5
Mid‐IR by Nonlinear Optical Frequency Conversion

This chapter reviews devices based on nonlinear optical (NLO) frequency downconversion that operate in the 2–20 μm mid‐IR region. It discusses the principles of nonlinear frequency conversion, gives a brief assessment of existing and emerging NLO materials suitable for mid‐IR, and reviews current approaches for obtaining tunable mid‐IR output. We start first with the techniques that involve three‐wave processes, namely those based on the second‐order nonlinearity χ(2).

5.1 Two Approaches to Frequency Downconversion Using Second‐order Nonlinearity

There are two basic techniques (Figure 5.1) for achieving broadly tunable mid‐IR output via frequency downconversion in a three‐wave process that relies upon second‐order nonlinearity χ(2) of optical materials, namely (i) using difference frequency generation (DFG) and (ii) using optical parametric oscillators (OPOs). The OPO variations include traveling‐wave optical parametric generators (OPGs) and optical parametric amplifiers (OPAs).

While both DFG and OPO are based on three‐photon processes, the main distinction between them is that in the former case one needs two pump sources (and at least one of them has to be tunable to achieve mid‐IR tunability) while in the latter case one uses just one pump laser source (and a “seed” wave in the case of an OPA).

In both DFG and OPO processes, the photon energy conservation dictates that

(5.1)equation

while the photon momentum conservation (also referred to as phase matching) requires that

(5.2)equation

Here, ω3, ω2, and ω1 are the so‐called pump, signal, and idler angular frequencies, such that ω3 > ω2 > ω1, and k3, k2, and k1 are, correspondingly, their wave vectors (such that their modules are |ki| = ωini/c, where i = 1, 2, 3 and ni is a refractive index).

Schematic diagram illustrating the two main techniques, DFG (top) and OPO (bottom), for using second-order nonlinearity of the materials for achieving tunable mid-IR output.

Figure 5.1 DFG and OPO – the two main techniques for using second‐order nonlinearity of the materials for achieving tunable mid‐IR output.

The momentum conservation condition is never satisfied in isotropic media (at least in the transparency range, where normal dispersion applies, i.e. refractive index n increases with the optical frequency ω). However, it can be fulfilled in birefringent crystals where one can use orthogonal polarizations so that the wavelength dispersion is compensated by polarization dispersion (so‐called birefringent or angular phase matching).

Alternatively, the photon momentum conservation can be satisfied in so‐called quasi‐phase‐matched (QPM) crystals. Quasi‐phase‐matching is obtained by periodic inversion of the crystalline orientation along the optical path. This periodicity of crystal orientation, and associated reversal of the sign of the optical nonlinearity, compensates for the wave‐vector mismatch because the relative phase is corrected at regular intervals [1].

The period of QPM grating Λ inside the crystal (Figure 5.2) should be such that

(5.3)equation

In this case, an additional k‐vector associated with orientation reversal periodicity (2π/Λ) compensates the mismatch between participating waves Δk = k3k2k1.

Quasi‐phase‐matching can be illustrated as follows: suppose one needs to drive a car from point A to point B, but the wheels of the car cannot be directed straight, but rather at +45° or −45°. In this case, the strategy is as follows: one starts going in the direction of 90° to the target and when the car makes a semicircle, steers the wheels from −45° to +45° position (or vice versa). The average travel distance will be increased by a factor of π/2, as compared to the straight line, the gas mileage will be reduced correspondingly by 2/π and that is exactly the factor by which the effective nonlinear coefficient is reduced in the QPM scenario, as compared to the true phase‐matched case. The key point is to flip the steering wheel at the right moment. Similarly, in the case of QPM crystals, the period of flipping the crystal orientation Λ needs to be precisely adjusted for a given three‐wave interaction.

Left: Schematic, line-angle diagram, and waveforms illustrating quasi-phase-matching. Right: Diagram displaying a wavy curve representing the path of a car from point A to point B.

Figure 5.2 Quasi‐phase‐matching. (a) In quasi‐phase‐matched (QPM) crystals with periodically flipped sign of the nonlinearity, the artificially created grating of crystal orientation reversal compensates for the wave‐vector mismatch. (b) Quasi‐phase‐matching is similar to the situation when one needs to drive a car from point A to point B, but the wheels of the car cannot be put straight, but rather at either +45° or at −45°. In this case, the strategy is to start going in the direction of 90° to the target and each time the car makes a semicircle, flip the wheels from +45° to −45°, or vice versa.

Unlike birefringent phase matching, quasi‐phase‐matching allows (i) to access the highest available nonlinear coefficient, like d33 in lithium niobate, and (ii) eliminates spatial (birefringent) walk‐off of the beams with orthogonal polarizations. This allows tight, so‐called noncritical, beam focusing. A typical example of QPM crystal is periodically poled lithium niobate (PPLN).

5.1.1 Difference Frequency Generation

DFG allows achieving compact mid‐IR sources based on frequency mixing the outputs of a pair of well‐developed near‐IR lasers. DFG is a cavityless, single‐pass wavelength conversion scheme. It only requires the input laser beams to be spatially and temporally overlapped within the nonlinear crystal. For example, lithium niobate (LiNbO3) is a low‐cost robust material, and periodic poling allows it to be tailored to phase‐match practically any three‐wave DFG interaction, limited only by the LiNbO3 absorption edge near λ ≈ 5 μm [2]. Both waveguide and bulk interaction geometries in PPLN have been used to mix near‐IR wavelengths to the mid‐IR. While DFG conversion efficiencies are higher in waveguides, bulk materials can handle higher power levels, are easier to fabricate, and are far less sensitive to the beam alignment. The DFG source tuning is achieved by changing the wavelength of either one of the near‐IR lasers, or both of them.

In the limit of small conversion efficiency and perfect phase matching between the interacting waves, the DFG power P1 is expressed by the following product:

where P2 and P3 are the powers of the two pump lasers (the one with a shorter wavelength, P3, is called the pump and another, P2, is called the signal). Thus, P1 is linear with respect to both P2 and P3 and one can introduce a normalized conversion efficiency η (in % per W) that depends on the effective nonlinear coefficient (deff), the output mid‐IR frequency (ω1), focusing strength, and interaction length (L) in the following way [3]:

(5.5)equation

Here, n is the average refractive index, and “area” is the effective cross section – a measure of how tightly the interacting beams are focused inside the crystal (the effective cross section can be very small in waveguides). The important term that determines DFG efficiency, images is referred to as the nonlinear optical figure of merit (NLO FOM) of a crystal. One can also see that the DFG conversion efficiency drastically drops on increasing the mid‐IR wavelength, due to the images term.

5.1.2 Optical Parametric Oscillators (OPOs)

Optical parametric oscillators offer extremely wide tunability, intrinsically limited only by the material transparency, require only a single‐pump laser, and typically have very high conversion efficiency from the pump, much larger than that of DFG. OPOs and other parametric devices such as OPGs and OPAs are the sources of choice when one needs the broadest continuous tunability (up to two or three octaves in frequency), high peak (>1 kW) or average (>1 W) power, and high (>50%) quantum conversion efficiency. This comes at a cost of the need for a resonant cavity (typically resonant for the signal wave), or having a high peak power (>1 MW) pump in the case of single‐pass devices such as OPG or OPA. Also, OPOs have a distinct oscillation threshold, while DFG devices do not (one always gets an output from a DFG system).

Parametric frequency downconversion in an OPO can be regarded as the inverse process of sum‐frequency generation with an NLO crystal that can be viewed as a catalyst that promotes decay of the pump photon into two smaller photons. The OPO tuning is achieved by changing the phase‐matching condition that in turn changes the ratio between the signal and idler photon energies, and thus tunes the output frequency. This can be accomplished by:

  • rotating the crystal (for birefringent NLO materials)
  • changing the inversion period (for QPM crystals)
  • changing the temperature of the crystal
  • tuning the pump wavelength.

Figure 5.3 depicts four basic types of optical parametric devices, namely (i) continuous wave (CW), (ii) pulsed nanosecond, (iii) synchronously pumped ultrafast, and (iv) traveling‐wave device such as OPG and OPA. Since parametric gain depends on the instantaneous intensity (power density) of the pump, CW OPOs (Figure 5.3a) require a high average power (typically >1 W) to reach the threshold; also, since in CW OPOs the gain is small, one requires a low‐loss cavity. In contrast, with nanosecond pumping (Figure 5.3b), the requirements for cavity loss are much less demanding (e.g. a simple “plano‐plano” short cavity is sufficient); the OPO threshold in this case can be achieved with modest pump pulse energies, in the microjoule to millijoule range, with correspondingly modest average pump powers [4]. With pico‐ or femtosecond pumping (Figure 5.3c), it is necessary to match the OPO roundtrip time to the repetition period of a pump laser – a synchronous pumping (or sync‐pumping). In this case, the OPO pulse after a full roundtrip will be amplified again by the next incoming pump pulse. Due to the short duration of mode‐locked pump pulses, the OPO threshold can be low in terms of the average power, around 100 mW or less. Finally, OPAs and OPGs are single‐pass devices that do not require a cavity (Figure 5.3d). As a pump, they typically use short‐duration (ps‐ or fs‐range) intense pulses with the input pump power density of at least 100 MW/cm2. The difference between an OPA and an OPG is that, in the former case, there is a weak “seed” pulse (characteristically at the signal wave frequency ω2) that is injected along with a strong pump at ω3, while in the OPG it is the quantum noise that serves as a seed for parametric amplification. In both cases, a single‐pass gain in an NLO crystal is very high (~106 or more, an exponential process) and a substantial fraction (>10%) of the pump is converted into the signal and the idler waves.

For perfect phase matching between the interacting waves and in the limit of no pump depletion, a single‐pass parametric gain for a seed pulse Pin (at either signal or idler wave) in the presence of a pump field at ω3 is expressed [3] as

(5.6)equation

where L is the length of a nonlinear crystal and Γ is the gain increment given by:

Here, Ipump is the pump laser intensity (power density), ω1 and ω2 are idler and signal frequencies, λ1 and λ2 are corresponding wavelengths (such that ωi = 2πc/λi), deff is the effective nonlinear coefficient, and n is the average refractive index. Again, the important term, which determines the gain, is the NLO FOM images.

Image described by caption and surrounding text.

Figure 5.3 Four basic types of optical parametric devices. (a) Continuous‐wave OPO. (b) Pulsed nanosecond OPO. (c) Sync‐pumped ultrafast OPO. (d) Traveling‐wave device such as OPG and OPA.

In fact, a single‐pass parametric gain varies drastically with pump intensity. For example, in the low‐gain limit (ΓL ≪̸ 1),

so that the incremental increase in power is

and is proportional to the pump power density. On the other hand, in the high‐gain limit (ΓL ≫̸ 1), the cosh function is approximated by an exponent and

so that G grows exponentially with the pump field.

As a numerical example, consider PPLN with an effective NLO coefficient deff = 14 pm/V suitable for mid‐IR range [5], with λ3 = 1.06 μm (pump wavelength), λ2 = 1.57 μm (signal), and λ1 = 3.3 μm (idler). At a CW pump power of 10 W, a beam diameter of 100 μm (assume for simplicity a flat‐top uniform intensity distribution for all three waves), which corresponds to a pump power density of 0.13 MW/cm2, and for a 30‐mm‐long crystal, we get from (5.8) to (5.9) an incremental increase in power of only 13% per single pass. For ultrafast pump pulses, though, the situation drastically changes. At 1‐ps pulse duration, 100‐μm flat‐top beam diameter, at only 5‐mm‐long crystal, and a modest pulse energy of 1 μJ, corresponding to the pump power density of 13 GW/cm2, a single‐pass parametric gain reaches G = 1.7·1016.

5.1.3 Brief Review of χ(2) Nonlinear Crystals for Mid‐IR

Table 5.1 compares linear and NLO properties of selected χ(2) NLO crystals that are most suitable for mid‐IR applications. The crystals are divided into three categories in this table:

  • Periodically poled (PP) QPM oxides, such as lithium niobate (LiNbO3) or potassium titanyl phosphate (KTiOPO4).
  • Birefringent crystals, some of which have been known for quite a long time (e.g. AGS, AGSe, ZGP, CGA, CdSe, and GaSe) [12], while the others have been developed quite recently, such as cadmium silicon phosphide (CSP).
  • Newly developed QPM semiconductors with periodic inversion of crystalline orientation, such as orientation‐patterned gallium arsenide (OP‐GaAs), gallium phosphide (OP‐GaP), zinc selenide (OP‐ZnSe), and gallium nitride (OP‐GaN).

5.1.3.1 Periodically Poled Oxides

PP oxides achieved their maturity in the mid‐1990s and are widely used because of their comparatively low cost and high nonlinear coefficient d33. The advent of the first practical QPM material – PPLN – with the successful implementation of quasi‐phase‐matching by periodic inversion of its ferroelectric domains, represented a new paradigm in NLO materials development [1, 17]. No longer constrained by the strict requirements of birefringence phase matching, QPM enabled the use of a much broader range of materials, provided that a mechanism existed for creating the alternating QPM domain structure. In ferroelectric oxides, this was achieved by applying lithographic electrodes, followed by electric field poling of periodic domains of alternating polarity. QPM offered the advantages of noncritical phase matching, polarization flexibility, and engineerable functionality.

Potassium titanyl phosphate, KTiOPO4 (KTP), and its isomorphs, belong to another class of crystals that are quite attractive for periodic poling [7]. KTP possesses high laser damage threshold, and has low susceptibility for the so‐called photorefractive effect. (The photorefractive effect is an unwanted altering of the refractive index in the crystal caused by parasitically generated visible light, characteristic of PPLN and PPLT.) For KTP, the nonlinear coefficient d33 is about 2/3 of that of LiNbO3. Periodic poling of the isomorphs of KTP like KTiOAsO4 (KTA) and RbTiOAsO4 (RTA) has also been successfully implemented. The KTA and RTA crystals have been widely used in mid‐IR OPOs, primarily because of their slightly wider infrared transmission.

Generally, ferroelectric NLO oxide crystals such as lithium niobate, lithium tantalite, or KTP and its analogs limit the longwave operation to 4–5 μm due to the onset of multi‐phonon absorption.

5.1.3.2 Birefringent Crystals

Among birefringent NLO crystals for mid‐IR listed in Table 5.1, ZGP is currently the most robust and is the material of choice for 2‐μm‐pumped optical parametric oscillators. ZGP has a very high nonlinear coefficient of 75 pm/V (see Table 5.1) and the highest thermal conductivity (35 W/mK) of any bulk birefringent crystal that is phase‐matchable for 2‐μm pump. ZGP allows achieving OPO tunability from 2.5 to >10 μm. Improved polishing and antireflection coatings enable laser damage thresholds in excess of 4 J/cm2 for pulsed (20 ns, 10 kHz) illumination. Very high mid‐IR (3–5 μm) output powers in excess of 30 W have been achieved with ZGP in the pulse‐periodic mode [18].

CdSiP2 (CSP) is a new NLO crystal suitable for laser frequency conversion in the 1–7 μm spectral range [19]. The material is particularly promising due to its high nonlinear coefficient of 84.5 pm/V [9], which is the largest of any new phase‐matchable inorganic crystal grown in the past four decades. Most importantly, with its bandgap of 2.45 eV, CSP allows 1.064‐μm OPO pumping without the onset of two‐photon absorption. Its thermal conductivity (13.6 W/mK) is higher than that of existing longwave infrared NLO crystals suitable for 1‐μm pump, most notable AgGaS2 that has thermal conductivity of only 1.4 W/mK.

Table 5.1 Linear and nonlinear optical properties of selected second‐order nonlinear crystals suitable for mid‐IR applications.

Crystal Transparency range (μm) deff (pm/V)a Ave. ref. index NLO FOM images with respect to PPLN Ref.
Periodically poled oxides
PP LN (LiNbO3) 0.4–5.5 (2/π)d33 = (2/π)·22 = 14 2.12 1 [5]
PP LT (LiTaO3) 0.35–4.5 (2/π)d33 = (2/π)·10.7 = 6.8 2.11 0.24 [6]
PP KTP (KTiOPO4) 0.35–4.3 (2/π)d33 = (2/π)·16.9 = 10.8 1.8 1.0 [7]
PP KTA (KTiOAsO4) 0.35–5.3 (2/π)d33 = (2/π)·16.2 = 10.3 1.8 0.9 [7]
PP RTA (RbTiOAsO4) 0.35–5.3 (2/π)d33 = (2/π)·15.8 = 10.1 1.8 0.9 [7]
Birefringent
AGS (AgGaS2) 0.47–13 d36 = 12 2.4 0.5 [8]
AGSe (AgGaSe2) 0.71–19 d36 = 33 2.65 2.8 [8]
CSP (CdSiP2) 0.65–7 d36 = 84.5 3.05 12.2 [9]
LIS (LiInS2) 0.4–12 d31 = 7.2; d24 = 5.9 2.1 0.23 [10]
LGS (LiGaS2) 0.33–11.6 d31 = 5.7; d24 = 5.2 2.1 0.16 [11]
CdSe 0.75–25 d31 = 18 2.46 1.1 [8]
ZGP (ZnGeP2) 0.74–12 d36 = 75 3.13 8.9 [12]
GaSe 0.65–20 d22 = 54 2.73 7 [12]
CGA (CdGeAs2) 2.4–18 d36 = 236 3.6 65 [12]
Orientation patterned, cubic
OP‐GaAs 0.9–17 (2/π) images d14 = (2/π) images·94 = 69.1 3.3 6.5 [13]
OP‐GaP 0.57–13 (2/π) images d14 = (2/π) images·37 = 27.2 3.05 1.27 [6]
OP‐GaP 0.57–13 (2/π) images d14 = (2/π) images·35 = 25.7 3.05 1.13 b
OP‐ZnSe 0.55–20 (2/π) images d14 = (2/π) images ·30 = 22.1 2.44 1.6 [14] c
OP‐ZnSe 0.55–20 (2/π) images d14 = (2/π) images ·20 = 14.7 2.44 0.7 d
Orientation patterned, hexagonal
OP‐GaN 0.37–7 (2/π)d33 = (2/π)·16.5 = 12.1 2.3 0.6 [15]

a The (2/π) reduction factor accounts for the effect of quasi‐phase‐matching [1]. For cubic‐symmetry crystals (e.g. GaAs), there is an additional images factor, which appears when deff is maximized (for example, when all polarizations are along <111>) and deff = images d14. Listed here are nonlinear coefficients that are best suited for mid‐IR. They might be lower than those for the visible and near‐IR because of the dispersion [16].

b Mid‐IR d14 coefficient of GaP was measured in my group at University of Central Florida, to be 35 ± 2 pm/V, via direct comparison of SHG efficiency between quasi‐phase‐matched OP‐GaP and OP‐GaAs crystals in a frequency‐doubling process using nanosecond pulses at λ = 4.7 μm, and referencing to the known d14 of GaAs (94 pm/V).

c d14 of ZnSe was measured in this work via SHG of 1.3 μm.

d Mid‐IR d14 coefficient of ZnSe was measured in my group at University of Central Florida, to be 20 ± 2 pm/V, via direct comparison of SHG efficiency between a single‐coherence‐length (~100 μm) 110‐cut ZnSe and a single‐coherence‐length (~45 μm) 110‐cut GaAs in a frequency‐doubling process using nanosecond pulses at λ = 4.7 μm, and referencing to the known d14 of GaAs (94 pm/V).

Gallium selenide (GaSe) deserves a special place in the family of birefringent NLO crystals. It is a two‐dimensional III–VI semiconductor with layered structure and weak (van der Waals) interlayer coupling. Although the crystal's mechanical softness makes it difficult to cut and polish it along arbitrary directions (GaSe can only be cleaved perpendicular to its c‐axis, along 001 planes), the crystal has unique advantages including: enormous birefringence (Δn ~ 0.3) that makes it phase‐matchable for numerous three‐wave processes (including THz generation), broad transparency range (0.65–20 μm) with extremely low optical losses (<0.1 cm−1 at 1–15 μm), large NLO coefficient (54 pm/V), and large bandgap (2 eV) that minimizes two‐photon absorption at λ > 1.25 μm. Here is a tribute to GaSe crystal – a list of the most prominent NLO applications of GaSe in the mid‐IR as well as in the longwave IR (terahertz) spectral regions:

  • broadband frequency doubling of 6–12 μm radiation [20]
  • DFG in the CW regime (8.8–15 μm) [21]
  • DFG with picosecond pulses (4–18 μm) [22]
  • DFG with femtosecond pulses (3–20 μm) [23]
  • optical parametric generation (OPG) with continuous tunability from 3 to 19 μm [24]
  • generation of frequency combs in the 4–17 μm spectral range via DFG [2527]
  • generation of single‐cycle electromagnetic transients (center wavelength from 7 to 3000 μm) via optical rectification [28, 29]
  • generation of broadly tunable (0.2–5.3 THz) terahertz radiation by nanosecond DFG [30]
  • high (up to 23rd order) harmonic generation from 10‐μm mid‐IR pulses [31]
  • coherent electro‐optic detection of mid‐IR and THz waves [32]

5.1.3.3 Emerging QPM Nonlinear Optical Materials

Zinc blende semiconductors such as GaAs and GaP are particularly appealing for mid‐IR nonlinear‐optical frequency conversion – due to their deep infrared transparency, high second‐order nonlinearity, superior thermal conductivity, and high surface damage threshold. Their cubic crystal structure lacks birefringence; therefore, they require spatial modulation of the optical nonlinearity via quasi‐phase‐matching. Zinc‐blende‐type semiconductors are not ferroelectric and no techniques analogous to electric field poling in LiNbO3 exist for inducing domain flipping in already grown crystal.

OP‐GaAs was the first realization of a practical QPM semiconductor. This was done using all‐epitaxial processing pioneered independently at Stanford University [33] and at the University of Tokyo [34]. The present approach for making OP‐GaAs relies on polar‐on‐nonpolar molecular beam epitaxy (MBE) whereby a thin Ge layer is deposited on a GaAs substrate and the subsequent GaAs layer – under the proper growth conditions – has an inverted orientation relative to the substrate. This layer is then lithographically patterned, etched back to the original substrate, and then simultaneously regrown with the opposite substrate polarity – first by MBE and then by high‐growth‐rate hydride vapor phase epitaxy (HVPE) – to produce thick (>500 μm) QPM structures suitable for bulk NLO applications [18,3537]. As we will see further in this chapter, numerous mid‐IR devices based on the OP‐GaAs have been demonstrated from CW to femtosecond in the time domain, and from mid‐IR to THz in frequency.

Although the OP‐GaAs exhibited outstanding performance with 2 μm pumping, it cannot be pumped at wavelengths below 1.8 μm due to the onset of two‐photon absorption and the resulting accumulation of free carriers. By contrast, gallium phosphide (GaP) – another III–V zinc blende semiconductor – has wider bandgap and small two‐photon absorption at 1 μm. The growth and processing of OP‐GaP was a direct extension of the OP‐GaAs process, where phosphorus was substituted for arsenic and silicon for germanium as the lattice‐matched nonpolar layer. As a result, the usable OP‐GaP samples with up to 1‐mm QPM layer thickness were grown at BAE Systems and a number of mid‐IR devices ranging from CW to ultrafast were demonstrated [18].

5.1.3.3.1 Zinc Selenide as an NLO Candidate

ZnSe, a II–VI semiconductor with zinc blende symmetry, is an excellent candidate for optical frequency conversion because of its high (~20 pm/V) nonlinear susceptibility, high optical damage threshold, good mechanical properties, and outstanding transparency range, from 0.55 to 20 μm. Kanner et al. reported the first results on the growth of QPM OP‐ZnSe [38]. In order to achieve orientation patterning, ZnSe films were grown on patterned GaAs templates. Because ZnSe has the same (zinc blende) crystal structure as GaAs, and is almost lattice‐matched to GaAs, it became possible to grow thick (>750 μm) films of ZnSe on top of OP‐GaAs template and preserve its orientation patterning. Initial results on frequency doubling of 1.6‐μm radiation as well as CO2 laser radiation, and DFG near 9 μm have been demonstrated [38].

5.1.3.3.2 Gallium Nitride as an NLO Candidate

A III–V semiconductor GaN has a wurtzite crystal structure. Its diagonal nonlinear‐optical coefficient (d33 ~ 16 pm/V) is of a similar magnitude to that of PPLN, but GaN has broader transparency range, from 0.37 to 7 μm. Because of its uniquely high thermal conductivity (220 W/mK), high laser damage threshold, wide bandgap (3.4 eV), and thus the ability to use standard 1‐μm lasers as a pump, it is well suited for high‐power mid‐IR NLO applications [15, 39, 40]. Although GaN has not yet reached maturity as an NLO material, several groups have demonstrated periodic inversion of the GaN polarity to achieve quasi‐phase‐matching (with GaN grown on both sapphire and GaN substrates) [4144]. Also, second harmonic generation from 1.66 μm was demonstrated in QPM GaN [41].

Petrov [11] published a detailed overview of the available non‐oxide NLO materials, emphasizing most recent developments of both birefringent (angle‐tuned) and QPM crystals. Also, a review paper by Schunemann et al. [18] highlights advances in the growth of the birefringent crystals ZnGeP2 and CdSiP2, as well as all‐epitaxial growth of orientation‐patterned semiconductors gallium arsenide (OP‐GaAs) and gallium phosphide (OP‐GaP).

