} 0 & \frac{k}{\beta_0} & 0 & i & 0 & i \\ \frac{k}{\rho_0} & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & -i & 0 & 0 \\ -ia_1(t)\omega_{01}^2 & 0 & i\omega_{01}^2 & -2i\gamma_1 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & -i \\ -ia_2(t)\omega_{02}^2 & 0 & 0 & 0 & i\omega_{02}^2 & -2i\gamma_2 \end{pmatrix} \tag*{(14)}

with ψ=(p,v,M1,−∂tM1,M2,−∂tM2)T\psi=(p,v,M_1,-\partial_tM_1,M_2,-\partial_tM_2)^T. Since now the modulation period becomes 2Tm2T_m, and there are 3 different phases in one modulation period: A, B and C, corresponding to resonant strength pairs (a1A,a2A)(a_{1A}, a_{2A}), (a1B,a2B)(a_{1B}, a_{2B}) and (a1C,a2C)(a_{1C}, a_{2C}). The modulation cycle follows the ABAC sequence, with each phase occupying Tm/2T_m/2 giving rise to the Floquet modes: ψ(t+2Tm)=e−iω^CTm/2e−iω^ATm/2e−iω^BTm/2e−iω^ATm/2ψ(t)\psi(t + 2T_m) = e^{-i\hat{\omega}_C T_m/2}e^{-i\hat{\omega}_A T_m/2}e^{-i\hat{\omega}_B T_m/2}e^{-i\hat{\omega}_A T_m/2}\psi(t). The band struciioooo9ikl uture can be obtained by solving the secular equation:

det⁡[e−iΩTmI6−e−iω^CTm/2e−iω^ATm/2e−iω^BTm/2e−iω^ATm/2]=0.(15)\det\left[e^{-i\Omega T_m}I_6-e^{-i\hat{\omega}_C T_m/2}e^{-i\hat{\omega}_A T_m/2}e^{-i\hat{\omega}_B T_m/2}e^{-i\hat{\omega}_A T_m/2}\right]=0. \tag*{(15)}

Fig 4(d) in the main text plotted the band structure of phononic time crystal according to S9 with modulation cycle following the ABAC sequence.

Temporal effective medium for nondispersive and dispersive phononic time crystal

In main text, the nondispersive phononic time crystal is experimentally realized by dispersive time-varying metamaterial with discrete resonating meta-atoms. However, their k-gaps matching each other is limited by the low modulating depth. In fact, the temporal effective medium of nondispersive and dispersive phononic time crystal has to abide by different average methods [1]. For dispersive case, the effective compressibility can be expressed as:

βeff=ξβA+(1−ξ)βB,(16)\beta_{\mathrm{eff}}=\xi\beta_A+(1-\xi)\beta_B, \tag*{(16)}

while the nondispersive case has the effective compressibility by

1βeff=ξβA+(1−ξ)βB,(17)\frac{1}{\beta_{\mathrm{eff}}}=\frac{\xi}{\beta_A}+\frac{(1-\xi)}{\beta_B}, \tag*{(17)}

where βA\beta_A and βB\beta_B are the static compressibilities of material A and B, ξ\xi is the duty cycle. Thus, the k-gap center kck_c of phononic time crystal vary with different temporal effective medium, which can be expressed as:

kc=k02βeff(18)k_c = \frac{k_0}{2}\sqrt{\beta_{\mathrm{eff}}} \tag*{(18)}

where wave number k0k_0 correspond to the time modulation defined as: km=2π/(c0Tm)k_m = 2\pi/(c_0T_m), where c0c_0 is the speed of sound in air. It means that the k-gap center will move with changing the effective compressibility. Thus, the k-gap center of the phononic time crystal separates for nondispersive and dispersive case due to different temporal average rules. For the dispersive phononic time crystal, balancing gain and loss through time domain, the effective compressibility always equal to 1, then kc=k02k_c = \frac{k_0}{2} in Figure 1(e) and kc=k02k_c = \frac{k_0}{2} and k04\frac{k_0}{4} in Figure 4(d). Furthermore, the effective acoustic impedance of the phononic time crystal can be engineered to approach that of air, thereby eliminating reflections at the spatial boundaries between air and the phononic time crystal. However, for the nondispersive case, modulating between βA=0.88\beta_A = 0.88 and βB=1.12\beta_B = 1.12 with duty cycle ξ=0.5\xi= 0.5, then the kc=k02∗0.993k_c = \frac{k_0}{2} * 0.993, thus the k-gap center slightly left shift in Figure 1(b), this effect is more obvious for greater modulating depth in Figure 4(c).

Stability analysis for phononic time crystals in the experiment

Since our meta-atoms involves feedback and time-varying resonant strength, the whole system (including all the atoms) is time dependent and frequency conversion occurs, we need stability analysis to make sure that the system is working in the stable regime. In this section, we develop a multiband harmonic model to determine the eigenmode (stability) of the time-dependent system. As the modulation strength and resonating frequency is modulated periodically, all the signals (from the detector and speaker) can be written in the Fourier series expansion with a set of harmonics fn=f+n/Tmf_n = f + n/T_m, where nn is an integer and $1/T_m$ is the modulation frequency. The detector and speaker signal can be expanded as p(xi,t)=∑npine−ifntp(x_i,t) = \sum_n p_{in}e^{-if_nt}, si(t)=∑nsine−ifnts_i(t) = \sum_n s_{in}e^{-if_nt} where $i represents the ii-th atom. The modulation of resonating strength and resonating frequency can be written as g(t)=∑ngne−int/Tmg(t)=\sum_n g_n e^{-int/T_m}, where f0=4.7 kHzf_0=4.7\,\mathrm{kHz}. Then, the response function Y can be written in terms of Fourier component (labeled by n harmonics): Ynn′i=ifngn−n′f02−fn2−2iγfnY^i_{nn'}=\frac{if_ng_{n-n'}}{f_0^2-f_n^2-2i\gamma f_n} and it connects the detector and speaker signal by sin=Ynn′ipin′s_{in}=Y^i_{nn'}p_{in'}. Since the convolution generates the secondary radiation and propagates in waveguide and then is detected again, there is a loop transfer function from detector DiD_i to DjD_j by T{jn},{in′}(ω)=eik0(fn)∣xj−xi∣Ynn′i=eik0(fn)∣xj−xi∣ifngn−n′f02−fn2−2iγfn\mathcal{T}_{\{jn\},\{in'\}}(\omega)=e^{ik_0(f_n)|x_j-x_i|}Y^i_{nn'}=e^{ik_0(f_n)|x_j-x_i|}\frac{if_ng_{n-n'}}{f_0^2-f_n^2-2i\gamma f_n} where ∣xj−xi∣|x_j-x_i| is the distance from speaker ii to detector jj (assuming DiD_i and SiS_i are in the same position since they are collocated). The T{jn},{in′}\mathcal{T}_{\{jn\},\{in'\}} is a square matrix and a stable system means all the poles the of the system response matrix, i.e. zeros of Det⁡(I−T{jn},{in′})\operatorname{Det}(I-\mathcal{T}_{\{jn\},\{in'\}}) have to lie in the lower complex plane. The number of harmonics (a total of 7 in the following analysis) is taken to be enough to have converging numerical results. For the time-varying acoustic system in Figure 2 (modulating resonant strength gg between gA=100 Hzg_A=100\,\mathrm{Hz} and gB=−100 Hzg_B=-100\,\mathrm{Hz}, Δg=200 Hz\Delta g=200\,\mathrm{Hz}), we plot the Det⁡(I−T{jn},{in′})\operatorname{Det}(I-\mathcal{T}_{\{jn\},\{in'\}}) in the complex frequency plane, as shown in Figure S1. The minimum value of Det⁡(I−T{jn},{in′})\operatorname{Det}(I-\mathcal{T}_{\{jn\},\{in'\}}) in the blue dot indicates the pole of the system, around the frequency of (4690−i140) Hz(4690-\mathrm{i}140)\,\mathrm{Hz}, in the lower half plane. This means the system is working in the stable regime. When we increase the modulation depth, the system poles will gradually touch the real frequency axes and make the system become unstable. Figure S1 plots the pole diagram where one of the poles just touch the real frequency axes, corresponding to the maximum modulation depth Δg=14.56∗200 Hz\Delta g=14.56*200\,\mathrm{Hz} (Δa=31.05×10−2\Delta a=31.05\times10^{-2}). However, in the experiment, we only obtain the threshold at Δa=10.79×10−2\Delta a=10.79\times10^{-2} due to the intrinsic response of speaker and time delay of microcontroller making the system easier to become unstable.

