Wang Zhaoa,
Yi-sheng Huangb,
Zhou-bin Linb,
Bo Weic,
Feng-wu Wanga,
Mai Xua,
Xing Zhaoa,
Qing-hua Zhenga and
Wei-wei Zhou*a
aAnhui Key Laboratory of Low Temperature Co-fired Materials, School of Electronic Engineering, Huainan Normal University, Huainan, Anhui 232038, PR China. E-mail: wwz9829@126.com
bKey Laboratory of Optoelectronic Materials Chemistry and Physics, Fujian Institute of Research on The Structure of Matter, Chinese Academy of Sciences, Fuzhou, Fujian 350002, PR China
cJiangsu Key Laboratory of Advanced Functional Materials, School of Chemistry and Materials Engineering, Changshu Institute of Technology, Changshu, Jiangsu 215500, PR China
First published on 10th April 2015
Thin mica-like Yb3+:CsGd(MoO4)2 crystals, with a perfect (100) cleavage plane, have been grown by a top seeded solution growth method from a flux of Cs2Mo3O10. The cleavage habit has been found to be closely connected with the layered structure. The Stark-level positions of Yb3+ have been deduced from the excitation and emission spectra at 77 K with the aid of the Raman spectrum, indicating the largest ground-state splitting among the double molybdates and double tungstates ever reported except for Yb3+:LiLu(WO4)2. The emission cross-sections have been determined by the combinatorial reciprocity method and the Füchtbauer–Ladenburg formula. The wavelength dependence of the gain cross-section predicts a broad tuning range and potential sub-100 fs laser pulse generation. Taking into account the cleavage behavior, the crystal is particularly suitable for microchip lasers, in which thin platelets of gain media can be prepared by a simple cleavage technique.
To our knowledge, the literatures on the DMo and DW crystals with M = Rb and Cs are still limited. Most likely, the difficulties (e.g., incongruent melting, structural transformation and cleavage) in the growth of single crystals with centimeter dimensions hamper the spectroscopic analysis and laser applications.10,11 Our group have presented the growth and spectroscopic characterization of Nd3+:CsLa(WO4)2 crystal,11 an interesting crystal with the ordered tetragonal structure. Unfortunately, we failed to obtain the nominal Yb3+:CsLa(WO4)2 crystal, since Yb3+ could hardly incorporate into the crystal lattice.
CsGd(MoO4)2 (CGM) crystallizes in the monoclinic system with the space group P2/c and the cell parameters: a = 9.5289(2) Å, b = 5.0823(2) Å, c = 8.0563(7) Å, β = 91.246(6)°, and Z = 2.12 The CGM features a highly anisotropic layered crystal structure (Fig. 1a), which gives rise to some interesting phenomena as a dipolar magnet.13 The Eu3+:CGM red-emitting phosphor is also reported to be attractive for potential applications in the white light-emitting diode devices.12 From the structural point of view, CGM offers a highly distorted cationic site suitable for lasing ions, especially for Yb3+. Yb3+ lasers operate in a quasi-three-level scheme, since the fundamental and terminal laser levels belong to the same 2F7/2 manifold. In order to limit the thermal population of the terminal level, the crystal-field splitting of the 2F7/2 manifold ΔE as large as possible is desirable. Considering the ΔE is sensitive to the crystal field, one important strategy to extend the ΔE is to employ hosts that accommodate Yb3+ with low site symmetry and different Yb–O distances.14 Similar to the famous KGd(WO4)2,15 Gd3+ occupies an eight-coordinated site (Wyckoff position: 2e) of CGM with C2 symmetry and Gd–O distances ranging from 2.269 Å to 2.704 Å (Fig. 1b). The distortion degree of the [GdO8] polyhedron in CGM is even greater than that in KGd(WO4)2 with the minimum Gd–O distance of 2.271 Å and maximum of 2.650 Å. Such a highly distorted local environment is expected to produce an intensive crystal-field strengths and thus large ΔE of Yb3+. Moreover, the absence of the structural disorder or multisite character in CGM predicts high optical cross-sections and moderate bandwidths. Based on the above consideration, we investigated the crystal growth and the spectroscopic characteristics of the Yb3+:CGM crystal so as to evaluate its feasibility for laser applications.
Fig. 1 The photographs of (a) crystal structure, (b) coordination environment of Gd3+ and (c) a typical (100) cleavage slice of the Yb3+:CGM crystal. |
A piece of 0.8 mm thick cleavage plate with its main faces parallel to the (100) planes (Fig. 1c), was exfoliated from the as-grown crystal by a knife and directly used for the spectral measurements without polishing. The absorption spectra were recorded by a UV/VIS/NIR spectrometer (Lambda900, Perkin–Elmer). The 77- and 300 K excitation and fluorescence spectra as well as the decay curves were carried out on a fluorescence spectrophotometer (FLS920, Edinburgh) equipped with a 450 W continuous xenon lamp and a microsecond pulsed flashlamp.
So far the typical thickness of the CGM crystal has been reported to be 1 mm at most.13,17 Much effort about crystal growth method and technique is still required to obtain crystals with large size. It should be noticed that the cleavage yields an interesting cleavage slice with smooth and parallel surfaces, as shown in Fig. 1c. The unprocessed cleavage slice can be used directly as a gain medium for microchip lasers.18,19 A microchip laser consists of a small piece of laser medium polished flat and parallel on two sides, and the cavity mirrors deposited directly onto the medium faces.20 The conventional fabrication processing includes slicing (to submillimeter thick wafers), polishing, coating and at last cutting (into 1 mm-square pieces). Each piece serves as a complete laser cavity. The cleavage technique is simpler than the conventional machining processing, since the slicing and polishing are dispensable for cleavage microchips. Effective laser operations have been realized in the unprocessed cleavage microchips of Nd3+:LaB3O6 (ref. 18) and Tm3+:BaGd2(MoO4)4.19
The absorption spectrum enables us to estimate the radiative lifetime τr of the 2F5/2 upper laser level by the following formula:14
(1) |
The fluorescence decay curves were recorded under pulsed excitation at 976 nm and emission at 1039 nm, as shown in Fig. 4. It is universally known that the reabsorption and total internal reflection will elongate remarkably the measured fluorescence lifetime τf, especially for Yb3+ characterized with a large degree of overlap between absorption and emission. Not surprisingly, the τf (581 μs) for the bulk sample is significantly longer than the τr (235 μs). In the past few years, the powder method has been proved to be a simple and efficient approach to more accurately measure the intrinsic lifetime of the 2F5/2 manifold.7,8,22,23 The bulk sample is ground to fine powder and dispersed in the monochlorobenzene fluid (n = 1.52). Detailed experimental procedure has been provided elsewhere.8 In principle, a fluid with a refractive index close to that of the crystal (n ≈ 2) is desirable in order to eliminate the total internal reflection. But the fluids with high n are toxic. So the monochlorobenzene (n = 1.52) has been chosen for refractive-index matching in this work. As expected, the powder method yields much shorter τf (274 μs) than that for the bulk sample (581 μs). The quantum yield, defined as η = τf/τr, is slightly greater than 1. One possible explanation is that the powder method cannot eliminate the reabsorption and total internal reflection completely, resulting in the overestimated τf.22,23 Anyway, the result indicates high η for the Yb3+:CGM crystal, which implies that nonradiative relaxation is rather inefficient.
The low-temperature excitation and emission spectra are recorded at 77 K to reduce the spectral broadening and further determine the energies of Stark levels, as illustrated in Fig. 5. The common excitation and emission lines at 976 nm (10246 cm−1) are readily assigned to be resonant zero-phonon line transition between the lowest Stark levels of the 2F7/2 and 2F5/2 manifolds (1 ↔ 5). The other two well-resolved excitation lines at 962 nm (10395 cm−1) and 929 nm (10764 cm−1) are attributed to the electronic transitions 1 → 6 and 1 → 7, respectively. Different from the excitation spectrum, the emission spectrum displays more transition lines than expected. Obviously, some spectral lines in the 980–1030 nm spectral region, should belong to the vibronic sidebands originating in the strong electron-phonon interaction of Yb3+ with the lattice vibrations.23,24 Given that the phonon sidebands should appear at longer wavelength than the corresponding electron transition in the emission spectrum,23 these lines can only be associated with the 5 → 1, 5 → 2 or 5 → 3 electronic transitions. Keeping in mind the hypothesis that the Raman spectrum should reflect vibronic structures accompanying each pure electronic transition, the Raman spectrum is introduced to aid the interpretation of the electronic levels.23 Fig. 6 compares the excitation and emission spectra with the Raman spectrum, with an aim to distinguish the vibronic transitions from pure electronic transitions. These spectra are adjusted to the same energy scale. The energy origins of the excitation and emission spectra are taken at 5 ↔ 1 transition, while three energy onsets of Raman spectrum are selected to identify the possible vibronic sidebands from the 5 → 1, 5 → 2 and 5 → 3 transitions, respectively. The common spectral lines in the emission and Raman spectra are marked by asterisks in Fig. 6 and interpreted as the vibronic sidebands, while the remaining lines which are present in the emission spectrum but absent in the Raman spectrum, belong to the electronic transitions.
Eventually, the sequence of Yb3+ energy levels is established as follows: 2F7/2 = 0, 255, 306, 612 cm−1, and 2F5/2 = 10246, 10395 and 10764 cm−1, as sketched in the inset of Fig. 6. To our knowledge, the overall splitting for the 2F7/2 manifold (ΔE = 612 cm−1) is the largest among the DMo and DW ever reported except Yb3+:LiLu(WO4)2 (ΔE = 749 cm−1).25 The high ΔE stems from a strong crystal field characterized by a low point symmetry (C2) and a large distribution of Gd(Yb)–O distances, which is beneficial to limit thermal population of the terminal laser level and reduce the threshold pumping power.
Our assignment is verified by the ‘barycenter plot’ method proposed by E. Antic-Fidancev.26 In principle, the energy separation between the 2F7/2 and 2F5/2 manifolds should be constant whatever the matrix. In other words, the 2F5/2 barycenter as a function of the 2F7/2 barycenter should obey a linear relationship with a slope of unity.22–24 Fig. 7 presents the evolution of the 2F5/2 barycenter with the 2F7/2 barycenter for various Yb3+-doped crystals. The representative point for the title crystal is well-aligned with the theoretical line, confirming our interpretation.
Knowledge of the Stark levels makes it possible to estimate the emission cross-section σem from the absorption spectrum in accordance with the reciprocity method (RM):27
σRMem(λ) = σabs(λ)(Zl/Zu)exp[(Ezl − hc/λ)/kBT] | (2) |
Zk = ∑dkexp(−Ek/kBT) | (3) |
The RM is valid only in the vicinity of the fundamental transition where there is significant absorption.14 The absorption is so low at long-wavelength wing of the absorption spectrum that even a noise will be magnified exponentially in terms of eqn (2). The RM is no longer reliable at longer wavelengths, and this is the reason why the σRMem increases abnormally at λ > 1025 nm in Fig. 3. The σRMem in the long-wavelength range can be corrected by the Füchtbauer–Ladenburg (FL) formula:28
(4) |
The FL relationship also has its limitations: it is valid only in the spectral region where there is no reabsorption.13 It should be noted that the reabsorption decreases the fluorescence intensity, resulting in an underestimate of the integral in eqn (4). For this reason, the eqn (4) is rewritten as:29
σFLem(λ) = Cλ5I(λ) | (5) |
(6) |
The combination of the RM and FL methods can circumvent their shortcomings and describe precisely the wavelength dependence of the σem in the whole spectral region, as shown in Fig. 3. The σRMem is more reliable at λ < 1007 nm, while σFLem is more accurate at λ > 1007 nm. The σem are equal to 3.6 × 10−20 at 1007 nm and 1.0 × 10−20 cm2 at 1038 nm, respectively. The corrected σem is subsequently used to calculate the gain cross-section σg(λ), which is defined as:
σg(λ) = βσem(λ) − (1 − β)σabs(λ) | (7) |
(8) |
(9) |
Imin = βmin(λext)Isat(λpump) | (10) |
If λpump = 976 nm and τf = 274 μs are selected, the Isat are calculated to be 19.0 kW cm−2. Then the Imin are determined to be 2.9 kW cm−2 for λext = 1007 nm and 0.8 kW cm−2 for λext = 1038 nm, respectively.
The important structural and spectroscopic parameters of the Yb3+:CGM crystal are summarized in Table 1 and compared with those of other Yb3+-doped DMo and DW crystals. From the spectroscopic point of view, the Yb3+:CGM crystal is situated at the intermediate position between the disordered tetragonal and ordered monoclinic DMo and DW crystals. Until now, sub-100 fs laser pulses have already been demonstrated in Yb3+:KGd(WO4)2 (78 fs),34 Yb3+:NaY(WO4)2 (53 fs)35 and Yb3+:NaY(MoO4)2 (91 fs).36 The production of ultrafast pulses is known to depend critically on the optical bandwidths of gain media.31,37 From this prospective, the disordered tetragonal DMo and DW are more advantageous to deliver shorter pulse duration and higher peak power as compared with the ordered monoclinic ones. But the former suffers from low optical cross-sections and hence a low gain. A tradeoff between optical bandwidths and emission cross-sections is achieved in the case of the Yb3+:CGM crystal. In addition, the crystal is suitable for LD pumping at 976 nm due to broad absorption bandwidth (FWHM = 8.1 nm), and operates in a quasi-three-level scheme with large ground-state splitting (ΔE = 612 cm−1). And lastly, an unprocessed cleavage slice can be used directly for microchip lasers, simplifying fabrication processes. The unique combination of above-listed advantages proposes the potential application of the Yb3+:CGM crystal in the LD-pumped microchip laser systems capable to produce the tunable and sub-100 fs pulsed laser.
Crystals | NGM | NGW | KGW | CGM |
---|---|---|---|---|
a Notes: Czochralski (Cz.), NaGd(MoO4)2 (NGM), NaGd(WO4)2 (NGW), KGd(WO4)2 (KGW). | ||||
Growth method | Cz. | Cz. | TSSG | TSSG |
Space group | I41/a | I | C2/c | P2/c |
Multisite or disorder | √ | √ | × | × |
Symmetry (Gd) | S4 | S4 | C2 | C2 |
ΔE (2F7/2) (cm−1) | <500 | 492 | 535 | 612 |
λabs (nm) | 975 | 975 | 981 | 976 |
Δλabs (nm) | 50 | 50 (π) | 3.7 | 8.1 |
43 (σ) | ||||
σabs (10−20 cm2) | 2.2 (π) | 1.78 (π) | 12.0 | 3.9 |
1.36 (σ) | ||||
λem (nm) | 1000 | 1000 | 1023 | 1007 |
1038 | ||||
Δλ*em (nm) | 43 | 40 (π) | 25 | 39 |
σem (10−20 cm2) | 2.7 (π) | 1.89 (π) | 2.8 | 3.6 |
1.4 (σ) | 0.75 (σ) | 1.0 | ||
τf (μs) | 280 | 320 | 317 | 274 |
Δtmin (fs) | 24 | 26 | 44 | 27 |
Isat (kW cm−2) | 33.1 (π) | 35.7 (π) | 5.3 | 19.0 |
Reference | 4, 7 | 32, 33 | 9, 31 | This work |
This journal is © The Royal Society of Chemistry 2015 |