5.2 Continuous‐wave (CW) Regime

5.2.1 DFG of CW Radiation

Specific advantages of the DFG approach for generating mid‐IR include: readily available room‐temperature single‐frequency near‐IR “pump” and “signal” laser sources, no need for cryogenic cooling, and high mid‐IR beam quality. For example, the outputs of well‐developed telecom‐range narrow‐linewidth diode‐ or fiber‐lasers can be fiber‐coupled and mixed in a nonlinear crystal. In addition, coherence properties of the DFG output are inherited from those of the pump lasers. Therefore, narrow‐linewidth (1 kHz to 1 MHz) mid‐IR output can be produced via DFG using appropriate pump and signal inputs.

The best DFG results, in terms of efficiency, are usually obtained in QPM crystals. In the early work on using PP crystals for frequency conversion, Sanders et al. [45] demonstrated a broadly tunable mid‐IR source based on mixing the outputs of two wavelength‐tunable single‐spatial‐mode laser diodes at around 780 and 980 nm in bulk PPLN crystal fabricated by electric field poling of a 0.5‐mm thick z‐cut LiNbO3 wafer. Coherent mid‐IR radiation was generated in a 7.8‐mm‐long crystal over the range 3.6–4.3 μm with mid‐IR power of 6 μW. A more efficient PPLN DFG system (Figure 5.4) was based upon mixing the outputs of a distributed feedback (DFB) diode laser at 1562 nm amplified in an erbium fiber amplifier (EFA) to increase the optical power to 500 mW (with 360‐kHz linewidth) and a DFB fiber laser at 1083 nm amplified in an ytterbium fiber amplifier (YFA) to raise the power to 800 mW (100‐kHz linewidth). Tunable narrow‐linewidth output with 0.4‐mW power was demonstrated near 3.5‐μm wavelength, corresponding to DFG normalized conversion efficiency η = 0.1%/W [46, 47]. Also, the authors performed sensitive and selective spectroscopic detection of formaldehyde with this DFG source near 3.53 μm.

Image described by caption.

Figure 5.4 Experimental setup for the 3.5‐μm DFG source based on mixing the amplified outputs of two DFB lasers in PPLN. DL, diode laser; ISO, optical isolator; PC, polarization controller; WDM, wavelength division multiplexer.

Source: based on figure 3 of [46], with permission of OSA, The Optical Society.

As much as 3.5 W of CW power at 3.4 μm (linewidth ~1.5 nm) was obtained in a 5‐cm long PPLN crystal by difference frequency mixing of 1.064 and 1.55 μm linearly polarized fiber lasers, with correspondingly 43.3 and 31 W of CW power [48]. This is one of the highest CW DFG powers demonstrated so far.

In an alternative DFG approach utilizing a ridge‐type PPLN waveguide, the conversion efficiency can be more than 100 times higher than in bulk PPLN crystals, resulting in tens of milliwatts of DFG power [4952]. This improvement comes from the fact that the beam size is no longer limited by diffraction and is kept at a very small size over the whole length of the crystal. Using direct‐bonded PPLN with an integral ridge waveguide structure, an 11‐mW pump wave (1047 nm), and a 66‐mW signal wave (1550 nm), Tadanaga and coworkers at NTT corporation produced 0.26 mW of radiation at 3.3 μm with a normalized conversion efficiency of η = 40%/W [49]. Asobe et al. [52] achieved highly efficient DFG using a QPM Zn:LiNbO3 waveguide fabricated by direct bonding. The DFG performance was measured using a 1.064 mm pump generated with a laser diode and amplified in an ytterbium‐doped fiber amplifier (YDFA) and a signal generated by a 1.55‐μm external cavity tunable laser diode and an erbium‐doped fiber amplifier (EDFA). CW signal and pump powers of 558 and 444 mW, respectively, were injected into the ridge waveguide (11 μm thick, 17 μm wide, and 38 mm long) and a 65‐mW mid‐IR output was obtained at 3.4 μm, corresponding to the normalized conversion efficiency η = 26%/W. The reported DFG source could be tunable over 10 nm, determined by its phase‐matching bandwidth, by scanning the signal wavelength [52].

Image described by caption and surrounding text.

Figure 5.5 Schematic of the experimental DFG setup based on OP‐GaAs to produce 0.5‐mW 7.6–8.2 μm output.

Source: based on figure 1 of [55], with permission of OSA, The Optical Society.

Development of new QPM nonlinear crystals with deeper IR transparency, such as GaAs, potentially allows extending tunability range of DFG sources to λ > 10 μm. DFG of 8‐μm radiation using orientation‐patterned GaAs was first demonstrated by Levi et al. in a 19‐mm‐long OP‐GaAs crystal [53]. The pump source – a fiber‐coupled CW DFB laser diode (1.306–1.314 μm, 3.3 mW) – was mixed with a signal source – an external‐cavity diode laser (1.51–1.58 μm), amplified in an EDFA to 787 mW – to generate 7.9‐μm idler radiation with 38‐nW power. Later, the same team produced several microwatts of the DFG output in OP‐GaAs that was tunable in the broad range of 7–9 μm (span 300 cm−1) by mixing the outputs of tunable 1.3‐μm (80 mW) and fixed‐wavelength 1.55‐μm (2 W) lasers [54]. The authors also used the DFG system to perform cavity ring‐down spectroscopy of N2O gas. In terms of the average DFG power, the above result was improved by Vasilyev et al. [55], who generated a tunable (7.6–8.2 μm) single‐frequency DFG output with 0.5‐mW power from a 33 mm‐long OP‐GaAs crystal. The DFG source was pumped by a tunable single‐frequency external‐cavity diode laser, amplified by an EDFA (9 W at 1.55 μm), and a custom Tm‐doped fiber laser (0.5 W at 1.93 μm, Figure 5.5). The DFG output wavelength was tuned by simultaneous tuning of the diode laser wavelength and of the OP‐GaAs crystal temperature. The authors tested spectroscopic capabilities of their DFG source by measuring methane absorption spectra near 7.65 μm [55].

Thus, difference frequency mixing of telecom sources in QPM crystals such as PPLN and OP‐GaAs has been shown to be an effective method for the generation of broadly tunable spectroscopic‐grade output in the mid‐IR. The main results on CW DFG are summarized in Table 5.2.

5.2.2 CW OPOs

Due to a non‐dissipative nature of the parametric conversion process, optical parametric oscillators can be very efficient converters from the near‐IR to mid‐IR, mostly limited by the Stokes (quantum defect) limit – e.g. by the photon energy difference between the pump and the idler wave. The OPO gain originates from coherently driven nonlinear polarization, and its long‐wavelength tuning cutoff is set only by the transparency of the nonlinear medium (OPOs can produce an idler wave even in the THz range, below the band of phonon resonances [56]).

Table 5.2 Summary of CW DFG in QPM crystals.

DFG structure DFG wavelength (μm) Pump and signal DFG CW power and normalized efficiency Ref.
PPLN, bulk
PPLN, L = 7.8 mm 3.6–4.3 780 nm (180 mW)
980 nm (400 mW)
6 μW, η = 0.008%/W [45]
PPLN, L = 50 mm 3.5 1083 nm (800 mW)
1562 nm (500 mW)
0.4 mW, η = 0.1%/W [46, 47]
PPLN, L = 50 mm 3.4 1064.6 nm (43.3 W)
1549.8 nm (31 W)
3.55 W, η = 0.26%/W [48]
PPLN, waveguides
PPLN ridge WG, direct bonded, L = 50 mm 3.2 1047 nm (11 mW)
1550 nm (66 mW)
260 μW, η = 40%/W [49]
PPLN ridge WG, direct bonded, L = 50 mm 3.36 1064 nm (46 mW)
1558 nm (7 mW)
146 μW, η = 45%/W [51]
PPLN ridge WG, direct bonded, L = 50 mm 3.52 1083 nm (318 mW)
1562 nm (503 mW)
15 mW, η = 9.4%/W [50]
Zn:PPLN ridge WG, direct bonded, L = 38 mm 3.4 1064 nm (444 mW)
1550 nm (558 mW)
65 mW, η = 26%/W [52]
OP‐GaAs, bulk
OP‐GaAs, L = 19 mm 7.9 ~1300 nm (11 mW)
~1550 nm (66 mW)
38 nW, η = 0.005%/W [53]
OP‐GaAs, L = 19 mm 7–9 ~1300 nm (80 mW)
~1550 nm (2 W)
A few microwatts, η = 0.002%/W [54]
OP‐GaAs, L = 33 mm 7.6–8.2 ~1550 nm (9 W)
~1930 nm (0.5 W)
0.5 mW, η = 0.003%/W [55]

In an optical parametric oscillator, at least one of the two generated waves (signal or idler) resonates in a cavity. Typically, it is the shorter‐wavelength “signal” wave (this scenario is easier to implement from a practical point of view). The oscillation starts from quantum noise if the gain for the resonant field, provided by the incident pump light, overcomes the round‐trip loss. As compared to singly resonant OPOs (SROs), much lower pump thresholds can be achieved in doubly resonant (with signal and idler or signal and pump resonating) or even triply resonant (signal, idler, and pump resonating) OPOs. However, the latter two versions are more difficult to implement because of tight constrains, e.g. on the cavity length, imposed by the requirement for multiple resonances. The first operation of a CW OPO was demonstrated in a cavity that was resonant for both signal and idler waves [57].

Fabrication of PPLN in the early 1990s tremendously stimulated development of OPOs. In 1996, Bosenberg et al. demonstrated the first CW QPM OPO, which promptly revealed the exceptional performance of QPM materials. The OPO was resonant at a signal wave in a two‐mirror linear [58] or in a four‐mirror ring bow‐tie cavity configuration [59] and was based on a 50‐mm‐long PPLN crystal as a nonlinear medium. The pump source for this device was a 1064‐nm Nd:YAG laser with 13.5 W of available power. The oscillation threshold power was 2.9 W in a linear cavity and 3.6 W (pump power density ~55 kW/cm2) in a ring cavity (Figure 5.6a). The milestones of their research were (i) obtaining the idler output power of 3.55 W at λ = 3.25 μm, which corresponds to 80% of the quantum‐limited performance, (ii) observing 93% pump depletion (Figure 5.6b), (iii) achieving the OPO idler tuning over the 3.24–3.95 μm spectral range by using PPLN crystal with multiple grating periods, and (iv) maintaining single‐longitudinal‐mode oscillation in a ring cavity containing an intracavity etalon.

Image described by caption.

Figure 5.6 (a) Schematic of a ring‐cavity CW OPO based on PPLN crystal [59]. The two curved mirrors had 100‐mm radius of curvature, and the remaining two mirrors were flat. The PPLN crystal was 50 mm long and had a grating period of 29.75 μm. (b) Pump depletion and idler output versus pump input for the ring cavity operating at an idler wavelength of 3.25 μm.

Source: reproduced from figures 1 and 2 of [59], with permission of OSA, The Optical Society.

Kumar et al. demonstrated a high‐power CW optical parametric oscillator that was pumped by a single‐frequency ytterbium fiber laser (1064 nm, maximum power 28.6 W) and produced the total output power of 17.5 W, out of which 9.8 W were in the near‐IR “signal” at λs = 1.63 μm and 7.7 W in the “idler” wave at λi = 3.07 μm [60]. Figure 5.7 shows the experimental setup and how the signal/idler outputs change with the signal‐wave outcoupling, while the inset shows variation of the OPO threshold pump power. The maximum overall power of 17.5 W was reached at the optimal output coupling of ∼3.8%, corresponding to the OPO conversion efficiency of 61%. Under this condition the pump depletion reached 69.4%. In the absence of the output coupling for the signal wave, 8.6 W of the idler power was generated at λ = 3.06 μm, with pump‐to‐idler conversion efficiency of 30%, and pump threshold of 3.6 W. In the idler branch, the OPO was tunable from 2.8 to 3.2 μm by changing the PPLN temperature [60].

Image described by caption and surrounding text.

Figure 5.7 (a) Experimental setup for the Yb‐fiber‐laser‐pumped ring‐cavity PPLN OPO with a total (signal + idler) power of 17.5 W. FI, Faraday isolator; λ/2, half‐wave plate; PBS, polarizing beamsplitter; L, lens; M1–M4, dichroic mirrors. The mirror M4 was used as an output coupler (OC) for the resonating signal wave across 1.6–1.7 μm. (b) Variation of the extracted signal (1.63 μm) and idler (3.07 μm) power with OC transmission. Inset: Threshold pump power versus OC transmission.

Source: reproduced from figures 1 and 2 of [60], with permission of Springer.

The same research team demonstrated a high‐power CW OPO based on periodically poled lithium tantalate (PPLT), LiTaO3, crystal. Despite its twice smaller nonlinear coefficient, as compared to the lithium niobate (see Table 5.1), the crystal has an increased resistance to photorefractive damage and higher thermal conductivity, along with increased optical damage threshold, thus making it attractive for multi‐watt mid‐IR generation. The OPO pump source was a single‐frequency Yb fiber laser delivering up to 30 W at 1064 nm. A 30‐mm‐long MgO‐doped stoichiometric PP LiTaO3 (MgO:sPPLT) was used in this setup as a gain element [61]. The singly resonant OPO (SRO) was tunable from 3.03 to 3.46 μm and generated 5.5 W of CW power at 3.1–3.3 μm with the pump depletion up to 64%.

Henderson et al. reported a broadly tunable single‐frequency (<1 MHz linewidth) CW OPO [62]. The OPO had a four‐mirror ring cavity, utilized an 80‐mm‐long PPLN crystal, and was driven by a narrow‐linewidth (50 kHz) all‐fiber pump source at 1083 nm. The oscillation threshold as low as 780 mW was measured at λ = 2.8 μm idler wave; the maximum pump depletion was 85% at a pump power of 2.8 W. At this input level, 750 mW of idler was produced at λ = 2.8 μm in a near‐diffraction‐limited beam. The overall tunability was 2.65–3.2 μm achieved by changing the QPM period of PPLN containing multiple grating periods. By fine‐tuning only the pump wavelength, mode‐hope free tuning range of 60 GHz (Δλ ~ 2 nm) was achieved [62]. More results on broadly (>1000 nm) and continuously tunable mid‐IR single longitudinal mode (SLM) CW optical parametric oscillators based on PPLN can be found in [6366].

Several groups reported on widely tunable CW single‐frequency OPOs where the pump wave resonates, in addition to the OPO signal (or idler) wave. This pump‐enhanced SRO configuration is labeled as PE‐SRO [6770]. Rihan et al. demonstrated a PE‐SRO pumped by a single‐frequency Ti:sapphire laser delivering 760 mW at 795 nm with an octave‐wide tunability of the idler wave (1.7–3.5 μm) with 20–50 mW of the output power. Thanks to the pump enhancement, a threshold as low as 110 mW was achieved [71] (see Table 5.3).

Until recently, the long wavelength range of CW OPOs was limited to around 5 μm – a limitation set by multi‐phonon absorption in oxide crystals such as lithium niobate, lithium tantalate, or potassium titanyl phosphate [74]. As an attempt to extend this limit to longer wavelengths, Schunemann and coauthors employed OP‐GaAs as a nonlinear medium and made the first experimental demonstration of CW OPO operation in GaAs, as well as the first CW OPO pumped at a laser wavelength longer than 1.55 μm [72]. The OPO was emitting near degeneracy, at λs ~ 3.8 μm and λi ~ 4.7 μm. The OP‐GaAs sample, 40‐mm‐long, with 1.7‐mm‐thick QPM grating structure, and QPM period 63.5 μm, was grown using low‐pressure HVPE with the bulk absorption loss of 0.004 cm−1 at 2.4 μm. The OPO pump was a Ho:YAG oscillator–amplifier system at λ = 2.09 μm, pumped in turn by thulium fiber lasers. The OPO resonator was configured as a bow‐tie ring (Figure 5.8). The maximum total output power (signal + idler) was 5.3 W at 24.7 W of pump power. The OPO threshold was 11.5 W, and maximum conversion efficiency (signal + idler) of 23.6% occurred at 1.8 times the threshold [72]. The high oscillation threshold can be explained by at least two reasons: (i) the OPO threshold scales as the product of the signal and the idler wavelengths (see Eq. 5.7) and thus goes up when the pump wavelength is shifted to longer wavelengths, (ii) the bulk OP‐GaAs losses at resonating wavelengths (3.8 and 4.7 μm) might be higher than those measured at 2.4 μm.

Table 5.3 Summary of CW mid‐IR OPOs.

OPO crystal Pump laser Wavelength tuning (μm) Other parameters Ref.
PPLN L = 50 mm 1.064 μm 13.5 W 3.24–3.95 SRO, idler power 3.55 W at 3.25 μm, threshold 3.6 W (ring cavity), quantum efficiency 80%, pump depletion 93% [58, 59]
PPLN L = 50 mm 1.064 μm 28.6 W 2.8–3.2 SRO, idler power 8.6 W at 3.06 μm, threshold 3.6 W, pump‐to‐idler conversion 30%, pump depletion 79% [60]
PPLT L = 30 mm 1.064 μm 30 W 2.8–3.2 SRO, idler power 5.5 W at 3.1–3.3 μm, threshold 17.5 W, pump‐to‐idler conversion 18.5%, pump depletion 64% [61]
PPLN L = 80 mm 1.083 μm 2.8 W 2.65–3.2 SRO, single frequency; idler power 750 mW at 2.8 μm; threshold 780 mW; quantum efficiency 69%; pump depletion 85%; continuous‐mode‐hop‐free scans of 60 GHz [62]
PPLN L = 50 mm 775–860 nm 6 W 2.5–4.4 SRO, single frequency; idler power 800 mW; threshold 1.5 W; pump depletion 80%; continuous‐mode‐hop‐free scans of 40 GHz by tuning the pump wavelength [66]
PPLN L = 50 mm 795 nm 760 mW 1.7–3.5 PE‐SRO, single frequency; idler power 20–50 mW, threshold 110 mW [71]
OP‐GaAs L = 40 mm 2.09 μm 24.7 W 3.8 and 4.7 Doubly resonant, signal + idler power 5.3 W, quantum efficiency 23.6%, threshold 11.5 W [72]
AgGaSe2 whispering gallery resonator 1.57 μm
12 mW
2–8 Triply resonant, 800 μW at 2.5 μm, 10 μW at 8 μm; quantum efficiency 12%, threshold 0.5–2 mW [73]
Schematic of the bow‐tie CW OP‐GaAs OPO system. The telescope, isolator, chopper, Ho:YAG pump laser, etc. are labeled.

Figure 5.8 Schematic of the bow‐tie CW OP‐GaAs OPO system.

Source: reproduced from figure 1 of [72], with permission of OSA, The Optical Society.

A totally new approach to CW OPOs is based on whispering gallery resonators (WGRs) where a millimeter‐size monolithic cavity guides light via total internal reflection. With small round‐trip losses for all interacting waves, a WGR OPO is intrinsically triply resonant and thresholds down to the microwatt level in lithium niobate can be reached [7577]. Meisenheimer et al. demonstrated a CW OPO emitting mid‐IR light at wavelengths up to 8 μm in a device that was based on a 3.5‐mm‐diameter WGR made of silver gallium selenide (AgGaSe2) [73].

Employing AgGaSe2 crystal significantly extends the tuning range of existing OPOs into the mid‐IR (Table 5.3). The pump at λ = 1.57 μm in [73] was provided by a DFB 12‐mW laser diode and coupled into the resonator via an evanescent field using a silicon prism (inset, Figure 5.9). The spatial distribution of the light field in the WGR is characterized by three mode numbers (m, p, q), one of which (m) is the longitudinal mode number, and the other two represent the transverse radial (q), and polar (p) mode numbers. Energy and angular momentum conservation laws dictate a particular relationship between these mode numbers. Figure 5.9 shows the signal output power at λs = 2.54 μm versus pump power dependence, where the output power of 0.8 mW was achieved at 12 mW pump. The corresponding idler power at λi = 4.11 μm was estimated to be 0.5 mW. For different combinations of the polar mode numbers p and crystal temperature, OPO tuning (signal and idler) in the whole range of 2–8 μm was produced with the output power ranging from 800 μW (at 2 μm) to 10 μW (at 8 μm).

Table 5.3 summarizes the main results on CW OPOs. A very good review of the progress in the development of CW optical parametric oscillators (including whispering‐gallery OPO devices) can be found in [74].

Graph of signal output power versus pump power displaying ascending circle markers. An inset displays a microresonator and a Si prism.

Figure 5.9 Signal output power at λs = 2.54 μm (with the corresponding idler wave at λi = 4.11 μm) versus pump power for the WGR pump‐mode with q‐number of 7. The inset shows a microresonator and an Si prism.

Source: reproduced from figure 3 of [73], with permission of OSA, The Optical Society.

5.3 Pulsed Regime

5.3.1 Pulsed DFG

In the pulsed regime, DFG conversion efficiency can be dramatically scaled up. As follows from (5.4), the peak power of DFG output is proportional to the product of the peak powers of the two pump beams. Hence, at the same average power, one can get much higher DFG conversion efficiency, as compared to CW regime (the enhancement factor scales as the inverse of the duty factor of the pump). In the pulsed DFG approach, it is not only that the two pump beams need to perfectly overlap in space, but they also need to overlap in time.

Since the DFG process has no threshold and requirements for optical losses in a crystal are not very demanding, the great selection of nonlinear crystals listed in Table 5.4, such as PPLN, AGS, AGSe, GaSe, and CGA, combined with a variety of pump laser sources, could be used for frequency mixing to achieve mid‐IR output. For example, broadband difference frequency mixing in AGS crystal with nanosecond pulses and with wavelengths extending to λ > 11 μm has been demonstrated in a number of early works [7881]. Seymour et al. demonstrated that the long‐wavelength limit of DFG output could be extended to 18.3 μm – with detectable power even beyond the two‐phonon absorption band of AGS [79]. Also, DFG from λ = 2.7 to 38.4 μm – well beyond multi‐phonon infrared resonances – was reached in GaSe with nanosecond pump pulses near λ ≈ 1 μm [82]. Continuously tunable 6.8–20.1‐μm output has been achieved via pulsed DFG in CdGeAs2 – a crystal with the highest NLO coefficient (236 pm/V) among all χ(2) materials in practical use. For a pump, the authors used the signal (4–5 μm) and the idler (6.5–9.5 μm) outputs of a ZnGeP2 OPG [83] (see Table 5.4).

Table 5.4 Summary of pulsed DFG sources.

DFG crystal DFG wavelength (μm) Pump and signal DFG parameters Ref.
Broadly tunable, angular phase matched
AGS L = 1.5–1.7 mm 4.6–12 694 nm (1.4 mJ)
737–817 nm (130 μJ)
4 nJ @ 11 μm (10 ns duration) [78]
AGS L = 2.8 mm 5.5–18.3 539–658 nm (19 μJ)
555–747 nm (23 μJ)
16 nJ @ 11.8 μm (4 ns duration) [79]
AGS L = 10 mm 5–11 1.064 μm (120 mJ)
1.18–1.35 μm (~12 mJ)
2 μJ @ 6 μm (12 ns, 10 Hz) [80]
AGS L = 10 mm 5–12 1.76–2.01 μm (18 mJ)
2.26–2.7 μm (17 mJ)
96 μJ @ 7.5 μm (8 ns, 30 Hz) [81]
GaSe L = 20 mm 2.7–38.4 1.064 μm (6 mJ)
1.09–1.75 μm (3–5 mJ)
12 μJ @ 5.9 μm (5 ns, 10 Hz) [82]
CGA L = 5.8 mm 6.8–20.1 4–5 μm (5 μJ)
6.5–9.5 μm (1.5 μJ)
0.3 μJ @ 10–13 μm (0.1 ns, 3 Hz) [83]
High energy per pulse
KTA L = 15 mm 3–5.3 785–886 nm (70 mJ) 1.064 μm (170 mJ) 0.5 mJ @ 4.4 μm, 0.3 mJ @ 5 μm (2 ns, 10 Hz) [84]
KTA six crystals in series, total L = 60 mm 3.14–4.81 1.065 μm (50 mJ) 1.37–1.61 μm (4 mJ) 1 mJ @ 3.5 μm, 0.4 mJ @ 4.5 μm, (2 ns, 20 Hz) [85]
High repetition rate
PPLN L = 50 mm 3.52 1.064 μm (19 μJ, 7.7 W ave.) 1.525 μm (10 mW CW) 1 W ave. power (2.5 ns, 400 kHz), 1 GHz linewidth, conv. eff. from pump 13.4% [86]
PPLN L = 50 mm 3.2–5.7 1.064 μm (30 μJ, 252 mW ave.) 1.5–1.6 μm (8 mW CW) 0.2–14 mW ave. (6 ns, 8.4 kHz), linewidth 154 MHz @ 2 mW, 195 GHz @ 14 mW [87]

High, millijoule‐level, mid‐IR DFG pulse energies could be obtained at low (~10 Hz) repetition rates. For example, Kung produced continuously tunable mid‐IR radiation from 3.0 to 5.3 μm by difference frequency mixing in a KTA crystal using amplified Ti:sapphire laser at 785–886 nm (pulse energy 60–70 mJ) and an Nd:YAG laser at 1064 nm (pulse energy 170 mJ) at 10 Hz repetition rate [84]. An energy of over 0.5 mJ in a 2‐ns pulse (30‐GHz linewidth) was achieved at λ = 4.4 μm, providing peak power of 250 kW. Similarly, Miyamoto et al. reported on the generation of nanosecond mid‐IR pulses with high energy (0.4–1 mJ/pulse), large tunability (from 3.1 to 4.8 μm), and narrow linewidth (1.4 GHz) using DFG in KTA. Five or six KTA crystals in series with the total length of 50–60 mm were used in this work [85]. DFG was obtained by mixing 50‐mJ pump pulses at 1065 nm with wavelength‐tunable (1368–1611 nm) 4‐mJ signal pulses. A narrow linewidth and good frequency reproducibility of the DFG output were confirmed by observing a ro‐vibrational absorption line of CO gas at 4.587 μm.

For DFG at high (>1 kHz) repetition rates in the 2–5 μm range – a region suitable for molecular spectroscopy – PPLN is one of the most appropriate materials due to its high conversion efficiency and commercial availability of crystals that can exceed 50 mm in length. Belden et al. generated >1 W of average DFG power in a narrow‐linewidth output at 3.52 μm using a 5 cm‐long PPLN crystal [86]. The pump was an all‐fiber laser source, producing 2.5‐ns pulses at 1064 nm at 400 kHz repetition rate, and 7.7 W average power, while the “signal” beam was from a fiber‐coupled DFB diode laser emitting 10 mW of CW power at 1525 nm. One should notice that the peak pump power was about million times stronger than the “signal” and the DFG operation was closer to the regime, characteristic of the OPA – with the “signal” beam serving as a seed. The threshold for the onset of the exponential behavior corresponding to (5.10) was at 2.5 W of pump power. Overall, the 1.064–3.52‐μm power conversion efficiency was as high as 13.4% (quantum efficiency 44%).

A 3–8 kHz repetition rate DFG system based on PPLN crystal with the average power of 0.2–14 mW designed for gas sensing via photoacoustic spectroscopy in the wavelength range 3.2–3.7 μm is described in [87]. The pump laser was a diode‐pumped pulsed high peak power passively Q‐switched Nd:YAG laser in a nonplanar ring oscillator (NPRO) configuration. This laser set the pulse repetition rate, pulse duration, and the linewidth (154 MHz) for the generated DFG output. The “signal” was a CW external cavity diode laser tunable between 1500 and 1600 nm, which set the DFG tuning range. As in [86], this DFG source was based on a combination of a pulsed pump laser and a CW signal laser. In this configuration, the exponential OPA process was considerably suppressed at low (<2 mW) output DFG power. At the higher average power (up to 14 mW), the OPA regime became dominant and the linewidth was broadened by more than 1000 times, to 195 GHz [87].

Table 5.4 gives a summary of pulsed DFG sources.

The reader can find more information on pulsed DFG in the reviews [11, 87].

5.3.2 Pulsed OPOs

To achieve OPO action with nanosecond (ns) pump pulses, a simple short‐length linear cavity consisting of two flat mirrors might be sufficient. The threshold of a pulsed SRO (resonating at either signal wave or idler wave) is determined by the two main factors:

  • Parametric gain should compensate the round‐trip loss in the cavity at the resonating wave.
  • Parametric gain should be high enough to overcome the so‐called “buildup” loss. The latter is related to the fact that for an ns pump, the intracavity OPO power needs to build up in the OPO cavity in a limited amount of time (typically 10–100 round trips) – from quantum noise to a detectable value [88].

Interestingly, the loss in the cavity plays a secondary role in determining the threshold of a pulsed OPO. In fact, the equivalent quantum noise “seed” power of an OPO is given by Yariv [3] and is expressed as the following product: photon energy × bandwidth. For a wavelength of ~3 μm (photon energy = 6.6·10−20 J) and an OPO bandwidth of ~1 cm−1 (3·1010 Hz), one gets the quantum noise seed power of ~2 nW. Assuming detectable peak power of ~1 W, one needs the total buildup of intracavity intensity of about 109. With typically ~30 roundtrips in a pulsed ns OPO, this means that a single‐pass OPO gain factor in (5.8) should be G ≈ 2, which is larger than a typical intracavity loss for the resonant wave.

5.3.2.1 Broadly Tunable Pulsed OPOs

Shortly after the invention of the laser, pulsed OPOs with their capability of producing broadband mid‐IR tuning, reaching several octaves, became important spectroscopic tools. As early as 1973, Hanna et al. had demonstrated an optical parametric oscillator based on a proustite (Ag3AsS3) crystal pumped by a Q‐switched neodymium (1.065 μm) laser that was tuned over a substantial (1.22–8.5 μm) portion of the infrared [89].

Two decades later, owing to the development of the poling technology in ferroelectrics, PPLN became one of the most widely used OPO nonlinear materials for generating tunable light, especially in the 1–5 μm spectral region. High nonlinear coefficient, good optical quality, and low absorption allowed dramatic reduction in OPO pump threshold and improvement of the conversion efficiency. The first nanosecond OPO based on PPLN crystal was reported by Myers et al. [90]. The authors used a 15‐mm‐long, 0.5‐mm‐thick PPLN crystal with a 31‐μm QPM grating period, pumped by a Q‐switched diode‐pumped 1.064‐μm Nd:YAG laser with 7–20 ns pulse duration and 100 Hz to 10 kHz repetition rate. The OPO resonator was a linear cavity with mirrors selected to resonate at the signal wave. With the focused pump laser beam size of 47 μm, the OPO pump threshold was 12 μJ, and the output was continuously tunable from 1.66 to 2.95 μm, with the crystal temperature varying from room temperature to 180 °C. To extend the tuning range of the PPLN OPO, the same team reported an OPO that used multiple grating sections on a single PPLN chip. The PPLN chip was 26‐mm long and consisted of 25 0.5‐mm‐wide gratings with QPM periods from 26 to 32 μm in 0.25‐μm steps (Figure 5.10a) [91]. The oscillation threshold was as low as 6 μJ with a 7‐ns pump pulse (energy fluence 0.09 J/cm2). For tuning, the PPLN crystal was translated across the beam so that it interacted with different grating sections, and a tunable output of 1.36–4.83 μm was achieved. The OPO tuning curve with respect to the grating period is shown in Figure 5.10b. At a pump repetition rate of 1 kHz and the average pump power 100 mW (100 μJ/pulse), the output power at 4 and 4.83 μm amounted to 6 and 2 mW correspondingly.

Silver gallium sulfide, AgGaS2 (AGS), is one of the few nonlinear materials that can be conveniently pumped by commercial 1‐μm lasers and provide very large tunability, well over an octave, with the idler wavelength stretching beyond 10 μm. The first OPO based on AGS crystal was reported in 1984 by Fan et al. [92]. A type‐I angle‐tuned OPO was pumped by a Q‐switched Nd:YAG laser (τ = 18 ns) and tuned from 1.4 to 4 μm. Subsequently, Vodopyanov et al. demonstrated an AGS OPO with even larger tuning range of 3.9–11.3 μm [93]. The singly resonant angle‐tuned OPO was formed by two flat mirrors and used a 20‐mm AGS crystal cut at 45.1° to the optical z‐axis for the type‐II phase matching. The crystal was pumped by 1.06‐μm pulses from a Nd:YAG laser with 12–100 ns pulse duration and yielded up to 0.37 mJ idler wave pulse energy at λ = 6 μm at 15 mJ pump. In the configuration with the recycling of the pump and idler beams in the second pass, the OPO pump threshold was 85 μJ. The OPO linewidth was ~1 cm−1 and the quantum conversion efficiency reached 22%. This device, with its tuning range of 3.9–11.3 μm (Figure 5.11), can be considered to be one of the longest wavelength OPOs pumped at 1 μm (only surpassed by a BGSe crystal, see later in this section). Nevertheless, the practical use of AGS is limited because of its comparatively low surface damage threshold (~0.2 J/cm2 for ns pulses).

Experimental setup for the pulsed OPO using multi‐grating PPLN crystal (a). Graph of wavelength versus grating period displaying a dotted and solid leftward-opening parabolic curve (b).

Figure 5.10 (a) Experimental setup for the pulsed OPO using multi‐grating PPLN crystal. For tuning, the PPLN crystal was translated through the resonator so the pump beam interacted with different grating sections. (b) The OPO wavelength tuning as a function of grating period, achieved by translation of the PPLN crystal through different grating sections.

Source: reproduced from figures 2 and 3 of [91], with permission of OSA, The Optical Society.

Because of its longwave IR cutoff (19 μm), silver gallium selenide, AgGaSe2, has been successful in providing longwave mid‐IR OPO tunability. As opposed to AGS, it is not phase‐matchable (at least using angular phase matching) for 1‐μm pumping; however, it phase matches with a variety of pump sources at λ > 1.3 μm. The first operation of an AGSe OPO was reported by Eckardt et al. with continuous tuning ranges of 6.7–6.9 μm (1.34‐μm neodymium laser pump) and 2.65–9.02 μm (2.05‐μm holmium laser pump) [94]. Quarles et al. demonstrated the full continuous tuning range between 2.49 and 12.05 μm with a 2.05‐μm holmium‐laser pump, with a single angle‐tuned AGSe crystal, and with the output OPO energy up to a few millijoules [95]. Continuously tunable IR output in the range 6.1–14.1 μm has been demonstrated by Chandra et al. [96] in a cascaded OPO based on angle‐tuned AGSe crystal pumped by 1.57‐μm pulses (the output of a KTP OPO). Energies of up to 1.2 mJ/pulse at 9 μm idler wave (quantum efficiency 23%) with bandwidths of 5 cm−1 were obtained using a 35‐mm‐long AGSe crystal and type‐I phase matching. The main limitation was the surface damage of the AGSe crystal starting at ~20 MW/cm2 intensity (0.12 J/cm2 fluence).

An extremely high nonlinear‐optical coefficient of a ZnGeP2 (ZGP) crystal dNL = 75 pm/V, with its NLO FOM d2/n3 (n is refractive index) almost nine times that of PPLN, combined with good optical, mechanical, thermal, and surface damage properties, favors a variety of nonlinear‐optical applications in the 2–12 μm range, including efficient high average power, high pulse energy, as well as broadly tunable OPOs. Despite its large (2 eV) bandgap, ZGP has a residual absorption tail in the near‐IR region. As a result, for efficient performance, the ZGP pump wavelength should be chosen at 2 μm or above. For example, 2‐μm holmium or 3‐μm erbium lasers are suitable candidates for this purpose. A ZGP OPO with 3.8–12.4 μm mid‐IR tunability and >1 mJ idler pulse energy was demonstrated in [97, 98]. A pump source was a Q‐switched erbium laser with λ = 2.8 μm (Er, Cr:YSGG) or 2.93 μm (Er, Cr, Tm:YAG), with 100‐ns pulse duration, 10‐Hz repetition rate, and 10‐mJ energy. An AR‐coated ZGP crystal was 20‐mm long and was cut for type‐I (θ0 = 49.5°) or for type‐II (θ0 = 70°) phase matching. The lowest OPO threshold was obtained in a flat–flat double‐pass OPO cavity configuration shown in the inset of Figure 5.12a. The OPO output was continuously tunable, via crystal angle tuning, from 3.8 to 12.4 μm (type‐I phase matching, Figure 5.12a) and from 4 to 10 μm (type‐II phase matching, Figure 5.12b) with a linewidth of 2–3 cm−1. Figure 5.12c shows the dependence of the type‐I OPO idler energy at λ = 8.1 μm as a function of the pump energy, where a 1‐mJ pulse energy was reached at 10.5 mJ of the pump. The OPO pumping threshold was less than 1 mJ/pulse in the whole 4–12 μm range and the quantum conversion efficiency reached 35%.

Graph of idler wavelength vs. phase-matching angle displaying a descending curve with circle markers. Inset: Graph of OPO threshold vs. idler wavelength with an ascending curve.

Figure 5.11 AGS type‐II OPO angular tuning curve (idler wave only). Solid line – theoretical tuning curve. Inset: OPO threshold fluence as a function of the idler wavelength.

Source: reproduced from figure 2 of [93], with permission of AIP, American Institute of Physics.

High quadratic nonlinearity of ZGP made it possible to demonstrate a record‐low‐threshold OPO with a wide tunability. The OPO was a tandem system, where a noncritically phase‐matched (NCPM) ZGP OPO was pumped by the idler wave output of a PPLN OPO, pumped in turn by a Nd:YAG laser (1.6 mJ, 20 ns, 1 kHz) [99]. The singly resonant ZGP OPO (Figure 5.13a) contained a 24‐mm θ = 90°‐cut ZGP crystal and was formed by a concave output coupler mirror M3 (highly reflective at signal and transmissive at the idler and the pump) and a gold reflector mirror M4, which was deposited directly into the polished flat surface of the ZGP crystal (the front surface of ZGP was AR‐coated). Tuning the PPLN OPO pump wavelength in the range 2.3–3.7 μm resulted in tuning the ZGP OPO output from 3.7 to 10.2 μm. The OPO tuning curves (versus pump wavelength) are shown in Figure 5.13b. Solid lines on this figure correspond to a theoretical prediction based on known dispersion data. The ZGP OPO idler output amounted to 25 μJ at 1 kHz and λ = 7 μm corresponding to photon conversion efficiency of 30%. The total conversion from the 1.064 μm laser to the 7 μm output amounted to 1.5%. At a pump beam size of w0 = 125 μm (at λ = 3.1 μm), which is close to a confocal focusing condition, the OPO pump threshold was remarkably low, 2 μJ [99]. This is the lowest OPO threshold reported so far for a singly resonant pulsed OPO.

The first optical parametric oscillator based on gallium arsenide was demonstrated in [100, 101]. The OPO utilized a QPM OP‐GaAs crystal, 11‐mm long, 5‐mm wide, and 0.5‐mm thick, with a domain reversal period of 61.2 μm. Tunable (1.75–2 μm) pulses from a Nd:YAG‐laser‐pumped PPLN OPO (10 Hz, 0–70 μJ, 6 ns) were used as a pump. A 13‐mm‐long OP‐GaAs OPO cavity (Figure 5.14) was formed by two flat mirrors M4–M5. The input–output mirror M4 was reflective at the signal wavelength and transmissive at the pump and the idler. In a single‐pass arrangement, mirror M5 was dielectric, identical to M4, while in a two‐pass arrangement, a flat gold mirror was used as M5 to reflect all three waves. Thus, the signal wave was resonated, whereas the pump and idler waves were recycled to have a second pass before leaving the cavity. Because of the symmetry of GaAs, the output polarizations of the OP‐GaAs OPO were orthogonal to that of the pump and were extracted by use of a ZnSe plate at Brewster's angle (Figure 5.14). The lowest OPO pump threshold (16 μJ) and the highest output were obtained in a two‐pass arrangement in the GaAs OPO. Figure 5.15a shows the OP‐GaAs OPO tuning curve (with respect to the pump wavelength). The OP‐GaAs crystal allowed mid‐IR tuning between 2 and 11 μm, which was limited by (i) the spectral range of the OPO mirrors and (ii) the onset of two‐photon absorption at <1.75 μm pump wavelength. Temperature tuning curves for two selected pump wavelengths are represented in Figure 5.15b. At the pump energy around 40–50 μJ, the OPO idler output was 3 μJ (λ = 7.9 μm), corresponding to the quantum conversion efficiency of 26%.

Image described by caption and surrounding text.

Figure 5.12 ZGP OPO with a wide (3.8–12.4 μm) mid‐IR tunability and more than 1 mJ pulse energy. (a) Angular tuning curve for the type‐I OPO. Inset: OPO schematic. The front OPO mirror M1 was transmissive for the pump and the idler and highly reflective (98%) for the signal. A gold rear mirror M2 was highly reflective (R > 98%) for the pump, signal, and idler. Thus, the signal wave resonated, while the pump and the idler were double‐passed. A dichroic beam splitter (BS) separated the incoming pump beam from the outcoming idler. (b) Angular tuning curve for the type‐II OPO. (c) Dependence of the type‐I OPO idler pulse energy as a function of the pump. The inset shows the OPO idler beam far‐field intensity distribution (the corresponding beam quality factor M2 ≈ 1.5).

Source: reproduced from figures 2 and 3 of [97], with permission of OSA, The Optical Society.

Image described by caption and surrounding text.

Figure 5.13 (a) Schematic of the tandem noncritically phase‐matched ZGP OPO. M1–M4, OPO mirrors; L, infrared focusing lenses; BS, beamsplitters. (b) Tuning curve for ZGP OPO, as a function of pump wavelength. The insets show longwave absorption coefficients for PPLN and ZGP and also the far‐field beam profile for the idler beam at λ = 6 μm.

Source: reproduced from figures 1 and 2 of [99], with permission of OSA, The Optical Society.

Image described by caption and surrounding text.

Figure 5.14 Schematic of the OP‐GaAs OPO. Tunable, 1.75–2‐μm, signal pulses from a three‐mirror PPLN OPO were used as a pump (a 22‐μm‐thick uncoated silicon etalon was used inside PPLN OPO cavity to reduce its linewidth to 5 cm−1). M1–M3, PPLN OPO mirrors; M4–M5, GaAs OPO mirrors. The filter F was used to block both the PPLN idler wave (2.3–2.6 μm) and the 1.06‐μm pump.

Source: reproduced from figure 1 of [100], with permission of OSA, The Optical Society.

Image described by caption and surrounding text.

Figure 5.15 (a) OP‐GaAs OPO tuning curve (t = 20 °C) with respect to the pump wavelength. (b) OP‐GaAs OPO temperature‐tuning curves for two selected pump wavelengths. Solid lines are calculated, based on data from [102]. Inset shows OPO spectral line shapes at different GaAs temperatures (pump 1.89 μm).

Source: reproduced from figures 3 and 4 of [101], with permission of SPIE.

The broadest continuous mid‐IR tunability from a GaAs system was reported in [103]. With the choice of the pump near λ ≈ 3 μm (from a PPLN OPO), the authors were able to achieve tunable output in the whole range of 4–14.2 μm with a linewidth of 2–6 cm−1, using a single OP‐GaAs crystal with a domain reversal period of 150 μm (Figure 5.16). The OPO output was tuned using (i) an intracavity diffraction grating and (ii) fine adjustment (within 160 nm) of the pump wavelength near 3 μm. The OPO was operating at 2 kHz repetition rate. Its threshold, in terms of pump pulse energy, was approximately 25 μJ and the output average power reached 14 mW at 6‐μm wavelength [103].

CSP has a transparency window extending from 1 to 6.5 μm [104]. Prominently, when pumped at 1.064 μm, CSP allows parametric generation under noncritical (θ = 90°) phase matching with a large nonlinearity (see Table 5.1), providing mid‐IR idler radiation near 6 μm, which is important for medical applications. The main problem yet to be solved with this crystal is the residual absorption close to the bandgap, which is not intrinsic. For example, at 1.064 μm, the bulk absorption is ~0.2 cm−1. The first OPO operation of CSP was demonstrated by Petrov et al. in the nanosecond noncritical (θ = 90°) singly resonant regime, pumped at 1.064 μm [105]. The CSP sample had a length of 8 mm and the OPO cavity consisted of two plane mirrors separated by 9.5 mm. With the signal/idler wavelengths of 1.285/6.193 μm, the pump threshold was 1.8 mJ. The maximum idler energy was 0.47 mJ (10 Hz repetition rate), at an incident pump energy of 21.4 mJ. This gives an idler conversion efficiency of 2.2% and quantum conversion efficiency of 12.8%. The CSP damage threshold (0.22 J/cm2), in terms of pulse energy fluence, was similar to that of AgGaS2.

Image described by caption.

Figure 5.16 (a) Schematic of the 4–14 μm OP‐GaAs OPO. Tunable pump pulses (near ~3 μm) are directed into an OP‐GaAs crystal. The L‐shaped OPO cavity is formed by: (i) diffraction grating, (ii) dichroic mirror DM that reflects (R > 98%) the signal at 4–6 μm and transmits the 3‐μm pump and idler at 6–14 μm, and (iii) metallic mirror M. The lens L is introduced to make a stable cavity and expand the beam at the grating. (b) The OPO tuning curve. The wavelength was tuned by (i) an intracavity diffraction grating and by (ii) fine adjustment (within 160 nm) of the pump wavelength. Solid line: theoretical tuning curve. The vertical dashed lines “A” and “B” represent turning points of the tuning curve. Also shown are diffraction grating angles for several resonating signal wavelengths. The inset shows a far‐field beam profile at λ ≈ 8 μm.

Source: reproduced from figure 1 of [103], with permission of OSA, The Optical Society.

Marchev et al. used a 21.4‐mm‐long θ = 90°‐cut CSP crystal pumped by 8‐ns pulses at 1.064 μm in an OPG arrangement (without a cavity but with weak residual reflections from the parallel crystal surfaces), in a double‐pass configuration for pump, signal, and idler (Figure 5.17) [106]. A 45° ZnSe bending mirror (BM) was used for separating the pump radiation from the signal and idler outputs, and a silver‐coated mirror – to retroreflect all the three waves for a second pass. The pumping threshold for the device was 213 μJ (0.23 MW/cm2 on‐axis intensity). At the maximum pump energy of 12 mJ (12.7 MW/cm2), the total output energy exceeded 4 mJ, from which 3.64 mJ were at 1.288 μm (signal) and 0.52 mJ at 6.125 μm (idler), with a quantum conversion efficiency of 34.7%. The beam propagation parameter M2 for the idler wave was between 7 and 8. At a repetition rate of 100 Hz, the average idler power was 52.3 mW at 6 μm.

Image described by caption.

Figure 5.17 Experimental setup of the CSP‐based OPG device. T, telescope; D, diaphragm; BM, bending mirror; TR, total reflector (metal mirror); F, long‐pass filter; P, polarizer; λ/2, half‐wave plate; D and DD, diaphragms.

Source: reproduced from figure 1 of [106], with permission of OSA, The Optical Society.

An extra‐wide mid‐IR tunability (2.7–17 μm) under 1.064 μm pumping was demonstrated in a newly developed chalcogenide crystal BGSe (BaGa4Se7) with monoclinic symmetry [107]. The crystal has an excellent optical quality and is transparent from 0.8 to 15 μm (at 0.3 cm−1 absorption level). A singly resonant nanosecond OPO had a linear cavity consisting of a flat input–output coupler and a flat gold‐coated total rear reflector, which ensures recycling of the pump and a double pass for the nonresonant idler prior to its extraction from the cavity. Pumping via a 45° ZnSe BM, highly transmitting for the signal and idler, ensured separation of the input and output waves. The pump source was a diode‐pumped Nd:YAG master oscillator–power amplifier (MOPA) system operating at 10 Hz, with 8 ns pulse duration and 63 mJ pulse energy. The OPO angle tuning curves for type‐I and type‐II phase matching are shown in Figure 5.18. The best results at 63‐mJ pump include: unprecedented tuning range, from 2.7 to 17 μm achieved with a single crystal cut; 4.7‐mJ pulse energy obtained for the λ = 5.3 μm idler wave (pump‐to‐idler conversion efficiency of 7.5%); and 3.7‐mJ idler output at λ = 7.2 μm (pump‐to‐idler conversion efficiency of 5.9%; quantum conversion efficiency of 40%) [107].

Image described by caption and surrounding text.

Figure 5.18 Experimental angle tuning curves for type I and type II BGSe‐based OPO pumped at 1.064 μm. Solid curves: calculated.

Source: reproduced from figure 3 of [107], with permission of OSA, The Optical Society.

5.3.2.2 Narrow‐linewidth Pulsed OPOs

Because of the limited buildup time in pulsed OPOs, it is more challenging to achieve a narrow‐linewidth output in such devices, as compared to their CW counterparts. Richman et al. have developed a PPLN‐based pulsed mid‐IR OPO that resonates the signal wave and uses just one intracavity etalon to restrict lasing to an SLM of the resonator cavity [108]. The OPO ring resonator (Figure 5.19) consisted of three mirrors. The two dichroic mirrors on either side of the PPLN crystal (25 mm long, 0.5 mm high, and 19 mm wide) transmitted the pump and idler beams and reflected the signal beam. The third mirror outcoupled 10% of the signal beam. The airspace etalon was made of one convex and one concave mirror, with a 95% reflective coating for the signal wave and each with a curvature that matches the phase‐front curvature of the ring‐cavity TEM00 mode. The etalon with mirror spacing of 357 μm (free spectral range 14 cm−1) had a finesse of 60 and the insertion loss of 30%. The pump was a narrow‐linewidth injection‐seeded Q‐switched laser at 1.06 μm and 1‐kHz repetition rate, and its output was focused to a 240‐μm spot size in the PPLN crystal. Both the etalon mirror spacing and the OPO resonator length were adjusted by piezoelectric translators (PZTs), so that it was possible to tune the frequency continuously over 10 cm−1 without using motorized parts. Translation of a multi‐grating PPLN wafer allowed access to any wavelength from 1.45 to 1.8 μm (signal) and from 2.6 to 4 μm (idler). Up to 18 μJ/pulse in the idler beam and up to 15 μJ/pulse in the signal beam were produced with only 200‐μJ pump energy. The measured single‐mode OPO linewidth was 0.005 cm−1.

Image described by caption and surrounding text.

Figure 5.19 Schematic of a pulsed single‐longitudinal‐mode tunable PPLN OPO. The ring resonator consists of three mirrors. The two flat mirrors on either side of the PPLN crystal transmit the pump and idler beams and reflect the signal beam. The third (curved) mirror outcouples 10% of the signal beam for spectral characterization. The airspace etalon is made of two lenses, each with a 95% reflective coating and each with a curvature that nearly matches the phase‐front curvature of the ring‐cavity TEM00 mode. The right‐hand dichroic mirror and the curved mirror are on a single translation stage driven by a PZT to control the cavity length.