Image not extracted

(a)

Image not extracted

(b)

Figure S1. (a) Pole diagram of time-varying meta-atoms in Fig.2 (b) Pole diagram of time-varying meta-atoms with threshold touching the real axes.

References

  1. [1]Else, D. V., Bauer, B. & Nayak, C. Floquet Time Crystals. Phys. Rev. Lett. 117, 090402, (2016).DOI
  2. [2]Lyubarov, M. et al. Amplified emission and lasing in photonic time crystals. Science 377, 425-428, (2022).DOI
  3. [3]Emanuele, G. et al. Photonics of time-varying media. Adv. Photonics 4, 014002, (2022).DOI
  4. [4]Yin, S., Galiffi, E. & Alù, A. Floquet metamaterials. eLight 2, 8, (2022).DOI
  5. [5]Asgari, M. M. et al. Theory and applications of photonic time crystals: a tutorial. Adv. Opt. Photon. 16, 958, (2024).DOI
  6. [6]Yang, Y. et al. Topologically Protected Edge States in Time Photonic Crystals with Chiral Symmetry. ACS Photonics, (2025).arxiv.org/abs/2501.08546
  7. [7]Oudich, M., Deng, Y., Tao, M. & Jing, Y. Space-time phononic crystals with anomalous topological edge states. Phys. Rev. Res. 1, 033069, (2019).
  8. [8]Lustig, E., Sharabi, Y. & Segev, M. Topological aspects of photonic time crystals. Optica 5, 1390-1395, (2018).
  9. [9]Wang, B. et al. Observation of Photonic Topological Floquet Time Crystals. Laser Photonics Rev. 16, 2100469, (2022).DOI
  10. [10]Sharabi, Y., Lustig, E. & Segev, M. Disordered Photonic Time Crystals. Phys. Rev. Lett. 126, 163902, (2021).DOI
  11. [11]Dikopoltsev, A. et al. Light emission by free electrons in photonic time-crystals. Proc. Natl. Acad. Sci. U. S. A. 119, e2119705119, (2022).DOI
  12. [12]Pan, Y., Cohen, M.-I. & Segev, M. Superluminal k-Gap Solitons in Nonlinear Photonic Time Crystals. Phys. Rev. Lett. 130, 233801, (2023).DOI
  13. [13]Reyes-Ayona, J. R. & Halevi, P. Observation of genuine wave vector (k or β) gap in a dynamic transmission line and temporal photonic crystals. Appl. Phys. Lett. 107, 074101, (2015).DOI
  14. [14]Park, J. et al. Revealing non-Hermitian band structure of photonic Floquet media. Sci. Adv. 8, eabo6220, (2022).
  15. [15]Wang, X. et al. Metasurface-based realization of photonic time crystals. Sci. Adv. 9, eadg7541, (2023).DOI
  16. [16]Alam, M. Z., De Leon, I. & Boyd, R. W. Large optical nonlinearity of indium tin oxide in its epsilon-near-zero region. Science 352, 795-797, (2016).DOI
  17. [17]Bohn, J., Luk, T. S., Horsley, S. & Hendry, E. Spatiotemporal refraction of light in an epsilon-near-zero indium tin oxide layer: frequency shifting effects arising from interfaces. Optica 8, 1532-1537, (2021).DOI
  18. [18]Zhou, Y., Alam, M.Z., Karimi, M. et al. Broadband frequency translation through time refraction in an epsilon-near-zero material. Nat Commun 11, 2180 (2020).DOI
  19. [19]Caspani, L. et al. Enhanced Nonlinear Refractive Index in ε-Near-Zero Materials. Phys. Rev. Lett. 116, 233901, (2016).DOI
  20. [20]Tirole, R., Vezzoli, S., Galiffi, E. et al. Double-slit time diffraction at optical frequencies. Nat. Phys. 19, 999–1002 (2023).DOI
  21. [21]Lustig, E. et al. Time-refraction optics with single cycle modulation. Nanophotonics 12, 2221-2230, (2023).DOI
  22. [22]Wang, X., Garg, P., Mirmoosa, M.S. et al. Expanding momentum bandgaps in photonic time crystals through resonances. Nat. Photon. 19, 149–155 (2025).DOI
  23. [23]Wang, Y. et al. Observation of Nonreciprocal Wave Propagation in a Dynamic Phononic Lattice. Phys. Rev. Lett. 121, 194301, (2018).DOI
  24. [24]Trainiti, G. et al. Time-Periodic Stiffness Modulation in Elastic Metamaterials for Selective Wave Filtering: Theory and Experiment. Phys. Rev. Lett. 122, 124301, (2019).DOI
  25. [25]Kim, B. L., Chong, C., Hajarolasvadi, S., Wang, Y. & Daraio, C. Dynamics of time-modulated, nonlinear phononic lattices. Phys. Rev. E 107, 034211, (2023).DOI
  26. [26]Appfel, B., Wildeman, S., Eddi, A. & Fort, E. Experimental Implementation of Wave Propagation in Disordered Time-Varying Media. Phys. Rev. Lett. 128, 094503, (2022).
  27. [27]Cheng, Z. et al. Observation of π/2 Modes in an Acoustic Floquet System. Phys. Rev. Lett. 129, 254301, (2022).DOI
  28. [28]Shen, C., Zhu, X., Li, J. & Cummer, S. A. Nonreciprocal acoustic transmission in space-time modulated coupled resonators. Phys. Rev. B 100, 054302, (2019).DOI
  29. [29]Chen, Z. et al. Efficient nonreciprocal mode transitions in spatiotemporally modulated acoustic metamaterials. Sci. Adv. 7, eabj1198, (2021).DOI
  30. [30]Cho, C., Wen, X., Park, N. & Li, J. Digitally virtualized atoms for acoustic metamaterials. Nat Commun. 11, 251, (2020).
  31. [31]Wen, X., Zhu, X., Wu, H. W. & Li, J. Realizing spatiotemporal effective media for acoustic metamaterials. Phys. Rev. B 104, L060304, (2021).DOI
  32. [32]Zhu, X., Wu, H.-W., Zhuo, Y., Liu, Z. & Li, J. Effective medium for time-varying frequency-dispersive acoustic metamaterials. Phys. Rev. B 108, 104303, (2023).DOI
  33. [33]Wen, X. et al. Unidirectional amplification with acoustic non-Hermitian space–time varying metamaterial. Commun. Phys. 5, 18, (2022).DOI
  34. [34]Feng, F., Wang, N. & Wang, G. P. Temporal transfer matrix method for Lorentzian dispersive time-varying media. Appl. Phys. Lett. 124, 101701, (2024).DOI
  35. [35]Zhao, D., Xiao, M., Ling, C. W., Chan, C. T. & Fung, K. H. Topological interface modes in local resonant acoustic systems. Phys. Rev. B 98, 014110, (2018).DOI
  36. [36]Deymier, P. A. Acoustic Metamaterials and Phononic Crystals Vol. 173 (Springer Science & Business Media, 2013).

Paper details

vlnc:W_x8nhkd4h
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 11

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents