Optical networks on chip (ONoCs) offer the advantages of large bandwidth and low latency in optical data transmission and are extensively utilized in artificial intelligence, high-speed communications, and optical sensing [1–3]. However, the development of ONoC is hindered by the scarcity of multi-spectral and multi-functional optical modulators [4,5]. Particularly, with the advent of fifth generation (5G) [6–8] and sixth generation (6G) [9–11] communications, the need for such modulators that cover both microwave and terahertz (THz) ranges has become a critical and challenging issue [5,12]. Metasurfaces enhance light-matter interactions within a minimal volume, serving as an essential tool for achieving precise and complex light field modulation in the absence of integrated functional optical devices [13,14]. Current metasurfaces, combined with optically controlled natural materials, can dynamically modulate the amplitude [15], frequency [16], phase [17], polarization [18], and wavefront [19] of electromagnetic waves, applied in dynamic notch filters, phase modulators, tunable perfect absorbers, polarization converters, and holography [20–24]. However, most studies on metasurfaces that cover both microwave and THz ranges can only dynamically control one spectrum, with the functionalities of the other remaining static, thus failing to achieve dual-frequency-range modulation [25,26]. To address the spectral broadening challenges introduced by 6G technology, further research is required on integrated optical elements capable of dynamically modulating functions across both microwave and THz ranges.

On the other hand, externally controlled light-sensitive functional materials are an indispensable part of all-optical modulators [27–29]. In the microwave range, functional materials of active modulators are generally light-responsive materials or components [30,31]. Components are usually photodiodes, varactors, and photoresistors [32]. The light-responsive materials are usually photosensitive materials such as silicon (Si) and graphene [33–35]. In the THz range, functional materials are generally light-responsive materials such as graphene [36], Si [37], perovskite, and vanadium dioxide (${\mathrm{VO}}_2$) [38,39]. Among them, the perovskite films gain attention due to their odd-order nonlinear polarization properties in THz range and absorption characteristics in microwave range [40–43]. Additionally, photosensitive Si plays an important role in the field of modulators in the microwave and THz ranges [44]. Although the dynamic modulation of microwave or THz range using different control materials has been extensively studied. The need for dynamic modulators capable of handling both microwave and THz ranges remains a significant challenge because their ranges are three orders of magnitude different and the interaction between electromagnetic waves and matter is vastly different. Thus, there is a pressing need for specially designed metasurfaces integrated with functional optical control materials to realize dual-band tunable optical modulators of microwave and THz ranges to meet the new requirements of communication technology as 5G advances to 6G.

In this paper, we introduce an innovative all-optical dual-frequency-range modulator, a first of its kind. This modulator integrates a planar nested multiscale metasurface with ${\text{MAPbBr}}_3$ film and Si islands to enable all-optical modulation of both microwave and THz ranges. By utilizing visible light and near-infrared light pumping, it can dynamically switch between transmission and shielding states across both range types. We also elucidate the modulation principles of this all-optical modulator by studying the current distribution and near-field distribution of the electric field. Additionally, the modulator maintains a certain visible and infrared transmittance. Our modulator offers new insights into the design of multi wave and multifunctional integrated optical components.

The schematic diagram of the dual-frequency-range modulator is shown in Fig. 1(a). The modulator has four states. When there is no light pumped, both microwaves and THz waves are in the transmission state, allowing the transmission of information as depicted in Figs. 1(c) and 1(e). When only light1 is pumped, the modulator shields microwaves, as indicated in Fig. 1(b), while the THz wave is not transmitted, as shown in Fig. 1(e). When only light2 is pumped, the modulator allows microwave transmission and shields THz waves, as shown in Figs. 1(b) and 1(c). When light1 and light2 are pumped at the same time, the microwave and THz wave are shielded, as shown in Figs. 1(b) and 1(d).

Figure 2 illustrates the structure diagram of the dual-frequency-range modulator. Figure 2(a) depicts the metal-only basic unit, labeled as structure S1, which forms the orange section of the larger structure, labeled as structure L, shown in Fig. 2(b). Figure 2(c) displays the small basic Si-hybrid structure, labeled as structure S2. Controlled by light1, conductivity of Si on structure S2 is modulable, influencing the overall structural performance and forming the blue section of structure L. The top is covered with a perovskite film to realize the THz band amplitude modulation function of the metasurface. The film’s conductivity can be modulated by light2, affecting the structure’s properties. The geometric parameters are $m_1=4.96\mathrm{mm}$, $m_2=1.28\mathrm{mm}$, $m_3=0.4\mathrm{mm}$, $m_4=4\mathrm{mm}$, $m_5=0.32\mathrm{mm}$, $m_6=0.08\mathrm{mm}$, $P_{x1}=P_{y1}=7.04\mathrm{mm}$, $t_1=24\mathrm{\mu m}$, $t_2=10\mathrm{\mu m}$, $t_3=6\mathrm{\mu m}$, $t_4=4\mathrm{\mu m}$, and $P_{x2}=P_{y2}=80\mathrm{\mu m}$.

As depicted in Figs. 2(a) and 2(c), the two smaller basic units comprise four split-ring resonators (SRRs) and four metallic wires each. We use the Computer Simulation Technology (CST) Microwave Studio to obtain the simulation results and current distribution of the modulator to analyze the modulator modulation principle. In the THz range, structure S1 generates a transparent peak due to electromagnetically induced transparency (EIT) resonance. As shown in Fig. 3(a), when the THz wave in the $y$-polarized direction is incident on the plane, the SRRs act as a bright mode, directly interacting with the incident wave. A resonance peak occurs at 0.98 THz, which is generated by LC resonance. The current distribution diagram is shown in Fig. 3(c). This interaction results in resonance exhibiting strong electric field constraints within the small capacitance gaps of the SRRs, rendering it highly sensitive to the surrounding light environment, which is suitable for application in THz-related modulation devices. The metal wire surrounding structure S1 forms a CRR, and its THz transmittance curve is shown in Fig. 3(b). The individual CRR does not resonate within the THz range. As shown in Fig. 3(d), the metal wire parallel to the resonance direction behaves as the quasi-dark mode, generating dipole resonance through its interaction with the incident THz wave. The current along the $y$-axis is visible in Fig. 3(d), but the resonance intensity is weaker than that produced by the SRRs. The metal wire perpendicular to the resonance direction is the dark mode, unable to resonate directly with the THz wave, resulting in no current along the $x$-axis in Fig. 3(d). At the same time, the current of the metal wire along the $y$-axis direction is suppressed by the metal wire along the $x$-axis direction, so that it is converted from the bright mode to the quasi-dark mode.

The total THz transmittance of structure S1 as a function of frequency is shown in Fig. 4(a). Under THz wave excitation, EIT resonance occurs in structure S1, creating a transmission window with a peak transmittance of 74.13% at 0.66 THz. The corresponding current distribution is depicted in Fig. 4(b). Compared to the current distributions of the SRRs and CRR individually, the overall current intensity in structure S1 increases. This is due to the SRRs’ bright mode generating a strong current, which induces a magnetic field that excites currents in both the quasi-dark and dark mode metal wires along the $y$-axis. However, this also induces a magnetic field that alters the current direction in the SRRs. Consequently, the current directions in the SRR and CRR metal strips are opposite, resulting in the EIT effect. As the conductivity of the ${\text{MAPbBr}}_3$ film increases, the transmission window gradually diminishes, and the EIT resonance effect fades, ultimately reaching a shielding state with a transmittance of only 3.24%, as shown in Fig. 4(b). The current distribution in this state is shown in Fig. 4(c). Compared to a conductivity of 0 S/m, the overall current intensity significantly decreases, and the current direction on the SRRs is the same as that on the CRR, causing the disappearance of the EIT effect and the transmission window.

For structure S2, in the THz range, when the conductivity of Si is 0 S/m, the metal wires cannot form a CRR, and only the metal wire along the $y$-axis is excited by the THz wave to produce LC resonance, as shown in Fig. 5(a). A resonance peak is observed at 0.80 THz in the metal wire, and the corresponding current distribution is shown in Fig. 5(c), where the metal wire along the $y$-axis acts as the bright mode. The current is present in the $y$-axis direction, while no current flows in the metal wire along the $x$-axis. The metal wire in the $x$-axis direction is not connected to the metal wire in the $y$-axis direction, which cannot suppress its resonance and cannot make it become a quasi-dark mode. As analyzed in structure S1, the SRRs generate resonance at 0.98 THz as a bright mode. The THz transmittance curve for structure S2 with a Si conductivity of 0 S/m is shown in Fig. 6(a). The maximum transmittance is 24.78% at 0.73 THz, indicating a weak modulation effect. The current distributions of ${\text{MAPbBr}}_3$ with conductivities of 0 and 6000 S/m are shown in Figs. 6(b) and 6(c). It is apparent that the current directions in the SRRs and metal wire are aligned, indicating the absence of EIT resonance. This weak modulation effect is not caused by the creation and disappearance of EIT resonance but by LC resonance.

When the conductivity of Si is 50,000 S/m, the metal wires reconnect to form a CRR; the THz transmittance of the structure S2 peripheral metal wire changing with frequency is shown in Fig. 5(b), similar to the CRR in structure S1. The dark mode metal wire along the $x$-axis suppresses the current in the metal wire along the $y$-axis, transitioning it from the bright mode to the quasi-dark mode, thereby suppressing the resonance in the $y$-direction. The corresponding current distribution is shown in Fig. 5(d), where the current along the $y$-axis is weaker than in Fig. 3(d). When the conductivity of Si reaches 50,000 S/m, the THz transmittance curve, shown in Fig. 6(d), peaks at 84.29% at 0.58 THz. As the conductivity of ${\text{MAPbBr}}_3$ increases, the transmission window gradually closes, and the EIT resonance effect disappears. Finally, it is in the shielding state, and the transmittance is only 1.28%. The current distributions of ${\text{MAPbBr}}_3$ at 0 and 6000 S/m are shown in Figs. 6(e) and 6(f). It can be seen from Figs. 6(e) and 6(f) that the current direction of the two diagrams is similar to that of structure S1, that is, the current direction of the metal strip on the SRRs and the metal strip on the CRR is opposite as the conductivity of the ${\text{MAPbBr}}_3$ film is 0 S/m, resulting in the EIT effect. As the conductivity of the ${\text{MAPbBr}}_3$ film increases to 6000 S/m, the current direction of the metal strip on the SRRs and the CRR metal strip becomes the same, which proves that the EIT effect disappears.

In the microwave range, structure S2, with a period of 80 μm, does not resonate with the microwave, and the change of transmittance with frequency is basically a straight line. As the conductivity of Si increases, the transmittance gradually decreases by approximately 15%, as shown in Fig. 7(b). For structure S1, although its structural period is also 80 μm, its four metal strips form a CRR. The outer ring of the CRR has the same size as the structural period, and the metal strips of multiple small structures S1 are connected, forming a large metal grid. This can be approximated as a conductive film with poor conductivity, causing the reflection between the microwave and structure S1 to dominate, leading to an excellent shielding effect with a transmittance of only 0.02%, as shown in Fig. 7(c).

Structure S1 and structure S2 finally form a large structure L, where the center is the Jerusalem structure, and the edges are composed of CRR, with the outer ring of CRR matching the unit structure size, as shown in Fig. 2(b). When the conductivity of Si is 0 S/m, structure S1 primarily transmits microwaves, as shown in Fig. 2(b), while structure S2 serves as a shielding element. In the large structure L, the position of structure S2 can be approximated as a conductive film with low conductivity. The curve of microwave transmittance of large structure L with frequency is shown in Fig. 7(a). When the conductivity of Si is 0 S/m, the structure can generate a transmission peak with a transmittance of 51.27% at 9.92 GHz, induced by LC resonance. When the conductivity of Si is 800 S/m, the Si in structure S2 exhibits some conductive properties. Combined with the strong shielding effect of structure S1 on microwaves, the overall large structure can be approximated as a layer of conductive film with low conductivity, as shown in Fig. 7(a). Consequently, the final transmittance is reduced to only 5%. The current distribution of the metasurface for both cases is shown in Figs. 7(d) and 7(e). When the conductivity of Si is 0 and 800 S/m, the current directions in the middle and outer sections are opposite and remain unchanged with the variation of Si conductivity. The overall current intensity is small. This indicates that there is a weak EIT effect in this structure. However, the resonance peak and control effect are primarily governed by LC resonance rather than the EIT effect.

To validate our design, we fabricated samples on a commercial silicon-on-sapphire (SoS) chip. The SoS wafer is an epitaxial Si with a thickness of 700 nm grown on a $497\mathrm{\mu }\mathrm{m}$ R-plane sapphire substrate, with the Si resistance exceeding $1000\mathrm{\Omega }$. Initially, the Si thickness was reduced to 500 nm using reactive ion etching (RIE). Subsequently, the position and shape of the Si islands were defined via UV lithography, followed by the selective removal of undesired Si regions through RIE. This process obtained the designed Si islands. The UV lithography alignment was then employed to determine the placement and contours of the Cu structures, ensuring alignment with the previously processed Si islands. Cu was deposited to a thickness of 200 nm using magnetron sputtering, with excess metal later removed to realize the Cu structures. Figure 2(d) illustrates the overall structure of the hybrid metasurface as captured by a confocal microscope, and the details of the corresponding position amplification are shown in Fig. 2(e). Microscopic examination confirmed that the fabricated samples were complete, with the Si islands precisely corresponding to the Cu structures, thereby meeting the design requirements.

The test results of the all-optical dual-frequency-range modulator at 8–18 GHz using a microwave vector network analyzer are presented in Fig. 8. In the experiments, we employed 639 nm visible light (light1) and 1064 nm near-infrared light (light2) to pump the samples, with the results depicted in Figs. 8(a) and 8(b), respectively. When there is no pumped light, the sample forms an LC resonance at the microwave range, forming a transmission peak with a maximum transmittance of 55.52% at 10.05 GHz. As the intensity of light1 increases, the microwave transmittance progressively decreases, reaching a minimum of 3.31% at the highest light intensity, and the average transmittance is only 2.91%, showing a shielding state. According to the definition of transmittance modulation depth (MD) $\eta =(T_{\mathrm{EIT}\mathrm{nolight}}-T_{\mathrm{EIT}\mathrm{light}})/T_{\mathrm{EIT}\mathrm{nolight}}$, the modulation depth at 10.05 GHz is 94.03%. It is basically consistent with simulation results of the modulator in the microwave range shown in Fig. 7 obtained by simulation, which proves the accuracy of the above analysis. Conversely, when the sample is pumped by light2, the obtained transmission rate curve remains essentially unchanged, indicating that light2 has no effect on the GHz range.

The THz-TDS detection system was employed to test the all-optical dual-frequency-range modulator in the 0.1–1.1 THz range, as illustrated in Fig. 9. In the experiments, we utilized light1 and light2 to pump the samples, as depicted in Figs. 9(a) and 9(b), respectively. When there is no light pump, the interaction between the THz waves and the sample resulted in an energy exchange, producing a broadband transmission effect with transmittance exceeding 50% between 0.54 and 0.93 THz, peaking at 65.33%. As the intensity of light2 increased, the transmittance of the THz waves gradually decreased, culminating in a minimum transmittance of 10.62% at maximum light intensity, indicating a shielding state. Similarly, when pumped with maximum intensity light1, the sample exhibited a broadband transmission effect greater than 50% from 0.40 to 0.95 THz. This situation not only extended the transmission bandwidth but also increased the peak transmittance to 80.63%. However, an increase in light2 intensity again gradually reduced the THz wave transmittance, reaching a shielding state with only 7.44% transmittance at the highest intensity, corresponding to a modulation depth of 90.77%. It is basically consistent with simulation results of the modulator in THz range shown in Fig. 5 obtained by CST simulation, which proves the accuracy of the above analysis.

To achieve dual-frequency-range modulation of microwave and THz ranges, two materials, Si and ${\text{MAPbBr}}_3$, are used. The bandgap of Si is 1.12 eV, and the pump energy of 1064 nm is 1.17 eV, which is close to Si’s bandgap. Si has a small absorption coefficient, and this energy does not significantly alter its conductivity. Moreover, the maximum intensity of the 1064 nm laser used in this experiment is only 1100 mW, which is insufficient to induce nonlinear effects in Si. On the other hand, the pump energy of 639 nm (2.00 eV) exceeds Si’s bandgap, allowing linear absorption in both microwave and THz ranges, leading to a change in Si’s conductivity, as shown in Fig. 10(a). The ${\text{MAPbBr}}_3$ film exhibits excellent nonlinear properties due to its unique structure, enabling strong absorption of photons with energy lower than its bandgap. The bandgap of ${\text{MAPbBr}}_3$ is 2.3 eV, and the 1064 nm pump light satisfies the condition for nonlinear generation. ${\text{MAPbBr}}_3$ has a high three-photon absorption coefficient, which allows for effective adjustment of its conductivity in the THz range, as shown in Fig. 10(b). However, the energy of 639 nm is less than the bandgap width of ${\text{MAPbBr}}_3$, which is unable to produce linear absorption, and its nonlinear absorption coefficient is also small. Consequently, the pump light of 639 nm is unable to alter the conductivity of the perovskite film. At the same time, ${\text{MAPbBr}}_3$ only has absorption in the microwave range, and its conductivity remains unaffected by external stimuli in that range.

To understand the resonance mechanism of the structure, we studied the near-field distribution of the all-optical dual-frequency-range modulator as shown in Fig. 11. Figures 11(a)–11(d) depict the microwave near-field distribution at varying Si conductivities. As shown in Fig. 11(a), when the Si conductivity is 0 S/m, a strong electric field is generated at the interface between the Jerusalem structure and the metal frame. With increasing Si conductivity, the electric field intensity diminishes until it disappears, indicating that the microwave transmission peak results from LC resonance. As Si conductivity increases, the capacitance in the LC resonance state decreases, leading to the disappearance of the resonance and the transmission peak. Figures 11(e)–11(h) show the electric field distribution for structure S1 in the THz wave at varying ${\text{MAPbBr}}_3$ film conductivities. Initially, when ${\text{MAPbBr}}_3$ conductivity is 0 S/m, the electric field mainly concentrates near the SRR gap, which indicates that the LC resonance generated by the SRR of structure S1 is dominant at this time. As ${\text{MAPbBr}}_3$ conductivity increases, the electric field intensity progressively weakens and eventually vanishes, indicating a gradual reduction and ultimate disappearance of EIT resonance. Figures 11(i)–11(p) show the electric field distribution for structure S2 in the THz wave, with conductivity of Si at 0 S/m shown in Figs. 11(i)–11(l). Here, strong resonant electric fields localize at the four corners, with a weaker electric field at the SRRs gap, demonstrating that the dipole resonance generated by the metal wire along the electric field direction in structure S2 is dominant. The EIT resonance at this time is weak. As the conductivity of the ${\text{MAPbBr}}_3$ film increases, the electric field at the four corners decreases and eventually disappears. This leads to the disappearance of the dipole resonance and the associated transmission peak. With conductivity of Si at 50,000 S/m shown in Figs. 11(m)–11(p), strong resonant electric fields localize at the four corners, with a weaker electric field at the SRRs gap, demonstrating that the dipole resonance generated by the metal wire along the electric field direction in structure S2 is dominant. The EIT resonance at this time is dominated by the dipole resonance. As the conductivity of the ${\text{MAPbBr}}_3$ film increases, the electric field at the four corners decreases and eventually disappears. This leads to the disappearance of the final double EIT resonance and the associated transmission peak. From the above analysis, it can be concluded that the generation and disappearance of microwave LC resonance can be controlled by modulating the conductivity of the Si, enabling the switching between transmission and shielding states for the microwave range. Similarly, by controlling the conductivity of the ${\text{MAPbBr}}_3$ film, the generation and disappearance of the EIT and LC resonance in the THz range can be achieved, enabling the switching between transmission and shielding states for the THz range.

In addition to enabling the switching between multiple states in both microwave and THz ranges, the modulator also exhibits a certain optical transmittance. The transmission spectrum of the test sample in the 200–2500 nm range is shown in Fig. 12. The structure demonstrates optical transmittance visible to the naked eye, with an average transmittance of 54.74%. The optical transmission characteristics of the sample are attributed to the fact that the wavelength of the optical range is much smaller than the structural period, allowing light waves to pass through the sample. However, due to the high proportion of metal structure and the intrinsic light absorption of the ${\text{MAPbBr}}_3$ film, the final average transmittance is limited to 54.74%.

In summary, for the first time, we proposed and experimentally verified an all-optical dual-frequency-range modulator. The modulator is based on a planar nested multiscale metasurface combined with ${\text{MAPbBr}}_3$ and Si. It demonstrates amplitude modulation capabilities for microwave and THz ranges, achieving maximum modulation depths of 94.03% and 90.77%, respectively, when pumped with visible and infrared light. This dual-frequency-range modulation results from the synergistic effects of nested multiscale structures of the metasurface, the odd-order nonlinear polarization properties of perovskite in the THz range, and the linear absorption properties of Si in the microwave range. Analysis of the current distribution and near-field electric field distribution reveals that microwave range modulation is achieved by controlling LC resonance intensity, while THz range modulation is modulated through alterations in EIT and LC resonance intensities. Additionally, the structure maintains a 54.74% transmittance in the visible-infrared band. This all-optical dual-frequency-range modulator offers potential for all-optical coding metasurfaces, thereby providing new insights into the development of optical components.