Source: reproduced from figure 1 of [108], with permission of OSA, The Optical Society.

Ganikhanov et al. demonstrated a narrow‐linewidth and broadly tunable output from a singly resonant (idler wave) nanosecond OPO based on type‐II ZGP crystal pumped at 2.55 μm. With the OPO cavity containing a diffraction grating and an Si etalon, the authors achieved mid‐IR tunability from 3.7 to 8 μm with the output energies 10–200 μJ/pulse, and with the output linewidth of 0.1 cm−1, corresponding to three axial cavity modes [109]. It was also demonstrated that a narrow‐linewidth idler OPO output can be achieved even using a comparatively broadband pump.

The concept of a doubly resonant nested‐cavity OPO enables achieving an SLM emission from a pulsed OPO without using any intracavity etalons or resorting to injection seeding. In addition, one can achieve very low oscillation thresholds of a few microjoules [110, 111]. Traditionally it was understood that a doubly resonant condition precludes continuous frequency tuning, since it is satisfied only at particular wavelengths. A nested‐cavity doubly resonant OPO solves this problem. In such an OPO (Figure 5.20) the inner mirrors (M2, M3) are deposited onto the nonlinear crystal faces whereas the external mirrors (M1, M4) are mounted on two PZT actuators for fine‐tuning of the lengths, such that the signal wave oscillates between M1 and M3, while the idler between M2 and M4 (alternatively, it can be a three‐mirror design). The mode spacing is chosen to be slightly different for the signal and the idler, so that only one pair of signal–idler modes oscillates. By fine‐tuning the PZT actuators, tunable mode‐hop‐free SLM oscillation can be achieved over >100 GHz range. Low‐threshold nested‐cavity OPOs based on PPLN crystals pumped by compact passively Q‐switched single‐frequency microlasers of fiber‐based lasers operating at high (4.8–100 kHz) repetition rates have been demonstrated and used for various spectroscopic applications in the range from 2 to 4.3 μm [112, 113].

Image described by caption and surrounding text.

Figure 5.20 (a) Schematic of a nested dual‐cavity doubly resonant OPO for SLM operation. M1–M3 form the “signal” cavity, and M2–M4 form the “idler” cavity. (b) Mode picture for the signal and idler waves. Since the mode spacing is different for the signal and the idler waves, only one pair of signal–idler modes oscillates.

Source: reproduced from figure 16 of [110], with permission of OSA, The Optical Society.

Clément et al. reported a nanosecond doubly resonant nested‐cavity OPO based on 10‐mm‐long OP‐GaAs emitting a tunable single‐frequency radiation in the longwave IR region [114]. The OPO was pumped by an SLM Tm:YAP microlaser emitting pulses at λ = 1.94 μm, with a maximum output energy of 170 μJ, pulse duration of 36 ns, and a repetition rate of 100 Hz. The OPO pump threshold energy was 10 μJ. With the OP‐GaAs QPM period of 72.6 μm, SLM wavelength tuning over the range of 10.3–10.9 μm was obtained by varying the crystal temperature. Based on this device, a differential absorption LIDAR was demonstrated by carrying out detection of ammonia vapor around λ = 10.4 μm [114].

The summary of pulsed narrow‐band OPOs can be found in Table 5.5.

Table 5.5 Summary of pulsed nanosecond OPOs.

OPO crystal OPO wavelength (μm) Pump OPO parameters Ref.
Broadly tunable
PPLN 1.36–4.83 1.064 μm (100 μJ, 7 ns, 1 kHz) 6 μJ @ 4 μm; 2 μJ @ 4.8 μm [91]
AGS 3.9–11.3 1.064 μm (15 mJ, 12 ns, 10 Hz) 0.37 mJ @ 6 μm [93]
AGSe 6.1–14.1 1.57 μm (30 mJ, 6 ns, 5 Hz) 1.2 mJ @ 9 μm [96]
ZGP 3.8–12.4 2.93 μm (10 mJ, 100 ns, 10 Hz) 1.2 mJ @ 6.6 μm; 1 mJ @ 8.1 μm [97]
OP‐GaAs 2–11 1.75–2 μm (60 μJ, 6 ns, 10 Hz) 3 μJ @ 7.9 μm [101]
OP‐GaAs 4–14 ~3 μm (120 μJ, 20 ns, 2 kHz) 7 μJ @ 6 μm [103]
BGSe 2.7–17 1.064 μm (63 mJ, 8 ns, 10 Hz) 4.7 mJ @ 5.3 μm; 3.7 mJ @ 7.2 μm [107]
Narrow linewidth
PPLN ~3 1.064 μm (200 μJ, 10 ns, 1 kHz) 18 μJ, 0.01 cm−1 [108]
PPLN 3.8–4.3 1.064 μm (16 μJ, 9.8 ns, 4.8 kHz) Nested‐cavity SLM OPO, 0.3 μJ @ 3.9 μm [112]
PPLN 3.3–3.5 1.064 μm (4–10 μJ, 1 μs, 40–100 kHz) Nested‐cavity SLM OPO, 0.5 μJ @ 3.5 μm [113]
ZGP 3.7–8 2.55 μm (12 mJ, 12 ns, 10 Hz) 10 μJ @ 3.7 μm, 0.1 cm−1; 10 μJ @ 8 μm, 0.1 cm−1 [109]
OP‐GaAs 10.3–10.9 1938.5 μm (170 μJ, 36 ns, 100 Hz) Nested‐cavity SLM OPO, 2 μJ/pulse [114]
High average power
PPKTP 1.72 and 2.76 1.064 μm (7.2 W, 20 kHz, 5.8 ns) 2 W (sig + idler) [7]
KTP 2.13 1.064 μm (135 W, 20 kHz, 40 ns) 53 W in two polarizations [115]
ZGP 3.67 and 4.67 2.05 μm (20.1 W, 10 kHz, 11 ns) 10.1 W (sig + idler), conv. eff. 50.2% [116]
ZGP 3.7–4.1 4.4–4.8 2.13 μm (25 W, 20 kHz, 40 ns) 14 W (sig + idler) [115]
ZGP 3.8 and 4.6 2.1 μm (37.7 W, 32 ns, 45 kHz) 22 W (sig + idler), M2 ≈ 1.4, conv. eff. 58% (slope 75%) [117]
ZGP 4.3 2 μm (7.2 W, 5 kHz) 2.53 W, cascaded intracavity OPO [118]
ZGP Broadband 3.5–5 μm 2.09 μm (43 W, 50 ns, 35 kHz) 27 W (slope eff. 67%); 99 W (25% duty cycle) [119]
OP‐GaAs Broadband 3.5–5 μm 2.09 μm (6.1 W, 65 ns, 20 kHz) 2.85 W ave. power (sig + idler) [120]
OP‐GaAs 3.6 and 4.4 ≈2 μm, 60 W ave. power 18 W ave. power (sig + idler) [18]
OP‐GaAs 10.6 1.95 μm (12 W, 160 ns, 50 kHz) 800 mW ave. power, conv. eff. 6.8%, quant. eff. 36.8% [121]
High pulse energy
KTP L = 15 mm 2.6–3.2 1.064 μm (145 mJ, 5 ns, 10 Hz) 17 mJ @ 2.9 μm [122]
LiNbO3 L = 50 mm 2.13 1.064 μm (600 mJ, 15 ns) 300 mJ [123]
PPLN
PPLT L = 40 mm
1.89 and 2.43
1.84 and 2.52
1.064 μm (196 mJ, 10 ns, 30 Hz) 124 mJ (sig + idler)
118 mJ (sig + idler)
[124]
ZGP L = 25 mm 6.9–9.9 2.8 μm (25 mJ,50 ns, 10 Hz) 2.4 mJ @ 6.9 μm, 0.7 mJ @ 9.9 μm [125]
ZGP RISTRA L = 10 mm 3.4 2.05 μm (55 mJ,14 ns, 500 Hz) 10 mJ, near‐diffraction‐limited [126]
ZGP RISTRA Broadband 3.5–5 μm 2.053 μm (45.6 mJ,100 Hz) 23.8 mJ (sig + idler) [127]
ZGP MOPA Broadband 3.5–5 μm 2.05 μm (500 mJ,15 ns, 1 Hz) 212 mJ (sig + idler) [128]

5.3.2.3 High Average Power OPOs

5.3.2.3.1 Based on PPKTP and KTP

Among the family of PP oxide crystals, large‐aperture PPKTP crystals are well suited for generating average powers exceeding 1 W in the pulsed nanosecond regime. Peltz et al. have reported a high average power OPO using a 3‐mm‐thick PPKTP crystal [7]. A diode‐pumped Nd:YVO4 laser system (1.064 μm, 5.8 ns, 10–20 kHz) served as a pump. At a repetition rate of 20 kHz and 7.2 W pump power, the total signal + idler output power (at correspondingly 1.72 and 2.76 μm) reached 2 W.

Bulk KTP crystals enable much higher average powers, as compared to PPKTP, due to their larger apertures. Cheung et al. [115] used a diode array‐pumped Nd:YAG MOPA system with the repetition rate of 20 kHz, pulse duration of 40 ns, and the average power of 135 W, as a pump source for an OPO based on bulk KTP crystals. A degenerate OPO (λ ≈ 2.13 μm) used six KTP crystals, 3 × 3 × 6 mm in size, cut for type‐II phase matching in a walk‐off compensated configuration. The average power of 53 W (in two orthogonal polarizations) has been achieved with 43% conversion efficiency.

5.3.2.3.2 Based on ZGP

The ZGP crystal is extremely suitable for scaling to high average OPO powers in the longwave mid‐IR range, since in addition to high quadratic nonlinearity it exhibits excellent thermal and mechanical properties. Typically, high‐power ZGP OPOs are based on frequency downconversion of 2‐μm lasers.

Wu et al. [118] demonstrated a coupled tandem approach, where a 2‐μm output of a KTP OPO pumped a ZGP OPO. Not only was the ZGP OPO placed within the cavity of the KTP OPO, but also the KTP OPO was placed within the cavity of a diode‐pumped Nd:YALO laser (Figure 5.21). The Nd:YALO laser was able to generate ~58 W of the output power at λ ~ 1 μm at a pulse repetition rate of 5 kHz. The output from the ZGP OPO was tunable over the range of 3–6 μm, with the maximum output power of 2.53 W achieved at 4.3 μm, at approximately 580 W of diode pump power.

Image described by caption and surrounding text.

Figure 5.21 Schematic diagram of the coupled tandem KTP OPO–ZGP OPO. The Nd:YALO laser cavity (the pump) was formed by a high reflector M1 and an output coupler M2. The KTP OPO utilized four type‐II, 51°‐cut, diffusion‐bonded walk‐off‐compensated KTP crystals (5 × 5 × 8 mm), placed between mirrors M3 and M1. A ZGP OPO cavity was formed by mirrors M4 and M5 and used type I, 53°‐cut ZGP with 5 × 5 × 10 mm dimensions.

Source: reproduced from figure 1 of [118], with permission of OSA, The Optical Society.

The first ZGP OPO with an average output power exceeding 10 W was reported by Budni et al. [116]. The OPO used a 14‐mm‐long type‐I ZGP crystal and was pumped at 10 kHz by 11‐ns, λ = 2.05 μm pulses from a diode‐pumped Q‐switched Ho,Tm:YLF‐laser operating at T = 77 K. At the maximum pump drive level of 20.1 W incident onto the ZGP crystal, the output average power (signal at 3.67 μm plus idler at 4.67 μm) reached 10.1 W, corresponding to a conversion efficiency of 50.2%. It is important that the operation of this OPO at maximum power level was well below the ZGP damage threshold.

Cheung et al. used the high‐power 2.13‐μm output obtained from the KTP OPO described above for pumping a ZGP OPO. A polarizer was used to separate the two orthogonally polarized 2.13‐μm beams from the KTP OPO (approximately 25 W in each beam), and each beam was sent into a separate ZGP OPO producing broadband output in the 3–5 μm band. With two ZGP OPOs running simultaneously, 13 and 11 W were obtained with the beam quality factor M2 ≈ 4 [115].

Lippert et al. have demonstrated a 22‐W output from a single ZGP OPO [117]. The OPO was pumped by a pulsed 2.1‐μm Ho:YAG laser (pumped in turn by a Tm‐fiber laser) with 32‐ns pulse duration and 45‐kHz repetition frequency. An innovative V‐shaped three‐mirror ring OPO design was used to achieve an excellent beam quality. The ring is in the noncritical plane of the crystal, as shown in Figure 5.22. Compared to other ring resonators, the V‐shaped three‐mirror ring cavity has the advantages of short round‐trip time and also that a single ZGP crystal can be used for two passes. The crystal was tuned by rotation in the orthogonal plane. With 37.7 W of Ho:YAG pump, the authors obtained 22 W of average power in the 3–5 μm range (3.8‐μm signal plus 4.6‐μm idler), with a beam quality factor of M2 ≈ 1.4. The absolute conversion efficiency was 58%, and the slope efficiency was 75%. The output slope showed no sign of roll‐off at high power, which indicates that the high efficiency can be maintained at even higher pump power.

Schematic diagram illustrating a V-shaped 3-mirror ring resonator of a high-power ZGP OPO with two passes through the same crystal and angle tuning about an axis in the plane of the right.

Figure 5.22 V‐shaped 3‐mirror ring resonator of a high‐power ZGP OPO with two passes through the same crystal and angle tuning about an axis in the plane of the ring.

Source: reproduced from figure 1 of [117], with permission of OSA, The Optical Society.

Hemming et al. have generated the highest reported output power from a mid‐IR ZGP OPO. The pump was a thulium‐fiber‐laser‐pumped Q‐switched Ho:YAG laser (2.09 μm) operating at 35‐kHz pulse repetition rate, with up to 60 W of output power. The system produced 27 W of the output power in the 3–5 μm band with the beam quality factor M2 = 4 when operating continuously in a repetitively Q‐switched mode. However, when the OPO operated at a duty cycle of 25%, the average output power during the “open” cycle reached 99 W [119].

5.3.2.3.3 Based on OP‐GaAs

In the early work on OP‐GaAs, Kieleck et al. reported a high‐efficiency high repetition rate mid‐IR OPO utilizing OP‐GaAs pumped by a 2.09‐μm Ho:YAG laser [120]. The OP‐GaAs crystal was 20 mm long, 5 mm wide, and 450 μm thick, and had 63‐μm grating period. Up to 2.85 W are obtained in the 3–5 μm band for 6.1 W of pump power at 20 kHz repetition rate, corresponding to an optical‐to‐optical conversion efficiency of 46.5%, with an OPO pump threshold of 1 W. According to the authors, the OP‐GaAs OPO performance, in terms of the average power, was only limited by the thickness of the crystal. Thanks to the further progress in all‐epitaxial processing of OP‐GaAs, namely increasing crystal's thickness in the direction of the epitaxial growth and improving the quality of orientation patterning, it became possible to demonstrate 18 W of the average power (signal + idler) from an OP‐GaAs OPO pumped by a pulsed high repetition rate Tm‐fiber laser (λ ≈ 2 μm), with the signal and idler wavelengths of 3.6 and 4.4 μm, respectively [18].

On a longer‐wavelength side, an optical parametric oscillator based on OP‐GaAs generating an idler output at around 10.6 μm with relatively high average power was reported by Wueppen et al. [121]. The system used an SLM (spectral width <100 MHz) thulium fiber laser at 1.95‐μm wavelength, 50‐kHz repetition rate, and 150‐ns pulse duration as a pump. With a signal‐resonant bow‐tie OPO cavity and an OP‐GaAs crystal that was 40 mm long, 5 mm wide, and 1.3 mm thick, the average idler output power of more than 800 mW (pulse energy 16 μJ) was obtained at 10.6 μm, which corresponds to a quantum conversion efficiency of 36.8% [121].

5.3.2.4 High Pulse Energy OPOs

Large‐aperture bulk oxide crystals such as lithium niobate and its family, as well as KTP and its family are well suited for producing high pulse energies. Vysniauskas et al. [122] demonstrated a high‐energy OPO based on angular‐tuned KTA crystal. The OPO was pumped by a Q‐switched Nd:YAG laser with 5 ns pulse duration and 10 Hz repetition rate. The pump beam had a diameter of 4 mm and was double‐passed through the 15‐mm‐long KTA crystal. The OPO was singly resonant for the idler wave and had an unstable resonator cavity to improve the spatial beam distribution. The idler beam tunability over the 2.6–3.2 μm range was achieved with the pulse energy of 17 mJ at 2.9 μm for a pump pulse energy of 145 mJ. Even higher mid‐IR OPO output energies were reported by Mennerat and Kupecek [123]. An OPO was based on a 50 mm‐long, 15 mm‐aperture, 47°‐cut LiNbO3 crystal and was pumped with 15‐ns pulses at 1.064 μm. The maximum output energy of 300 mJ (signal + idler) was achieved near OPO degeneracy (2.13 μm) at a 600 mJ pump.

Scaling a nanosecond OPO to high pulse energies entails increasing the beam diameters to avoid surface damage. The OPO cavity, however, should be kept short for high efficiency and low threshold. As a result, the cavity Fresnel‐number (D2/λL, where D is the beam diameter and L is the cavity length) increases, which worsens the spatial coherence across the beam. Hansson et al. [129] used an unstable OPO resonator to improve the beam quality of a PPRTA‐based OPO (Figure 5.23). An unstable resonator effectively filters out the OPO modes with high spatial frequency components by a combination of laser‐mode magnification and feedback of only the lowest‐order spatial modes. The authors demonstrated an improvement in M2 – the beam quality parameter – by a factor of 3 for a λ = 3.3‐μm OPO idler wave, as compared to that of a plane‐parallel OPO resonator.

Schematic diagram illustrating PP RTA-based OPO with an unstable resonator cavity. The input coupler is located at the left of the PP RTA, while output coupler is at the right.

Figure 5.23 PP RTA‐based OPO with an unstable resonator cavity.

Source: reproduced from figure 1 of [129], with permission of OSA, The Optical Society.

Over the past decade, QPM ferroelectric PPLN and PPLT crystals with large thickness along the poling z‐axis, exceeding 5 mm, have become available, thanks to the maturity of high‐voltage poling technology [124, 130, 131]. This progress allowed producing high‐pulse‐energy outputs from OPOs based on QPM crystals, while benefiting from their engineerable phase matching and access to the highest nonlinear coefficients. Ishizuki et al. produced high‐energy output from an OPO based on congruent MgO‐doped PPLT (MgO–PPLT) crystal [124]. The authors used 196‐mJ pump pulses at 1064 nm with 10‐ns duration at 30 Hz repetition rate and generated 118 mJ total output in the OPO signal (1.84 μm) plus idler (2.52 μm) waves. Such high pulse energies from a QPM crystal became possible because of the availability of large‐aperture (5 × 16 mm) PPLT structure with 40‐mm length [124]. The same team also demonstrated an OPO based on large‐aperture MgO‐doped PPLN (MgO–PPLN) crystal using the same setup, including OPO mirrors, cavity, and the pump source. With the wavelength of signal and idler waves of 1.89 and 2.43 μm, the total OPO output (signal + idler) was measured to be a 124 mJ/pulse at 193‐mJ pump pulse energy [124]. The authors argue that despite its lower second‐order nonlinearity, lithium tantalate has higher thermal conductivity and shorter absorption edge in the UV region, compared to lithium niobate, and produces similar OPO outputs in terms of energy per pulse.

OPOs based on the ZGP crystal are widely used for generating tunable IR radiation in the whole of the 2.5–12‐μm region. ZGP crystal possesses a set of unique properties such as high nonlinearity (d36 = 75 ± 8 pm/V), high thermal conductivity of 35–36 W/(mK) (higher than YAG crystal), and relatively high hardness 5.5 (Mohs scale). Hence, it is a remarkably promising material for generating high‐energy mid‐IR pulses. Allik et al. [125] reported a high‐energy‐per‐pulse ZGP optical parametric oscillator pumped by a 2.8 μm Er,Cr:YSGG laser with a 10‐Hz repetition rate, 25‐mJ energy, and 50‐ns duration. A 25‐mm‐long ZGP crystal was cut at θ = 65° for type‐II phase matching. The OPO yielded idler output in the forward direction of 0.7–2.4 mJ/pulse, in the wavelength range 6.9–9.9 μm. The quantum conversion efficiency reached 29% at the OPO linewidth of typically 4 cm−1.

By exploring a nonplanar OPO cavity containing a ZGP crystal, Dergachev et al. reported on a near‐diffraction‐limited output at the OPO signal wave at λ = 3.4 μm with pulse energy of 10 mJ at a repetition rate of 500 Hz [126]. As a pump source, the authors utilized a 2‐μm, Ho:YLF MOPA system producing >55 mJ pulse energy at 500 Hz. (The Ho:YLF MOPA was pumped in turn by a 100‐W Tm‐fiber laser at 1940 nm.) The OPO used a 10‐mm‐long ZGP crystal, which was AR‐coated to minimize reflections at the pump, signal, and idler wavelengths. The OPO resonator, shown in Figure 5.24, was based on a four‐mirror nonplanar image‐rotating ring cavity, known as Rotated Image Singly Resonant Twisted RectAngle (RISTRA) [132]. This cavity design was selected because the intracavity image rotation generates highly symmetric, high‐quality output beams under operating conditions that would otherwise lead to very poor beam quality. Image rotation is particularly effective for improving beam quality in high‐energy ns OPOs, where the ratio of beam diameter to cavity length results in very large Fresnel numbers.

Image described by caption and surrounding text.

Figure 5.24 Geometric configuration of a nonplanar cavity for a high‐pulse‐energy Rotated Image Singly Resonant Twisted RectAngle (RISTRA) ZGP‐based OPO. The inset shows the far‐field beam profile.

Source: adapted from figure 4a of [126], with permission of OSA, The Optical Society.

Using a ZGP‐based RISTRA OPO, Stöppler et al. demonstrated an overall OPO pulse energy of 23.8 mJ in the broadband 3–5 μm output. As a pump, the authors used 45.6‐mJ pulses from a fiber‐pumped oscillator–amplifier system based on Ho3+:LuLiF4 active medium (λ = 2.053 μm), at a repetition rate of 100 Hz [127]. Haakestad et al. reported broadband (3.5–5 μm) energetic pulses produced via nonlinear conversion in a near‐degenerate ZGP‐based MOPA system, pumped by 0.5‐J pulses from a Q‐switched cryogenic Ho:YLF oscillator at 2.05 μm [128]. A singly resonant master OPO had a V‐shaped three‐mirror ring resonator and contained two ZGP crystals, while the power amplifier used from one to three large‐aperture ZGP crystals (Figure 5.25). Pulses with up to 212 mJ energy at 1 Hz repetition rate were obtained, with pulse duration of 15 ns and beam quality M2 = 3. These are the highest‐energy pulses generated in this wavelength region by a nanosecond solid‐state laser source. The main advantages of the V‐shaped three‐mirror ring‐cavity OPO used in this work are: (i) two‐pass pumping without pump feedback into the laser, (ii) simple alignment, (iii) compactness, and (iv) reduced fluence at crystal surfaces because the forward and backward propagating beams were not overlapping.

5.3.2.5 Waveguide OPOs

Singly resonant pulsed OPO operation (τ ≈ 100 ns) has been demonstrated in a PPLN‐based waveguide. The OPO was pumped at 760 nm and had a very low oscillation threshold, 1.6 W in terms of the peak power (pump pulse energy 0.16 μJ) [133]. The idler tuning range of 1.18–2.080 μm was achieved by tuning the pump wavelength from 756 to 772 nm, with the idler peak power of 220 mW (corresponding to the pulse energy 22 nJ).

Image described by caption.

Figure 5.25 Schematic of the ZGP‐based master oscillator–power amplifier system. A Q‐switched cryogenic Ho:YLF oscillator provided 0.5‐J pulses at λ = 2.05 μm and 1 Hz repetition rate. The pump beam for the OPO was obtained using one reflection from a CaF2 wedge, while the transmitted pump was directed to the OPA. The ring‐cavity OPO used two 6‐mm‐long ZGP crystals; the number of ZGP crystals in the OPA varied between one and three. PBS, polarizing beam splitter; λ/2, half‐wave plate at 2.05 μm.

Source: reproduced from figure 1 of [128], with permission of OSA, The Optical Society.

The first demonstration of the OPO operation in a GaAs‐based waveguide was reported by Oron et al. [134]. A 13‐mm‐long waveguide was based on an OP‐GaAs with 12 × 3 μm2 cross section. A monolithic OPO cavity was formed by dielectric facet coating. With a pulsed pump near 2 μm (25 ns, 10 kHz), the OPO peak threshold power was 7 W (pulse energy 175 nJ). At the pump peak power of 11.6 W, the overall OPO output peak power of 0.6 W (pulse energy 15 nJ) was generated in the signal (3.6 μm) plus idler (4.5 μm) waves. In a quasi‐CW pump mode (duty factor 5%, chopping rate 1 kHz), the OPO pump threshold was 5.7 W. At the maximum available pump power of 6.6 W, the detected OPO signal power was a few milliwatts. The authors explain the low power output by the fact that the pump power was too close to the OPO threshold to enable efficient energy conversion [134].

The results on the pulsed nanosecond OPOs are summarized in Table 5.5.

5.4 Regime of Ultrashort (ps and fs) Pulses

5.4.1 Ultrafast DFG

Downconversion of near‐IR sources by DFG remains one of the most common approaches for getting ultrafast pulses in the mid‐IR. DFG is attractive because it offers a straightforward solution with a single‐pass geometry. In the early work on ultrafast DFG, Dahinten et al. generated nearly bandwidth‐limited mid‐IR pulses of 1 ps duration, tunable between 4 and 18 μm [22]. The mid‐IR output was generated via difference frequency mixing of Nd:glass laser pulses and pulses from an infrared dye laser, pumped by the same Nd laser. Tuning between 4 and 18 μm was achieved by various combinations of laser dyes and nonlinear crystals (AgGaS2 and GaSe). The energy of the mid‐IR pulses reached a few microjoules, with the photon conversion efficiency ~2%. Ehret and Schneider generated mid‐IR via difference frequency mixing of the signal and idler output of an optical parametric oscillator, pumped by a Ti:sapphire laser (λ = 815 nm) at a repetition rate of 76 MHz. AgGaS2 (2‐mm thick) and GaSe (1‐mm thick) crystals were applied as nonlinear media [135]. GaSe had a larger tuning range and was more efficient in the whole spectral range than AgGaS2. The average IR power at 8.5 μm was 2 mW with GaSe crystal and 1.3 mW with AgGaS2. Kaindl et al. reported on a mid‐IR light source that provides femtosecond pulses on a microjoule energy scale, broadly tunable in the 3–20‐μm wavelength range with pulse durations as short as 50 fs at 5 μm [23]. The pulses were generated by phase‐matched difference‐frequency mixing in a 1‐mm‐thick GaSe using near‐IR signal and idler pulses from a parametric device based on a 1‐kHz Ti:sapphire amplifier system.

In the past decade, an approach based on generating mid‐IR ultrafast pulses via DFG with well‐established ultrafast Yb, Er, and Tm‐doped fibers has gained much popularity. Erny et al. reported on DFG of mid‐IR femtosecond pulses tunable in the 3.2–4.8 μm range from a two‐branch mode‐locked Er‐doped fiber source operating at a repetition rate of 82 MHz. DFG was achieved via nonlinear mixing, in a 2‐mm‐thick MgO:PPLN crystal, of 170‐mW, 65‐fs pump pulses at a fixed wavelength of 1.58 μm with 11.5‐mW, 40‐fs pulses that were tunable between 1.05 and 1.18 μm and were produced in a highly nonlinear fiber with a core diameter of 3.7 μm [136]. The average DFG power of 1.07 mW, produced at 3.6 μm, corresponds to the quantum efficiency of ~30%, if counted with respect to the shorter‐wave pump component.

Winters et al. presented an approach to femtosecond mid‐IR DFG that was based on frequency redshifted solitons and where (as in the previous example) the pump and signal DFG pulses were derived from the same 1.55‐μm Er‐fiber laser [137]. A spectrally shifted pulse was created through intra‐pulse Raman scattering of femtosecond optical solitons propagating in anomalous‐dispersion single‐mode polarization‐maintaining optical fiber (L = 25 m, mode‐field diameter 10.5 μm). The soliton redshift had linear dependence on the injected pump power (Figure 5.26a); thus, the authors were able to generate a signal pulse that was detuned from the pump by a controllable amount, up to more than 1000 cm−1. This allowed generating mid‐IR pulses over the range of 9.7–14.9 μm using a 1‐mm‐thick GaSe or a 5‐mm‐thick AgGaSe2 crystal (Figure 5.26b). According to the authors, AgGaSe2 produced higher DFG output powers than GaSe (possibly because of larger crystal length). Yao et al. performed DFG by mixing the outputs from two photonic crystal fibers (PCF) pumped by the same Yb‐fiber laser (1.035 μm, 1.3 W, 300 fs, 40 MHz). Facilitated by self‐phase modulation, the output spectrum of the first PCF possesses two dominant outermost peaks that can be extended to 970 and 1092 nm. Spectral tuning was realized by varying the coupled‐in power, which alters the wavelength. The second PCF with two closely spaced zero dispersion wavelengths around the laser wavelength was used to generate intense Stokes pulses between 1.24 and 1.26 μm. The two sets of pulses were mixed in an AgGaS2 crystal, which resulted in producing mid‐IR pulses tunable from 4.2 to 9 μm with a maximum average power of 640 μW at 4.5 μm [138].

Image described by caption.

Figure 5.26 (a) Spectra of the frequency‐shifted solitons, along with the pump spectra, obtained through intra‐pulse Raman scattering in a single‐mode fiber at different launched 1.55‐μm pump powers. (b) Normalized DFG spectra obtained with AgGaSe2 and GaSe crystals.

Source: reproduced from figures 1 and 3 of [137], with permission of OSA, The Optical Society.

Through DFG in a 2‐mm‐long fanned‐out OP‐GaAs, Phillips et al. demonstrated a mid‐IR output tunable from 6.7 to 12.7 μm. A Tm‐doped‐fiber oscillator–amplifier system was used to generate 150‐fs pulses at λ ≈ 1.95 μm. These pulses were utilized to generate a comparatively broadband output centered at λ ≈ 2.5 μm, through Raman soliton self‐frequency shift in a fluoride fiber, followed by mid‐IR DFG in the OP‐GaAs crystal [139]. The concept is illustrated in Figure 5.27. By lateral translation of the fan‐out OP‐GaAs (and thus changing the period of the quasi‐phase‐matching grating), a tuning range of 6.7–12.7 μm was achieved with up to 1.3 mW of the average power at ~9 μm.

Setup of a Tm‐fiber laser‐based 6.7–12.7 μm DFG system. The Tm fiber amplifier, PBS, variable attenuator, fluoride fiber, pump combiner, DSF, SMF, etc. are labeled.

Figure 5.27 Setup of a Tm‐fiber laser‐based tunable 6.7–12.7 μm DFG system. After the Tm amplifier, the pulses from one arm were coupled into a single‐mode fluoride fiber in order to facilitate Raman soliton self‐frequency shift to ~2.5 μm. Subsequently, the pulses were recombined with the second‐arm 1.95‐µm beam in the OP‐GaAs to produce DFG. SMF, single‐mode fiber; DSF, dispersion‐shifted fiber; PBS, polarizing beam splitter; LPF, long‐pass filter.

Source: reproduced from figure 1 of [139], with permission of OSA, The Optical Society.

A high‐power mid‐IR DFG source was reported by Zhu et al. [140]. The researchers used intense ultrashort pulses centered at ~1.05 and 1.55 μm, seeded by a common 250‐MHz mode‐locked Er‐fiber oscillator. Half of the output from the oscillator was amplified by an EDFA to about 450 mW and coupled into a highly nonlinear fiber to broaden the spectrum and to generate 1.05‐μm pulses through the process of self‐phase modulation. These seed pulses were amplified by a YDFA to 1.2 W. The other half of the output from the Er‐fiber oscillator was amplified by the second EDFA to about 450 mW (center wavelength 1.55 μm, pulse duration 60 fs). The DFG occurred in a MgO:PPLN crystal, and up to 120 mW of the output power in a broad spectrum spanning from 2.9 to 3.6 μm was produced [140]. Based on a similar approach, even higher DFG powers were produced near 3 μm wavelength, namely 150 mW in a broadband output with a spectral bandwidth from 2.7 to 3.45 μm [141]. By frequency mixing amplified 1.05‐ and 1.55‐μm fs pulses seeded by the same oscillator (the average power after amplification was correspondingly 4 W and 140 mW) and by using a 3‐mm‐long MgO:PPLN crystal, Cruz et al. produced a DFG output at 100 MHz repetition rate and center wavelength tunable from 2.6 to 5.2 μm. The highest DFG average power of 500 mW was reported for an output with an instantaneous span of 2.8–3.5 μm [142].

At longer waves, a high‐power ultrafast DFG system was employed by Beutler et al. [143]. The authors used an AgGaSe2 crystal for DFG between the signal and idler waves of an OPO, which was synchronously pumped by a Yb‐fiber oscillator–amplifier system (1.032 μm, 7.8 W). In the picosecond regime, continuous 5–18 μm tuning was achieved at 80 MHz repetition rate with average power of 140 mW at 6 μm. In the femtosecond regime at 53 MHz repetition rate, similar tunability (5–17 μm) was achieved, with the average power of 69 mW at 6 μm.

5.4.2 Intra‐pulse DFG (Optical Rectification)

Mid‐IR radiation can be generated by mixing spectral components within the same broadband near‐IR pulse. This process is referred to as an optical rectification, or self‐mixing, or intrapulse DFG. In the time domain, a few‐cycle pulse creates nonlinear polarization, resulting in forward emission of long‐wavelength light having an optical period on the order of the pump pulse duration. In the frequency domain, this corresponds to DFG between the spectral components within the same pump pulse [144, 145]. Thus, there is no need for a second near‐IR pulse. However, the shortest mid‐IR wavelength that can be produced via optical rectification is limited by the spectral span of the pump pulse, therefore extremely short (few‐optical‐cycle) pulses with 10–20 fs pulse duration are typically used for mid‐IR optical rectification.

Using phase‐matched optical rectification in GaSe crystal with broadband 20‐fs pulses from a mode‐locked Ti:sapphire oscillator, Kaindl et al. generated femtosecond pulses tunable from 9 to 18 μm at 88‐MHz repetition rate [146]. Direct measurements of the spectrum and the pulse duration at center λ ≈ 11.5 μm demonstrated nearly bandwidth‐limited pulses of 140‐fs duration. Huber et al. produced bandwidth‐limited infrared pulses as short as 50 fs using phase‐matched optical rectification of 10‐fs 780‐nm laser pulses in a thin (90‐μm) GaSe crystal. The central frequency of the transients was continuously tunable by changing the GaSe phase‐matching angle over a wide interval of wavelengths extending from 7 μm all the way to the far‐IR domain, λ = 3000 μm [28].

Recently, intra‐pulse DFG using 1.03‐μm pump pulses was employed to generate high‐average‐power mid‐IR pulses. The pump system included a 90‐W average power thin‐disc Yb:YAG oscillator followed by a nonlinear compression stage to obtain 50 W of the average power in 19‐fs‐long pulses at 100‐MHz repetition rate. A 1‐mm‐thick LiGaS2 (LGS) crystal was used to generate mid‐IR radiation via self‐mixing. The generated spectrum spanned from 6.7 to 18 μm (at −30 dB level) [147]. The duration of the mid‐IR pulse, measured by electro‐optical sampling, was 66 fs, which corresponds to about two cycles of the mid‐IR electric field at the center wavelength of ~11 μm. The generated average mid‐IR power was 100 mW, corresponding to ~0.1% conversion efficiency from the original (uncompressed) 1.03‐μm pump.

Using another high‐power laser source – a mode‐locked Ho:YAG thin‐disk oscillator (λ ≈ 2 μm) with 18.7 W average power and a repetition rate of 77 MHz, whose output pulses were subsequently compressed to 15 fs duration in a nonlinear fiber – Zhang et al. demonstrated, via intra‐pulse DFG in a GaSe crystal, a broadband longwave mid‐IR output with the spectral span of 4.4–20 μm (at −30 dB level), with an average power of 24 mW [148].

Vasilyev et al. [149] showed that relatively long, λ ~ 2.5 μm, central wavelength of a few‐cycle Cr2+:ZnS driving source (20‐fs pulse duration, 6 W average power, 78 MHz repetition rate) enabled the use of a highly nonlinear ZGP crystal for IDFG with high conversion efficiency (>3%) and output power of 148 mW, with a spectral span of 5.8–12.5 μm. An even broader spectrum (although at a smaller, 13 mW, output power) was achieved in GaSe crystal with the same pump: 4.3–16.6 μm for type I and 5.8–17.6 μm for type‐II phase matching.

The main results on ultrafast DFG (including IDFG) are summarized in Table 5.6. Additional information can be found in the comprehensive review [11] and also in Section.

Table 5.6 Summary of the ultrafast downconversion systems.

Nonlinear crystal Tuning range (μm) Pump Output parameters Ref.
Ultrafast DFG
AgGaS2;GaSe 4–10; 6–18 1064 μm, 1 mJ, 2 ps and 1.1–1.4 μm, 1 Hz 1 nJ to 3 μJ/pulse, 1 ps, quant. eff. 2% [22]
GaSe 5.3–18 1.35–1.6 μm, 300 mW, and 2.05–1.65 μm, 230 mW, 76 MHz 2 mW at 8.5 μm, 300 fs, quant. eff. 3.3% [135]
GaSe 3–20 1.26–1.54 μm, 50 μJ and 2.2–1.66 μm, 30 μJ, 1 kHz 1 μJ, 54 fs at 5.5 μm, quant. eff. 10% [23]
PPLN 3.2–4.8 1.58 μm, 170 mW, 65 fs and 1.05–1.18 μm, 11.5 mW, 40 fs, 82 MHz 1.07 mW at 3.6 μm, quant. eff. 30% [136]
AgGaSe2 9.7–14.9 1.55 μm, 135 mW and 1.7–1.85 μm, 37 MHz 1.5 μW, 420 fs [137]
AgGaS2 4.2–9 0.97–1.092 μm, and 1.24–1.26 μm, 50 mW, 300 fs, 40 MHz 640 μW at 4.5 μm [138]
OP‐GaAs 6.7–12.7 1.95 μm, 430 mW, 150 fs and ~2.5 μm, 30 mW, 72 MHz 1.3 mW at ~9 μm [139]
GaSe 8–14 1.55 μm, 550 mW, 50 fs and 1.76–1.93 μm, 100–250 mW, 84 fs, 250 MHz 4 mW at 7.8 μm [150]
GaSe 4–17 1.55 μm, 360 mW, 100 fs and SC 1.7–2.3 μm, 160 mW, 40 MHz 1 mW [27]
GaSe AgGaSe2 10.5–16.5 1.56–1.62 μm and 1.8–1.86 μm, total 1.45 W, 250 fs, 42 MHz 4.3 mW at 13 μm [151]
GaSe 16–20 1.04 μm, 1.1 W and 1.105 μm, 100 mW, 500 fs, 50 MHz 1.5 mW at 18 μm [152]
PPLN 2.9–3.6 1.05 μm, 1.2 W, 90 fs and 1.55 μm, 450 mW, 60 fs, 250 MHz 120 mW, broadband [140]
PPLN 2.7–3.45 1.048 μm, 4 W, 210 fs and 1.57 μm, 140 mW, 60 fs, 100 MHz 150 mW, broadband [141]
PPLN 2.7–4.2 1.04 μm, 2.3 W, 150 fs and 1.3–1.7 μm, 125 MHz 237 mW at ~3.3 μm [153]
PPLN 2.6–5.2 1.048 μm, 4 W, 210 fs and 1.57 μm, 140 mW, 60 fs, 100 MHz 500 mW at ~3.15 μm [142]
AgGaSe2 5–18 1.38–1.98 μm, 2 W and 4.1–2.2 μm, 1.3 W, 2.1–2.6 ps, 80 MHz 140 mW at 6 μm [143]
Ultrafast intra‐pulse DFG (optical rectification)
GaSe 9–18 0.83 μm, 100 mW, 20 fs, Ave. power 1 μW, 140 fs at 11.5 μm [146]
GaSe Broadband 9−12 0.8 μm, 500 mW, 10 fs 10 μW [154]
LGS Broadband 6.7–18 1.03 μm, 50 W, 19 fs, 100 MHz 103 mW [147]
GaSe Broadband 4.4–20 2 μm, 18.7 W, 15 fs, 77 MHz 24 mW [148]
GaSe Broadband 6–18 2 μm, 32 W, 16 fs, 1.25 MHz 450 mW [155]
GaSe Broadband 4.3–16.6; 5.8–17.6 2.5 μm, 5.9 W, 20 fs, 78 MHz 13 mW [149]
ZGP Broadband 5.8–12.5 2.5 μm, 4.5 W, 20 fs, 78 MHz 148 mW [149]
Picosecond OPOs
KTP 1.54 and 3.47 1.064 μm, 29 W, 7 ps, 83 MHz 6.4 W at 3.47 μm quant. eff. 72.5% [156]
PPLN 2.13; 2.32 1.064 μm, 30.8 W, 37 ps, 103 MHz 20.2 W at 2.13 μm 10.5 W at 2.32 μm quant. eff. 76% [157]
PPLN 3.06–4.16 1.064 μm, 16 W, 21 ps, 81 MHz 4.6 W at 3.33 μm quant. eff. 90% [158]
PPLN 2.3–3.5 1.035 μm, 11 W, 150 ps, 1 MHz 1.5 W (1.5 μJ at 1 MHz) quant. eff. 43% [159]
CSP 6.09–6.58 1.064 μm, 600 mW; 1‐μs macropulses at 20 Hz filled with 8.6‐ps micropulses at 450 MHz 30 mW (1.5 mJ in macropulse) quant. eff. 29.5% [160]
Femtosecond OPOs
PPLN 1.7–5.4 790–815 nm, 850 mW, 90 ps, 81 MHz 20 mW at 5.4 μm [161]
PPLN 2.18–3.73 790 nm, 1 W, 20 fs, 100 MHz 33 mW at 3.72 μm, min.duration 33 fs at 2.7 μm [162]
AgGaSe2 4.1–7.9 1.55 μm, 400 mW, 120 fs, 82 MHz 35 mW at 4.55 μm, 22 mW at 5.25 μm [163]
CSP 6.54–7.19 1.029 μm, 3.7 W, 560 fs, 43 MHz 110 mW at 7.05 μm, quant. eff. 20% [164]
CSP 6.32–7.06 1.015–1.074 μm, 0.7–1 W, 140 fs, 80 MHz 32 mW at 6.8 μm, quant. eff. 22% [165]
OP‐GaP 5–12 1.04 μm, 150 fs, 101 MHz 55 mW (5.4 μm), 7.5 mW (11.8 μm) [166]
OPGs
AGS 1.2–10 1.064 μm (10 mJ, 20 ps) Quant. eff. 0.1–10%, threshold 3 GW/cm2 [167]
ZGP 3.9–10 2.8 μm (3 mJ, 100 ps, 3 Hz) Quant. eff. 18%, threshold 0.09 GW/cm2 [168]
ZGP 5–11 3.15 μm (60 μJ, 2.7 ps, 10 Hz) Quant. eff. 20%, threshold 0.1 GW/cm2 [169]
CdSe 3.6–4.38–13 2.8 μm (3 mJ, 100 ps, 3 Hz) Quant. eff. 10%, threshold 0.47 GW/cm2 [170]
GaSe 3–19 2.8 μm (3 mJ, 100 ps, 3 Hz) Quant. eff. 5%, threshold 1.1 GW/cm2 [168]
CSP 6.15–6.73 1.064 μm (2.1 mJ, 20 ps, 5 Hz) Quant. eff. 8.6%, threshold 0.4 GW/cm2 [171]
OPAs
KTP, OPCPA 3.9 1.03 μm (250 mJ, 70 ps, 20 Hz) 8 mJ, 83 fs [172]
ZGP two‐stage OPA 5 2.05 μm (1.6 mJ, 60 fs, 100 Hz) 0.2 mJ, 450 fs [173]
ZGP, OPCPA 7 2 μm (40 mJ, 70 ps, 100 Hz) 0.2 mJ, 180 fs [174]

5.4.3 Ultrafast OPOs

5.4.3.1 Picosecond Mode

The use of ultrashort pump pulses with an SRO operating in the synchronously pumped regime (that is, when the cavity length of the OPO exactly matches the repetition rate of the pump) can dramatically reduce the OPO threshold, down to <100 mW, in terms of the average pump power, thanks to the high peak power of ultrashort pulses that effectively determines the parametric gain [175].

In the early work on mid‐IR OPOs, an efficient and highly stable optical parametric oscillator with both high peak and high average power based on a KTA crystal synchronously pumped by a diode‐pumped mode‐locked Nd:YVO4 oscillator–amplifier system was achieved by Ruffing et al. [156]. The pump laser delivered 7‐ps‐long pulses at 1.064 μm with a repetition rate of 83 MHz and an average power of 29 W. The synchronously pumped OPO had a folded signal‐resonant linear resonator (Figure 5.28) and contained a 15‐mm‐long KTA crystal, cut for type‐II noncritical phase matching. The cavity‐length detuning tolerance was a few 100 μm. The OPO operated at a fixed signal–idler pair of wavelengths; the average power for the signal wave (1.54 μm) was 14.6 W, while for the idler wave (3.47 μm), the power was 6.4 W. The total OPO output of 21 W corresponds to the optical conversion efficiency of 72.5%.

By the use of a 50‐mm‐long MgO‐doped PPLN nonlinear crystal and an Nd:YVO4 mode‐locked laser as a pump (pulse duration 37 ps, repetition rate 103 MHz, average power 30.8 W), Qin et al. obtained 20.2 W of average power at 2.128 μm from a picosecond optical parametric oscillator operating at degeneracy [157]. The OPO pulse duration was measured to be 29 ps, and the optical conversion efficiency was 66%. In the nondegenerate mode of operation, a signal (1.97 μm) power of 12.8 W and an idler (2.32 μm) power of 10.5 W were generated, corresponding to a conversion efficiency of as high as 76%.

An elegant way of generating high‐energy, mid‐IR picosecond pulses at 1 MHz repetition rate was demonstrated in [159]. The authors used an OPO cavity with a length that was a small fraction (1/193) of that required to match the pump repetition rate of 1 MHz. The OPO, based on an MgO‐doped PPLN crystal, was pumped by a fiber MOPA system delivering 11‐μJ, 150‐ps pulses at 1.035 μm. For the 1.55‐m‐long OPO cavity, the resonating near‐IR signal pulses had a repetition rate that was the 193rd harmonic of the 1‐MHz pump. In contrast, the repetition frequency of the nonresonant idler wave, tunable from 2.3 to 3.5 μm, was the same as that of the pump, since it was only generated in the presence of both the pump and the signal pulses. The authors demonstrated OPO idler pulses with energies as high as 1.5 μJ, with the quantum efficiency up to 43%.

Image described by caption and surrounding text.

Figure 5.28 A picosecond optical parametric oscillator based on a KTA crystal, synchronously pumped by a mode‐locked Nd:YVO4 oscillator–amplifier system.

Source: reproduced from figure 1 of [156], with permission of Springer.

5.4.3.2 Femtosecond Mode

In 1997, Burr et al. demonstrated an efficient high‐repetition‐rate femtosecond optical parametric oscillator that was broadly tunable in the mid‐IR [161, 176]. The ring‐cavity OPO in [161] was based on PPLN and was synchronously pumped at a repetition rate of 81 MHz by a mode‐locked Ti:sapphire laser with 90‐fs duration, average power of 850 mW, and tunable (790–815 nm) central wavelength. A small noncollinear angle between the pump and the signal was introduced (Figure 5.29), so that the long‐wavelength idler beam was extracted at an angle to the resonating signal wave. The OPO cavity contained a prism sequence for group velocity dispersion compensation. (In the case of the femtosecond pump, the OPO spectral span is large enough to worry about dispersion of the roundtrip travel time for different spectral components of the resonating wave.) The noncollinear angle between pump and signal, measured internal to the crystal, varied from 1.0° to 1.6°, and PPLN length varied between 0.3 and 0.8 mm, so that the idler was tunable from 1.7 μm to beyond 5.4 μm, with a maximum average power of 200 mW near degeneracy (~1.6 μm), and more than 20 mW at λ = 5.4 μm. Interferometric autocorrelation (Figure 5.29) was used to characterize the mid‐IR idler pulses, which typically had duration of 125 fs. Near degeneracy, the OPO had a pumping threshold as low as 65 mW of average pump power and the pump depletion was as high as 85%.

A PPLN‐based optical parametric oscillator delivering extremely short (few‐optical‐cycle) mid‐IR pulses was reported in [162]. The OPO was synchronously pumped by 20‐fs Ti:sapphire laser pulses. By using a PPLN crystal length as short as 250 μm, and by careful dispersion management, near‐transform‐limited, few‐cycle idler pulses tunable across the 2.18–3.73 μm range have been generated, with 33‐fs duration at λ = 2.7 μm.

Using AgGaSe2, a crystal with much deeper IR cutoff, as compared to PPLN, Marzenell et al. demonstrated operation of an ultrafast OPO that was tunable from 2 to 8 μm with the exception of a small gap near 3–4 μm [163]. To ensure a reduced two‐photon absorption in the AgGaSe2 crystal, the OPO was pumped at 1.55 μm (the output of a Ti:sapphire laser‐pumped OPO). Femtosecond pulses were generated between 1.93 and 2.49 μm (signal wave), and between 4.1 and 7.9 μm (idler wave) with the average output power of 67 mW (at 2.35 μm) and 35 mW (at 4.55 μm). The signal and idler pulse durations were measured to be, respectively, 230–520 and 300–640 fs. For the low‐loss cavity (small output coupling), the OPO threshold was <100 mW.

Image described by caption.

Figure 5.29 Femtosecond mid‐IR PPLN OPO synchronously pumped by a Ti:sapphire laser. A prism sequence was used for group velocity dispersion compensation. The inset shows: (a) idler spectrum (solid curve) and background transmission in air (dashed curve); (b) interferometric autocorrelation of the idler at λ ~ 5.4 μm.

Source: the inset is reproduced from figure 2 of [161], with permission of OSA, The Optical Society.

Kumar et al. reported a femtosecond OPO, based on a CSP crystal. The OPO was pumped near 1‐μm wavelength and delivered idler wave tunability in the range 6.3–7.2 μm [164, 165], which is important for spectroscopic and also medical applications (6.45 μm coincides with the absorption peak of amide II protein of human tissue). The maximum OPO average power reached 100 mW at λ = 7.05 μm.

Maidment et al. reported a broadly tunable femtosecond optical parametric oscillator based on a new semiconductor nonlinear material, OP‐GaP, which enables the production of femtosecond pulses covering a large portion (5–12 μm) of the long‐wavelength mid‐IR region [166]. The pump source was a 1.04‐μm Yb laser delivering 150‐fs pulses at 101 MHz repetition frequency. To obtain broad spectral coverage, seven 1‐mm‐long OP‐GaP crystals, with orientation‐reversal periods variable from Λ = 21.5 to Λ = 34.0 μm were used in this experiment. The idler wave spectra, centered from 5.4 to 11.8 μm, are presented in Figure 5.30, together with the average power data for each crystal. The output average power was 55 mW at 5.4 μm and 7.5 mW at 11.8 μm [166].

Image described by caption.

Figure 5.30 (a) Schematic of the synchronously pumped singly resonant femtosecond OPO based on the OP‐GaP crystal. (b) Idler spectra obtained with seven orientation‐reversal periods of the OP‐GaP. Diamonds and squares indicate the measured average power.

Source: reproduced from figures 3 and 4 of [166], with permission of OSA, The Optical Society.

The main results on ultrafast OPOs are summarized in Table 5.6. Some additional information on ultrafast optical parametric sources can be found in the reviews [11, 177], and also in Sections and.

5.4.4 Ultrafast OPGs

The principle of operation of a traveling‐wave “superfluorescent” OPG is based on a single‐pass high‐gain (>1010) amplification of quantum noise in a nonlinear crystal pumped by intense short laser pulses.

The main characteristics of the traveling‐wave OPGs are:

  • Simplicity (no cavity needed).
  • Ability to produce high peak power outputs (>1 MW) in the form of a single pulse.
  • Broad tunability, restricted only by the phase matching and crystal transparency.
  • No buildup time. This allows generating synchronized, independently tunable pulses from different OPGs pumped by the same laser, which is attractive for time‐resolved pump‐probe spectroscopy.
  • High pump power density, typically >1 GW/cm2, is needed to achieve an OPG threshold.

In their early work, Elsaesser et al. reported a traveling‐wave OPG operating over the whole range of 1.2–8 μm based on proustite (Ag3AsS3) crystal pumped by single pulses of Nd:YAG laser radiation (λ = 1.06 μm, pulse duration 21 ps) with a threshold pump intensity 6 GW/cm2 [178]. The energy conversion efficiency amounted to 10−4–10−2, and the OPG‐pulse spectral bandwidth was 10–40 cm−1, with OPG‐pulse duration being 8 ps. The OPG conversion efficiency was significantly improved (to 10−3–10−1) when AGS crystals were used. Two AGS crystals of 1.5 and 3 cm in length were placed in series and were pumped by picosecond Nd:YAG laser radiation [167]. The output was tunable in the range of 1.2–10 μm and the pumping threshold was 3 GW/cm2.

Subsequently, using λ = 2.8‐μm 100‐ps pump pulses from a mode‐locked Er,Cr:YSGG laser, traveling‐wave OPGs with a number of crystals (ZGP, CdSe, and GaSe) were demonstrated [24, 168, 170, 179]. The ZGP crystal has shown the best performance in terms of the smallest pump threshold, which was only 0.09 GW/cm2 for a 4‐cm‐long crystal, with the OPG tunability from 3.9 to 10 μm and quantum conversion efficiency that reached 18% (see Table 5.6) [168]. Using the same pump source and a CdSe crystal, the tuning range was 3.6–4.3 μm (signal wave) and 8–13 μm (idler wave) [170]. It has also been shown that in the important spectral range of 8–12 μm, CdSe was superior to ZGP in GaSe, in the sense of larger conversion efficiency and narrower linewidth.

When pumped by λ = 2.8‐μm pulses, an OPG based on GaSe crystal holds the record for the broadest continuous tunability from 3.3 to 19 μm. This was achieved in a double‐pass OPG geometry with a single z‐cut angle‐tuned GaSe crystal [168]. Despite the fact that GaSe can be cleaved only along the (001) plane (z‐cut, θ = 0°), its extremely large birefringence (Δn ~ 0.35) allows almost any conceivable three‐wave interaction in its transparency range to be phase‐matched. The GaSe crystal was 14‐mm long and an elliptical focusing with an aspect ratio 1:20 was used for the 2.8‐μm 100‐ps 3‐mJ pump beam – in order to keep the beam size sufficiently large in the walk‐off plane (the walk‐off amounted to 0.8 mm for the internal phase‐matching angle of θ = 12°). The OPG threshold intensity was 1.1 GW/cm2. At the pump intensity ~5 GW/cm2, the quantum conversion efficiency was 5% in the 4–11 μm range (and declined at longer wavelengths) [168, 179]. Figure 5.31 shows experimental tuning curves for the type I and type II GaSe OPG.

Image described by caption and surrounding text.

Figure 5.31 OPG tuning curves obtained with a single z‐cut GaSe crystal pumped at 2.8 μm. Vertical bars represent experimental linewidths. Solid curves – calculated tuning curves. Inset: transmission spectrum of an uncoated crystal.

Source: reproduced from figure 3 of [179], with permission of OSA, The Optical Society.

The main disadvantage of OPGs is high divergence – the result of the lack of the optical cavity. For example, both collinear and noncollinear three‐wave interactions may occur in the same pumping geometry. This can also result in spectral broadening (especially for type‐I phase matching). A solution to this problem, however, may be a double‐crystal or a double‐pass scheme with a single crystal, where the second pass filters out the off‐axis components of the OPG beam produced in the first pass. Very often an OPG is used as a “seed” source for a high‐power ultrafast OPA system, as in [173].

The summary of the main OPG results can be found in Table 5.6.

5.4.5 Ultrafast OPAs

Ultrafast pulses of mid‐IR coherent radiation play an increasingly important role in such fields as strong‐field physics and generation of coherent X‐rays by high harmonic generation. An OPA approach allows producing high peak power (up to >100 GW) femtosecond pulses with superior spatial, temporal, and spectral characteristics. This is especially true when optical parametric chirped pulse amplification (OPCPA) is exploited. The concept of chirped‐pulse amplification was first developed for laser amplifiers, but it was soon realized that it is also suitable for OPAs. First, the seed pulse is temporally stretched – to 100‐ps–1‐ns duration. This makes it possible to apply much higher pump energies from comparatively simple Q‐switched lasers operating in the nanosecond regime and obtain much higher amplified pulse energies. After optical parametric amplification, the chirped mid‐IR pulse is compressed (e.g. with a diffraction grating pair), in some occasions down to few optical cycles.

Andriukaitis et al. demonstrated a compact 20‐Hz repetition‐rate mid‐IR OPCPA system operating at a central wavelength of 3.9 μm and delivering 8‐mJ pulses of 83 fs pulse duration (<7 optical cycles) [172]. The three OPAs in series were based on KTP crystals and were driven by 190‐fs pulses at 1030 nm. The OPA seed pulse was generated via white‐light generation in a 6‐mm‐long YAG crystal using a 2‐μJ portion of the pump. The energy in the signal wave (at ~1.4 μm) after the final OPA stage was 65 μJ. Finally, the OPCPA section, consisting of two stages using 10‐mm‐long KTA crystals, pumped by an Nd system with 70‐ps‐long pulses with energies 250 mJ, produced 22‐mJ signal (at 1.4 μm) and 13‐mJ idler (at 3.9 μm) pulses. Subsequently, the idler pulses were compressed to an 83‐fs duration [172].

Sanchez et al. used an OPCPA architecture to yield <8 optical‐cycle duration pulses at a longer (7‐μm) central wavelength at 100 Hz repetition rate [174]. The mid‐IR seed at λ = 7 μm for the OPA chain was generated in a CSP crystal through DFG of the femtosecond outputs of two mutually coherent Er‐ and Tm‐fiber lasers. An OPCPA was pumped by optically synchronized Ho:YLF laser pulses at 2 μm (40 mJ, 11 ps). The stretched, picojoule‐level seed pulses (of 6‐ps duration) were amplified in a chain of three consecutive noncollinear ZGP OPA stages (with ZGP crystal lengths of 5, 5, and 3 mm respectively) and subsequently compressed to get 7‐μm pulses with 0.2‐mJ energy and 180‐fs duration. In addition, the system produced intrinsically phase‐stable pulses – in terms of their carrier‐envelope phase [174].

The other selected results of ultrafast OPAs are listed in Table 5.6.

5.5 Raman Frequency Converters

A Raman frequency converter is a coherent light source where an amplifier medium is based on stimulated Raman scattering (SRS), a parametric third‐order NLO process [180]. SRS has been successfully utilized for several decades as a means to generate longer wavelengths from relatively mature laser sources [181]. The Raman‐active medium can be either a bulk crystal, an optical fiber, a waveguide, a liquid, or a gas. Raman frequency conversion can be achieved both as a single‐pass super‐radiant SRS emission or as output from an oscillator (often referred to as a “Raman laser”). In fibers or waveguides, the long interaction length makes it easy to exceed the oscillation threshold, even in the CW mode [182]. Thanks to a broad Raman gain contour in glasses and other solids, Raman oscillators can be made tunable over hundreds of wavenumbers, even with a fixed‐wavelength pump. Also, the ability to cascade several Raman Stokes orders in the same device opens up the possibility of obtaining mid‐wavelengths from a significantly shorter wavelength pump.

5.5.1 Crystalline Raman Converters

In general, mid‐IR Raman conversion is challenging, as compared to the visible and near‐IR range – due to a reduced Raman gain, which is inversely proportional to the wavelength [180], and also due to limited IR transparency of materials. Currently, the most common Raman crystals include SrWO4, KGdWO4, BaWO4, PbWO4, Ba(NO3)2, YVO4, GdVO4, LiIO3, and diamond. Among these crystals, BaWO4 is particularly attractive due to its respectable optical properties and high Raman gain. When pumped with 10‐ns pulses at λ = 1.56 μm, multistage SRS to the first, second, third, and fourth Stokes‐shifted components with the new wavelengths of, respectively, 1.8, 2.2, 2.75, and 3.7 μm were observed, with pump thresholds in the range of 2.7–8 mJ [183]. Also, using BaWO4 crystal, a diode‐end‐pumped actively Q‐switched intracavity Raman laser with Tm,Ho:GdVO4 (2.053 μm) laser gain medium has been demonstrated [184]. At the first Stokes wavelength of 2.53 μm, the average output power of 186 mW (1 kHz, 7.8 ns) was obtained at an 802‐nm diode pump power of 2.8 W, corresponding to a diode‐to‐Stokes optical conversion efficiency of 6.6%. The Raman lasing threshold was 2 W with respect to the diode power.

5.5.2 Fiber Raman Converters

Single‐pass cascaded Raman generation was demonstrated by Rakich et al. [185] as a practical and efficient means of power transfer from telecommunications wavelengths to the mid‐IR through use of conventional silica fibers and amplifiers. Silica fibers were shown to facilitate 37% efficient Raman power conversion from 1.53 to 2.15 μm and 16% from 1.53 to 2.41 μm band, using 2‐ns 680‐kHz‐repetition‐rate pulses from an all‐fiber laser source. At high powers, anomalous dispersion and Kerr nonlinearities give rise to modulation instabilities, yielding significant spectral broadening and pulse distortion. For this reason, normal dispersion fibers are more desirable for controlled and maximally efficient cascaded Raman at the nanosecond time scale. The authors in [185] have shown that normal fiber dispersion can be obtained over the entire silica transparency window through proper choice of fiber cutoff wavelength λc and numerical aperture. The spectral evolution, with the pump power increase, of the resulting cascaded Raman process is summarized by the intensity map of Figure 5.32. As the laser power is increased, significant and controlled spectral redshifts take place through higher‐order cascaded Raman power transfer, where the fundamental 1.531‐μm pump wavelength is shifted in 14.7‐THz (490‐cm−1) steps to 1.64, 1.78, 1.94, 2.14, and 2.41 μm wavelength bands. While the spectral width of each successive order does broaden, the generated spectral bands are very clean, showing negligible power transfer to continuum. Despite the high material losses of silica at 2.41 μm, a strong fifth Raman order is formed, producing significant (16%) power transfer from the pump.

Image described by caption and surrounding text.

Figure 5.32 Cascaded Raman process in a silica fiber: an intensity map shows the measured fraction of the pump power (1.531 μm) and the Stokes orders at 1.64, 1.78, 1.94, 2.14, and 2.41 μm, as the pump laser power is increased.

Source: reproduced from figure 3a of [185], with permission of OSA, The Optical Society.

More information on fiber‐based Raman converters can be found in Section.

5.5.3 Silicon Raman Converters

With the Raman frequency shift of 520 cm−1, silicon Raman lasers are considered as potential candidates for covering the technologically important mid‐IR region of 2.2–6.5 μm, that is below the multiphonon absorption in silicon (it becomes significant above 6.5 μm). High Raman gain coefficient combined with good crystal quality, high thermal conductivity, and high optical damage threshold renders silicon a very attractive Raman medium, even when compared to the very best Raman crystals [186, 187]. Silicon‐based photonic devices have an additional advantage due to their compatibility with the manufacturing infrastructure of silicon electronics, which offers the possibility of creating integrated low‐loss waveguides, microcavities, and resonators.

The first silicon Raman laser was reported in the near‐IR by Boyraz and Jalali [188]. The laser consisted of a silicon gain medium – a 2‐cm‐long rib waveguide – incorporated into a fiber loop cavity that was synchronously pumped with 30‐ps 1540‐nm pulses at a 25 MHz repetition rate and produced output pulses at the Stokes wavelength of 1675 nm. A lasing threshold was at 9 W of peak power, and the slope efficiency was 8.5%. This result was followed by the demonstration of the first CW silicon Raman laser (λpump = 1550 nm, λRaman = 1686 nm), with the pump threshold of 180 mW [189]. Here, the authors incorporated reverse biased PIN diodes along the waveguides to remove free carriers created by two‐photon absorption in silicon, resulting in additional severe losses. Takahashi et al. reported a CW near‐IR Raman silicon laser (λpump = 1428 nm, λRaman = 1543 nm) using a high‐quality‐factor nanocavity, yielding a device with a cavity size of less than 10 μm and a very low lasing threshold of 1 μW [190]. Using a silicon rib waveguide in a ring‐cavity “racetrack” configuration, Rong et al. obtained, with a 1550‐nm pump, cascaded Raman lasing at 1686 nm (first Stokes) and 1848 nm (second Stokes), with the threshold of appearance of the second Stokes of 120 mW, with respect to the coupled pump power, and with an output power at the second Stokes of 5 mW [191].

In the mid‐IR, Raghunathan et al. demonstrated a silicon Raman amplifier, where an input signal at 3.39 μm wavelength was amplified by 12 dB in a 2.5‐cm‐long bulk silicon sample. The sample was pumped with 5‐ns‐long pulses at 2.88 μm with a peak pump intensity of 217 MW/cm2 (close to the damage threshold of the sample surface) [192]. Parametric amplification and parametric generation of mid‐IR light based on another third‐order nonlinear effect, namely four‐wave mixing (FWM) based on Kerr nonlinearity, were demonstrated independently by two groups using silicon waveguides and pump wavelengths near one‐half the bandgap energy (E ≈ 0.55 eV, λ ≈ 2.2 μm), so that there is no parasitic two‐photon absorption [193, 194]. In [193], a parametric gain of 25.4 dB was obtained in a 4‐mm‐long silicon chip using 2‐ps, 28‐W peak power pulses.

Using a silicon microresonator, Griffith et al. demonstrated near‐octave spanning, coherent mid‐IR frequency comb generation based on the interplay between FWM and coherent generation of Stokes and anti‐Stokes frequencies via SRS in silicon [195]. The silicon microresonator was pumped with a monochromatic OPO source at 2.6 μm with 180 mW of power and produced a broadband output spanning 2.46–4.28 μm. The authors of [195] note that silicon suffers from three‐photon absorption (3PA) in the 2.2–3.3 μm wavelength range, and the generated photocarriers with long lifetimes can cause significant mid‐IR absorption. To mitigate this, the silicon microresonators were embedded in an integrated PIN diode to enable extraction of the generated free carriers. When a reverse‐bias voltage of −12 V was applied to the PIN junction, carriers were swept out of the diode depletion region, thus considerably improving the device performance.

5.5.4 Diamond Raman Converters

Diamond has several properties well suited for nonlinear mid‐IR generation including wide transparency window, exceptional thermal conductivity, high damage threshold, high Raman gain coefficient, and an absence, due to its crystal symmetry, of the first‐order IR lattice absorption. The large bandgap of diamond (5.4 eV) ensures that there is no two‐photon absorption and a large Raman shift frequency (1332 cm−1) allows a big wavelength shift from the pump wavelength. However, there are two mid‐IR absorption bands in bulk diamond. The three‐phonon absorption band extends from 2.5 to 3.75 μm and the stronger two‐phonon band is prominent from 3.75 to 6 μm.

Sabella et al. reported a pulsed mid‐IR diamond Raman laser with the output that was tuned from 3.38 to 3.80 μm (that is, in the local minima between the two multi‐phonon absorption bands) [196]. The 21‐mm‐long cavity contained an 8‐mm‐long uncoated diamond crystal, grown by chemical vapor deposition (CVD). As a pump, the authors used a tunable OPO in the vicinity of 2.5 μm with up to 1 mJ/pulse (10 Hz, 5 ns) and a linewidth of 0.55 cm−1, which was less than the Raman linewidth of diamond (1.5 cm−1). The Raman laser tuning of 3.38–3.80 μm was obtained through varying the pump wavelength in the range 2.33–2.52 μm. The highest pulse energy obtained was 80 μJ at 3.7 μm, which corresponds to 15% conversion efficiency (22% quantum efficiency) with respect to the pump entering the crystal, with a threshold near 250 μJ/pulse.

Latawiec et al. demonstrated a CW diamond Raman laser operating near 2‐μm wavelength. The authors used a high‐quality‐factor (Q) “racetrack” diamond microresonator (path length 600 μm, waveguide width 800 nm, height 700 nm) embedded in silica on a silicon chip. By tuning the telecom‐range pumping wavelength (λ ∼ 1.6 μm), the output was discretely tunable over a ∼ 100 nm bandwidth around the Stokes wavelength of 2 μm with the output power of 250 μW. The CW operation pump threshold power was 85 mW [197].

5.5.5 Other Raman Converters

Whispering gallery mode (WGM) optical microcavities, such as microspheres, microdisks, and microtoroids, offer good possibilities for SRS emission, thanks to their high quality factors and small mode volume. In the near‐IR range, SRS threshold powers as low as 74 and 15 μW were observed, respectively in silica microtoroids [198] and CaF2 microdisks [199]. Chalcogenide glasses such as As2S3 and As2Se3 are very advantageous for SRS emission. As2S3 has a high Raman gain coefficient, almost 100 times that of silica, and has a transparency window that extends up to 6 μm. Vanier et al. presented the first observation of SRS in high‐quality‐factor 40‐μm‐diameter As2S3 microspheres fabricated using CO2 laser melting [200]. Raman laser emission (in both forward and backward direction) was observed with the coupled 1550‐nm pump power threshold of 13 μW, with internal conversion efficiency of 10.7%. Although the Raman emission wavelength was still in the near‐IR (at 1640.7 nm), considering a respectable mid‐IR transparency of As2S3, microspheres made of this material can be regarded as good candidates for mid‐IR Raman generation, e.g. via cascaded SRS process.

Using stimulated backward Raman scattering in solid para‐hydrogen at T = 4 K (Raman shift 4149.7 cm−1, linewidth 7 MHz), Kuyanov et al. developed a single‐pass Raman converter operating in the mid‐IR range with continuous tunability of 4.4–8 μm [201]. The device was pumped by a tunable near‐IR optical parametric oscillator (OPO) with an output energy of 20 mJ, 7‐ns pulse duration, and 20‐Hz repetition rate. The output wavelength tuning (4.4–8 μm, spectral linewidth 0.4 cm−1) was achieved by tuning the pump in the range 1.56–1.85 μm. The mid‐IR output energies varied from 1.7 mJ at 4.4 μm to 120 μJ at 8 μm, which corresponds to quantum efficiencies of 53 and 8%, respectively. The threshold for the SRS generation was between 2 and 8 mJ; also, the backward‐propagating beam contained the dominant part of the SRS radiation.

Gladyshev et al. demonstrated a Raman source at λ = 4.4 μm based on silica hollow core fiber (HCF) filled with hydrogen gas [202]. The hollow core diameter was 77 μm and the fiber cladding was formed by 10 non‐contacting silica capillaries with 1.15‐μm‐thick walls (a “revolver” design, Figure 5.33). The calculated optical losses for the fundamental modes for the pump (1.56 μm) and the Stokes (4.4 μm) wavelengths were, respectively, 0.0025 and 0.92 dB/m. A 15‐m segment of the revolver HCF was filled with molecular hydrogen at a pressure of 30 atm. The fiber ends were hermetically glued into miniature gas cells with sapphire windows for light in‐ and outcoupling. Under pumping by a pulsed erbium fiber laser (1.56 μm, 2 ns, 25 kHz, 1.2 W) in a single‐pass geometry, the Raman quantum conversion efficiency reached 15%, corresponding to 30 mW of the average power (0.6 kW peak) at λ = 4.4 μm [202].

Image described by caption.

Figure 5.33 Schematic of the 4.4‐μm Raman source based on a silica hollow core fiber (HCF). L1, L2, incoupling fused silica lenses; W1, W2, sapphire windows of the gas cells; L3, ZnSe collimating lens. The inset shows an electron‐microscope image of the fiber cross section.

Source: reproduced from figure 1 of [202], with permission of Turpion, Ltd and Kvantovaya Elektronika.

5.6 Summary

Difference frequency generators conveniently convert radiation of near‐IR lasers to the mid‐IR, preserving coherence properties of the pump (e.g. narrow linewidth) and are widely used in spectroscopy owing to the broad and continuous tuning. The main drawback of DFG sources is typically low conversion efficiency and also the need for two driving laser sources. OPOs, on the other hand, need only one pump laser, are more efficient (the quantum conversion efficiency from the pump can be as high as 80%), and can produce high average power (>10 W) or high pulse energy (>100 mJ). However, this comes at the expense of the requirement for a resonant cavity. As for nonlinear crystals, QPM materials, such as PPLN and PPLT, as well as PPKTP crystal and its family, serve as a workhorse for frequency conversion in the whole wavelength range of 1–5 μm, thanks to their robustness, high nonlinearity, and affordable cost. OP‐GaAs and OP‐GaP, grown by epitaxy, are remarkably robust and are being used in a number of longwave (>5 μm) applications. OP‐GaAs serves (similar to bulk ZGP) in many high‐power OPO applications in the whole range between 3 and 10 μm. The newly developed nonlinear crystal CSP allows direct downconversion from a 1‐μm pump to the 6‐μm output under noncritical phase matching. Also, Raman frequency converters based on the well‐developed silicon technology have the potential for covering the technologically important region of 2–6.5 μm.

References

  1. 1 Fejer, M.M., Magel, G.A., Jundt, D.H., and Byer, R.L. (1992). Quasi‐phase‐matched 2nd harmonic‐generation: tuning and tolerances. IEEE J. Quantum Electron. 28: 2631.
  2. 2 Lim, E., Hertz, H., Bortz, M., and Fejer, M. (1991). Infrared radiation generated by quasi‐phasematched difference‐frequency‐mixing in a periodically‐poled lithium niobate waveguide. Appl. Phys. Lett. 59: 2207.
  3. 3 Yariv, A. (1989). Quantum Electronics, 3e. New York: Wiley.
  4. 4 Vodopyanov, K.L. (2003). Pulsed mid‐IR optical parametric oscillators. In: Solid‐State Mid‐Infrared Laser Sources (ed. I.T. Sorokina and K.L. Vodopyanov), Topics Appl. Phys., vol. 89. Berlin: Springer.
  5. 5 Myers, L.E. and Bosenberg, W.R. (1997). Periodically poled lithium niobate and quasi‐phase‐matched optical parametric oscillators. IEEE J. Quantum Electron. 33: 1663.
  6. 6 Shoji, I., Kondo, T., Kitamoto, A., Shirane, M., and Ito, R. (1997). Absolute scale of second‐order nonlinear‐optical coefficients. J. Opt. Soc. Am. B 14: 2268.
  7. 7 Peltz, M., Bäderl, U., Borsutzky, A., Wallenstein, R., Hellström, J., Karlsson, H., Pasiskevicius, V., and Laurell, F. (2001). Optical parametric oscillators for high pulse energy and high average power operation based on large aperture periodically poled KTP and RTA. Appl. Phys. B 73: 663.
  8. 8 Roberts, D.A. (1992). Simplified characterization of uniaxial and biaxial nonlinear optical crystals: a plea for standardization of nomenclature and conventions. IEEE J. Quantum Electron. 28: 2057.
  9. 9 Petrov, V., Noack, F., Tunchev, I., Schunemann, P., and Zawilski, K. (2009). The nonlinear coefficient d36 of CdSiP2. Proc. SPIE 7197: 719721.
  10. 10 Fossier, S., Salaün, S., Mangin, J., Bidault, O., Thenot, I., Zondy, J.J., Chen, W., Rotermund, F., Petrov, V., Petrov, P., Henningsen, J., Yelisseyev, A., Isaenko, L., Lobanov, S., Balachninaite, O., Slekys, G., and Sirutkaitis, V. (2004). Optical, vibrational, thermal, electrical, damage and phase‐matching properties of lithium thioindate. J. Opt. Soc. Am. B 21: 1981.
  11. 11 Petrov, V. (2015). Frequency down‐conversion of solid‐state laser sources to the mid‐infrared spectral range using non‐oxide nonlinear crystals. Prog. Quantum Electron. 42: 1.
  12. 12 Nikogosyan, D.N. (1997). Properties of Optical and Laser‐Related Materials. Chichester, NY: Wiley.
  13. 13 Skauli, T., Vodopyanov, K.L., Pinguet, T.J., Schober, A., Levi, O., Eyres, L.A., Fejer, M.M., Harris, J.S., Gerard, B., Becouarn, L., Lallier, E., and Arisholm, G. (2002). Measurement of the nonlinear coefficient of orientation‐patterned GaAs and demonstration of highly efficient second‐harmonic generation. Opt. Lett. 27: 628.
  14. 14 Wagner, H.P., Kühnelt, M., Langbein, W., and Hvam, J.M. (1998). Dispersion of the second‐order nonlinear susceptibility in ZnTe, ZnSe, and ZnS. Phys. Rev. B 58: 494.
  15. 15 Zhang, H.Y., He, X.H., Shih, Y.H., Schurman, M., Feng, Z.C., and Stall, R.A. (1996). Study of nonlinear optical effects in GaN:Mg epitaxial film. Appl. Phys. Lett. 69: 2953.
  16. 16 Miller, R.C. (1964). Optical second harmonic generation in piezoelectric crystals. Appl. Phys. Lett. 5: 17.
  17. 17 Byer, R.L. (1997). Quasi‐phasematched nonlinear interactions and devices. J. Nonlinear Opt. Phys. Mater. 6: 549.
  18. 18 Schunemann, P.G., Zawilski, K.T., Pomeranz, L.A., Creeden, D.J., and Budni, P.A. (2016). Advances in nonlinear optical crystals for mid‐infrared coherent sources. J. Opt. Soc. Am. 33: D36.
  19. 19 Schunemann, P.G. and Zawilski, K.T. (2013). Nonlinear optical CdSiP2 crystal and producing method and devices therefrom. U.S. patent 8,379,296, 19 February 2013.
  20. 20 Auerhammer, J.M. and Eliel, E.R. (1996). Frequency doubling of mid‐infrared radiation in gallium selenide. Opt. Lett. 21: 773.
  21. 21 Eckhoff, W.C., Putnam, R.S., Wang, S.X., Curl, R.F., and Tittel, F.K. (1996). A continuously tunable long‐wavelength cw IR source for high‐resolution spectroscopy and trace‐gas detection. Appl. Phys. B 63: 437.
  22. 22 Dahinten, T., Plodereder, U., Seilmeier, A., Vodopyanov, K.L., Allakhverdiev, K.R., and Ibragimov, Z.A. (1993). Infrared pulses of 1 picosecond duration tunable between 4 μm and 18 μm. IEEE J. Quantum Electron. 29: 2245.
  23. 23 Kaindl, R.A., Wurm, M., Reimann, K., Hamm, P., Weiner, A.M., and Woerner, M. (2000). Generation, shaping, and characterization of intense femtosecond pulses tunable from 3 to 20 μm. J. Opt. Soc. Am. B 17: 2086.
  24. 24 Vodopyanov, K.L. (1993). Parametric generation of tunable infrared radiation in ZnGeP2 and GaSe pumped at 3 μm. J. Opt. Soc. Am. B 10: 1723.
  25. 25 Keilmann, F., Gohle, C., and Holzwarth, R. (2004). Time‐domain mid‐infrared frequency‐comb spectrometer. Opt. Lett. 29: 1542.
  26. 26 Ruehl, A., Gambetta, A., Hartl, I., Fermann, M.E., Eikema, K.S.E., and Marangoni, M. (2012). Widely‐tunable mid‐infrared frequency comb source based on difference frequency generation. Opt. Lett. 37: 2232.
  27. 27 Keilmann, F. and Amarie, S. (2012). Mid‐infrared frequency comb spanning an octave based on an Er fiber laser and difference‐frequency generation. J. Infrared Milli. Terahz Waves 33: 479.
  28. 28 Huber, R., Brodschelm, A., Tauser, F., and Leitenstorfer, A. (2000). Generation and field‐resolved detection of femtosecond electromagnetic pulses tunable up to 41 THz. Appl. Phys. Lett. 76: 3191.
  29. 29 Reimann, K., Smith, R.P., Weiner, A.M., Elsaesser, T., and Woerner, M. (2003). Direct field‐resolved detection of terahertz transients with amplitudes of megavolts per centimeter. Opt. Lett. 28: 471.
  30. 30 Shi, W., Ding, Y.J., Fernelius, N., and Vodopyanov, K.L. (2002). Efficient, tunable, and coherent 0.18–5.27‐THz source based on GaSe crystal. Opt. Lett. 27: 1454.
  31. 31 Schubert, O., Hohenleutner, M., Langer, F., Urbanek, B., Lange, C., Huttner, U., Golde, D., Meier, T., Kira, M., Koch, S.W., and Huber, R. (2014). Sub‐cycle control of terahertz high‐harmonic generation by dynamical Bloch oscillations. Nat. Photon. 8: 119.
  32. 32 Kubler, C., Huber, R., Tubel, S., and Leitenstorfer, A. (2004). Ultrabroadband detection of multi‐terahertz field transients with GaSe electro‐optic sensors: approaching the near infrared. Appl. Phys. Lett. 85: 3360.
  33. 33 Ebert, C.B., Eyres, L.A., Fejer, M.M., and Harris, J.S. (1999). MBE growth of antiphase GaAs films using GaAs/Ge/GaAs heteroepitaxy. J. Cryst. Growth 201: 187.
  34. 34 Koh, S., Kondo, T., Ebihara, M., Ishiwada, T., Sawada, H., Ichinose, H., Shoji, I., and Ito, R. (1999). GaAs/Ge/GaAs sublattice reversal epitaxy on GaAs (100) and (111) substrates for nonlinear optical devices. Jpn. J. Appl. Phys. 38: L508.
  35. 35 Eyres, L.A., Tourreau, P.J., Pinguet, T.J., Ebert, C.B., Lallier, E., Harris, J.S., Fejer, M.M., and Gerard, B. (2000). Quasi‐phasematched frequency conversion in thick all‐epitaxial, orientation‐patterned GaAs films, Technical Digest. In: Advanced Solid State Lasers. Washington, DC: Optical Society of America, Paper TuA2.
  36. 36 Lynch, C., Bliss, D.F., Zens, T., Lin, A., Harris, J.S., Kuo, P.S., and Fejer, M.M. (2008). Growth of mm‐thick orientation‐patterned GaAs for IR and THz generation. J. Cryst. Growth 310: 241.
  37. 37 Schunemann, P.G. and Setzler, S.D. (2011). Future directions in quasi‐phasematched semiconductors for mid‐infrared lasers. Proc. SPIE 7917: 79171F.
  38. 38 Kanner, G.S., Marable, M.L., Singh, N.B., Berghmans, A., Kahler, D., Wagner, B., Lin, A., Fejer, M.M., Harris, J.S., and Schepler, K.L. (2009). Optical probes of orientation‐patterned ZnSe quasi‐phase‐matched devices. Opt. Eng. 48: 114201.
  39. 39 Miragliotta, J., Wickenden, D.K., Kistenmacher, T.J., and Bryden, W.A. (1993). Linear‐ and nonlinear‐optical properties of GaN thin films. J. Opt. Soc. Am. B 10: 1447.
  40. 40 Abe, M., Sato, H., Shoji, I., Suda, J., Yoshimura, M., Kitaoka, Y., Mori, Y., and Kondo, T. (2010). Accurate measurement of quadratic nonlinear‐optical coefficients of gallium nitride. J. Opt. Soc. Am. B 27: 2026.
  41. 41 Chowdhury, A., Ng, H.M., Bhardwaj, M., and Weimann, N.G. (2003). Second‐harmonic generation in periodically poled GaN. Appl. Phys. Lett. 83: 1077.
  42. 42 Mita, S., Collazo, R., and Sitar, Z. (2009). Fabrication of a GaN lateral polarity junction by metalorganic chemical vapor deposition. J. Cryst. Growth 311: 3044.
  43. 43 Katayama, R., Kuge, Y., Onabe, K., Matsushita, T., and Kondo, T. (2006). Complementary analyses on the local polarity in lateral polarity‐inverted GaN heterostructure on sapphire (0001) substrate. Appl. Phys. Lett. 89: 231910.
  44. 44 Hite, J., Twigg, M., Mastro, M., Freitas, J. Jr., Meyer, J., Vurgaftman, I., O’Connor, S., Condon, N., Kub, F., Bowman, S., and Eddy, C. Jr. (2012). Development of periodically oriented gallium nitride for non‐linear optics. Opt. Mater. Express 2: 1203.
  45. 45 Sanders, S., Lang, R.J., Myers, L.E., Fejer, M.M., and Byer, R.L. (1996). Broadly tunable mid‐IR radiation source based on difference frequency mixing of high power wavelength‐tunable laser diodes in bulk periodically poled LiNbO3. Electron. Lett. 32: 218.
  46. 46 Weibring, P., Richter, D., Walega, J.G., and Fried, A. (2007). First demonstration of a high performance difference frequency spectrometer on airborne platforms. Opt. Express 15: 13476.
  47. 47 Weibring, P., Richter, D., Fried, A., Walega, J.G., and Dyroff, C. (2006). Ultra‐high‐precision mid‐IR spectrometer II: system description and spectroscopic performance. Appl. Phys. B 85: 207.
  48. 48 Guha, S., Barnes, J.O., and Gonzalez, L.P. (2014). Multiwatt‐level continuous‐wave midwave infrared generation using difference frequency mixing in periodically poled MgO‐doped lithium niobate. Opt. Lett. 39: 5018.
  49. 49 Tadanaga, O., Yanagawa, T., Nishida, Y., Miyazawa, H., Magari, K., Asobe, M., and Suzuki, H. (2006). Efficient 3‐μm difference frequency generation using direct‐bonded quasi‐phase‐matched LiNbO3 ridge waveguides. Appl. Phys. Lett. 88: 061101.
  50. 50 Richter, D., Weibring, P., Fried, A., Tadanaga, O., Nishida, Y., Asobe, M., and Suzuki, H. (2007). High‐power, tunable difference frequency generation source for absorption spectroscopy based on a ridge waveguide periodically poled lithium niobate crystal. Opt. Express 15: 564.
  51. 51 Denzer, W., Hancock, G., Hutchinson, A., Munday, M., Peverall, R., and Ritchie, G.A.D. (2007). Mid‐infrared generation and spectroscopy with a PPLN ridge waveguide. Appl. Phys. B 86: 437.
  52. 52 Asobe, M., Tadanaga, O., Yanagawa, T., Umeki, T., Nishida, Y., and Suzuki, H. (2008). High‐power mid‐infrared wavelength generation using difference frequency generation in damage‐resistant Zn:LiNbO3 waveguide. Electron. Lett. 44: 288.
  53. 53 Levi, O., Pinguet, T.J., Skauli, T., Eyres, L.A., Parameswaran, K.R., Harris, J.S. Jr., Fejer, M.M., Kulp, T.J., Bisson, S.E., Gerard, B., Lallier, E., and Becouarn, L. (2002). Difference frequency generation of 8‐μm radiation in orientation‐patterned GaAs. Opt. Lett. 27: 2091.
  54. 54 Bisson, S.E., Kulp, T.J., Levi, O., Harris, J.S., and Fejer, M.M. (2006). Long‐wave IR chemical sensing based on difference frequency generation in orientation‐patterned GaAs. Appl. Phys. B 85: 199.
  55. 55 Vasilyev, S., Schiller, S., Nevsky, A., Grisard, A., Faye, D., Lallier, E., Zhang, Z., Boyland, A.J., Sahu, J.K., Ibsen, M., and Clarkson, W.A. (2008). Broadly tunable single‐frequency cw mid‐infrared source with milliwatt‐level output based on difference‐frequency generation in orientation‐patterned GaAs. Opt. Lett. 33: 1413.
  56. 56 Sowade, R., Breunig, I., Mayorga, I.C., Kiessling, J., Tulea, C., Dierolf, V., and Buse, K. (2009). Continuous‐wave optical parametric terahertz source. Opt. Express 17: 22303.
  57. 57 Smith, R.G., Geusic, J.E., Levinstein, H., Rubin, J.J., Singh, S., and Van Uiter, L. (1968). Continuous optical parametric oscillation in Ba2NaNb5O15. Appl. Phys. Lett. 12: 308.
  58. 58 Bosenberg, W.R., Drobshoff, A., Alexander, J.I., Myers, L.E., and Byer, R.L. (1996). Continuous‐wave singly resonant optical parametric oscillator based on periodically poled LiNbO3. Opt. Lett. 21: 713.
  59. 59 Bosenberg, W.R., Drobshoff, A., Alexander, J.I., Myers, L.E., and Byer, R.L. (1996). 93% pump depletion, 3.5‐W continuous‐wave, singly resonant optical parametric oscillator. Opt. Lett. 21: 1336.
  60. 60 Kumar, S.C., Das, R., Samanta, G., and Ebrahim‐Zadeh, M. (2011). Optimally‐output‐coupled, 17.5 W, fiber‐laser‐pumped continuous‐wave optical parametric oscillator. Appl. Phys. B 102: 31.
  61. 61 Kumar, S.C. and Ebrahim‐Zadeh, M. (2011). High‐power, continuous‐wave, mid‐infrared optical parametric oscillator based on MgO:sPPLT. Opt. Lett. 36: 2578.
  62. 62 Henderson, A. and Stafford, R. (2006). Low threshold, singly‐resonant CW OPO pumped by an all‐fiber pump source. Opt. Express 14: 767.
  63. 63 Popp, A., Müller, F., Kühnemann, F., Schiller, S., von Basum, G., Dahnke, H., Hering, P., and Mürtz, M. (2002). Ultra‐sensitive mid‐infrared cavity leak‐out spectroscopy using a cw optical parametric oscillator. Appl. Phys. B 75: 751.
  64. 64 Ngai, A.K.Y., Persijn, S.T., von Basum, G., and Harren, F.J.M. (2006). Automatically tunable continuous‐wave optical parametric oscillator for high‐resolution spectroscopy and sensitive trace‐gas detection. Appl. Phys. B 85: 173.
  65. 65 Lindsay, I., Adhimoolam, B., Groß, P., Klein, M., and Boller, K. (2005). 110 GHz rapid, continuous tuning from an optical parametric oscillator pumped by a fiber‐amplified DBR diode laser. Opt. Express 13: 1234.
  66. 66 Siltanen, M., Vainio, M., and Halonen, L. (2010). Pump‐tunable continuous‐wave singly resonant optical parametric oscillator from 2.5 to 4.4 μm. Opt. Express 18: 14087.
  67. 67 Scheidt, M., Beier, B., Boller, K.‐J., and Wallenstein, R. (1997). Frequency‐stable operation of a diode‐pumped continuous‐wave RbTiOAsO4 optical parametric oscillator. Opt. Lett. 22: 1287.
  68. 68 Turnbull, G.A., McGloin, D., Lindsay, I.D., Ebrahimzadeh, M., and Dunn, M.H. (2000). Extended mode‐hop‐free tuning by use of a dual‐cavity, pump‐enhanced optical parametric oscillator. Opt. Lett. 25: 341.
  69. 69 Lindsay, I.D., Petriokes, C., Dunn, M.H., and Ebrahimzadeh, M. (2001). Continuous‐wave pump‐enhanced singly resonant optical parametric oscillator pumped by an extended‐cavity diode laser. Appl. Phys. Lett. 78: 871.
  70. 70 Stothard, D.J.M., Lindsay, I.D., and Dunn, M.H. (2004). Continuous‐wave pump‐enhanced optical parametric oscillator with ring resonator for wide and continuous tuning of single‐frequency radiation. Opt. Express 12: 502.
  71. 71 Rihan, A., Andrieux, E., Zanon‐Willette, T., Briaudeau, S., Himbert, M., and Zondy, J.‐J. (2011). A pump‐resonant signal‐resonant optical parametric oscillator for spectroscopic breath analysis. Appl. Phys. B 102: 367.
  72. 72 Schunemann, P.G., Pomeranz, L.A., Setzler, S.D., Jones, C.W., and Budni, P.A. (2013). CW mid‐IR OPO based on OP‐GaAs, Technical Digest. In: The European Conference on Lasers and Electro‐Optics. Washington, DC: Optical Society of America, Paper JSII_2_3.
  73. 73 Meisenheimer, S.‐K., Fürst, J.U., Buse, K., and Breunig, I. (2017). Continuous‐wave optical parametric oscillation tunable up to an 8 μm wavelength. Optica 4: 189.
  74. 74 Breunig, I., Haertle, D., and Buse, K. (2011). Continuous‐wave optical parametric oscillators: recent developments and prospects. Appl. Phys. B 105: 99.
  75. 75 Fürst, J.U., Strekalov, D.V., Elser, D., Aiello, A., Andersen, U.L., Marquardt, C., and Leuchs, G. (2010). Low‐threshold optical parametric oscillations in a whispering gallery mode resonator. Phys. Rev. Lett. 105: 263904.
  76. 76 Beckmann, T., Linnenbank, H., Steigerwald, H., Sturman, B., Haertle, D., Buse, K., and Breunig, I. (2011). Highly tunable low‐threshold optical parametric oscillation in radially poled whispering gallery resonators. Phys. Rev. Lett. 106: 143903.
  77. 77 Werner, C.S., Buse, K., and Breunig, I. (2015). Continuous‐wave whispering‐gallery optical parametric oscillator for high‐resolution spectroscopy. Opt. Lett. 40: 772.
  78. 78 Hanna, D.C., Rampal, V.V., and Smith, R.C. (1973). Tunable infrared down‐conversion in silver thiogallate. Opt. Commun. 8: 151.
  79. 79 Seymour, R.J. and Zernike, F. (1976). Infrared radiation tunable from 5.5 to 18.3 μm generated by mixing in AgGaS2. Appl. Phys. Lett. 29: 705.
  80. 80 Kato, K. (1984). High‐power difference‐frequency generation at 5–11 μm in AgGaS2. IEEE J. Quantum Electron. QE‐20: 698.
  81. 81 Haidar, S., Nakamura, K., Niwa, E., Masumoto, K., and Ito, H. (1999). Mid‐infrared (5–12‐μm) and limited (5.5–8.5‐μm) single‐knob tuning generated by difference‐frequency mixing in single‐crystal AgGaS2. Appl. Opt. 38: 1798.
  82. 82 Shi, W. and Ding, Y. (2004). A monochromatic and high‐power terahertz source tunable in the ranges of 2.7–38.4 and 58.2–3540 μm for variety of potential applications. Appl. Phys. Lett. 84: 1635.
  83. 83 Vodopyanov, K.L. and Schunemann, P.G. (1998). Efficient difference frequency generation of 7–20‐μm radiation in CdGeAs2. Opt. Lett. 23: 1096.
  84. 84 Kung, A.H. (1994). Narrowband mid‐infrared generation using KTiOAsO4. Appl. Phys. Lett. 65: 1082.
  85. 85 Miyamoto, Y., Hara, H., Masuda, T., Hiraki, T., Sasao, N., and Uetake, S. (2017). Injection‐seeded tunable mid‐infrared pulses generated by difference frequency mixing. Jpn. J. Appl. Phys. 56: 032101.
  86. 86 Belden, P., Chen, D.W., and Di Teodoro, F. (2015). Watt‐level, gigahertz‐linewidth difference‐frequency generation in PPLN pumped by a nanosecond‐pulse fiber laser source. Opt. Lett. 40: 958.
  87. 87 Fischer, C. and Sigrist, M.W. (2003). Mid‐IR difference frequency generation. In: Solid‐state Mid‐infrared Laser Sources (ed. I.T. Sorokina and K.L. Vodopyanov). Topics Appl. Phys., vol. 89. Berlin: Springer.
  88. 88 Brosnan, S.J. and Byer, R.L. (1979). Optical parametric oscillator threshold and linewidth studies. IEEE J. Quantum Electron. QE‐15: 415.
  89. 89 Hanna, D.C., Luther‐Davies, B., and Smith, R.C. (1973). Singly resonant proustite parametric oscillator tuned from 1.22 to 8.5 μm. Appl. Phys. Lett. 22: 440.
  90. 90 Myers, L.E., Eckardt, R.C., Fejer, M.M., Byer, R.L., Bosenberg, W.R., and Pierce, J.W. (1995). Quasi‐phase‐matched optical parametric oscillators in bulk periodically poled LiNbO3. J. Opt. Soc. Am. B 12: 2102.
  91. 91 Myers, L.E., Eckardt, R.C., Fejer, M.M., Byer, R.L., and Bosenberg, W.R. (1996). Multi‐grating quasi‐phase‐matched optical parametric oscillators in periodically poled LiNbO3. Opt. Lett. 21: 591.
  92. 92 Fan, Y.X., Eckardt, R.L., Byer, R.L., Route, R.K., and Feigelson, R.S. (1984). AgGaS2 infrared parametric oscillator. Appl. Phys. Lett. 45: 313.
  93. 93 Vodopyanov, K.L., Maffetone, J.P., Zwieback, I., and Ruderman, W. (1999). AgGaS2 optical parametric oscillator continuously tunable from 3.9 to 11.3 μm. Appl. Phys. Lett. 75: 1204.
  94. 94 Eckardt, R.C., Fan, Y.X., Byer, R.L., Marquardt, C.L., Storm, M.E., and Esterowitz, L. (1986). Broadly tunable infrared parametric oscillator using AgGaSe2. Appl. Phys. Lett. 49: 608.
  95. 95 Quarles, G.J., Marquardt, C.L., and Esterowitz, L. (1990). 2‐μm pumped AgGaSe2 with continuous tuning 2.49–12.05 μm. In: Proc. LEOS. Boston, MA, Paper ELT7.1.
  96. 96 Chandra, S., Allik, T.H., Catella, G., Utano, R., and Hutchinson, J.A. (1997). Continuously tunable, 6–14 μm silver‐gallium selenide optical parametric oscillator pumped at 1.57 μm. Appl. Phys. Lett. 71: 584.
  97. 97 Vodopyanov, K.L., Ganikhanov, F., Maffetone, J.P., Zwieback, I., and Ruderman, W. (2000). ZnGeP2 optical parametric oscillator with 3.8–12.4‐μm tunability. Opt. Lett. 25: 841.
  98. 98 Vodopyanov, K.L. (2001). OPOs target the longwave infrared. Laser Focus World 37: 225.
  99. 99 Vodopyanov, K.L. and Schunemann, P.G. (2003). Broadly tunable noncritically phase‐matched ZGP optical parametric oscillator with a 2‐microjoule pump threshold. Opt. Lett. 28: 441.
  100. 100 Vodopyanov, K.L., Levi, O., Kuo, P.S., Pinguet, T.J., Harris, J.S., Fejer, M.M., Gerard, B., Becouarn, L., and Lallier, E. (2004). Optical parametric oscillation in quasi‐phase‐matched GaAs. Opt. Lett. 29: 1912.
  101. 101 Vodopyanov, K.L., Levi, O., Kuo, P.S., Pinguet, T.J., Harris, J.S., Fejer, M.M., Gerard, B., Becouarn, L., and Lallier, E. (2004). Optical parametric oscillator based on orientation‐patterned GaAs. Proc. SPIE 5620: 63–69.
  102. 102 Skauli, T., Kuo, P.S., Vodopyanov, K.L., Pinguet, T.J., Levi, O., Eyres, L.A., Harris, J.S., and Fejer, M.M. (2003). Determination of GaAs refractive index and its temperature dependence, with application to quasi‐phasematched nonlinear optics. J. Appl. Phys. 94: 6447.
  103. 103 Vodopyanov, K.L., Makasyuk, I., and Schunemann, P.G. (2014). Grating tunable 4–14 μm GaAs optical parametric oscillator pumped at 3 μm. Opt. Express 22: 4131.
  104. 104 Zawilski, K.T., Schunemann, P.G., Pollak, T.C., Zelmon, D.E., Fernelius, N.C., and Hopkins, F.K. (2010). Growth and characterization of large CdSiP2 single crystals. J. Cryst. Growth 312: 1127.
  105. 105 Petrov, V., Schunemann, P.G., Zawilski, K.T., and Pollak, T.M. (2009). Noncritical singly resonant optical parametric oscillator operation near 6.2 μm based on a CdSiP2 crystal pumped at 1064 nm. Opt. Lett. 34: 2399.
  106. 106 Marchev, G., Tyazhev, A., Petrov, V., Schunemann, P.G., Zawilski, K.T., Stöppler, G., and Eichhorn, M. (2012). Optical parametric generation in CdSiP2 at 6.125 μm pumped by 8 ns long pulses at 1064 nm. Opt. Lett. 37: 740.
  107. 107 Kostyukova, N.Y., Boyko, A.A., Badikov, V., Badikov, D., Shevyrdyaeva, G., Panyutin, V., Marchev, G.M., Kolker, D.B., and Petrov, V. (2016). Widely tunable in the mid‐IR BaGa4Se7 optical parametric oscillator pumped at 1064 nm. Opt. Lett. 41: 3667.
  108. 108 Richman, B.A., Aniolek, K.W., Kulp, T.J., and Bisson, S.E. (2000). Continuously tunable, single‐longitudinal‐mode, pulsed mid‐infrared optical parametric oscillator based on periodically poled lithium niobate. J. Opt. Soc. Am. B 17: 1233.
  109. 109 Ganikhanov, F., Caughey, T., and Vodopyanov, K.L. (2001). Narrow‐linewidth middle‐infrared ZnGeP2 optical parametric oscillator. J. Opt. Soc. Am. B 18: 818.
  110. 110 Scherrer, B., Ribet, I., Godard, A., Rosencher, E., and Lefebvre, M. (2000). Dual‐cavity doubly resonant optical parametric oscillators: demonstration of pulsed single‐mode operation. J. Opt. Soc. Am. B 17: 1716.
  111. 111 Drag, C., Desormeaux, A., Lefebvre, M., and Rosencher, E. (2002). Entangled cavity optical parametric oscillator for mid‐infrared pulsed single‐longitudinal‐mode operation. Opt. Lett. 27: 1238.
  112. 112 Hardy, B., Berrou, A., Guilbaud, S., Raybaut, M., Godard, A., and Lefebvre, M. (2011). Compact, single‐frequency, doubly resonant optical parametric oscillator pumped in an achromatic phase‐adapted double‐pass geometry. Opt. Lett. 36: 678.
  113. 113 Barria, J.B., Roux, S., Dherbecourt, J.‐B., Raybaut, M., Melkonian, J.‐M., Godard, A., and Lefebvre, M. (2013). Microsecond fiber laser pumped, single‐frequency optical parametric oscillator for trace gas detection. Opt. Lett. 38: 2165.
  114. 114 Clément, Q., Melkonian, J.‐M., Dherbecourt, J.‐B., Raybaut, M., Grisard, A., Lallier, E., Gérard, B., Faure, B., Souhaité, G., and Godard, A. (2015). Longwave infrared, single‐frequency, tunable, pulsed optical parametric oscillator based on orientation‐patterned GaAs for gas sensing. Opt. Lett. 40: 2676.
  115. 115 Cheung, E., Palese, S., Injeyan, H., Hoefer, C., Hilyard, R., Komine, H., Berg, J., and Bosenberg, W. (1999). High power conversion to mid‐IR using KTP and ZGP OPOs, Technical Digest. In: Advanced Solid‐State Lasers. Washington, DC: Optical Society of America, Paper WC1.
  116. 116 Budni, P.A., Pomeranz, L.A., Lemons, M.L., Schunemann, P.G., Pollak, T.M., and Chicklis, E.P. (1998). 10 W mid‐IR holmium pumped ZnGeP2 OPO, Technical Digest. In: Advanced Solid State Lasers. Washington, DC: Optical Society of America, Paper FC1.
  117. 117 Lippert, E., Fonnum, H., Arisholm, G., and Stenersen, K. (2010). A 22‐watt mid‐infrared optical parametric oscillator with V‐shaped 3‐mirror ring resonator. Opt. Express 18: 26475.
  118. 118 Wu, R., Lai, K.S., Lau, W.E., Wong, H.F., Lim, Y.L., Lim, K.W., and Chia, L.C.L. (2002). A novel laser integrated with a coupled tandem OPO configuration, Technical Digest. In: Conference on Lasers and Electro‐Optics. Washington, DC: Optical Society of America, Paper CTuD6.
  119. 119 Hemming, A., Richards, J., Davidson, A., Carmody, N., Bennetts, S., Simakov, N., and Haub, J. (2013). 99‐W mid‐IR operation of a ZGP OPO at 25% duty cycle. Opt. Express 21: 10062.
  120. 120 Kieleck, C., Eichhorn, M., Hirth, A., Faye, D., and Lallier, E. (2009). High‐efficiency 20–50 kHz mid‐infrared orientation‐patterned GaAs optical parametric oscillator pumped by a 2‐μm holmium laser. Opt. Lett. 34: 262.
  121. 121 Wueppen, J., Nyga, S., Jungbluth, B., and Hoffmann, D. (2016). 1.95 μm‐pumped OP‐GaAs optical parametric oscillator with 10.6 μm idler wavelength. Opt. Lett. 41: 4225.
  122. 122 Vysniauskas, G., Burns, D., and Bente, E. (2002). Development of a nanosecond high energy KTA OPO system operating 2.9 μm, Technical Digest. In: Conference on Lasers and Electro‐Optics. Washington, DC: Optical Society of America, Paper CWA28.
  123. 123 Mennerat, G. and Kupecek, P. (1998). High‐energy narrow‐linewidth tunable source in the mid infrared, Technical Digest. In: Advanced Solid‐State Lasers. Washington, DC: Optical Society of America, Paper FC13.
  124. 124 Ishizuki, H. and Taira, T. (2010). High energy quasi‐phase matched optical parametric oscillation using mg‐doped congruent LiTaO3 crystal. Opt. Express 18: 253.
  125. 125 Allik, T.H., Chandra, S., Rines, D.M., Schunemann, P.G., Hutchinson, J.A., and Utano, R. (1997). Tunable 7–12‐μm optical parametric oscillator using a Cr,Er:YSGG laser to pump CdSe and ZnGeP2 crystals. Opt. Lett. 22: 597.
  126. 126 Dergachev, A., Armstrong, D., Smith, A., Drake, T., and Dubois, M. (2007). 3.4‐μm ZGP RISTRA nanosecond optical parametric oscillator pumped by a 2.05‐μm ho:YLF MOPA system. Opt. Express 15: 14404.
  127. 127 Stöppler, G., Schellhorn, M., and Eichhorn, M. (2013). Ho3+:LLF MOPA pumped RISTRA ZGP OPO at 3–5 μm. Proc. SPIE 8604: 86040I.
  128. 128 Haakestad, M., Fonnum, H., and Lippert, E. (2014). Mid‐infrared source with 0.2 J pulse energy based on nonlinear conversion of Q‐switched pulses in ZnGeP2. Opt. Express 22: 8556.
  129. 129 Hansson, G., Karlsson, H., and Laurell, F. (2001). Unstable resonator optical parametric oscillator based on quasi‐phase‐matched RbTiOAsO4. Appl. Opt. 40: 5446.
  130. 130 Ishizuki, H. and Taira, T. (2005). High‐energy quasi‐phase‐matched optical parametric oscillation in a periodically poled MgO:LiNbO3 device with a 5 mm × 5 mm aperture. Opt. Lett. 30: 2918.
  131. 131 Saikawa, J., Miyazaki, M., Fujii, M., Ishizuki, H., and Taira, T. (2008). High‐energy, broadly tunable, narrow‐bandwidth mid‐infrared optical parametric system pumped by quasi‐phase‐matched devices. Opt. Lett. 33: 1699.
  132. 132 Smith, A.V. and Armstrong, D.J. (2002). Nanosecond optical parametric oscillator with 90° image rotation: design and performance. J. Opt. Soc. Am. B 19: 1801.
  133. 133 Arbore, M.A. and Fejer, M.M. (1997). Singly resonant optical parametric oscillation in periodically poled lithium niobate waveguides. Opt. Lett. 22: 151.
  134. 134 Oron, M.B., Blau, P., Pearl, S., and Katz, M. (2012). Optical parametric oscillation in orientation patterned GaAs waveguides. Proc. SPIE 8240: 82400C‐11.
  135. 135 Ehret, S. and Schneider, H. (1998). Generation of subpicosecond infrared pulses tunable between 5.2 μm and 18 μm at a repetition rate of 76 MHz. Appl. Phys. B 66: 27.
  136. 136 Erny, C., Moutzouris, K., Biegert, J., Kühlke, D., Adler, F., Leitenstorfer, A., and Keller, U. (2007). Mid‐infrared difference‐frequency generation of ultrashort pulses tunable between 3.2 and 4.8 μm from a compact fiber source. Opt. Lett. 32: 1138.
  137. 137 Winters, D.G., Schlup, P., and Bartels, R.A. (2010). Subpicosecond fiber‐based soliton‐tuned mid‐infrared source in the 9.7–14.9 μm wavelength region. Opt. Lett. 35: 2179.
  138. 138 Yao, Y. and Knox, W.H. (2013). Broadly tunable femtosecond mid‐infrared source based on dual photonic crystal fibers. Opt. Express 21: 26612.
  139. 139 Phillips, C.R., Jiang, J., Mohr, C., Lin, A.C., Langrock, C., Snure, M., Bliss, D., Zhu, M., Hartl, I., Harris, J.S., Fermann, M.E., and Fejer, M.M. (2012). Widely tunable midinfrared difference frequency generation in orientation‐patterned GaAs pumped with a femtosecond Tm‐fiber system. Opt. Lett. 37: 2928.
  140. 140 Zhu, F., Hundertmark, H., Kolomenskii, A.A., Strohaber, J., Holzwarth, R., and Schuessler, H.A. (2013). High‐power mid‐infrared frequency comb source based on a femtosecond Er:fiber oscillator. Opt. Lett. 38: 2360.
  141. 141 Meek, S.A., Poisson, A., Guelachvili, G., Hänsch, T.W., and Picqué, N. (2014). Fourier transform spectroscopy around 3 μm with a broad difference frequency comb. Appl. Phys. B 114: 573.
  142. 142 Cruz, F.C., Maser, D.L., Johnson, T., Ycas, G., Klose, A., Giorgetta, F.R., Coddington, I., and Diddams, S.A. (2015). Mid‐infrared optical frequency combs based on difference frequency generation for molecular spectroscopy. Opt. Express 23: 26814.
  143. 143 Beutler, M., Rimke, I., Büttner, E., Farinello, P., Agnesi, A., Badikov, V., Badikov, D., and Petrov, V. (2015). Difference‐frequency generation of ultrashort pulses in the mid‐IR using Yb‐fiber pump systems and AgGaSe2. Opt. Express 23: 2730.
  144. 144 Bonvalet, A., Joffre, M., Martin, J.L., and Migus, A. (1995). Generation of ultrabroadband femtosecond pulses in the mid‐infrared by optical rectification of 15 fs light pulses at 100 MHz repetition rate. Appl. Phys. Lett. 67: 2907.
  145. 145 Nahata, A., Weling, A.S., and Heinz, T.F. (1996). A wideband coherent terahertz spectroscopy system using optical rectification and electro‐optic sampling. Appl. Phys. Lett. 69: 2321.
  146. 146 Kaindl, R.A., Smith, D.C., Joschko, M., Hasselbeck, M.P., Woerner, M., and Elsaesser, T. (1998). Femtosecond infrared pulses tunable from 9 to 18 μm at an 88‐MHz repetition rate. Opt. Lett. 23: 861.
  147. 147 Pupeza, I., Sánchez, D., Zhang, J., Lilienfein, N., Seidel, M., Karpowicz, N., Paasch‐Colberg, T., Znakovskaya, I., Pescher, M., Schweinberger, W., Pervak, V., Fill, E., Pronin, O., Wei, Z., Krausz, F., Apolonski, A., and Biegert, J. (2015). High‐power sub‐two‐cycle mid‐infrared pulses at 100 MHz repetition rate. Nat. Photon. 9: 721.
  148. 148 Zhang, J., Mak, K., Nagl, N., Seidel, M., Bauer, D., Sutter, D., Pervak, V., Krausz, F., and Pronin, O. (2018). Multi‐mW, few‐cycle mid‐infrared continuum spanning from 500 to 2250 cm−1. Light Sci. Appl. 7: 17180.
  149. 149 Vasilyev, S., Moskalev, I.S., Smolski, V.O., Peppers, J.M., Mirov, M., Muraviev, A.V., Zawilski, K., Schunemann, P.G., Mirov, S.B., Vodopyanov, K.L., and Gapontsev, V.P. (2019). Super‐octave longwave mid‐infrared coherent transients produced by optical rectification of few‐cycle 2.5‐μm pulses. Optica 6: 111.
  150. 150 Gambetta, A., Coluccelli, N., Cassinerio, M., Gatti, D., Laporta, P., Galzerano, G., and Marangoni, M. (2013). Milliwatt‐level frequency combs in the 8–14 μm range via difference frequency generation from an Er:fiber oscillator. Opt. Lett. 38: 1155.
  151. 151 Hegenbarth, R., Steinmann, A., Sarkisov, S., and Giessen, H. (2012). Milliwatt‐level mid‐infrared (10.5–16.5 μm) difference frequency generation with a femtosecond dual‐signal‐wavelength optical parametric oscillator. Opt. Lett. 37: 3513.
  152. 152 Hajialamdari, M. and Strickland, D. (2012). Tunable mid‐infrared source from an ultrafast two‐color Yb:fiber chirped‐pulse amplifier. Opt. Lett. 37: 3570.
  153. 153 Soboń, G., Martynkien, T., Mergo, P., Rutkowski, L., and Foltynowicz, A. (2017). High‐power frequency comb source tunable from 2.7 to 4.2  μm based on difference frequency generation pumped by an Yb‐doped fiber laser. Opt. Lett. 42: 1748.
  154. 154 Schliesser, A., Brehm, M., and Keilmann, F. (2005). Frequency‐comb infrared spectrometer for rapid, remote chemical sensing. Opt. Express 13: 9029.
  155. 155 Gaida, C., Gebhardt, M., Huermann, T., Stutzki, F., Jauregui, C., Antonio‐Lopez, J., Schülzgen, A., Amezcua‐Correa, R., Tünnermann, A., Pupeza, I., and Limpert, J. (2018). Watt‐scale super‐octave mid‐infrared intrapulse difference frequency generation. Light Sci. Appl. 7: 94.
  156. 156 Ruffing, B., Nebel, A., and Wallenstein, R. (1998). All‐solid‐state cw mode‐locked picosecond KTiOAsO4 (KTA) optical parametric oscillator. Appl. Phys. B 67: 537.
  157. 157 Qin, Z., Xie, G., Ge, W., Yuan, P., and Qian, L. (2015). Over 20‐W mid‐infrared picosecond optical parametric oscillator. IEEE Photon. J. 7: 1400506.
  158. 158 Kokabee, O., Esteban‐Martin, A., and Ebrahim‐Zadeh, M. (2010). Efficient, high‐power, ytterbium‐fiber‐laser‐pumped picosecond optical parametric oscillator. Opt. Lett. 35: 3210.
  159. 159 Xu, L., Chan, H.‐Y., Alam, S.‐U., Richardson, D.J., and Shepherd, D.P. (2015). Fiber‐laser‐pumped, high‐energy, mid‐IR, picosecond optical parametric oscillator with a high‐harmonic cavity. Opt. Lett. 40: 3288.
  160. 160 Kumar, S.C., Agnesi, A., Dallocchio, P., Pirzio, F., Reali, G., Zawilski, K.T., Schunemann, P.G., and Ebrahim‐Zadeh, M. (2011). Compact, 1.5 mJ, 450 MHz, CdSiP2 picosecond optical parametric oscillator near 6.3 μm. Opt. Lett. 36: 3236.
  161. 161 Burr, K.C., Tang, C.L., Arbore, M.A., and Fejer, M.M. (1997). Broadly tunable mid‐infrared femtosecond optical parametric oscillator using all‐solid‐state‐pumped periodically poled lithium niobate. Opt. Lett. 22: 1458.
  162. 162 Kumar, S.C., Esteban‐Martin, A., Ideguchi, T., Yan, M., Holzner, S., Hänsch, T.W., Picqué, N., and Ebrahim‐Zadeh, M. (2014). Few‐cycle, broadband, mid‐infrared optical parametric oscillator pumped by a 20‐fs Ti:sapphire laser. Laser Photon. Rev. 91: 86.
  163. 163 Marzenell, S., Beigang, R., and Wallenstein, R. (1999). Synchronously pumped femtosecond optical parametric oscillator based on AgGaSe2 tunable from 2 μm to 8 μm. Appl. Phys. B 69: 423.
  164. 164 Kumar, S.C., Krauth, J., Steinmann, A., Zawilski, K.T., Schunemann, P.G., Giessen, H., and Ebrahim‐Zadeh, M. (2015). High‐power femtosecond mid‐infrared optical parametric oscillator at 7 μm based on CdSiP2. Opt. Lett. 40: 1398.
  165. 165 Kumar, S.C., Esteban‐Martin, A., Santana, A., Zawilski, K.T., Schunemann, P.G., and Ebrahim‐Zadeh, M. (2016). Pump‐tuned deep‐infrared femtosecond optical parametric oscillator across 6–7 μm based on CdSiP2. Opt. Lett. 41: 3355.
  166. 166 Maidment, L., Schunemann, P.G., and Reid, D.T. (2016). Molecular fingerprint‐region spectroscopy from 5 to 12 μm using an orientation‐patterned gallium phosphide optical parametric oscillator. Opt. Lett. 41: 4261.
  167. 167 Elsaesser, T., Seilmeier, A., Kaiser, W., Koidl, P., and Brandt, G. (1984). Parametric generation of tunable picosecond pulses in the medium infrared using AgGaS2 crystals. Appl. Phys. Lett. 44: 383.
  168. 168 Vodopyanov, K.L. and Chazapis, V. (1997). Extra‐wide tuning range optical parametric generator. Opt. Commun. 135: 98.
  169. 169 Petrov, V., Tanaka, Y., and Suzuki, T. (1997). Parametric generation of 1‐ps pulses between 5 and 11 μm with a ZnGeP2 crystal. IEEE J. Quantum Electron. 33: 1749.
  170. 170 Vodopyanov, K.L. (1998). Megawatt peak power 8–13 μm CdSe optical parametric generator pumped at 2.8 μm. Opt. Commun. 150: 210.
  171. 171 Kumar, S.C., Jelínek, M., Baudisch, M., Zawilski, K.T., Schunemann, P.G., Kubeček, V., Biegert, J., and Ebrahim‐Zadeh, M. (2012). Tunable, high‐energy, mid‐infrared, picosecond optical parametric generator based on CdSiP2. Opt. Express 20: 15703.
  172. 172 Andriukaitis, G., Balčiūnas, T., Ališauskas, S., Pugžlys, A., Baltuška, A., Popmintchev, T., Chen, M.‐C., Murnane, M.M., and Kapteyn, H.C. (2011). 90 GW peak power few‐cycle mid‐infrared pulses from an optical parametric amplifier. Opt. Lett. 36: 2755.
  173. 173 Wandel, S., Lin, M.W., Yin, Y., Xu, G., and Jovanovic, I. (2016). Parametric generation and characterization of femtosecond mid‐infrared pulses in ZnGeP2. Opt. Express 24: 5287.
  174. 174 Sanchez, D., Hemmer, M., Baudisch, M., Cousin, S.L., Zawilski, K., Schunemann, P., Chalus, O., Simon‐Boisson, C., and Biegert, J. (2016). 7μm, ultrafast, sub‐millijoule‐level, mid‐infrared optical parametric chirped pulse amplifier pumped at 2 μm. Optica 3: 147.
  175. 175 McCarthy, M.J. and Hanna, D.C. (1993). All‐solid‐state synchronously pumped optical parametric oscillator. J. Opt. Soc. Am. B 10: 2180.
  176. 176 Burr, K.C., Tang, C.L., Arbore, M.A., and Fejer, M.M. (1997). High‐repetition‐rate femtosecond optical parametric oscillator based on periodically poled lithium niobate. Appl. Phys. Lett. 70: 3341.
  177. 177 Kumar, S.C., Schunemann, P.G., Zawilski, K.T., and Ebrahim‐Zadeh, M. (2016). Advances in ultrafast optical parametric sources for the mid‐infrared based on CdSiP2. J. Opt. Soc. Am. B 33: D44.
  178. 178 Elsaesser, T., Seilmeier, A., and Kaiser, W. (1983). Parametric generation of tunable picosecond pulses in proustite between 1.2 and 8 μm. Opt. Commun. 44: 293.
  179. 179 Vodopyanov, K.L. (1999). Mid‐IR optical parametric generator with extrawide (3–19 μm) tunability: applications for spectroscopy of two‐dimensional electrons in quantum wells. J. Opt. Soc. Am. B 16: 1579.
  180. 180 Shen, Y.R. (2003). The Principles of Nonlinear Optics. Hoboken, NJ: Wiley.
  181. 181 Stolen, R.H., Ippen, E.P., and Tynes, A.R. (1972). Raman oscillation in glass optical waveguide. Appl. Phys. Lett. 20: 62.
  182. 182 Agrawal, G.P. (2013). Nonlinear Fiber Optics, 5e. Oxford: Academic Press.
  183. 183 Basiev, T.T., Basieva, M.N., Doroshenko, M.E., Fedorov, V.V., Osiko, V.V., and Mirov, S.B. (2006). Stimulated Raman scattering in mid IR spectral range 2.31–2.75–3.7 μm in BaWO4 crystal under 1.9 and 1.56 μm pumping. Laser Phys. Lett. 3: 17.
  184. 184 Zhao, J., Zhang, X., Guo, X., Bao, X., Li, L., and Cui, J. (2013). Diode‐pumped actively Q‐switched Tm, Ho:GdVO4/BaWO4 intracavity Raman laser at 2533 nm. Opt. Lett. 38: 1206.
  185. 185 Rakich, P.T., Fink, Y., and Soljacic, M. (2008). Efficient mid‐IR spectral generation via spontaneous fifth‐order cascaded‐Raman amplification in silica fibers. Opt. Lett. 33: 1690.
  186. 186 Jalali, B., Raghunathan, V., Shori, R., Fathpour, S., Dimitropoulos, D., and Stafsudd, O. (2006). Prospects for silicon mid‐IR Raman lasers. IEEE J. Selected Topics Quantum Electron. 12: 1618.
  187. 187 Leuthold, J., Koos, C., and Freude, W. (2010). Nonlinear silicon photonics. Nat. Photon. 4: 535.
  188. 188 Boyraz, O. and Jalali, B. (2004). Demonstration of a silicon Raman laser. Opt. Express 12: 5269.
  189. 189 Rong, H., Jones, R., Liu, A., Cohen, O., Hak, D., Fang, A., and Paniccia, M. (2005). A continuous‐wave Raman silicon laser. Nature 433: 725.
  190. 190 Takahashi, Y., Inui, Y., Chihara, M., Asano, T., Terawaki, R., and Noda, S. (2013). A micrometre‐scale Raman silicon laser with a microwatt threshold. Nature 498: 470.
  191. 191 Rong, H., Xu, S., Cohen, O., Raday, O., Lee, M., Sih, V., and Paniccia, M. (2008). A cascaded silicon Raman laser. Nat. Photon. 2: 170.
  192. 192 Raghunathan, V., Borlaug, D., Rice, R.R., and Jalali, B. (2007). Demonstration of a mid‐infrared silicon Raman amplifier. Opt. Express 15: 14355.
  193. 193 Liu, X., Osgood Jr, R.M., Vlasov, Y.A., and Green, W.M.J. (2010). Mid‐infrared optical parametric amplifier using silicon nanophotonic waveguides, Nat. Photon. 4, 557 (2010).
  194. 194 Chavez Boggio, J.M., Zlatanovic, S., Gholami, F., Aparicio, J.M., Moro, S., Balch, K., Alic, N., and Radic, S. (2010). Short wavelength infrared frequency conversion in ultra‐compact fiber device. Nat. Photon. 4: 561.
  195. 195 Griffith, A.G., Yu, M., Okawachi, Y., Cardenas, J., Mohanty, A., Gaeta, A.L., and Lipson, M. (2016). Coherent mid‐infrared frequency combs in silicon‐microresonators in the presence of Raman effects. Opt. Express 24: 13044.
  196. 196 Sabella, A., Piper, J.A., and Mildren, R.P. (2014). Diamond Raman laser with continuously tunable output from 3.38 to 3.80 μm. Opt. Lett. 39: 4037.
  197. 197 Latawiec, P., Venkataraman, V., Burek, M.J., Hausmann, B.J.M., Bulu, I., and Lončar, M. (2015). On‐chip diamond Raman laser. Optica 2: 924.
  198. 198 Kippenberg, T.J., Spillane, S.M., Armani, D.K., and Vahala, K.J. (2004). Ultralow‐threshold microcavity Raman laser on a microelectronic chip. Opt. Lett. 29: 1224.
  199. 199 Grudinin, I.S. and Maleki, L. (2007). Ultralow‐threshold Raman lasing with CaF2 resonators. Opt. Lett. 32: 166.
  200. 200 Vanier, F., Rochette, M., Godbout, N., and Peter, Y.‐A. (2013). Raman lasing in As2S3 high‐Q whispering gallery mode resonators. Opt. Lett. 38: 4966.
  201. 201 Kuyanov, K.E., Momose, T., and Vilesov, A.F. (2004). Solid hydrogen Raman shifter for the mid‐infrared range 4.4–8 μm. Appl. Opt. 43: 6023.
  202. 202 Gladyshev, A.V., Kosolapov, A.F., Khudyakov, M.M., Yatsenko, Yu.P., Kolyadin, A.N., Krylov, A.A., Pryamikov, A.D., Biriukov, A.S., Likhachev, M.E., Bufetov, I.A., and Dianov, E.M. (2017). 4.4‐μm Raman laser based on hollow‐core silica fibre. Quantum Electron. 47: 491.
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
54.224.52.210