1. Introduction

Radiation in the frequency range from terahertz to infrared (THz-to-IR) has the ability to penetrate many materials and reveal their corresponding spectral signatures, which has attracted considerable interest in its potential applications [1, 2]. THz-to-IR radiation is widely used in materials science research, pharmaceutical compound analysis, biomedical imaging, security, and particle physics research [3–5]. A major impediment to the widespread commercialization of this technology is the lack of an economical, compact, broadband, and high-intensity radiation source. Routinely employed THz sources include nonlinear crystals, photoconductive antennas, and gas plasmas [6–8]. However, there is still a need for broadband THz sources able to operate with low energy ultrafast lasers. More recently, several research groups that have used femtosecond-laser irradiation of metal surfaces to generate THz radiation employed thin metal films deposited onto dielectric substrates with periodic structures [9–12].

We have demonstrated a prospective intense and ultra-broadband THz-to-IR source using metal film deposited on a substrate with randomly ordered nanoscale pores [13]. After irradiating the sample with a femtosecond laser pulse, a fraction of the absorbed laser energy is retained in the surface layer of the sample and then dissipates into the bulk sample as residual thermal energy [14–16]. The residual energy causes the temperature of the bulk sample to increase [17]. When the bulk sample reaches thermal equilibrium, it acts as a thermal radiation source. The radiation spectrum over a bandwidth from 0 THz to 150 THz emitted from our sample was previously reported in Ref [13]. and shows approximately 95% optical absorption. The maximum pump to THz to IR energy conversion efficiency for our metal film reached 2%, which resulted from a maximum emitted radiation power. Further up-scaling of the radiation power is only limited by the available pump laser energy.

In the configuration used in our previous work, the sample was placed in the beam path with the metal film surface facing the incident beam and the THz-to-IR emission was collected from the side of the sample that faced away from the beam. The optical absorption was dramatically enhanced owing to both an antireflection mechanism and the dissipation of excited surface plasmon polaritons into the metal surface [13, 18].

In the study that we present in this letter, THz-to-IR emission was observed not only with the laser directly incident on the metal-coated front side but, surprisingly, also with the laser passing through the supporting substrate to the rear side of the metal film. In both cases, THz-to-IR radiation was measured in both directions along the laser beam axis (forward and backward to beam propagation, bi-directional). The observed strong dependence of the THz-to-IR radiation power on the incidence angle and laser polarization implies the involvement of surface plasmons, either delocalized or localized at the air/metal and metal/substrate interfaces.

2. Sample description

The substrate of our sample was a commercially available anodic aluminum oxide membrane (Whatman, Germany). The 60-$\mu \text{m}$-thick membrane consisted of randomly arranged pore arrays with 200 nm average pore diameter and 50% pore density. A ruthenium film was deposited on the substrate by magnetron sputtering, thereby acquiring a nanoscale porous roughness resulting from the through-pore arrays of the substrate. We fabricated several samples with metal films of various nominal thicknesses (e.g., 10 nm, 30 nm, 50 nm, 70 nm, 100 nm, 150 nm, 200 nm, and 300 nm). The morphologies of the metal films were observed by scanning electron microscopy (SEM). Figure 1 shows the top view [Fig. 1(a)] and cross-sectional view [Fig. 1(b)] of the metal-film side and the top view [Fig. 1(c)] of the substrate side of a representative sample with a 150-nm-thick metal film. The images show that the metal surface has a variety of nanoscale structures, including nanoscale voids and nanoprotrusions. The metal surface topography of the films was similar for all the tested metal films, except for the average diameter of the voids, which decreased with increasing thickness.

 

Fig. 1 (a)–(c) SEM images of a sample with a 150-nm-thick metal film. (a) Top view of the metal surface. (b) Cross-sectional view of the metal surface. (c) Top view of the substrate surface. The darkest regions indicate the absence of metal. (d), (e) Schemes of the experimental setups used for measuring the incidence angle ($\theta$) dependence and in-plane angular ($\phi$) distribution of THz-to-IR radiation. (d) Front-side excitation; the laser is incident on the metal surface. (e) Rear-side excitation; the laser is incident on the substrate. The red arrow indicates the direction of the optical beam.

Download Full Size | PDF

3. Experimental results

We used a 1-kHz repetition-rate regeneratively amplified laser providing 1-mJ pulses with a 100-fs pulse duration centered at 800 nm. The maximum power of the laser incident on the sample was 700 mW. The samples were placed in the path of the optical beam. The THz-to-IR radiation was detected by a calibrated Golay cell equipped with a diamond input window (Microtech SN:220712-D), which had an approximately flat response over a broad spectral range (0–150 THz). The schematic diagrams of the experimental setups used for measuring the incidence angle dependence and in-plane angular distribution of the THz-to-IR radiation are shown in Fig. 1(d) (laser irradiation of the front side) and Fig. 1(e) (irradiation of the rear side). In the configuration used for measuring the incidence angle ($\theta$) dependence, the sample and detector were spatially fixed at an optimized position. A modified setup was used to measure the in-plane angular distribution of the radiation by rotating the Golay cell detector in the horizontal plane around the center of the sample. The quantity $\phi$ represents the angle between the sample-detector direction and that of the incident laser. The distance between the sample and the detector was fixed.

In our previous research, we have demonstrated that high power radiation in THz-to-IR frequency region emitted from our sample under femtosecond laser irradiation is due to thermal radiation mechanism [13]. We measured the in-plane angular distribution of the THz-to-IR thermal emission with the optical laser normally incident on the sample ($\theta =0$°). The angle $\phi$ ranged from −120° to 140° in steps of 5° and was limited by the available space of the setup. Figures 2(a) and 2(b) show representative results obtained using a sample coated with a 150-nm-thick metal film. The patterns of both forward and backward radiations are circles tangent to the sample surface that satisfy Lambert’s cosine law. Although the measurements were performed with a p-polarized laser, identical angular distribution patterns were obtained for arbitrary polarizations of the incident laser, exhibiting an axially symmetric distribution of the THz-to-IR emission around the laser propagation direction. In the case of front-side excitation, the forward radiation was stronger than the backward wave, as shown in Fig. 2(a). In contrast, for rear-side excitation, the forward radiation was weaker than the backward one [Fig. 2(b)].

 

Fig. 2 (a), (b) Angular distribution of the THz-to-IR emission from the sample with 150-nm-thick metal-film coating for (a) front-side and (b) rear-side laser irradiation. The red arrow indicates the direction of the optical beam. The solid lines are experimental data. The dashed lines are guides for the eye. (c), (d) The power of the forward (blue data points) and backward (red data points) emission versus the metal-film thickness for the laser irradiation of (c) the front side and (d) the rear side. Figure (c) also displays the absorptivity at 800 nm as a function of film thickness.

Download Full Size | PDF

Figure 2(c) displays the forward and backward radiation power versus the thickness of the metal film for front-side excitation. The forward radiation is always stronger than that emitted in the backward direction, except for the sample with the 10-nm-thick metal film, which displays almost equal intensities for the two directions. The maximum intensity of the forward radiation corresponds to a thickness of 100 nm. In Fig. 2(c), we also plotted the absorptivity at 800 nm as a function of film thickness. The dramatically enhanced absorptivity varies with the metal surface roughness, which is modified by the film thickness, and also exhibits the maximum value at 100 nm; this implies that stronger absorption of the femtosecond laser enables more efficient THz-to-IR generation in the forward direction. Here, we assume that the absorption in the substrate is negligible.

The maximum intensity of the backward radiation occurs near a thickness of 50 nm. In metal films of a thickness similar to the skin depth of the electric field, a strong coupling of the surface plasmons in both interfaces (air/metal and metal/substrate) is possible, resulting in simultaneous thermal emission from both interfaces [19, 20]. As demonstrated in our previous work, the strongest localized surface plasmon electric field (and, therefore, the most efficient thermal radiation source) exists at the metal/substrate interface [18]. Therefore, the main factor that determines the observed different optimal thickness for the forward and backward radiation directions is the interplay between the absorption of the emitted THz-to-IR radiation by the metal film itself and the “gain” obtained by exciting the film with 800-nm pulses. This effect results in an optimal THz-to-IR emission efficiency that occurs at a certain film thickness in the transmission geometry.

Figure 2(d) displays the forward and backward radiation power as a function of film thickness for rear-side excitation. In this case, the absorptivity is considered to be constant because the deposition of the metal film cannot affect the air/substrate interface. For the backward radiation, which has negligible absorption in the substrate, the “gain” of the metal film at the metal/substrate interface generates THz-to-IR emission monotonously and then the power of emission reaches saturation. The optimal thickness is 50 nm for the forward radiation, which has been demonstrated to be the optimal thickness for the interplay between the absorption and “gain” of metal film in Fig. 2(c).

Figure 3 shows the THz-to-IR radiation power emitted from a sample coated with a 150-nm-thick metal film as a function of optical pump energy for front-side and rear-side laser irradiation. The THz-to-IR power was measured in the forward direction ($\phi =0$) with the laser irradiating the sample at a normal incidence angle ($\theta =0$). In the case of front-side irradiation, the forward THz-to-IR radiation is consistently stronger compared with that observed during rear-side irradiation with the same laser parameters. The experimental results were fitted with power laws with exponents of 1.82 for the front-side excitation and 1.85 for the rear-side excitation. The similarity of the two exponents implies that the underlying thermal emission mechanisms are the same for both configurations.

 

Fig. 3 Power of THz-to-IR radiation versus optical pump energy for front-side (red) and rear-side (blue) laser irradiation. The lines are exponential fitting curves.

Download Full Size | PDF

To further demonstrate the involvement of surface plasmons, we measured the THz-to-IR power dependence of the excitation beam on the incidence angle for both p and s laser polarizations; the results are shown in Fig. 4. The measurements were performed using a sample coated with a 150-nm-thick metal film. The angle $\theta$ ranged from −60° to 60° with a step of 5° and with a normal incidence corresponding to an angle of 0°. The above range was limited by the available space of the setup. Figure 4(a), which displays the experimental results for front-side irradiation, shows that for an s-polarized laser beam, the THz radiation power decreases as the incidence angle increases. Because sample rotation increases the illuminated area of the sample and reduces the effective excitation intensity, the power of the THz-to-IR radiation is expected to scale as ${\cos }^n(\theta )$, where $n=1.82$ represents the order of the power dependence of the THz-to-IR emission process.

 

Fig. 4 Measured forward THz-to-IR radiation power as a function of the incidence angle for (a) front-side and (b) rear-side optical laser irradiation. The red symbols correspond to p-polarized laser and the blue ones to s-polarized laser. The blue lines represent the fitting function ${\cos }^n\theta$, where $n$ is the order of power dependence of the THz-to-IR radiation.

Download Full Size | PDF

However, for a p-polarized laser beam, the THz power increases with the incidence angle. The maximum radiation intensity is observed near an angle similar to the resonance angle that corresponds to the phase-matching condition for a grating. Our randomly roughened surface can be described in terms of its Fourier components of a two-dimensional grating. The phase-matching condition is obtained when the wave vector of the external field projected onto the grating combined with multiples of the grating wave vector matches the wave vector of the surface plasmons [21]. The distribution of the resonance angle in our case appears to be somewhat wider than the surface plasmon absorption resonance owing to the variety of nanoscale structures in the roughened metal surface and the broad spectrum of the applied femtosecond laser.

Moreover, surface plasmons do not normally couple to propagating electromagnetic waves in the case of quasi-random arranged corrugated surface [9]. In metal films with thickness similar to the skin depth of the electric field, strong coupling of the surface plasmons in both the air/metal and metal/substrate interfaces is possible; this results in the generation of surface plasmons, either delocalized or localized at both interfaces [9, 10]. Figure 4(b) shows the incidence angle dependence of the THz-to-IR radiation power in the case of rear-side excitation. For both p- and s-polarized lasers, the THz-to-IR power decreases with increasing incidence angle and is expected to scale as ${\cos }^n(\theta )$, where $n=1.85$ represents the order of the power dependence of the THz-to-IR emission process. This implies that localized surface plasmons predominate in the metal/substrate interface when the laser passing through the supporting substrate is incident onto the rear side of the metal film.

4. Conclusion

In summary, we demonstrated the production of THz-to-IR radiation from a metal film coated on a substrate with randomly ordered pore arrays using femtosecond-laser irradiation of the front and rear sides. In both cases, the radiation was observed in both the direction of the laser beam propagation and the reverse direction. The strong dependence of the THz-to-IR power on the incidence angle and laser polarization implies the involvement of surface plasmons, either delocalized or localized at the air/metal and metal/substrate interfaces.