Wideband terahertz systems, operating in the frequency range between the microwave and far-infrared region of the electromagnetic spectrum, have the potential to combine the best of two worlds. On the one hand, they offer hundreds of GHz of spectral bandwidth and sub-mm spatial resolution. On the other hand, they have the capability to see through dielectric materials that are opaque to infrared and visible light. These properties make them attractive for a variety of applications, including material identification [
1,
2], material characterization [
3,
4], and imaging [
5,
6]. However, while electronic terahertz systems already exhibit a high degree of integration [
7,
8], their instantaneous bandwidths have yet failed to exceed a few tens of GHz. Photonic, i.e., laser-driven, terahertz time-domain spectroscopy (THz TDS) systems on the other hand can already provide a few THz of bandwidth at the cost of a low degree of integration and high system complexity. These systems use trains of ultra-short infrared pulses that are photodetected to generate broadband terahertz pulse trains. On the receiver side, the terahertz pulses are sampled with a time-delayed version of the same infrared pulse train [
3]. State-of-the-art fiber-coupled THz TDS systems employ mode-locked fiber lasers at 1550 nm with repetition rates around 100 MHz as their driving light source. These systems exhibit bandwidths of several THz and can be shrunk to shoe box size [
9]. However, a remaining challenge is the limited range, which decreases drastically with increasing frequency due to the low terahertz transmit power [
10,
11]. Furthermore, a decrease in size is not possible because the fiber laser inhibits integration. While state-of-the-art THz TDS systems already provide acquisition rates of a few kHz, spatially resolved measurements are still slow due to the need for a mechanical translation or rotation of either the object under test or the emitter and detector. We aim to increase the range of THz TDS systems and enable high-speed non-mechanical steering of the emitted terahertz beam by using a linear array of
N coherently driven spatially distributed emitters instead of a single emitter. This way, both the directivity and—through spatial power combining—the total radiated power can be increased, yielding a possible increase of the equivalent isotropically radiated power by
. The emitted beam can be steered by adjusting the time delays between the optical signals at the photonic terahertz emitters. This concept is known from photonic communications and is commonly referred to as photonic true time delay beam steering [
12,
13,
14,
15,
16]. The delays introduced in the infrared domain are translated directly to the terahertz domain. In the terahertz range, it has been successfully demonstrated by Bauerschmidt et al. for frequency-domain, i.e., continuous-wave, spectroscopy using individually lensed fiber-coupled terahertz emitters and fiber-coupled free-space variable optical delay lines [
17]. In a more recent approach, Preu et al. have replaced the discrete emitters with a monolithically integrated array of antenna-integrated photodiodes [
18]. However, the fiber-coupled free-space variable optical delay lines are still bulky and mechanically sensitive. Likewise, a fiber-coupled beam steering network is prohibitively complex even for a modest number of array elements. The solution is the use of integrated optics for the photonic beam steering network. Most prominently, integrated optical ring resonators (ORRs) have been proposed by Meijerink et al. [
14] and successfully implemented by Zhuang et al. [
15] as variable optical delay units (ODUs) for X-band communications. More recently, ORRs have been demonstrated for photonic beam steering for W-band communications by Liu et al. [
16]. The group delay response of ORRs is intrinsically narrowband and periodic in the frequency domain. Although bandwidths of a few GHz have been reported by cascading several rings [
16], this is still orders of magnitude too small for THz TDS.
In this paper, we propose a concept that exploits the periodicity of the ORRs’ group delay response to circumvent their bandwidth limitation. We choose the optical perimeter of the ORRs so that their free spectral range (FSR) is identical to the FSR of the mode-locked laser that drives the THz TDS system. This way the laser modes can be aligned with the resonant frequencies of the ORR. Consequently, the ORR introduces identical delays to all laser modes. To keep the ORRs small enough to be realized in an integrated optical circuit, the FSR needs to be on the order of several tens of GHz. This condition is fulfilled by using a monolithic mode-locked laser diode (MLLD) in an approach we call ultra-high repetition rate terahertz time-domain spectroscopy (UHRR THz TDS). This approach has first been demonstrated by Merghem et al. [
19] and system-theoretically analyzed by the authors [
20]. Besides its suitability for photonic beam steering, UHRR THz TDS has the advantage that the light source has the potential for monolithic integration, is electrically pumped, supplies an optical power of around 100 mW without amplification, and operates at 1550 nm telecom wavelengths [
21,
22]. Another practical advantage of the ultra-high repetition rate is the fact that identical pulses are detected every few picoseconds. This alleviates the need for accurate matching of the transmit and receive path lengths and allows the use of a much shorter ODU in the receiver arm. Particularly in reflection mode, where the distance to the target may not be known, this is highly beneficial. The disadvantage of UHRR THz TDS is its comparatively low bandwidth of typically less than 2 THz and poor spectral resolution of a few tens of GHz [
19,
20,
23]. The bandwidth is primarily determined by the optical bandwidth of the MLLD and the low-pass behavior of the terahertz components. The spectral resolution is equal to the repetition rate of the MLLD.