betaFeedback

WorksPhysics and Astronomy

Redetermination of the second-order nonlinear optical coefficients of BaMgF4

Abstract

BaMgF4 (BMF) has the potential to serve as a platform material for vacuum-ultraviolet (VUV) nonlinear optics owing to its ferroelectricity and high UV transparency down to 130 nm. However, the second-order nonlinear optical coefficients reported for BMF span approximately one order of magnitude, hindering a reliable assessment of its potential for nonlinear optical applications. Here, we redetermined these coefficients by combining Maker fringe, wedge, and single-pass second-harmonic-generation measurements at a fundamental wavelength of 1064 nm. Our results establish that all the second-order nonlinear optical coefficients of BMF lie in the range of 0.01-0.04 pm/V. We determined d31 to be 0.035 pm/V, which is among the largest coefficients of BMF, while it has been underestimated relative to the other coefficients in previous studies. Although d31 is not the largest coefficient, the use of its long corresponding coherence length is beneficial for quasi-phase matching, particularly for VUV generation, where the fabrication of lower-order quasi-phase-matching structures is crucial.

Introduction

Coherent light sources in the vacuum-ultraviolet (VUV) spectral range below ~200 nm have recently attracted growing interest owing to industrial and scientific applications, such as semiconductor wafer inspection [1], Raman spectroscopy [2, 3], and angle-resolved photoemission spectroscopy (ARPES) [4]. In addition, a 186.75-nm light source is required for a recently proposed Sn2+\mathrm{Sn}^{2+}-based optical atomic clock [5], and a 148.4-nm light source for a 229Th\mathrm{^{229}Th} optical nuclear clock [6]. These next-generation precision clocks have motivated the development of narrow-linewidth laser sources at these wavelengths.

Several nonlinear materials have been investigated for these applications, such as SrB4O7\mathrm{SrB}_{4}\mathrm{O}_{7} (SBO) and KBe2BO3F2\mathrm{KBe}_{2}\mathrm{BO}_{3}\mathrm{F}_{2} (KBBF). SBO has a nonlinear optical (NLO) coefficient as large as 1.5 pm/V [7]. However, since SBO is not ferroelectric, a periodically-poled structure for quasi-phase-matching (QPM) cannot be introduced by applying an external electric field. Wavelength conversion in SBO has instead been achieved by random QPM based on the twinned domain structure formed in grown crystals [8, 9], and the wavelength conversion to 148.4 nm has been demonstrated [10]. However, controlling the formation of such domain structures is challenging. KBBF satisfies birefringent phase-matching (BPM) for second-harmonic generation (SHG) down to 161 nm [11, 12] and for sum-frequency generation down to 149.8 nm [13], close to its absorption edge. At 193 nm, more than 15 mW output has been achieved in continuous-wave operation [14]. The drawbacks of KBBF, however, are its toxic beryllium content and the need for a prism-coupling device to access the phase-matching angle, because of the challenge of growing sufficiently thick crystals that can be cut at the required angle. Many fluorooxoborate crystals have been studied to overcome these limitations [15]. A representative example, NH4B4O6FNH_4B_4O_6F (ABF), can be grown up to centimeter size and enables BPM down to 158.9 nm [16].

SHG based on barium magnesium fluoride (BaMgF4BaMgF_4, BMF) is a promising route for VUV wavelength conversion. BMF is a non-centrosymmetric, ferroelectric crystal, known for its wide transparency range extending from 130 nm to 13 μ\mum [17, 18]. Although the second-order NLO coefficients of BMF are relatively small, its ferroelectricity enables periodic domain inversion, while its high damage threshold and UV resistance allow operation at high optical intensities, enhancing the conversion efficiency. The damage threshold was reported to be at least ∼109\sim10^9 W/cm2^2 at 1064 nm [19], and its high resistance to UV irradiation was demonstrated by the absence of transmission degradation, after irradiation with more than 1.1×1091.1 \times10^9 pulses at 157 nm (10 ns, 2 mJ/cm2^2 per pulse) [20]. These properties make BMF a promising platform for all-solid-state coherent light sources via QPM, particularly in the VUV spectral range.

Wavelength conversion to 148.4 nm for 229{}^{229}Th optical nuclear clocks has recently been achieved with a periodically-poled BMF [21], but with a limited conversion efficiency due to the use of 9th^{\mathrm{th}}-order QPM; thus, further optimization is required to improve the conversion efficiency.

Accurate knowledge of the NLO coefficients is essential for optimizing the conversion efficiency and designing efficient QPM devices, as the SHG output scales quadratically with the effective NLO coefficients. BMF is an orthorhombic crystal that belongs to the point group mm2. The twofold-rotation polar axis is the crystallographic cc-axis (cc = 5.821 Å), and the aa- and bb-axes are assigned according to the convention of a<ba < b (aa = 4.126 Å, bb = 14.518 Å) [22]. This axis assignment has consistently been adopted in previous studies of nonlinear optical properties of BMF crystals [19, 23, 24]. The second-order NLO coefficient tensor has five nonzero elements. Using the contracted matrix dild_{il} (dil=1/2χil(2)d_{il} = 1/2\chi_{il}^{(2)}), the second-order nonlinear polarization for SHG is expressed as

\begin{pmatrix} P_x(2\omega)\\ P_y(2\omega)\\ P_z(2\omega) \end{pmatrix} = \varepsilon_0 \begin{pmatrix} 0 & 0 & 0 & 0 & d_{15} & 0\\ 0 & 0 & 0 & d_{24} & 0 & 0\\ d_{31} & d_{32} & d_{33} & 0 & 0 & 0 \end{pmatrix} \begin{pmatrix} E_x^2(\omega)\\ E_y^2(\omega)\\ E_z^2(\omega)\\ 2E_y(\omega)E_z(\omega)\\ 2E_z(\omega)E_x(\omega)\\ 2E_x(\omega)E_y(\omega) \end{pmatrix}. \tag*{(1)}\end{pmatrix}

Here, x,y,z≡a,b,cx,y,z \equiv a,b,c. The subscripts ω\omega and 2ω2\omega denote the fundamental and the second-harmonic (SH) frequencies, respectively. Assuming Kleinman symmetry, d31=d15d_{31} = d_{15} and d32=d24d_{32} = d_{24}. Table 1 summarizes three previous studies that reported the NLO coefficients of BMF, dild_{il}, at a fundamental wavelength of 1064 nm together with the corresponding coherence lengths, lcl_c [19, 23, 24]. The reported NLO coefficients, however, show a substantial discrepancy between the two studies from the 1970s [19, 23] and the later study in 2012 [24].

Table 1. Previously reported second-order NLO coefficients of BMF. In [19, 23], the NLO coefficients were reported in the relative scale with their references, and the uncertainties were reported in percentage. Here, the NLO coefficients are calculated using the representative values of d11d_{11} = 0.30 pm/V for α\alpha-quartz and d36d_{36} = 0.39 pm/V for KH2_2PO4_4 (KDP) [25]. The uncertainties are also calculated into absolute value. In [24], the NLO coefficients were scaled using LiNbO3_3 (d31d_{31} = 4.35 pm/V) as reference.

In this work, using high-quality BMF single crystals grown by the Czochralski technique, we determined its NLO coefficients and coherence lengths by combining wedge and Maker fringe measurements, using d11d_{11} (0.30 pm/V) of α\alpha-SiO2_2 (α\alpha-quartz) as the reference for absolute scaling of the NLO coefficients, rather than d31d_{31} (4.35 pm/V) of LiNbO3_3 used in [24]. The resulting coefficient ratio of d31/d32d_{31}/d_{32} was further supported by independent single-pass SHG experiments. Our results confirm that the NLO coefficients of BMF are in the range of 0.01–0.04 pm/V, which is consistent with the reports from the 1970s [19, 23], and show that the coefficients d31d_{31} and d15d_{15} were underestimated.

Ref.[19][23][24]
NLO reference [pm/V]α-quartz d11=0.30d_{11}=0.30KDP d36=0.39d_{36}=0.39LiNbO3_3 d31=4.35d_{31}=4.35
d31d_{31} [pm/V]0.021±0.0040.021\pm0.0040.022±0.0050.022\pm0.0050.150.15
d32d_{32} [pm/V]0.039±0.0080.039\pm0.0080.033±0.0040.033\pm0.0040.360.36
d33d_{33} [pm/V]0.015±0.0030.015\pm0.0030.009±0.0010.009\pm0.0010.120.12
d15d_{15} [pm/V]0.021±0.0040.021\pm0.004--
d24d_{24} [pm/V]0.021±0.0040.021\pm0.0040.024±0.0040.024\pm0.004-
lc31l_{c31} [μm]∞\infty665±133665\pm133380380
lc32l_{c32} [μm]9.69.610.9±0.210.9\pm0.21111
lc33l_{c33} [μm]29.329.335.7±1.835.7\pm1.83535
lc15l_{c15} [μm]23.323.325.8325.83*-
lc24l_{c24} [μm]170170222±33222\pm33-

Table 3 (PDF p. 3).

*Calculated value from refractive indices.

Rotational Maker fringe and wedge measurements

Crystal growth and sample preparation

We grew several BMF single crystals by the Czochralski technique and prepared aa- and bb-cut plane-parallel as well as wedged samples, used for measurements with rotational Maker fringe and wedge techniques. The detailed growth procedure is described in [26]. Note that our crystals grown under the same conditions were also used for the 148-nm generation [21]. As a reference for the absolute NLO coefficients, we used a Y-cut (cut perpendicularly to [01ˉ\bar{1}10]-axis) α-quartz sample with the NLO coefficient d11d_{11} known to be 0.30 pm/V [25].

Rotational Maker fringe measurement

Figure 1 illustrates the experimental setup. As a pump source, we used a Q-switched laser at 1064 nm, delivering 1-ns pulses with a pulse energy of ≈75 μJ at a repetition rate of 1 kHz (CryLaS, DSS1064-Q3). Its output was focused onto a sample mounted on motorized translation and rotation stages. The beam diameter was measured to be 360 μm × 420 μm at the sample position. The polarization of the fundamental light was controlled with a half-wave plate and a polarizer. The SHG signal of a selected polarization was detected using a photomultiplier tube (PMT, Hamamatsu, R7207-01) after the fundamental light was suppressed by a harmonic separator and a bandpass filter (BPF). To measure and compare SH signals at substantially different intensities, neutral density (ND) filters with known transmittances were used to attenuate the SH signals when necessary. To compensate for the peak-power fluctuation of the fundamental pulses, the SH signal from a β-BaB2_2O4_4 (BBO) crystal was simultaneously monitored using a photodetector (PD). All optics, except for BMF samples, were anti-reflection coated at the fundamental or SH wavelengths. The signals from PD and PMT were respectively averaged in boxcar integrators (Stanford Research Systems, SR250) for 100 pulses to improve the signal-to-noise ratio. The polarization angle of the fundamental light was chosen to isolate SH signals from different NLO coefficients. Figure 2 shows the BMF sample orientation and polarization setting in the Maker fringe measurements for the five NLO coefficients.

The SHG intensity, I2ωI_{2\omega}, is given by [23]

I2ω(θ)=dil2f(θ)Iω2sin⁡2Ψ,(2)I_{2\omega}(\theta) = d_{il}^{2} f(\theta) I_{\omega}^{2} \sin^{2}\Psi, \tag*{(2)}

where θ\theta is the angle of incidence, f(θ)f(\theta) is the envelope function (see Appendix § 1), and Ψ\Psi is half of the phase difference between the fundamental and SH waves, given by

Ψ=2πLλω(nωcos⁡θω′−n2ωcos⁡θ2ω′).(3)\Psi= \frac{2\pi L}{\lambda_{\omega}}\left(n_{\omega}\cos\theta_{\omega}^{\prime} - n_{2\omega}\cos\theta_{2\omega}^{\prime}\right). \tag*{(3)}

Here, LL is the sample thickness, λω\lambda_{\omega} is the fundamental wavelength, nn is the refractive index, and θ′\theta^{\prime} is the refraction angle. The detailed expressions of SHG intensity for isotropic and uniaxial crystals are given in [27], which we used to analyze the result of α\alpha-quartz reference sample, and those for biaxial crystals in point group mm2 are given in [23]. Equations (2) and (3) allow us to calculate the relative NLO coefficients by comparing the envelope intensities at the normal incidence, Ienv(0)=dij2f(0)Iω2I_{\mathrm{env}}(0) = d_{ij}^{2} f(0) I_{\omega}^{2}, as

dil,BMFd11,α-quartz=Ienv,BMF(0)fBMF(0) / Ienv,α-quartz(0)fα-quartz(0)(4)\frac{d_{il,\mathrm{BMF}}}{d_{11,\alpha\text{-quartz}}} = \sqrt{\frac{I_{\mathrm{env},\mathrm{BMF}}(0)}{f_{\mathrm{BMF}}(0)}} \, / \, \sqrt{\frac{I_{\mathrm{env},\alpha\text{-quartz}}(0)}{f_{\alpha\text{-quartz}}(0)}} \tag*{(4)}

To validate our experimental setup, we compared the two representative NLO references, d11d_{11} of α-quartz and d36d_{36} of KH2_2PO4_4 (KDP) crystals. We obtained an envelope intensity ratio of Ienv,α-quartz/Ienv,KDP=1.7±0.1I_{\mathrm{env},\alpha\text{-quartz}}/I_{\mathrm{env},\mathrm{KDP}} = 1.7 \pm0.1, resulting in a NLO coefficient ratio of d11,α-quartz/d36,KDP=0.77±0.03d_{11,\alpha\text{-quartz}}/d_{36,\mathrm{KDP}} = 0.77 \pm0.03. These values are consistent with previous reports [25, 28].

The obtained Maker fringes are shown in Figure 3. As reported in [23], we observed the type-I phase-matching in the bcbc-plane for the d31d_{31} measurement (Figure 3(a)), at an incident angle θ\theta of 13.8°. As the SHG intensity at the phase-matching peaks in d31d_{31} measurement was too high to be measured with PMT, we restricted the rotation angle to ∣θ∣<10∘|\theta| < 10^\circ (Figure 3(a)). To observe a sufficient number of fringes in the limited angle range, we used a ~10-mm thick sample.

Maker fringes of BMF for the NLO coefficients

Figure 3. Maker fringes of BMF for the NLO coefficients of (a) d31d_{31} (b) d32d_{32} (c) d33d_{33} (d) d15d_{15} (e) d24d_{24}. The vertical scale is normalized to the envelope peak of SHG intensity obtained from the reference α\alpha-quartz sample shown in (f). For each measurement, the best fit results are denoted in the respective figure: the sample thickness LL, with the correction ΔL\Delta L to the measured value, the envelope intensity at normal incidence Ienv(0)I_{\mathrm{env}}(0), and corrections for refractive indices Δn\Delta n.

In the d15d_{15} measurement (Figure 3(d)), the rotation angle was also restricted to ∣θ∣<10∘|\theta| < 10^\circ to avoid the detection of type-I phase-matching SH signal due to d31d_{31}. The intense phase-matched signal, despite being orthogonally polarized to the signal originating from d15d_{15} (cf. Figure 2(a) and (Figure 2d)), could not be completely eliminated even with three polarizers. Despite the restricted rotation angle, we still observed an unexpected Maker fringe shape as well as an offset. We attributed this to the interference with the SH signal from other NLO coefficients. In our analysis, we subtracted the offset and fitted the theoretical equation to the experimental data in the range of ∣θ∣<5∘|\theta| < 5^\circ. Note that a similar difficulty in measuring d15d_{15} due to the phase-matching peak was also reported in [23].

Experimental configuration in the Maker fringe measurements for (a) $d_{31}$, (b) $d_{32}$, (c) $d_{33}$, (d) $d_{15}$, (e) $d_{24}$ coefficients of BMF

Figure 2. Experimental configuration in the Maker fringe measurements for (a) d31d_{31}, (b) d32d_{32}, (c) d33d_{33}, (d) d15d_{15}, (e) d24d_{24} coefficients of BMF. For d33d_{33}, both a- and b-cut samples can be used.

For each Maker fringe, we first calculated the theoretical fringes using the refractive indices based on the reported Sellmeier equations [18]. However, the calculated fringes did not fit the experimental ones, suggesting the limited accuracy of the reported Sellmeier equations. The discrepancy was also observed in the type-I phase-matching angle in the bcbc-plane. Phase-matching was observed at θ=13.8∘\theta=13.8^\circ, whereas the Sellmeier equations predict it at θ=17.2∘\theta=17.2^\circ, corresponding to a refraction angle of θ′=11.7∘\theta'=11.7^\circ inside the crystal. Therefore, we introduced refractive-index correction terms, Δnω\Delta n_\omega and Δn2ω\Delta n_{2\omega}, as additional fitting parameters to be added to the refractive indices from the Sellmeier equations. The applied corrections are indicated in each panel of Figure 3. These correction terms were added to each principal refractive index and were on the order of 10−410^{-4}, suggesting that the precision of the reported Sellmeier equations is limited to this order. Furthermore, we slightly varied the sample thickness by a few micrometers for better fitting. The Maker fringes, particularly for d32d_{32} and d24d_{24} of BMF as well as d11d_{11} of α\alpha-quartz, are affected by multiple reflections of the fundamental light between the two parallel sample surfaces [29, 30]. The resulting interference oscillations were considered in estimating the uncertainties. The systematic overestimation of the NLO coefficients due to multiple reflections was negligible, as it is reported to be less than 5% for crystals with refractive indices smaller than 2, including BMF [31].

Determination of the coherence lengths by wedge technique

The determination of NLO coefficients from the rotational Maker fringe results requires the accurate knowledge of coherence lengths for normal incidence, lc(θ=0)l_c(\theta=0), which is defined as a propagation length over which the phase difference between the fundamental and SH waves, 2Ψ2\Psi, becomes π\pi, given as

lc(θ=0)=λω4∣nω−n2ω∣(5)l_c(\theta=0)=\frac{\lambda_\omega}{4\lvert n_\omega-n_{2\omega}\rvert} \tag*{(5)}

The envelope function of the Maker fringes f(θ)f(\theta) in Eqs. (2) and (4) involves the term 1/(nω2−n2ω2)21/(n_\omega^2-n_{2\omega}^2)^2 (see Appendix § 1), which is inversely proportional to lc2l_c^2. Accordingly, the calculated NLO coefficients are affected by the accuracy of the coherence lengths, as dil∝lc−1d_{il}\propto l_c^{-1}. However, the coherence lengths calculated from Eq. (5) could be highly sensitive to the refractive index errors, particularly when ∣nω−n2ω∣\lvert n_\omega-n_{2\omega}\rvert is on the order of 10−410^{-4}, comparable to the precision of the reported Sellmeier equations [18].

Therefore, we experimentally determined the coherence lengths by the wedge technique. The results are shown in Figure 4. The obtained fringes were fitted using Eq. (2) for normal incidence (θ=θ′=0\theta=\theta'=0), with nω−n2ωn_\omega-n_{2\omega} as a fitting parameter. For d31d_{31} and d24d_{24}, the corresponding coherence lengths were longer than 200 μ\mum. We therefore used three and two samples, respectively, with different thicknesses but the same wedge angle of 0.54±0.01∘0.54\pm0.01^\circ, to observe sufficient data enabling reliable sinusoidal fitting. Only for d15d_{15}, we could not obtain a periodic sin⁡2Ψ\sin^2\Psi-shaped curve. This is attributable to the SHG signal from other NLO coefficients. Since the coherence length for d15d_{15} could not be determined from the fringe, we adopted a value calculated from the refractive indices from the reported Sellmeier equations [18]. As the corresponding ∣nω−n2ω∣\lvert n_\omega-n_{2\omega}\rvert is on the order of 10−210^{-2}, the coherence length for d15d_{15} can be calculated with sufficient accuracy from the reported Sellmeier equations. Some fringes exhibit unexpected spikes, especially for d24d_{24}, which might be attributed to surface defects, but further investigation is required to clarify their origin. Note that no grain boundaries were observed in the samples used in the experiments.

Results of wedge measurements for $d_{31}$, $d_{32}$, $d_{33}$, $d_{15}$, and $d_{24}$

Figure 4. The results of wedge measurements for (a) d31d_{31} (b) d32d_{32} (c) d33d_{33} (d) d15d_{15} (e) d24d_{24}. The vertical scale is normalized to the peak envelope intensity of the rotational Maker fringe measurement measured for the reference α\alpha-quartz. The values of the peak intensity and Δ(nω−n2ω)\Delta(n_{\omega}-n_{2\omega}) obtained for the best fit are indicated in each figure.

The dependence of the coherence lengths on the errors of nω−n2ωn_{\omega}-n_{2\omega}, calculated from Eq. (5), is shown in Figure 5 as the black solid curves, together with the previously reported coherence lengths [19, 23] and our results. The coherence lengths for d31d_{31} and d24d_{24} are shown to be particularly sensitive to the errors in refractive indices.

Dependence of BMF coherence length on $n_{\omega}-n_{2\omega}$ for $d_{31}$, $d_{32}$, $d_{33}$, $d_{15}$, and $d_{24}$

Figure 5. Dependence of BMF coherence length lc(θ=0)l_c(\theta=0) on the value of nω−n2ωn_{\omega}-n_{2\omega} (solid line) for (a) d31d_{31} (b) d32d_{32} (c) d33d_{33} (d) d15d_{15} (e) d24d_{24}. The horizontal axis spans ±0.001\pm0.001 around the value calculated from the reported Sellmeier equations [18]. The previously reported coherence lengths [19, 23] and those obtained in this work are also shown, with the colored bands indicating uncertainties.

Determination of the second-order nonlinear optical coefficients

Using the determined coherence lengths, we calculated the NLO coefficients according to Eq. (4). The results are summarized in Table 2. The uncertainties in the NLO coefficients were calculated from those in the coherence lengths and the envelope intensities. The coefficient d15d_{15} is identified to be the largest. This value is consistent with the equality d31=d15d_{31}=d_{15} under Kleinman symmetry. The five coefficients are fairly consistent with those reported in the 1970s [19, 23], while they are significantly different from the latest report [24]. We discuss this difference in Section 4. It is noteworthy that the value of d31d_{31} is found to be ~1.5 times larger than reported in the earlier two papers and is comparable to d32d_{32}, which was previously reported to be the largest NLO coefficient of BMF. This result is further examined in the following section.

The Second-order NLO coefficients of BMF determined in this work. The coherence length and its uncertainty for d15d_{15} are calculated using the reported Sellmeier equations [18], assuming a deviation of ±0.001\pm0.001 in the value of nω−n2ωn_{\omega}-n_{2\omega} as shown in Figure 5(d).

dild_{il}NLO coefficient [pm/V]Coherence length [μm]
d31d_{31}0.035±0.0020.035\pm0.002580±10580\pm10
d32d_{32}0.038±0.0020.038\pm0.00210.9±0.210.9\pm0.2
d33d_{33}0.0115±0.00040.0115\pm0.000434.6±0.234.6\pm0.2
d15d_{15}0.040±0.0040.040\pm0.00424±224\pm2
d24d_{24}0.024±0.0010.024\pm0.001210±5210\pm5

Table 2.

Comparison of d31d_{31} and d32d_{32} via single-pass SHG

Contrary to the three previous reports on the NLO coefficients of BMF, which reported d31d_{31} to be approximately half of d32d_{32}, our results suggest that d31d_{31} is comparable to d32d_{32}. To further validate our results, we compared these two NLO coefficients via single-pass SHG experiments, under the type-I BPM in the bcbc-plane (deff=d31cos⁡θPM′d_{\mathrm{eff}}=d_{31}\cos\theta'_{\mathrm{PM}}, where θPM′\theta'_{\mathrm{PM}} is the phase-matching angle in the crystal) and under QPM using d32d_{32} without any temperature control. The measurements were performed in a similar setup shown in Figure 1, but the generated SH signal was detected using a power meter instead of the PMT. We also increased the repetition rate of the Q-switched laser to 2.5 kHz for higher average power, and reduced the beam spot diameter to 70 μm × 110 μm for higher conversion efficiencies. The sample for BPM was bb-cut and 9.66 mm in thickness, while the one for QPM was a 5-mm-long aa-cut sample with a poled length of 4.01 mm. The periodic domain inversion was performed using a calligraphic poling technique [26]. The sample had 183 cycles of inverted domains with a nominal period of 21.896 μm\mu\mathrm{m} and ∼\sim40% duty ratio. The period, corresponding to 2lc322l_{c32}, was calculated from the reported Sellmeier equations [18]. The microscope image of the fabricated periodically-poled BMF is shown in Figure 6(a).

Experimental setup

Figure 1. Experimental setup. BPF: bandpass filter, PD: photo detector, PBS: polarizing beam splitter, λ/2\lambda/2: half-wave plate, ND filter: neutral density filter, PMT: photomultiplier tube

Figure 6. (a) Microscope image of a 300-μm\mu\mathrm{m} thick periodically-poled BMF crystal, which was selectively etched for visualizing the poled domains. The nominal poling period was 21.896 μm\mu\mathrm{m}, corresponding to the 1st1^{\mathrm{st}} order of QPM. (b) Comparison of SHG output average power between BPM using d31d_{31} and QPM using d32d_{32}.

Figure 6(b) shows the average SHG power versus the incident fundamental laser power for BPM and QPM. As expected, the SHG output power increased quadratically with the incident fundamental power. Although we aimed QPM at normal incidence, the QPM peak was observed at an incident angle of ∼9∘\sim9^\circ, indicating the domain period being not exactly equal to the actual 2lc322l_{c32}. This result suggests the required poling period of 21.78 μm\mu\mathrm{m}, corresponding to a shift in refractive indices of only ∼10−4\sim10^{-4}, which falls within the error bar in Figure 5(b). Thus, we conclude that the shift was due to the limited accuracy of the Sellmeier equations used. The peak from BPM was observed at an incident angle of 13.8°.

The SHG output power follows the Boyd-Kleinman equation [32], given by

P2ω=2ω2deff2Lkωπε0c3nω2n2ωPω2h(B,ξ),(6)P_{2\omega}=\frac{2\omega^{2}d_{\mathrm{eff}}^{2}Lk_{\omega}}{\pi\varepsilon_{0}c^{3}n_{\omega}^{2}n_{2\omega}}P_{\omega}^{2}h(B,\xi), \tag*{(6)}

where Pω,2ωP_{\omega,2\omega} are the peak power of the fundamental and SH wave, kωk_{\omega} is the wavenumber of the fundamental, and h(B,ξ)h(B,\xi) is the Boyd-Kleinman factor. Under loose focusing conditions, this factor is approximated as

h(B,ξ)≃ξ(1−t2/12+t4/120−t6/1344+⋯ ).(7)h(B,\xi)\simeq\xi(1-t^{2}/12+t^{4}/120-t^{6}/1344+\cdots). \tag*{(7)}

Here B=ρLkω/2B=\rho\sqrt{Lk_{\omega}}/2, ξ=L/b\xi=L/b, b=r2kωb=r^{2}k_{\omega}, and t=2B2ξt=2B\sqrt{2\xi}, where ρ\rho is the walk-off angle, LL is the nonlinear interaction length, corresponding to the crystal thickness for BPM or the poled length for QPM, and rr is the beam radius in the plane of spatial walk-off. The effective NLO coefficient for the BPM is deff=d31cos⁡θPM′d_{\mathrm{eff}}=d_{31}\cos\theta'_{\mathrm{PM}}, whereas that for QPM is scaled by a factor of SS, such that deff=d32Sd_{\mathrm{eff}}=d_{32}S. Assuming perfect periodicity with a duty ratio of DD, SS is 2sin⁡(πD)/π2\sin(\pi D)/\pi [33]. The walk-off angles between the fundamental and the SH wave at the corresponding angles of incidence were calculated to be 0.22° for BPM and 0.16° for QPM, respectively.

Taking these effects into account, the ratio of d31/d32d_{31}/d_{32} from the single-pass SHG measurement was calculated to be ≈0.84\approx0.84, in fairly good agreement with the ratio obtained from the Maker fringe and the wedge measurements: 0.92±\pm0.07. These values are clearly larger than the ratios from previously reported coefficients: 0.53 [19], 0.67 [23], and 0.41 [24].

Discussion

Comparison with previous studies

The NLO coefficients reported in 2012 [24] are approximately one order of magnitude larger than those reported in the other studies, as well as in our work. Their Maker fringes do not agree with the theoretically expected ones. In particular, the measured fringes for d31d_{31} shows no BPM feature. Besides, the use of LiNbO3\mathrm{LiNbO}_{3} (d31=4.35d_{31}=4.35 pm/V) as a reference to determine absolute NLO coefficients could have caused errors, because the NLO coefficients of BMF are found to be two orders of magnitude smaller than d31d_{31} of LiNbO3\mathrm{LiNbO}_{3}, making a direct comparison and scaling challenging. They state that their values are closer to the first-principles calculations, which yielded d32=0.055d_{32}=0.055 pm/V and d33=0.190d_{33}=0.190 pm/V [34]. However, only the reported d33=0.12d_{33}=0.12 pm/V is of the same order of magnitude as the calculation.

The difference in d31d_{31} between our work and the earlier two reports in the 1970s [19, 23] may be attributed to the uncertainty in the coherence length used in the analysis. Bergman et al. did not manage to measure it quantitatively and put it as infinity [19]. Bechthold et al. also overestimated the coherence length by 15% [23], which could result in 15% underestimation of d31d_{31}. In addition, the long coherence lengths make it challenging to reliably determine d31d_{31}, particularly for thin samples. The number of observable fringes is limited for thin samples, and the Maker fringes are affected by peaks arising from BPM. These factors complicate the analysis of the fringes. To mitigate these challenges, we used thick samples (>10 mm) and restricted our analysis to a narrow angular range to avoid the BPM peaks.

Future application of BMF for VUV nonlinear optics

The larger d31d_{31} value obtained in this work, although still not the largest of the five, is favorable for realizing wavelength conversion into the VUV region. Figure 7(a) shows the estimated wavelength dependence of the NLO coefficients. Assuming a constant Miller’s delta [35], d31d_{31} is roughly estimated to be 1.4 times larger, for example at a SHG wavelength of 180 nm, than the value determined at 532 nm. The use of d31d_{31} offers one of the longest coherence lengths, or poling periods for QPM, as shown in Figure 7(b). The smaller poling periods required at shorter wavelengths are one of the current technological obstacles in efficient VUV generation. Although the differences in coherence lengths among the corresponding coefficients decrease at shorter wavelengths (cf. Figure 7(b)), the coherence length associated with d31d_{31} is still 1.3 times longer than that associated with d32d_{32} for SHG to 180 nm. The longer coherence length allows the use of a lower QPM order, relaxing the fabrication requirements while retaining the advantage of the relatively large d31d_{31}. In fact, the first demonstration of 148-nm generation using periodically-poled BMF used d31d_{31} owing to the long poling period required [21].

Changes of the NLO coefficients and coherence lengths of BMF

Figure 7. (a) Changes of the NLO coefficients of BMF in the short wavelength region, assuming the constant Miller’s delta. (b) Changes of the coherence lengths.

Conclusion

We redetermined the second-order NLO coefficients of BMF. Our rotational Maker fringe and wedge measurements confirmed that the NLO coefficients of BMF are in the range of 0.01–0.04 pm/V, consistent with the two previous reports from the 1970s [19, 23]. Contrary to these reports, however, we suggest that d31d_{31} and d15d_{15} are comparable to d32d_{32}, which was previously reported to be the largest. The scale of d31d_{31} was also supported by our single-pass SHG experiments. The key to a reliable determination of NLO coefficients would be using a reference crystal possessing an NLO coefficient comparable with investigated coefficients, ensuring a sufficient number of fringes for robust fitting, and avoiding phase-matching peaks. The long coherence length associated with d31d_{31} allows for larger QPM periods than those required for d32d_{32}, facilitating the fabrication of lower-order QPM devices while retaining the advantage of the moderately large optical nonlinearity. To achieve high conversion efficiency in VUV generation, an additional enhancement mechanism would still be necessary, such as an external enhancement cavity [21], a whispering-gallery-mode resonator [36], or a QPM waveguide [37]. Our results provide useful guidelines for the design and fabrication of periodically-poled BMF devices for VUV generation.

Maker fringe envelope functions for orthorhombic crystals with mm2 point group {#sec:Appendix 1}

For crystals belonging to point group mm2, such as BMF, the theoretical equations for the Maker fringe measurements are given in [23]. The envelope function f(θ)f(\theta) in Eq. (2) is denoted as

f(θ)=1024π3c(4cos⁡2θw(ω)+cos⁡θ)2w(ω)n1(2ω)2cos⁡θ+v(2ω)2(v(2ω)−w(ω))2(v(2ω)+w(ω))(v(2ω)+n1(2ω)2cos⁡θ)3f(\theta)=\frac{1024\pi^3}{c}\left(\frac{4\cos^2\theta}{w^{(\omega)}+\cos\theta}\right)^2\frac{w^{(\omega)}n_1^{(2\omega)2}\cos\theta+v^{(2\omega)2}}{\left(v^{(2\omega)}-w^{(\omega)}\right)^2\left(v^{(2\omega)}+w^{(\omega)}\right)\left(v^{(2\omega)}+n_1^{(2\omega)2}\cos\theta\right)^3}

for d31d_{31}, d32d_{32},

f(θ)=1024π3c(4cos⁡2θw(ω)+cos⁡θ)2w(2ω)(w(ω)+cos⁡θ)(w(2ω)−w(ω))2(w(2ω)+w(ω))(w(2ω)+cos⁡θ)3f(\theta)=\frac{1024\pi^3}{c}\left(\frac{4\cos^2\theta}{w^{(\omega)}+\cos\theta}\right)^2\frac{w^{(2\omega)}\left(w^{(\omega)}+\cos\theta\right)}{\left(w^{(2\omega)}-w^{(\omega)}\right)^2\left(w^{(2\omega)}+w^{(\omega)}\right)\left(w^{(2\omega)}+\cos\theta\right)^3}

for d33d_{33}, and

f(θ)=1024π3c(4v(ω)cos⁡2θ(v(ω)+n1(ω)2cos⁡θ)(w(ω)+cos⁡θ))2f(\theta)=\frac{1024\pi^3}{c}\left(\frac{4v^{(\omega)}\cos^2\theta}{\left(v^{(\omega)}+n_1^{(\omega)2}\cos\theta\right)\left(w^{(\omega)}+\cos\theta\right)}\right)^2
×w(2ω)[(v(ω)+w(ω))/2+cos⁡θ](w(2ω)−(v(ω)+w(ω))/2)2(w(2ω)+(v(ω)+w(ω))/2)(w(2ω)+cos⁡θ)3{}\times\frac{w^{(2\omega)}\left[\left(v^{(\omega)}+w^{(\omega)}\right)/2+\cos\theta\right]}{\left(w^{(2\omega)}-\left(v^{(\omega)}+w^{(\omega)}\right)/2\right)^2\left(w^{(2\omega)}+\left(v^{(\omega)}+w^{(\omega)}\right)/2\right)\left(w^{(2\omega)}+\cos\theta\right)^3}

for d15d_{15}, d24d_{24}. Here, n1n_1 is the refractive index for the light polarized parallel to the plane of incidence and the crystal surface. Both vv and ww are equivalent to ncos⁡θ′n\cos\theta'. Note that vv is used for the s-polarized beam and ww for the p-polarized beam.

Dependence of the coherence lengths on wavelengths {#sec:Appendix 2}

The coherence lengths of BMF for different wavelengths calculated by Eq. (5) are shown in Figure 8. The longest coherence length changes from lc31l_{c31} to lc24l_{c24} at SHG wavelengths shorter than 516 nm, as type-II non-critical BPM occurs at 476 nm.

Dependence of the coherence lengths on SHG wavelengths

Figure 8. Dependence of the coherence lengths on SHG wavelengths. Type-II non-critical BPM occurs at 476 nm associated with d24d_{24}, while type-I non-critical BPM occurs at 567 nm associated with d31d_{31}.

Funding. German Federal Ministry of Research, Technology and Space (13N16924); Leibniz Association (J183/2023).

Acknowledgement. We are grateful to Prof. Ichiro Shoji (Chuo University, Japan) for valuable discussions.

Disclosures. The authors declare no conflicts of interest.

Data availability. Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

References

  1. [1]Z. Zhang, Y. Ma, C. Niu, K. Li, K. Wang, C. Wu, and Y. Yu, "Development of all-solid-state ultraviolet lasers," J. Laser Appl. 36, 041202 (2024).
  2. [2]I. K. Lednev, V. V. Ermolenko, W. He, and M. Xu, "Deep-UV Raman spectrometer tunable between 193 and 205 nm for structural characterization of proteins," Anal. Bioanal. Chem. 381, 431–437 (2005).
  3. [3]F. Fan, Z. Feng, and C. Li, "UV Raman spectroscopic study on the synthesis mechanism and assembly of molecular sieves," Chem. Soc. Rev. (2010).
  4. [4]Q.-J. Peng, N. Zong, S.-J. Zhang, Z.-M. Wang, F. Yang, F.-F. Zhang, Z.-Y. Xu, and X.-J. Zhou, "DUV/VUV all-solid-state lasers: Twenty years of progress and the future," IEEE J. Sel. Top. Quantum Electron. 24, 1–12 (2018).DOI
  5. [5]D. R. Leib­randt, S. G. Porsev, C. Cheung, and M. S. Safronova, "Prospects of a thousand-ion Sn²⁺ Coulomb-crystal clock with sub-10⁻¹⁹ inaccuracy," Nat. Commun. 15, 5663 (2024).DOI
  6. [6]E. Pek and C. Tamm, "Nuclear laser spectroscopy of the 3.5 eV transition in Th-229," Europhys. Lett. 61, 181–186 (2003).
  7. [7]V. Petrov, F. Noack, D. Shen, F. Pan, G. Shen, X. Wang, R. Komatsu, and V. Alex, "Application of the nonlinear crystal SrBa₄O₇ for ultrafast diagnostics converting to wavelengths as short as 125 nm," Opt. Lett. 29, 373–375 (2004).
  8. [8]A. I. Zaitsev, A. S. Aleksandrovsky, A. D. Vasiliev, and A. V. Zamkov, "Domain structure in strontium tetraborate single crystal," J. Cryst. Growth 310, 1–4 (2008).DOI
  9. [9]P. Trabs, F. Noack, A. S. Aleksandrovsky, A. I. Zaitsev, and V. Petrov, "Generation of coherent radiation in the vacuum ultraviolet using randomly quasi-phase-matched strontium tetraborate," Optics letters 41, 618–621 (2016).
  10. [10]V. Lal, M. V. Okhapkin, J. Tiedau, N. Irwin, V. Petrov, and E. Pek, "Continuous-wave laser source at the 148 nm nuclear transition of Th-229," Optica 12, 1971 (2025).
  11. [11]C. Chen, G. Wang, X. Wang, Y. Zhu, Z. Xu, T. Kanai, and S. Watanabe, "Improved sellmeier equations and phase-matching characteristics in deep-ultraviolet region of KBe₂BO₃F₂ crystal," IEEE J. Quantum Electron. 44, 617–621 (2008).DOI
  12. [12]S.-B. Dai, M. Chen, S.-J. Zhang, Z.-M. Wang, F.-F. Zhang, F. Yang, Z.-C. Wang, N. Zong, L.-J. Liu, X.-Y. Wang, J.-Y. Zhang, Y. Bo, D.-F. Cui, Q.-J. Peng, R.-K. Li, C.-T. Chen, and D.-X. Xu, "12 14 mW deep-ultraviolet laser at 165 nm by eight-harmonic generation of a 1319 nm Nd:YAG laser in KBBF," Laser Phys. Lett. 13, 035401 (2016).
  13. [13]T. Nakazato, I. Ito, Y. Kobayashi, X. Wang, C. Chen, and S. Watanabe, "Phase-matched frequency conversion below 150 nm in KBe₂BO₃F₂," Opt. Express 24, 17149–17158 (2016).DOI
  14. [14]M. Scholz, D. Opalevs, P. Leisching, W. Kaenders, G. Wang, X. Wang, R. Li, and C. Chen, "A bright continuous-wave laser source at 193 nm," Appl. Phys. Lett. 103, 051114 (2013).DOI
  15. [15]H. Su, Z. Yan, X. Hou, and M. Zhang, "Fluoroborates: A precious treasure of deep-ultraviolet nonlinear optical materials," Jiegou Huaxue 42, 100027 (2023).DOI
  16. [16]F. Zhang, Z. Chen, C. Cui, Z. Yang, M. Mutailipu, F. Li, X. Hou, X. Long, and S. Pan, "Vacuum ultraviolet second-harmonic generation in NH₄B₄O₆F crystal," Nature 650, 97–101 (2026).DOI
  17. [17]K. Shimamura, E. G. Villora, H. Zeng, M. Nakamura, S. Takekawa, and K. Kitamura, "Ferroelectric properties and poling of BaMgF₄ for ultraviolet all solid-state lasers," Appl. Phys. Lett. 89, 8–11 (2006).
