1. Introduction

High power orange lasers within the 600 nm spectral range assume a significant role for various potential applications in biomedicine [1], spectroscopy [2], and laser displays [3] etc. For example some of the most regular fluorescent markers frequently used for fluorescent based medical imaging have their depletion wavelengths in this range [4]. Until now, no traditional diode lasers have been shown to emit near 600 nm, mainly due to the absence of appropriate direct band gap materials [5]. Despite noteworthy advancement towards achieving a 599 nm emission using GalnP material, the approach utilized may not lead to broad bandwidths, thus restricting their potential applications [6]. Conventionally, short wavelengths in the visible range are usually produced through nonlinear three-wave mixing processes such as second-harmonic generation (SHG), or sum-frequency generation (SFG). On the account of the advancement in quasi-phase matching (QPM) technology, periodically poled nonlinear crystals like lithium niobate have been exploited for compressing ultrashort pulses [7,8], and wavelength conversion [9]. In addition, QPM allows access to the highest nonlinear coefficient (d₃₃) and is free of walk-off that offers a unique opportunity to exploit long interaction lengths convenient for attaining practical powers. Conventional approaches to implementing QPM have been well studied [10–14].

Among the different techniques for generating continuous wave (CW) light sources near the 600 nm wavelength region include, the use of resonant cavities in SFG [15,16], frequency doubling of 1064 nm lasers [14], tunable output in Cr:Forsterite [15], and avalanche upconversion in Pr, Yb:BaY₂F₈ [16]. Moreover, a chirped periodically poled lithium niobate has been used to generate ~50 nm wide-band radiation in the 770 nm wavelength region [17]. A similar technique was also used to generate a tunable orange radiation from 601 nm to 604 nm in quasi-periodically poled lithium tantalate superlattice [18]. Optical parametric oscillators (OPOs) are also attractive for generating visible radiation, for example, Mieth et al utilized MgO:PPLN-based OPO to deliver tunable orange light exceeding 1 W [19]. Such OPOs, although very useful in many applications, e.g. high resolution Doppler-free spectroscopy, they are generally bulky [20]

Although non-uniform structures have many spatial vectors suited to broadband applications, they have fixed grating parameters. Despite this limitation, by subjecting a step-chirped grating to a longitudinal temperature variation (temperature gradient), the phase matching bandwidth can be enhanced. The temperature gradient created along the crystal length modifies the refractive index in each step (with different uniform period). This method can viably induce a chirp in each grating section, and hence introduces an adjustability to the otherwise fixed grating properties of the step-chirped structure. Furthermore, step-chirped QPM gratings are relatively easy to fabricate in comparison to chirped QPM structures which usually do not achieve the desired smooth period change. This simple temperature gradient technique has been demonstrated in broadband generation of near-IR wavelengths in PPLN crystal previously [21]. Recently, we also reported a triple-wavelength orange laser, in which a step-chirped MgO:PPLN structure was used in single-pass mode [22]. In contrast to the commonly used single-pass nonlinear conversion scheme, double-pass approach has been shown to significantly enhance the conversion efficiency [23].

In this work, we demonstrate double-pass SFG by using thermal gradient in a step-chirped MgO:PPLN (ᴧ = 10.1-10.4 µm) which resulted in a marked increase of the average power and bandwidth. The generated broadband orange light could find prospective biomedical and spectroscopic applications.

2. Theoretical considerations

A temperature gradient along the propagation direction (z-axis) of the MgO:PPLN crystal creates a longitudinal modulation of the refractive index of the pump, signal and idler wavelengths. This modulation, in turn, induces a varying wave vector mismatch along the direction of propagation. Theoretically, the temperature distribution T(z) and wave vector mismatch Δk(z) are defined as [24,25],

where T(0) and T(L) are the temperatures at the input and output ends of the MgO:PPLN sample, L is the effective interaction length, $k_j=2\pi n_j/{\lambda }_j$ is the wave vector at the corresponding frequency${\omega }_j$, $j=1,2,3$,${\lambda }_j$ is the wavelength in vacuum,$n_j$ is the refractive index (extraordinary) at that wavelength that can be calculated using the Sellmeier equation [26], and $\Lambda (z)$ is the effective QPM-grating period. From Eq. (1), the temperature difference between the crystal ends can be defined as $\Delta T=T(0)-T(L)$. Thermal expansion of different PPLN periods at different temperatures can cause small deviations from $\Lambda$, given as [21]

where ${\Lambda }_0$is the period at $25$ᵒC, ${\Lambda }_n$ is the period of the n^th domain at temperature $T_n$, $\alpha =15.4\times {10}^{-6}$ᵒC⁻¹ and $\beta =5.3\times {10}^{-9}$ᵒC⁻² are the thermal expansion coefficients of lithium niobate taken from Ref [25]. However, the influence of thermal expansion on the effective grating period is one order of magnitude lower than that on the change in the refractive index, and therefore, this can be ignored [27]. Figure 1 illustrates the simulated output spectra for different temperature gradients for the designed MgO:PPLN crystal obtained by solving a system of slowly varying envelope coupled-mode equations for three-wave mixing following the method detailed in Ref [28]. Figure 1 demonstrates that the temperature gradient in the MgO:PPLN crystal influences the bandwidth of the generated output. The blue line represents the expected output spectra of a step-chirped QPM structure with four periods of equal lengths for perfect phase matching at 25 ᵒC (ΔT = 0 ᵒC). The black and red lines represent the expected spectra after subjecting the step-chirped MgO:PPLN crystal ends to temperature differences of 10 ᵒC (ΔT = 10 ᵒC) and 20 ᵒC (ΔT = 20 ᵒC), respectively, showing improved bandwidth. Beyond a temperature difference of 20 ᵒC, the bandwidth decreases with a general shift towards higher wavelengths, as shown with the green dotted line (ΔT = 30 ᵒC) in Fig. 1. Once various points along the crystal length attain distinctive temperatures (temperature gradient), perfect phase-matching of different incoming wavelengths of the broadband signal can be achieved at those points by appropriately adjusting the temperature at crystal ends. Consequently, all the interacting wavelengths that simultaneously satisfy the energy ($1/{\lambda }_3=1/{\lambda }_1+1/{\lambda }_2$) and momentum conservation ($\Delta k$) conditions will be upconverted, thus a significant enhancement of the upconverted wavelength bandwidth can be expected. Therefore, it is conceivable that the bandwidth of the generated output can be enhanced by controlling the temperature gradient of the crystal.

 

Fig. 1 Simulated output spectra for different temperature gradients in the MgO:PPLN crystal.

Download Full Size | PDF

3. Experimental configuration

Figure 2 shows the experimental setup of our double-pass SFG.

 

Fig. 2 Schematic of the experimental setup. ASE, amplified spontaneous emission; LD, laser diode; WDM, wave division multiplexer; CL, collimator lens; L, lens; GP, glass plate; M₁: concave mirror; M₂, plano mirror; θ, angle between first and second pass; O₁ and O₂, ovens; T₁ and T₂, temperatures at crystal ends.

Download Full Size | PDF

The MgO:PPLN crystal was 1 mm-thick and 40 mm-long (z-cut). It had four sections with grating periods of 10.1, 10.2, 10.3, and 10.4 µm each of them 10 mm-long, intended to fulfil the phase-matching condition ($1/{\lambda }_3=1/{\lambda }_1+1/{\lambda }_2$) for the SFG process. We used the well-established standard electric poling technique to fabricate the step-chirped QPM device on an optical grade MgO:LN wafer [29]. After poling, we etched the cleaned crystal in a solution of hydrofluoric acid (HF), polished the end faces, and anti-reflective (AR, R<0.2%) coated at 974 nm, 600 nm, and in 1525-1565 nm wavelengths. The MgO:PPLN crystal was subjected to a temperature gradient to modify the refractive indexes of each segment and hence, widen the input wavelength acceptance range. The beam of a laser diode (LD) pump (P₁) (974 nm (λ₁), 0.20 nm spectral width) was combined with a C-band (1525-1565 nm) amplified spontaneous emission (ASE) source (${\lambda }_2$) signal (P₂) using a standard 980/1550 nm wavelength division multiplexer (WDM) coupler with 0.26 and 0.28 dB insertion losses at 980 and 1550 nm, respectively. The collimated combined fundamental (using a C-lens CL) was focused into the nonlinear MgO:PPLN crystal by the plano-convex lens L (f = 125 mm). The second-pass was realized using a highly reflective concave mirror (M₁, R = 75 mm), coated for all the interacting wavelengths and set to a slight inclination angle θ ($\cong 18mrad$). Not only does the small inclination angle increase the effective interaction period of the second-pass, it is also expected to causes a decrease in the SFG bandwidth at 600 nm band. This effect is due to a deviation from the optimum phase matching conditions at wavelengths within the acceptance bandwidth of the second-pass. A 10 mm-thick AR coated glass plate (GP) placed between M₁ and the crystal compensates for the phase shift resulting from dispersion in air. The GP is adjusted such that the net phase shift between the fundamental and the generated beam is a multiple of 2π and remains constant across a wide wavelength range. Another plano mirror M₂ deflects the output beam for final evaluation The required domain period was calculated using the Sellmeier equation for MgO-doped lithium niobate crystal at 25 ᵒC [26].

The measured input powers of P₁ = 1 W and P₂ = 0.2 W were noted before the crystal input. The output power was measured with power meter (Ophir-VEGA, Ophir-Spiricon Inc.) and the spectra was detected with a fiber spectrometer (Resolution: 0.8 nm, BIM-6001, Brolight). The parasitic wavelengths at the output were eliminated using appropriate filters. In order to apply a temperature gradient to our MgO:PPLN crystal, the thermo-stabilized ovens O₁ and O₂ (model TCS-100, CTL photonics) were directly attached near the input and output facets, respectively with thermal contact lengths of ~5 mm. This causes a near linear temperature distribution in the crystal. For the remainder of this paper, the temperature difference between the two thermo-stabilized ovens will be represented by ΔT = T₁-T₂, where T₁ and T₂ are the temperatures of the MgO:PPLN crystal at the input and output ends, respectively.

4. Experimental results and discussion

Figure 3(a) shows the measured orange light spectrum obtained with the crystal at room temperature (T₁ = T₂ = 25 ᵒC, ΔT = 0 ᵒC). We observed an instantaneous quadruple-wavelength output at 595.4, 597.2, 598.6, and 599.6 nm, corresponding to a wavelength range of ~4.2 nm. The average full width at half-maximum (FWHM) bandwidth of each peak is ~0.78 nm. Here, the step-chirped grating structure allows the interaction between λ₁ and four wavelengths of the ASE source that fulfill the phase-matching condition. To setup a temperature variation along the crystal, and henceforth, force heat flow through the crystal, we adjusted the temperature of the input facet to a higher value T₁ and that of the output facet to a lower value T₂. A nearly linear distribution of heat was achieved along the crystal, creating a temperature gradient and consequently, multiple phase-matching conditions were realized. These conditions imply that the SFG process spreads to other wavelengths and effectively increases the input wavelength acceptance range of the crystal. As a result of the latter, the output spectrum of the quadruple-wavelength operation is re-shaped to exhibit instantaneous broad bandwidth. Therefore, when the temperature difference was set to ΔT = 10 ᵒC (T₁~35 ᵒC, T₂~25ᵒC), as is shown in Fig. 3(b), a broad output spectrum with a FWHM bandwidth of ~4.5 nm was achieved. This temperature difference also cause a slight symmetric broadening in the short and long wavelength directions. However, the output spectrum was not smooth and a decrease in the output power was evident, which is consistent with the theoretical predictions shown in Fig. 1. Upon increasing the temperature difference to ΔT = 20 ᵒC (T₁~52 ᵒC, T₂~32 ᵒC), a maximum bandwidth of ~7.2 nm (FWHM) was measured as shown in Fig. 3(c). This corresponds to bandwidth broadening by ~2.7 nm when the temperature difference is increased by 10 ᵒC, again with a central peak that is slightly symmetric on either side. Moreover, this temperature gradient condition was found to yield maximum bandwidth.

 

Fig. 3 Measured sum-frequency generation (SFG) spectra for MgO:PPLN crystal at (a) room temperature, (b) ΔT = 10 ᵒC, and (c) ΔT = 20 ᵒC.

Download Full Size | PDF

The spectral re-shaping and bandwidth broadening depicted in Figs. 3(b) and 3(c) is a result of increased number of the spectral components of the ASE source being quasi-phase-matched in a different section of the crystal, prompted by the slow variation of phase-(mis)matching due to the varying temperature-gradient-induced chirp. It would be more attractive to make the spectra shown in Fig. 3(c) to be more flat-topped by heating up the crystal at multiple locations. One possible effect of this approach is that the thermal gradient at each section of the crystal (with a different period) can be precisely adjusted to be more linear and bandwidth enhancement may also occur. Such a high power broadband laser can be potentially useful for various applications e.g. fluorescent-based medical imaging. Figure 4 depicts the measured SFG power (P_SFG) versus the product of P₁ and P₂. Without applying a temperature gradient, the maximum quadruple-wavelength orange light output power of 129 mW was obtained. On this point, we increased P₁ in steps to a maximum value of 1 W and fixed P₂ at a maximum of 0.2 W. No saturation was observed at this input power. An overall nonlinear conversion efficiency was obtained to be ~65%/W, estimated using the relation ${\eta }_{SFG}=P_{SFG}/\left(P_1{P}_2\right)$ for ΔT = 0 ᵒC. This value decreased by ~1/3 when SFG bandwidth was maximum (at ΔT = 20 ᵒC). The significant reduction in the conversion efficiency can be attributed to the fact that for a given temperature gradient profile, each point along the crystal length corresponds to a particular value, thus reducing the effective interaction length. However, substantial improvement can be expected with optimized focusing and second-pass conditions. A similar result of unprecedented SFG efficiency of 50% was demonstrated in Ref [30], where a maximum of 5.53 W was achieved in a two-crystal cascade configuration by frequency mixing two diode lasers with input powers of 5.0 W and 6.1 W.

 

Fig. 4 Measured sum-frequency generation (SFG) output power versus the product of input fundamental power at ΔT = 0 ᵒC.

Download Full Size | PDF

We also investigated the output spectrum at higher temperature gradients while keeping the output end at T₂ = 32 ᵒC (since it corresponds to the broadest bandwidth). In Fig. 5, we plot the spectra for several values of ΔT. We noticed different spectral shapes when T₁ was increased from 62 ᵒC to 82 ᵒC (Fig. 5), and a general shift in the peaks towards longer wavelengths. The reduction in bandwidth acceptance range for higher ΔT (Figs. 5(a)-5(c)) could be because of thermally-induced phase-mismatching at longer${\lambda }_2$wavelengths. However, the observed spectra for ΔT = 30 ᵒC in Fig. 5 (a) is inconsistent with the theoretical simulation shown in Fig. 1, and therefore more detailed experimental analysis is required to confirm this observation for higher temperature gradients. Nevertheless, the temperature gradient technique not only influences the bandwidth of the SFG output, but it also re-shapes the output spectrum.

 

Fig. 5 Measured output sum-frequency generation (SFG) spectra for higher temperature gradients.

Download Full Size | PDF

We measured other characteristics of the orange laser including the beam quality and stability. The spatial profile was measured using Beamgage (Ophir-Spiricon, Inc.) laser beam analyzer at a distance of ~20 cm from the output face of the crystal. The horizontal and vertical M² factors were measured to be ~1.7 and ~2.4 respectively under maximum output power. The degradation of the beam quality could be attributable to the use of non-achromatic lens which led to different focal positions inside the MgO:PPLN crystal.

Figure 6 shows the output power stability of the orange light at maximum power operation recorded over a period of 30 minutes using the StarLab 2.11 software from Ophir, Incl., indicating <3% fluctuation about 129 mW. This fluctuation can be attributed to the stability of both input lasers, fluctuation in temperature of the MgO:PPLN crystal, air currents and mechanical vibrations of the laboratory environment.

 

Fig. 6 Fluctuation of the generated orange light measured at maximum output power.

Download Full Size | PDF

5. Conclusion

In conclusion, a highly-efficient double-pass sum-frequency generation of orange laser has been demonstrated in a step-chirped MgO-doped periodically poled lithium niobate (MgO:PPLN). A maximum orange output power of 129 mW, equivalent to ~65%/W conversion efficiency was achieved at room temperature operation. By subjecting the end faces of the MgO:PPLN crystal to a temperature difference of ΔT = 20 ᵒC, a broadband orange radiation with ~7.2 nm bandwidth was achieved. This approach of combining the temperature-gradient technique and double-pass scheme not only offers the advantage of broadening the input wavelength acceptance, due to temperature-gradient-induced chirp, but also scales the power yield of sum-frequency generation.