Photonics Research, Volume. 11, Issue 7, 1203(2023)

High Q and sub-wavelength THz electric field confinement in ultrastrongly coupled THz resonators Editors' Pick

Simon Messelot, Solen Coeymans, Jérôme Tignon, Sukhdeep Dhillon, and Juliette Mangeney*
Author Affiliations
  • Laboratoire de Physique de l’Ecole Normale Supérieure, Ecole normale supérieure, PSL University, Sorbonne Université, Université Paris Diderot, Sorbonne Paris Cité, CNRS, 24 rue Lhomond, 75005 Paris, France
  • show less

    The control of light–matter coupling at the single electron level is currently a subject of growing interest for the development of novel quantum devices and for studies and applications of quantum electrodynamics. In the terahertz (THz) spectral range, this raises the particular and difficult challenge of building electromagnetic resonators that can conciliate low mode volume and high quality factor. Here, we report on hybrid THz cavities based on ultrastrong coupling between a Tamm cavity and an LC circuit metamaterial and show that they can combine high quality factors of up to Q=37 with a deep-subwavelength mode volume of V=3.2×10-4λ3. Our theoretical and experimental analysis of the coupled mode properties reveals that, in general, the ultrastrong coupling between a metamaterial and a Fabry–Perot cavity is an effective tool to almost completely suppress radiative losses and, thus, ultimately limit the total losses to the losses in the metallic layer. These Tamm cavity-LC metamaterial coupled resonators open a route toward the development of single photon THz emitters and detectors and to the exploration of ultrastrong THz light–matter coupling with a high degree of coherence in the few to single electron limit.

    1. INTRODUCTION

    The control of light–matter coupling by embedding matter into a photonic resonator is an essential ingredient for the development of single photon devices and cavity quantum electrodynamic studies and applications. The two parameters of photonic resonators that play a key role in light–matter coupling are the quality factor Q, which is inversely proportional to the photon decay rate of the cavity, and the mode volume V, which quantifies the confinement of the electric field. In many cases, photonic resonators associating strong electric field confinement and a high Q are highly desired. For instance, in the weak-coupling regime, the intensity of light–matter interaction is enhanced by the ratio Q/V, as described by Purcell effect [1]. Also, as the light–matter coupling constant g scales with Ne/V, with Ne the number of emitters in the matter part collectively involved in the interaction [24], small V are needed to reach the ultrastrong and deep-strong coupling regimes at the single electron level [59]. Moreover, the in-depth study of the ultrastrong coupling regime requires a high degree of coherence of the light–matter coupling [1014], quantified by the cooperativity C=4g2Q/(ω0γ), with γ being the matter non-radiative decay rate and ω0 being the photonic mode angular frequency, which implies a high Q. Therefore, a current challenge in the development of the light–matter coupling platform is to build photonic resonators that conciliate a small V with a high Q.

    The ultrastrong coupling regime in high cooperativity systems has been recently studied using cyclotron resonances in a high-mobility two-dimensional electron gas coupled to various types of terahertz (THz) photonic resonators [15]. For example, using a high-Q THz Fabry–Perot cavity made of distributed Bragg reflectors (DBRs; Q>180), Zhang et al. reported g/ω00.1 with C>300 and showed that these unique conditions increase cyclotron resonance lifetime via the suppression of superradiant decay [16]. More recently, in a similar THz Fabry–Perot cavity, Li et al. reported g/ω00.36 with a record-high cooperativity C=3513, allowing the observation of a vacuum Bloch–Siegert shift in ultra-narrow Landau polaritons [17]. Also, Mavrona et al. demonstrated a cooperativity of C=57 and g/ω00.175 using a Fabry–Perot cavity based on weakly transmitting hole patterned metal layer [18].

    However, in these works based on high-Q Fabry–Perot cavities, the light–matter coupling involves a very large number of electrons Ne due to the large V, limited by diffraction to the order of (λ2)3. A current challenge to explore advanced quantum phenomena such as the fermionic Rabi model for coupled systems is to reduce the number of electrons involved in the collective interaction while keeping a high coherence in the system. To this end, using a single complementary LC circuit THz resonator, Rajabali et al. recently achieved g/ω00.33 with C=94 implying an estimated optically active electron number of only 30,000, and even of 2000 with C=14 [19]. Such a small number of electrons are obtained thanks to the confinement in truly all space dimensions of the electromagnetic mode, which leads to a deep sub-wavelength mode volume V, the cooperativity remaining nevertheless moderate due to the low Q factor typical of LC circuit resonators (Q=11).

    Potential candidates that have recently emerged to simultaneously achieve a high Q and a strong electric field confinement are hybrid THz resonators based on Fabry–Perot cavities strongly coupled to LC circuit metamaterials. In these hybrid resonators, the strength of the coupling is given by the resonator-to-resonator coupling constant G defined by analogy to the light–matter coupling constant g. The ultrastrong coupling between a THz LC circuit resonator and a THz Fabry–Perot cavity was first reported by Meng et al. [20]. The authors demonstrated nonlocal collective interaction of spatially separated metamaterial layers mediated by the cavity photons [20] and also the potential of these hybrid THz resonators for phase-sensitive THz detection of chemical and biological materials [21]. More recently, they have demonstrated ultrastrong coupling between the effective magnetic dipole moments of metallic complementary metasurfaces and the magnetic field of a cavity mode [22]. Also, Jeanin et al. reported an hybrid THz cavity based on the coupling between LC circuit resonators and a dipolar antenna to engineer the radiative coupling properties and reach unity absorption [23].

    Here, we report on hybrid resonators based on the coupling between a THz Tamm cavity and a THz LC circuit metamaterial. The LC circuit resonators are directly patterned on the metallic mirror of the Tamm cavity and are designed to concentrate the THz electric field. Using transmittance measurements and temporal coupled mode theory, we analyze the properties of the coupled modes and demonstrate the reduced coupling constant G/ω00.1, showing that the two resonators are in the ultrastrong coupling regime. We further describe how the properties of the coupled modes emerge from the properties of the uncoupled metamaterial and the Fabry–Perot resonators. We show high quality factors, Q=37, explained by the quasi-total suppression of the radiative losses provided by the Tamm cavity and small V, in the (23)×104λ3 range, resulting from the principal quality of LC metamaterials. Our study demonstrates that, by combining high Q and small V, Tamm cavity-LC metamaterial coupled cavities fulfill the challenging requirement to conciliate strong THz electric field confinement with high quality factor, which is promising for the exploration of ultrastrong THz light–matter coupling with a high degree of coherence in the limit of a few electrons and for the realization of single photon THz emitters and detectors.

    2. STRONG COUPLING BETWEEN TAMM CAVITIES AND LC METAMATERIALS

    Our dual-cavity THz resonator is based on a hybrid THz Tamm cavity and an LC metamaterial structure and is presented in Fig. 1(a). The Tamm cavity [24,25], on its own, consists of a DBR composed of two high-resistivity silicon wafers of thickness 66 μm (3λ4) separated by a vacuum layer of thickness 75 μm (λ4) and a 100-nm-thick metallic gold mirror deposited on the top silicon wafer of the DBR [26]. The calculated reflection spectrum of this Tamm cavity, obtained by the finite element method (FEM, COMSOL Multiphysics), exhibits a Tamm mode at 0.98 THz and a resonance linewidth of ΓTamm=0.098  THz, which gives a quality factor Q=100 [see Fig. 1(b)]. The isolated LC metamaterial, whose base unit cell is presented in Fig. 1(a), is basically a gold layer including patterned holes giving its resonance properties. This pattern should not be mistaken for a complementary split ring resonator that is efficient for concentrating the magnetic field. The model we present concentrates the electric field at its center, similar to conventional split ring resonator, which is crucial for coupling it to matter systems involving the electric dipole interaction, such as in the common two-dimensional electron gas. For this reason, we call this pattern grid split ring resonators (GSRRs). The calculated transmission spectrum of an isolated GSRR metamaterial of dimensions w=4.4  μm, l=4.4  μm, and L=16.5  μm, shows an LC metamaterial resonant mode at 1 THz with a linewidth ΓLC=0.222  THz, resulting in a quality factor Q=4.5 [see Fig. 1(b)]. The hybrid THz resonator structures were fabricated by manually stacking commercially available thin silicon wafers and precision machined thin metal strips opening vacuum gaps, creating the THz DBR. The stacking is realized in a custom-made sample holder dedicated to maintaining the metal spacers positions and holding the stack together by applying pressure. Prior to stacking, the LC metamaterials were patterned on the top silicon wafer by performing a simple optical lithography and the deposition of a 3/100 nm chromium/gold metallic mirror via thermal evaporation under vacuum, followed by a lift-off.

    (a) Representation of the Tamm cavity/LC metamaterial coupled resonators structure and unit cell pattern of the LC metamaterial. (b) Reflection and transmission spectra of the uncoupled Tamm cavity and LC metamaterial, respectively, alongside their representations (inset).

    Figure 1.(a) Representation of the Tamm cavity/LC metamaterial coupled resonators structure and unit cell pattern of the LC metamaterial. (b) Reflection and transmission spectra of the uncoupled Tamm cavity and LC metamaterial, respectively, alongside their representations (inset).

    We fabricated nine different Tamm cavity-LC metamaterial coupled resonators, tuning the uncoupled resonance frequencies of the LC metamaterial fLC by applying an homothetic transformation to the w, l, and L dimensions of the LC circuit resonator at a fixed metamaterial period p=65  μm. We probe their optical properties in a transmission experiment using a Bruker Vertex 70v FTIR under vacuum, a Globar blackbody source, and a helium-cooled bolometer detector. Figure 2(a) reports the measured transmission spectra of the Tamm cavity-LC metamaterial coupled resonators for different fLC (represented by the colour dots). We observe for each transmission curve two well-separated resonance peaks that form two distinct lower and upper frequency coupled modes. These modes are the result of the coupling between the Tamm cavity and the LC metamaterial. Around the frequency matching between the Tamm cavity mode and the LC metamaterial mode (dark green curve), the two peaks are approximately symmetric and separated by a splitting of 0.20 THz. Figure 2(b) shows the frequencies of the two measured resonance peaks corresponding to the frequency of the upper (red squares) and lower (blue squares) coupled modes as a function of fLC. We clearly observe the two distinct branches forming the anti-crossing pattern, similar to the behavior observed in the strong light–matter coupling regime, demonstrating the strong coupling between the Tamm cavity and the LC metamaterial.

    (a) Transmission spectra of the Tamm cavity-LC metamaterial coupled resonators for decreasing (bottom to top) LC resonance frequencies fLC (indicated by color circles; see legend box). The Tamm cavity resonance is fixed, about 0.96 THz (dotted vertical line). The curves are offset for clarity. (b) Resonance frequencies of the upper (red squares) and lower coupled mode (blue squares) as a function of the uncoupled LC resonance frequency fLC (diagonal dotted line). Horizontal dotted line, uncoupled Tamm cavity resonance frequency.

    Figure 2.(a) Transmission spectra of the Tamm cavity-LC metamaterial coupled resonators for decreasing (bottom to top) LC resonance frequencies fLC (indicated by color circles; see legend box). The Tamm cavity resonance is fixed, about 0.96 THz (dotted vertical line). The curves are offset for clarity. (b) Resonance frequencies of the upper (red squares) and lower coupled mode (blue squares) as a function of the uncoupled LC resonance frequency fLC (diagonal dotted line). Horizontal dotted line, uncoupled Tamm cavity resonance frequency.

    To quantitatively describe the experimentally observed trends, we use a quantum formalism similar to the Rabi Hamiltonian, the Hopfield model [27], which involves bosonic operators only. In a general picture involving two resonators, creation and annihilation operators a^, a^ and b^, b^ are associated with the photonic modes A and B. Here, A is the Tamm cavity mode, and B is the LC metamaterial mode. The Hamiltonian of the coupled resonators can then be expressed as H^=ωA(a^a^+12)+ωB(b^b^+12)+G(a^b^+a^b^a^b^a^b^),where G is defined here as an energy exchange rate between the two harmonic oscillators. The uncoupled eigenstates are written as |NA,NB with N, N the numbers of photons in modes A and B, respectively. Under the rotating wave approximation and considering for simplicity the subspace with one photon in the system, the eigenstates are two coupled photonic modes |+1 and |1 that are linear combinations of the uncoupled eigenstates |1A,0B and |0A,1B, |+1=cosθ|1A,0B+sinθ|0A,1B,|1=sinθ|1A,0Bcosθ|0A,1B,where tan(2θ)=2Gδ. θ varies from 0 to π2 with the resonator detuning δ=ωBωA describing the continuous transition from |1A,0B to |0A,1B, θ=π4 at zero detuning. The lower (−) and upper (+) resonance angular frequencies read ω±=ωA+ωB2±12(ωAωB)2+4G2.

    Note that the significant reduction of the resonator coupling constant G when the LC pattern size is reduced (i.e., fLC increases) is accounted for using an empirical linear model G=G0(1β(ωBωA)); see Appendix B for details and arguments for this description. We find a good quantitative agreement between experiment and theory [Eq. (3)] for G0.10  THz at frequency matching [solid black lines in Fig. 2(b)]. From this analysis, we extract a reduced coupling constant G/ω00.1 demonstrating the ultrastrong coupling regime between the Tamm cavity and the LC metamaterial.

    3. TEMPORAL COUPLED MODE THEORY FOR THE QUALITY FACTOR OF STRONGLY COUPLED RESONATORS

    The experimental transmission spectra also provide a measure of the quality factors of the hybrid Tamm cavity-LC metamaterial resonators that we investigate in this section. We report on Fig. 3(a) the evolution of the linewidth of the two resonance peaks Γ± as a function of fLC. The two curves exhibit a crossing at zero detuning (corresponding to fLC=0.98  THz) resulting from the exchange of properties between the Tamm cavity and the LC metamaterial in the coupled modes. At the crossing, we measure Γ+=0.061  THz and Γ=0.064  THz yielding the quality factors Q+=18 and Q=14 for the upper and lower coupled modes. Such values are surprisingly high, considering standard theory of coupled modes [28], which predicts linewidths Γ±=ΓTamm+ΓLC20.11  THz resulting in Q±,CM9 (i.e., Q±,CM2QLC since ΓTammΓLC).

    (a) Resonance peak linewidth of the two coupled modes from Lorentzian fit on data from Fig. 2(a). Dashed lines are guide for the eyes. (b) Schematic picture of the Tamm cavity and LC metamaterial directly on top, including radiative channels s1 and s2. Bottom, corresponding interaction scheme between the Tamm mode A and the LC metamaterial mode B including the relevant coupling rates.

    Figure 3.(a) Resonance peak linewidth of the two coupled modes from Lorentzian fit on data from Fig. 2(a). Dashed lines are guide for the eyes. (b) Schematic picture of the Tamm cavity and LC metamaterial directly on top, including radiative channels s1 and s2. Bottom, corresponding interaction scheme between the Tamm mode A and the LC metamaterial mode B including the relevant coupling rates.

    To understand the physical mechanism underlying these higher than expected Q values of the coupled modes, we use the temporal coupled mode theory [23,29] and analyze the different decay and coupling rates involved in the system. The complete study, which includes the full derivation, the expression of the S-parameters, and the expression of the resonator coupling constant G, is presented in Appendix A. We consider a set of two coupled planar resonators A and B in a series, exchanging energy radiatively with an input channel s1 (excited by a wave s+,1) and an output channel s2, respectively, as sketched in Fig. 3(b). The coupled equations of motion are {dAdt=(iωAΓA2)A+iCBCAGB+Γrad1,ACAs+,1dBdt=(iωBΓB2)B+iCACBGA,where A and B are the complex field amplitude of mode A and B, respectively [30], G is the resonator coupling constant, CA  (CB) defines the relation between the complex field amplitudes A(B) and the mode energy in the resonator WA=CA|A|2 (WB=CB|B|2), and ΓA (ΓB) is the amplitude decay rate of mode A (B) with ΓA=Γrad1,A+Γloss,A (ΓB=Γrad2,B+Γloss,B).

    Using this model, we derive the expressions of the reflection and transmission coefficients of the coupled resonator system (see Appendix A) and determine the analytical expression of the resonance linewidth of the upper and lower coupled modes, Γ±, in the ultrastrong coupling regime. Considering the Tamm cavity as A and the LC metamaterial as B, Γ± reads at zero detuning, Γ±=Γrad1,Tamm+Γrad2,LC+Γloss,Tamm+Γloss,LC2.

    In this equation, Γrad1,LC and Γrad2,Tamm vanish as the energy emitted by the LC metamaterial in the left direction is absorbed by the Tamm cavity and reciprocally [see Fig. 3(b)]. This coupling mechanism is responsible for the narrow resonance linewidths of the coupled modes observed experimentally in Fig. 3(a) compared to those predicted by the standard theory of coupled modes, which explains the large values of Q and Q+. It also highlights the strong link between the coupling strength and the radiative coupling properties of the LC metamaterial, which is the origin of the large value of G as Γrad1,LC is the dominant loss term of the uncoupled LC metamaterial, consistently with the expression of G reported in the Appendix C, which scales like Γrad1,LC.

    The resonance linewidth of the coupled modes can be further simplified. Since there is no way to distinguish Γloss,Tamm from Γloss,LC as they both result from ohmic losses in the metal layer, we note Γloss,metal=Γloss,Tamm+Γloss,LC. Moreover, the DBR mirror is highly reflective, leading to Γrad1,TammΓrad2,LC. The resonance linewidth of the coupled modes reduces then to Γ±Γrad2,LC+Γloss,metal2. This simplified expression highlights that Γ± can be ultimately reduced down to its fundamental limit Γloss,metal2 by suppressing the radiative losses Γrad2,LC. To this aim, we fabricate Tamm cavity-LC metamaterial coupled resonators and add a supplementary mirror at a distance of 75 μm over the LC metamaterial, which suppresses radiative losses in s2 at the expense of preventing the use of these hybrid cavities in transmission.

    Figure 4 shows the experimental reflection spectra of two Tamm cavity-LC metamaterial coupled resonators based on DBRs made of two and three silicon layers and, respectively, one and two vacuum layers, including the additional gold mirror blocking transmission. We observe two high contrast peaks at resonance frequencies 0.84 and 1.06 THz, with Q=25.2±1.6 and Q=32.6±1.2 for the two-silicon-layer-based structure and Q=35±6 and Q=37±5 for the three-silicon-layer-based structure. The uncertainty regarding the three layers’ cavity quality factor is attributed to the resolution of our setup (6 GHz) combined with limited peak contrast and non-negligible parasitic features in the spectrum. Increasing the number of silicon layers in the DBR increases its reflectivity and accordingly reduces the value of Γrad1,Tamm so that the quality factor increases. However, adding more than three layers reduces the peak contrast dramatically because of critical coupling arguments [26].

    Reflection spectra of a Tamm cavity resonant at approximately 0.95 THz coupled with an LC metamaterial resonant at 0.92 THz, including an additional mirror blocking the transmission. Blue, two-layer Tamm cavity with Q=25.2±1.6 and Q=32.6±1.2 for the lower and upper frequency coupled modes, respectively. Red, three-layer Tamm cavity with Q=35±6 and Q=37±5, respectively (0.2 offset for clarity). Solid black lines, Lorentzian fits. Quality factor errors are evaluated from fitting standard deviation.

    Figure 4.Reflection spectra of a Tamm cavity resonant at approximately 0.95 THz coupled with an LC metamaterial resonant at 0.92 THz, including an additional mirror blocking the transmission. Blue, two-layer Tamm cavity with Q=25.2±1.6 and Q=32.6±1.2 for the lower and upper frequency coupled modes, respectively. Red, three-layer Tamm cavity with Q=35±6 and Q=37±5, respectively (0.2 offset for clarity). Solid black lines, Lorentzian fits. Quality factor errors are evaluated from fitting standard deviation.

    These results show that our approach improves the quality factor Q by a factor 8 over the uncoupled LC metamaterial. Also, our analysis reveals that the Q enhancement results from the suppression of the dominant decay rate that is the radiation loss rate of the LC metamaterial so that the quality factor is almost ultimately limited by the losses in the metallic layer.

    4. MODE VOLUME OF STRONGLY COUPLED RESONATORS

    We now examine the mode volume V of the Tamm cavity-LC metamaterial coupled resonators. The unit cell pattern of the LC metamaterial is designed to concentrate the electric field in the central capacitive gap between the metallic tips, and we are interested in how this electric field confinement is modified by the coupling to the Tamm cavity. To this purpose, using the eigenfrequency solver of COMSOL Multiphysics, we compute the electromagnetic field distributions of the lower and upper frequency resonant modes of the coupled resonators. Simulations were run in a single period of the metamaterial bounded by orthogonal perfect electric and perfect magnetic conductor boundary conditions to create the array periodicity, and they are bounded by perfectly matched layers in the propagation direction. The gold layer is modeled using the refractive index of gold interpolated from data from Ref. [31]. Figure 5(a) displays the in-plane electric field amplitude enhancement of the upper coupled mode in the LC metamaterial plane for fLC=1.01  THz compared to an input plane wave of unity amplitude (see Appendix C for the electric field profile along the optical axis). We observe a large increase of the electric field between the metallic tips and around their edges, reaching a factor up to 20. From the computed electric field distributions, we calculate the effective mode volumes according to the definition used for dissipative resonators [32], V=(ϵE2μH2)d3r2ϵ(r0)E2(r0),where r0 is taken at the center of the LC metamaterial pattern in vacuum above the layer surface [ϵ(r0)=ϵ0]. This equation is appropriate in our case as the quality factors of the coupled resonators remain in the range of a few tens. For the following, we consider only the real part (V) of this complex expression. We also compute the mode volume of the uncoupled LC metamaterial on an infinite silicon substrate and obtain VLC1.1×104λ3.

    (a) Distribution of the electric field in the LC metamaterial plane over a single unit cell from FEM simulations for the upper frequency coupled mode at fLC=1.01 THz. The figure represents the electric field enhancement factor, i.e., the electric field norm for an input wave of amplitude 1. r0 lies at the center of this picture. (b) Mode volume of upper (red) and lower (blue) frequency coupled modes, normalized by the mode volume of the uncoupled LC metamaterial, as a function of the detuning, from FEM simulation. Dashed line, model from Eqs. (10) and (11). The mode volume values at zero detuning are 3.2×10−4λ3 for V+ and 2.0×10−4λ3 for V−.

    Figure 5.(a) Distribution of the electric field in the LC metamaterial plane over a single unit cell from FEM simulations for the upper frequency coupled mode at fLC=1.01  THz. The figure represents the electric field enhancement factor, i.e., the electric field norm for an input wave of amplitude 1. r0 lies at the center of this picture. (b) Mode volume of upper (red) and lower (blue) frequency coupled modes, normalized by the mode volume of the uncoupled LC metamaterial, as a function of the detuning, from FEM simulation. Dashed line, model from Eqs. (10) and (11). The mode volume values at zero detuning are 3.2×104λ3 for V+ and 2.0×104λ3 for V.

    Figure 5(b) reports the calculated volumes of the upper and lower coupled modes V+ and V, respectively, at their resonance angular frequencies ω+ and ω normalized by VLC for increasing values of fLC (i.e., increasing values of detuning δ2π) and a fixed Tamm cavity resonance frequency fA=1  THz. V+ and V show opposite evolution with fLC: V+ decreases, whereas V increases when fLC increases. We also observe that the two tendencies cross at a small non-zero detuning δ2π=0.02  THz. This crossover shows that there is an exchange of mode volume properties between the two resonators similar to the Q properties. This behavior results from the continuous transition of the upper coupled mode from a mode mainly given by the Tamm mode to an LC metamaterial mode for increasing resonator detuning corresponding to the continuous evolution of θ in Eq. (2), and inversely for the lower coupled mode. Thus, V+ (V) converges toward VLC at large positive (negative) detuning, i.e., fLC much higher (lower) than 1 THz. In addition, we note that the mode volume is increased by only a factor 2.3 compared to VLC at the curve crossing and by factors 2.1 and 2.5 for V+ and V, respectively, at zero detuning. The mode volumes at zero detuning are reduced down to 3.2×104λ3 for V+ and 2.0×104λ3 for V, revealing a high degree of subwavelength confinement of the coupled modes (details are available in Appendix A).

    These simulation results raise the question of the origin of the factor 2 in the increase of V± compared to VLC, and the robustness of this value to the designs of the metamaterials and of the Fabry–Perot cavity. To provide some insight, we model the evolution of V± by computing ±1|E(r)2|±1, where E=EA(r)+EB(r) is electric field operator, with EA(r)=iEA(afA(r)afA*(r)),EB(r)=iEB(bfB(r)bfB*(r)).

    Ej=ωj2ϵ0Vj is the RMS electric field amplitudes associated to vacuum fluctuations in modes j=A,B, with Vj defined at the central position r0. fj(r) is the electric field amplitude profile of the mode, normalized such that fj(r0)=1. We then estimate the mode volume of the coupled mode V+ using the usual expression involving the electric field only [33], V+=ϵ(r)|+1|E(r)2|+1|d3rϵ(r0)|+1|E(r0)2|+1|.

    We recognize in the numerator the expression of the electromagnetic energy in the mode, whose value for one photon is directly 3ω+2. In the denominator, we use the expressions of the coupled terms from Eq. (2), +1|E(r0)2|+1=32cos2θEA2+32sin2θEB2+2cosθsinθEAEB,where tan(2θ)=2Gδ. Then, the coupled mode volume V+ is expressed as V+=VBsin2θω+ωB11+23ωAωBVBVAcotθ+ωAωBVBVAcot2θ,and equivalently for V, V=VBcos2θωωB1123ωAωBVBVAtanθ+ωAωBVBVAtan2θ.

    The detailed calculation is reported in Appendix B. Considering resonator A stands for the Tamm cavity and resonator B for the LC metamaterial, we compare the simulation results in Fig. 5(b) with the calculated V± [Eqs. (10) and (11)] as a function of the resonator detuning and find quantitative agreement for the only free parameter α=VBVA=0.22 (dashed lines), validating our theoretical description. Note that we take into account the evolution of G with ωB using β=0.18THz1 determined from fitting the simulated resonance frequencies.

    In a simplified picture at small detuning, ω+ωωB, and since VBVA, the coupled mode volumes can be reduced to V+VB/sin2θ and VVB/cos2θ. These expressions describe the crossing of the curves at frequency matching (δ=0 and θ=π/4) with V±=2VB. Thus, our model reproduces the increase of V± by a factor of 2 compared to VLC observed at zero detuning and, importantly, highlights that the mode volumes of the coupled resonator systems are dominated by the metamaterial mode volume at small detuning. The factor 2 can be interpreted as the dilution of the photon electromagnetic energy in the two resonators. Achieving coupled mode volumes of only 2 times, the mode volume of the uncoupled LC metamaterials is then a general result that can be transposed to other coupled systems based on metamaterials and Fabry–Perot cavity. Additionally, the refined model from Eqs. (10) and (11) captures the asymmetry of the V± curves at large detuning, due to the variation of the resonator coupling constant G and the contribution of the mode A. Also, our model reproduces the 0.02 THz shift of the mode volume curve crossing due to the symmetric or anti-symmetric contribution of the Tamm mode (mode A), which results in E+(r0)E(r0) at frequency matching. The deviation for V at negative detuning is attributed to the much larger wavelength for f.

    The present analysis demonstrates that the main strength of LC metamaterials, which is their extremely small mode volume, is preserved despite their coupling to the Fabry–Perot cavity with a very large mode volume, which underlines the main interest of coupling these two distinct types of photonic resonators.

    5. HYBRID RESONATORS FOR THz LIGHT–MATTER INTERACTION

    The properties of our hybrid THz cavities based on the coupling of LC metamaterials with Tamm cavities are very promising for the development of THz light–matter coupling platforms with a high degree of coherence. Indeed, these hybrid cavities conciliate quality factors of up to a few tens, provided by the Tamm cavity, with sub-wavelength electric field confinement, provided by the LC metamaterial, as summarized in Table 1.

    Comparison of V and Q (theoretical and experimental) performances for the LC metamaterial alone, coupled with a 2 silicon layers Tamm cavity, and coupled with a 3 silicon layers Tamm cavity + additional mirrora

     LC MMTamm(2L)-LCTamm(3L)-LC + Mir.
    V1.1×104λ33.2×104λ33.2×104λ3
    Qtheo4.53286
    Qexp/1837
    (QtheoV)enh.13.18.3
    (QexpV)enh.11.93.6

    Q/V values are normalized to the values of the uncoupled LC metamaterial to show the enhancement due to the coupling and we considered the upper frequency coupled mode; the Tamm(2L)-LC data are from Fig. 2(a) and the Tamm(3L)-LC + Mir. data are from Fig. 4.

    The lower values of the Q that we measured compared to those predicted are most likely due to the significantly lower conductivity of the chromium–gold layer we used for fabrication compared to the pure gold layer we consider in the simulation, which opens up opportunities for future improvements.

    6. CONCLUSION

    In conclusion, we have experimentally and theoretically investigated original hybrid THz cavities based on the ultrastrong coupling between a THz Tamm cavity and an LC circuit metamaterial. We have demonstrated Q of up to 37 associated with V as low as 3.2×104λ3. Our work shows that the principal qualities of resonators, which are a high Q for Tamm cavity and a low V for LC circuit metamaterials, act beneficially in the presence of the coupling. Compared to the uncoupled LC metamaterial, Q is enhanced by a factor of 8 due to the coupling with a Tamm cavity, while V is increased by only a factor of 2 as a result of the energy dilution within the two resonators. Thus, investigating light–matter interaction using these hybrid THz cavities should provide a high degree of coherence due to the enhancement of the cooperativity C of at least 3.8, opening interesting perspectives for the development of single photon THz emitters and detectors [34] and quantum technology applications. Also this hybrid system compares favorably to uncoupled LC metamaterials for the study of light–matter interaction in the ultrastrong coupling regime with quantum material having low non-radiative decay rates γ, as the reduction of g given by 1/V remains small compared to the reduction of the resonator linewidth κ. Furthermore, our analysis revealed the suppression of the radiative losses of the LC metamaterials when embedded in the Tamm cavity so that the quality factor is ultimately limited by the ohmic losses of the metallic layer. This property opens vast possibilities in the development of new LC metamaterial patterns previously disregarded due to large radiative losses and renews the interest of investigating low-loss metamaterials based on superconducting materials [35,36] to reach very high Q values. Additionally, our theoretical description of the variations of the mode volume due to the coupling evidences the generality of this concept: possible metamaterials are not limited to the hole array LC circuit we presented but could also include direct split ring resonators similar to Ref. [20] as well as micro-antennas.

    Acknowledgment

    Acknowledgment. The authors thank Jean-Francois Lampin and Jean-Michel Gerard for fruitful discussion on the strong coupling regime and cavity.

    APPENDIX A: TEMPORAL COUPLED MODE THEORY FOR S-PARAMETERS OF COUPLED RESONATOR SYSTEMS AND EVALUATION OF THE COUPLING CONSTANT G

    We present here the study of a system of two coupled resonators using the temporal coupled mode theory (temporal CMT), which is used to describe extensively the properties of single resonators in Refs. [23,29]. We will detail here the equations associated with this model to provide the expressions of reflection and transmission coefficients. In particular, we will show which parameters influence the value of the resonator coupling constant G.

    Equation of Motion of the Two Coupled Resonators

    We consider a set of two coupled resonators A and B to which two resonant modes at uncoupled angular frequencies ωA and ωB are associated. The field amplitudes in the respective modes are noted as A and B. These resonators are associated here “in-line,” i.e., resonator A is coupled to an input propagation channel s1 and resonator B to an output propagation channel s2, and transmission is only possible by passing through both resonators A and B (see Fig. 6). This corresponds to the scenario of the Tamm cavity coupled to a resonant LC circuit metamaterial deposited on its top mirror. We note that κi* is the incoming wave s+,μ to resonant mode i coupling constant, ti is the resonant mode i to outgoing wave s,μ coupling constant, and γi=1τloss,i is the amplitude non-radiative decay rate. We use notation ti and ri as these coefficients identify to the amplitude transmission and reflection coefficients in typical optical resonators such as Fabry–Perot cavities.

    Schematic coupling scheme between resonators A and B, as well as input and output propagation channels s1 and s2.

    Figure 6.Schematic coupling scheme between resonators A and B, as well as input and output propagation channels s1 and s2.

    Schematic pictures of the resonators under study, including input illumination “1” for (a) the LC circuit metamaterial on an infinite silicon substrate and (b) the Tamm cavity coupled to the LC circuit metamaterial directly on top. Bottom, corresponding interaction scheme between the Tamm mode A and the LC circuit metamaterial mode B including the relevant coupling rates.

    Figure 8.Schematic pictures of the resonators under study, including input illumination “1” for (a) the LC circuit metamaterial on an infinite silicon substrate and (b) the Tamm cavity coupled to the LC circuit metamaterial directly on top. Bottom, corresponding interaction scheme between the Tamm mode A and the LC circuit metamaterial mode B including the relevant coupling rates.

    Theoretical reflection and transmission spectra of the coupled resonator system from Eqs. (A23) and (A24). We used fA=fB=1 THz, G=0.1 THz, ΓA=0.01 THz, ΓB=0.05 THz, Γrad1,A=0.008 THz, and Γrad2,B=0.045 THz.

    Figure 7.Theoretical reflection and transmission spectra of the coupled resonator system from Eqs. (A23) and (A24). We used fA=fB=1  THz, G=0.1  THz, ΓA=0.01  THz, ΓB=0.05  THz, Γrad1,A=0.008  THz, and Γrad2,B=0.045  THz.

    Evolution of the radiative coupling rate to the substrate Γrad1,LC and of the resonator coupling constant G from FEM simulations. (a) Linear scale; (b) log scale.

    Figure 9.Evolution of the radiative coupling rate to the substrate Γrad1,LC and of the resonator coupling constant G from FEM simulations. (a) Linear scale; (b) log scale.

    Comparison of the radiative coupling rate to the substrate of a CSRR metamaterial Γrad1,LC versus the product G2CTammnSub from FEM simulations, showing excellent empirical agreement.

    Figure 10.Comparison of the radiative coupling rate to the substrate of a CSRR metamaterial Γrad1,LC versus the product G2CTammnSub from FEM simulations, showing excellent empirical agreement.

    APPENDIX B: EVOLUTION OF Γrad1,LC VERSUS fLC AND APPROXIMATE MODEL FOR THE EVOLUTION OF G

    In Fig. 2(b) of the main text, we observe in the anti-crossing of the upper and lower resonance frequencies an asymmetry between negative and positive detunings δ. We attribute this feature to the decrease of the resonator coupling constant G when the uncoupled LC resonance frequency fLC increases, which results from the reduction of the LC circuit pattern size. We estimate in this section the expected variations of G with fLC using GΓrad1,LC as demonstrated in the previous section.

    We can estimate Γrad1,LC using FEM simulations from COMSOL Multiphysics. In order to measure Γrad1,LC only, we simulate the LC metamaterial on an infinite silicon substrate, excited with an input wave from the silicon side, and include an additional mirror on top of the structure to prevent transmission of the structure so that Γrad2,LC=0. Also, the metal layer is modeled as a perfect electric conductor to ensure Γloss,LC=0. The resulting system presents no transmission and no absorption so that the power coefficients are trivial with R=1 and T=A=0, preventing the simple measurement of Γrad1,LC. We, thus, include artificially a small amount of losses by modeling the additional mirror as a gold layer with finite conductivity so that we can directly measure the resonance linewidth Γrad1,LC on the non-zero (of the order of 0.1%) absorbance peak.

    Figure 11(a) represents the resulting spectra for different LC pattern sizes corresponding to different resonance frequencies fLC (legend box). We observe an important reduction of the peak width when fLC increases.

    (a) Resonance peak of an uncoupled LC metamaterial on an infinite silicon substrate with transmission blocked for increasing fLC from FEM simulations. (b) Radiative coupling rate to the substrate of the LC metamaterial Γrad1,LC as a function of fLC, deduced from (a).

    Figure 11.(a) Resonance peak of an uncoupled LC metamaterial on an infinite silicon substrate with transmission blocked for increasing fLC from FEM simulations. (b) Radiative coupling rate to the substrate of the LC metamaterial Γrad1,LC as a function of fLC, deduced from (a).

    We observe a clear linear evolution of Γrad1,LC with fLC in Fig. 11(b), which we fit with a linear function in the form Γrad1,LC=Γ0(1β(fLCf0)), f0=1  THz, and we find Γ0=0.1389  THz and β=2.796  THz1. Finally, since GΓrad1,LC, we have GG0=Γrad1,LCΓ0, and we can propose a linear expansion for G around ωA2π=1  THz in the form G=G0(1β(ωBωA)) with β=12β2π=0.22  THz1.

    In our measurement presented in Fig. 2(b) of the main manuscript, the variability we observe prevents the precise determination of β, and we find β0.13±0.11THz1 from curve fitting using Eq. (3) of the main manuscript and the linear model for G, which remains a consistent value. This model explains the reason for the asymmetry we observe in Fig. 2(b) of the main manuscript.

    APPENDIX C: ELECTRIC FIELD PROFILE ALONG THE OPTICAL AXIS

    We present in Fig. 12 the electric field profile inside the cavity for the higher and lower frequency coupled modes, with and without an additional mirror, from FEM simulations. We primarily notice the large electric field concentration in the z=0 plane corresponding to the LC metamaterial plane. Also, the different scales for the electric field are mainly explained by the quality factor increase in the presence of an additional mirror. Finally, the additional zero of the electric field around z=20  μm for f+ is revealing of the symmetric (f) or anti-symmetric (f+) coupling between the Tamm and LC metamaterial modes.

    Electric field enhancement along the optical axis for the higher (red) and lower (blue) resonant coupled modes, without (left) and with (right) the additional mirror. The cut axes intersect the metamaterial plane at the center of the metamaterial unit cell. The cavity is illuminated from the left with a plane wave of unity amplitude, the LC metamaterial is located at z=0, and the additional mirror is located at z=+75 μm.

    Figure 12.Electric field enhancement along the optical axis for the higher (red) and lower (blue) resonant coupled modes, without (left) and with (right) the additional mirror. The cut axes intersect the metamaterial plane at the center of the metamaterial unit cell. The cavity is illuminated from the left with a plane wave of unity amplitude, the LC metamaterial is located at z=0, and the additional mirror is located at z=+75  μm.

    APPENDIX D: ADDITIONAL FIGURES FOR THE EVOLUTION OF THE MODE VOLUME FROM FEM SIMULATIONS

    We simulate the resonant modes of the coupled Tamm-LC metamaterial (GSRR) structure and of the uncoupled LC metamaterial on an infinite silicon substrate using the eigenfrequency solver of COMSOL Multiphysics. Simulations were run in a single period of the metamaterial bounded by orthogonal perfect electric and perfect magnetic conductor boundary conditions to create the array periodicity, bounded by perfectly matched layers in the propagation direction. Since the quality factors of the coupled resonators are rather low, in the 10 to 50 range, we use for the mode volume the improved formula from Ref. [32] as follows: V=(ϵE2μH2)d3r2ϵ(r0)E2(r0),where r0 is taken at the center of the metamaterial pattern in vacuum above the surface [ϵ(r0)=ϵ0]. From this complex expression, we kept only the real part (V). The mode volumes of the uncoupled metamaterial (VGSRR) and of the upper (V+) and lower (V) frequency coupled modes as a function of the detuning (δ2π) are represented in Fig. 13(a). Figure 13(b) represents the mode volumes VGSRR, V+, and V normalized by the typical volume λ3, where λ is taken at the actual resonance frequency, i.e., λ(ω) is larger than λ(ω+) for a given detuning δ.

    (a) Mode volume of the uncoupled metamaterial (black) and of the upper (red) and lower (blue) frequency coupled modes, computed from COMSOL simulation. (b) Same, normalized by the value of λ3 at the resonance frequency.

    Figure 13.(a) Mode volume of the uncoupled metamaterial (black) and of the upper (red) and lower (blue) frequency coupled modes, computed from COMSOL simulation. (b) Same, normalized by the value of λ3 at the resonance frequency.

    APPENDIX E: DERIVATION OF THE MODE VOLUME OF STRONGLY COUPLED RESONATORS

    In this section, we derive an expression for the mode volume of the coupled modes V± of two strongly coupled resonators A and B as a function of the mode volume of the uncoupled modes VA and VB.

    We consider two modes A and B forming a normal basis on which we can write the electric field in the coupled resonators as E(r,t)=modejαj(t)fj(r).

    αj(t) is the complex amplitude of the mode, accounting for the temporal evolution. fj(r) is a solution of the Helmholtz equation, i.e., is the electric field amplitude profile of the mode, normalized at a reference point r0 such that fj(r0)=1, with r0 being usually the point at which the electric field is maximum. In order to ensure coherence between definitions of the mode volume V and f, V should be defined as well at the position r0, so we write V=ϵ(r)|E(r)|2d3rϵ(r0)|E(r0)|2=ϵ(r)ϵ0|f(r)|2d3r.

    We recall here that this expression is a simplification meant for easier writing: the upper term gives the normalization by the electromagnetic energy and should include the contribution of the magnetic energy and of the time averaging. As they are responsible for factors 2 and 12, respectively, which cancels themselves out, this is usual for the definition of the mode volume.

    From the quantum theory of coupled resonators based on the Hopfield model presented in the main text, the one photon coupled states |±1 are the result of the hybridization of the one photon uncoupled states |1A,0B and |0A,1B, |+1=cosθ|1A,0B+sinθ|0A,1B,|1=sinθ|1A,0Bcosθ|0A,1B,with θ being defined by tan(2θ)=2Gδ, δ=ωBωA. In order to compute the mode volume of the coupled modes using this expression, we evaluate in the following the quantity |E(r)|2 as ±1|E(r)2|±1, using the electric field operator E=EA(r)+EB(r) with EA(r)=iEA(afA(r)afA*(r)),EB(r)=iEB(bfB(r)bfB*(r)),where Ej=ωj2ϵ0Vj are the RMS electric field amplitudes associated to vacuum fluctuations in modes j=A,B and Vj is the uncoupled mode volumes defined at the central position r0.

    First, we expand the expression of E(r)2, E(r)2=EA(r)2+EB(r)2+2EA(r)EB(r)=EA2(|fA(r)|2(1+2aa)fA(r)2aafA*(r)2aa)+EB2(|fB(r)|2(1+2bb)fB(r)2bbfB*(r)2bb)+2EAEB(fA(r)fB*(r)abfA(r)fB(r)ab+h.c.).

    Following the rotating wave approximation, only resonant terms have non-zero contributions to the expectation value ±1|E(r)2|±1. For the |+1 state, +1|E(r)2|+1=3cos2θEA2|fA(r)|2+3sin2θEB2|fB(r)|2+2sinθcosθEAEBRe(fAfB*).

    Let us now evaluate the up and bottom terms of the expression of the mode volume using this expression. The upper term gives the expression of the matrix element of the Hamiltonian operator for the state |+1, ϵ(r)+1|E(r)2|+1d3r=32cos2θωA+32sin2θωB+2sinθcosθωAωBVAVBϵr(r)Re(fAfB*)d3r,whose value is directly 32ω+ since |+1 is an eigenstate. The third term of the above expression is an overlap integral yielding the value of the resonator coupling constant G, if the magnetic contribution is additionally taken into account.

    Considering the bottom term, using Eq. (E5) evaluated at r0 and fA(r0)=fB(r0)=1, ϵ(r0)+1|E(r0)2|+1=32cos2θωAVA+32sin2θωBVB+sinθcosθ2ωAωBVAVB.

    We can finally propose an expression for the mode volumes V+ and V of the coupled modes in a symmetric form as follows: V+=ω+cos2θωAVA+sin2θωBVB+23sinθcosθωAωBVAVB,V=ωsin2θωAVA+cos2θωBVB23sinθcosθωAωBVAVB,from which are deduced the expressions of the main text.

    [1] E. M. Purcell. Spontaneous emission probabilities at radio frequencies. Phys. Rev., 69, 681(1946).

    [15] N. M. Peraca, S. T. Cundiff, M. Kira, A. Baydin, W. Gao, M. Bamba, J. Kono. Ultrastrong light–matter coupling in semiconductors. Semiconductors and Semimetals, 105, 89-151(2020).

    [30] H. A. Haus. Waves and Fields in Optoelectronics(1984).

    Tools

    Get Citation

    Copy Citation Text

    Simon Messelot, Solen Coeymans, Jérôme Tignon, Sukhdeep Dhillon, Juliette Mangeney. High Q and sub-wavelength THz electric field confinement in ultrastrongly coupled THz resonators[J]. Photonics Research, 2023, 11(7): 1203

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Optical Devices

    Received: Nov. 29, 2022

    Accepted: May. 2, 2023

    Published Online: Jun. 16, 2023

    The Author Email: Juliette Mangeney (juliette.mangeney@phys.ens.fr)

    DOI:10.1364/PRJ.482195

    Topics