  18. [18]E. Villora, K. Shimamura, K. Sumiya, and H. Ishibashi, "Birefringent- and quasi phase-matching with BaMgF₄ for vacuum-UV/UV and mid-IR all solid-state lasers," Opt. Express 17, 12362–12378 (2009).pubmed.ncbi.nlm.nih.gov/19654638
  19. [19]J. G. Bergman, G. R. Crane, and H. Guggenheim, "Linear and nonlinear optical properties of ferroelectric BaMgF₄ and BaZnF₄," J. Appl. Phys. 46, 4645–4646 (1975).DOI
  20. [20]S. C. Buchter, T. Y. Fan, V. Liberman, J. J. Zayhowski, M. Rothschild, E. J. Mason, A. Cassanho, H. P. Jenssen, and J. H. Burnett, "Periodically poled BaMgF₄ for ultraviolet frequency generation," Opt. Lett. 26, 1693 (2001).pubmed.ncbi.nlm.nih.gov/18049702
  21. [21]K. Subramanian, H. Tanaka, S. J. Herr, N. K. Sah, G. Jha, F. Zacherl, S. A. Pradeep, V. Andriushkov, K. Zhang, D. Fenner, Y. Wang, M. Hugenschmidt, F. Kühnemann, G. G. M. Bonetti, S. Ui, M. Bickermann, C. Ma, X. Zheng, M. Zopf, B. Lommel, J. C. Müller, S. Hannig, D. Budker, F. Schmidt-Kaler, and L. von der Wense, "Generation of continuous-wave laser light at 148.4 nm using cavity-enhanced second-harmonic generation in BaMgF₄," Optica 13, 2089 (2026).
  22. [22]F. Gingl, "BaMgF₄ and Ba₂Mg₃F₁₀: New examples for structural relationships between hydrides and fluorides," Z. Anorg. Allg. Chem. 623, 705–709 (1997).DOI
  23. [23]P. S. Bechthold and S. Haussühl, "Nonlinear optical properties of orthorhombic barium formate and magnesium barium fluoride," Appl. Phys. 14, 403–410 (1977).
  24. [24]J. Chen, X. Chen, Y. Ma, Y. Zheng, A. Wu, H. Li, L. Jiang, and J. Xu, "Measurement of second-order nonlinear optical coefficients of BaMgF₄," J. Opt. Soc. Am. B 29, 665 (2012).DOI
  25. [25]I. Shoji, T. Kondo, A. Kitamoto, M. Shirane, and R. Ito, "Absolute scale of second-order nonlinear-optical coefficients," JOSA B 14, 2268–2294 (1997).DOI
  26. [26]S. Herr, H. Tanaka, I. Breunig, M. Bickermann, and F. Kuehnemann, "Fanout periodic poling of BaMgF₄ crystals," Opt. Mater. Express 13, 2158–2164 (2023).DOI
  27. [27]J. Jerphagnon and S. K. Kurtz, "Maker fringes: A detailed comparison of theory and experiment for isotropic and uniaxial crystals," J. Appl. Phys. 41, 1667–1681 (1970).DOI
  28. [28]J. Jerphagnon and S. K. Kurtz, "Optical nonlinear susceptibilities: Accurate relative values for quartz, ammonium dihydrogen phosphate, and potassium dihydrogen phosphate," Phys. Rev. 1, 1739–1744 (1970).DOI
  29. [29]M. Braun, F. Bauer, T. Vogtmann, and M. Schwoerer, "Precise second-harmonic generation Maker fringe measurements in single crystals of the diacetylene NP/4-MPU and evaluation by a second-harmonic generation theory in 4 × 4 matrix formulation and ray tracing," J. Opt. Soc. Am. B, JOSAB 14, 1699–1706 (1997).DOI
  30. [30]M. Abe, I. Shoji, J. Suda, and T. Kondo, "Comprehensive analysis of multiple-reflection effects on rotational Maker-fringe experiments," J. Opt. Soc. Am. B, JOSAB 25, 1616–1624 (2008).
  31. [31]I. Shoji, T. Kondo, and R. Ito, "Second-order nonlinear susceptibilities of various dielectric and semiconductor materials," Opt. Quantum Electron. 34, 797–833 (2002).DOI
  32. [32]G. D. Boyd and D. A. Kleinman, "Parametric interaction of focused Gaussian light beams," J. Appl. Phys. 39, 3597–3639 (1968).
  33. [33]M. M. Fejer, G. A. Magel, D. H. Jundt, and R. L. Byer, "Quasi-phase-matched second harmonic generation: tuning and tolerances," IEEE J. Quantum Electron. 28, 2631–2654 (1992).DOI
  34. [34]H. Huang, Z. S. Lin, L. Bai, R. He, and C. T. Chen, "Mechanism of the linear and nonlinear optical effects of SrAlF₅ and BaMgF₄ crystals," Solid State Commun. (2010).
  35. [35]R. C. Miller, "Optical second harmonic generation in piezoelectric crystals," Appl. Phys. Lett. 5, 17–19 (1964).DOI
  36. [36]A. Mrokon, H. Kraft, D. Shin, H. Tanaka, S. J. Herr, K. Buse, and I. Breunig, "Electrically induced resonance shifts of whispering gallery resonators made of barium magnesium fluoride," APL Photonics 10, (2025).
  37. [37]C. Wang, C. Langrock, A. Marandi, M. Jankowski, M. Zhang, B. Desiatov, and Lončar, "Ultrahigh-efficiency wavelength conversion in nanophotonic periodically poled lithium niobate waveguides," Optica 5, 1438–1441 (2018).DOI

Paper details

vlnc:W_cw92gx8u
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents