1. Introduction

Ceramic lasers, emerging from intensive research and development activities globally [1], hold a great promise in high-power laser related applications with much lower cost and higher flexibility [2,3] and have been envisioned as new players in the laser industry to reshape the market in near future [4]. Along this line, the research enthusiasm in developing self-Q switched lasers [5,6] with electro-optic (EO) single crystals [7,8,9] was rekindled lately in the exploration of ceramic gain media with excellent EO effects, especially in rare earth (RE) doped transparent lanthanum-modified lead zirconate titanate (PLZT) ceramics [10,11]. Their wide optical transmission window from visible to mid-wave IR regime, absence of water absorption bands [10], high RE solubility, and structural traits in between of single-crystal and glass materials make them ideal active media in constructing electrically tunable lasers and amplifiers [12,13] with new features [14]. Powered by the matured microfabrication technology in PLZT and similar series, this family is promising in issuing multifunctional, monolithic, photonic systems [15,16]. In this work, strong optical amplification and initial lasing action in Nd³⁺ doped PLZT (Nd³⁺:PLZT) plates are reported by pumping the plates with a pulsed diode laser at 805 nm. These groundbreaking works have demonstrated the EO ceramic gain medium family can indeed serve as optically active materials and would usher in a broad spectrum of applications, including development of self Q-switched ceramic lasers, zero loss optical devices, and other multifunctional photonic components and systems. With a model of continuous trap distribution and possible stimulated processes involved, an intriguing slowly trailing-off feature observed both in the optical amplification and the lasing action is analyzed satisfactorily and direct potential uses of it are given.

2. Experimental and discussion

The specimens used in the works reported in this paper consisted of 65mol% lead zirconate plus 35 mol% lead titanate and 10mol% lanthanum, i.e. PLZT (10/65/35), to which 1.0 to 2.5 mol% Nd³⁺ cations had been added. The origins of the components were PbO, La₂O₃, ZrO₂, TiO₂, and Nd₂O₃, respectively. All oxides were purchased from Fisher Scientific, with purity 99.85% or higher. A typical sintering process of the ceramic is detailed in [11]. Figure 1 shows typical spectra of room temperature ground state absorbance and of photoluminescence obtained in a 1.0 mol% Nd³⁺:PLZT plate (2.0 mm in thickness). The absorbance spectrum was measured by a spectrophotometer (Perkin-Elmer, Lamda 9). A number of spectral peaks were observed (refer to Fig. 1(a)). The most interested absorption peak in this work was the one at 801.5 nm, which is close to the center wavelength at 805 nm of the pumping source applied in this work. The absorption coefficient of the 1.0 mol% Nd³⁺:PLZT plate was 2.96 cm⁻¹ at 801.5 nm. The full width at half maximum (FWHM) of the peak in Nd³⁺:PLZT was 16.0 nm, 3.2 times broader than that obtained in Nd³⁺ doped crystalline yttrium vanadate (Nd³⁺:YVO₄) slabs under the same experimental condition, and about 15.4 times broader than that reported in both Nd³⁺:YAG single crystals and ceramics [17]. For direct comparison, the corresponding peak in a 1.0 wt.% Nd³⁺:YVO₄ slab, purchased from Casix Inc., was normalized and then exhibited in Fig. 1a with the dashed line. It is seen that around 6.5 nm peak shift was seen between the Nd³⁺:PLZT and Nd³⁺:YVO₄ samples. In respect with light emission, the photoluminescence (PL) (shown in Fig. 1(b)) was measured with a monochromator (McPherson, model 78A-3) and a cw diode laser as the excitation source (LDI 820). A PbS detector was used at the exit of the monochromator to convert the photoluminescence signal to an electrical one. Three-emission peaks at 913.4, 1064.4 and 1343.4 nm were observed that correspond to the ⁴F_3/2→⁴I_9/2, ⁴F_3/2→⁴I_11/2 and ⁴F_3/2→⁴I_13/2 transitions, respectively. Among these emission peaks, the most interested is the one centered at 1064.4 nm, with FWHM of 35.6 nm, which is about 22.7 times broader than that obtained in the single crystalline Nd³⁺:YVO₄ slab. As given against the PL peak in the Nd³⁺:PLZT plate, the peak in Nd³⁺:YVO₄ is centered at 1064.8 nm, with 1.5 nm FWHM. To see the peak width ratio clearly, the PL peaks of transition ⁴F_3/2→⁴I_11/2 was replotted in the inset of Fig. 1(b). It is seen that only 0.4 nm PL peak shift is observed between the Nd³⁺:PLZT and Nd³⁺:YVO₄ specimens. Since no appreciable absorbance and PL peak shift was seen among all Nd³⁺:PLZT plates, only the data obtained in the 1.0 mol % Nd³⁺:PLZT plates, employed both in the optical amplification and lasing action experiments, were presented in this paper. The remarkable spectral broadening in the Nd³⁺:PLZT ceramics, owing to the polycrystalline nature and slight composition variation from grain to grain in this naturally disordered PLZT family, resembles them with glass gain media. These traits have marked out this new category of gain media ideal for making tunable lasers and optical amplifiers, empowered by their accompanying decent EO property, which will be described in the following.

 

Fig. 1 Absorption and PL properties of Nd³⁺:PLZT samples: a) A typical absorption spectrum in a 1.0 mol% Nd³⁺:PLZT plate (dot-dashed line depicts the absorption peak of a 1.0 mol% Nd³⁺:YVO₄ sample at 808 nm for comparison); b) A typical PL spectrum obtained in a 1.0mol% Nd³⁺:PLZT plate (dot-dashed line shows the PL peak at 1064.8 nm in a 1.0 mol% Nd³⁺:YVO₄ plate for comparison).

Download Full Size | PDF

The electric field induced phase retardation was utilized to measure the EO coefficients in the Nd³⁺:PLZT plates with 0 to 2.5 mol % doping percentage throughout their entire optical window. The corresponding quadratic EO coefficients were calculated accordingly (see [10] and [11]]) and shown in Fig. 2 . In measuring the EO coefficients, one of the PLZT slabs was placed between two cross polarizers with relatively high transmission for each waveband. The polarizations of the polarizers are symmetrical to the applied field, forming +45° and −45° angle with the direction of the electric field, respectively. The transmitted light intensities increased with the applied field, because the slabs served as variable wave plates. The quadratic EO coefficients in Nd³⁺:PLZT plates measured at wavelength 532 nm with different doping percentage are exhibited in Fig. 2(a), and a typical EO induced retardation curves are shown in the inset of Fig. 2(a) for 1.0 mol% Nd³⁺:PLZT. It is seen that the introduction of Nd³⁺ cation into PLZT has resulted in noticeable reduction in EO coefficients, though the measured EO coefficient in the highest doping sample (2.5 mol%) remains still decent values, which are sufficient for many applications. These decent EO coefficients are no doubt highly desirable in designing and developing self Q-switched ceramic lasers, zero loss optical devices, and other multifunctional optical components. To see the full potential of this novel gain medium in the entire transparent window, the dependence of the EO coefficient versus wavelength in a 1.0 mol% Nd³⁺:PLZT plate was measured and depicted in Fig. 2(b). The thickness of the specimen was 1.44 mm in thickness. It is encouraging to see that as high as 0.2×10⁻¹⁶ m²/V² EO coefficient was obtained at 5.5 µm, only about 30% reduction compared with that in the visible regime. In addition, the inset in Fig. 2(b) depicts the entire transparent window of pure PLZT from 400 nm to 7.0 µm. Its excellent transmission in such a broad window, especially in short- and mid-wave IR regimes makes it an excellent candidate in many EO related applications nowadays. Finally, to get a key number in optical amplification and lasing action, by measuring transmission of the specimen, we estimated the upper limit of the propagation loss coefficient, which was 0.06 cm⁻¹ at 1064.4 nm. This number included contributions of both scattering and absorption. Actually, this relatively low propagation loss coefficient is one of the key factors in leading to the optical amplification and the lasing action that will be reported as follows.

 

Fig. 2 EO properties and Curie temperature of Nd³⁺:PLZT samples with various doping concentrations throughout their entire transparent window. a) EO coefficient versus Nd³⁺ cation concentration; b) EO coefficient versus wavelength for 1.0mol% Nd³⁺ doped sample.

Download Full Size | PDF

Single-pass gain was measured in a thin 1.0 mol%Nd³⁺:PLZT plate and exhibited in Fig. 3(a) . One can see that as high as 540% single-pass gain was observed at 1064.4 nm in a 2.0 mm plate (the full dimension of the specimen used was 4.0 mm × 4.0 mm × 2.0 mm). The specimen was prepared by lapping, polishing, and then AR coating treatments (reflectivity R<0.2% at 1064 nm, and R<0.5% at 808 nm).

 

Fig. 3 (a) Single-pass gain versus seed power when the pumping power was fixed at 5.0 W and the crossing angle at 155°. The inset is a dynamic curve recorded with an HP Infinium oscilloscope for the end-pumping geometry; (b) Schematic diagram of energy levels that illustrates the continuous trap distribution [19,21]: ST-shallow trap level; TAT-thermally active level at energy E_a; TDT-thermally disconnected traps; RC-recombination center.

Download Full Size | PDF

A solid-state laser (Coherent, DPSS 1064) was used as the seed source, in which the center wavelength was 1064.4 nm and the FWHM was 0.4 nm. The seed source was collimated by a two-lens set and attenuated by neutral density filters, and then detected with a Newport photodetector (818-SL, response time <2 µs) connected to a lock-in amplifier for a weak seed light or to a Hewlett Packard Infinium digital oscilloscope. A fiber-pigtailed high-power laser diode (Apollo Instruments, S-30-806-6) of 805-nm center wavelength was employed to pump the specimen at its pulse mode. The crossing angle between the pumping beam and the seed beam was 155°. The pumping beam was 3.0 mm in diameter and the seed beam 1.0 mm to ensure a good coupling from the pump to seed beam. The pumping power was 5.0 W. The experimental setup was described elsewhere in details [16]. One can see from Fig. 3(a) that net single-pass gain was achieved in the entire range of 0.42 µW to 107 µW, although it decreased with the increase of the seed power. It is worth noting that this is nearly an exponential dependence and will be discussed in more details later.

More strikingly, the response time (in order of seconds) in the gain dynamics was deemed as very long, compared with the typical photoluminescence lifetime (in order of hundreds of μs) obtained in the Nd³⁺:PLZT ceramic plate. For direct comparison, the transmitted seed power and pumping power were read directly with two Newport photodetectors (818-SL, response time <2 µs) connected to a four channel Hewlett Packard Infinium digital oscilloscope. A typical dynamics thus recorded is shown in the inset of Fig. 3 (a). At beginning, with the seed beam on and the pump beam off, the reading for the seed beam was 0.25 scales (2.0 µW). When the pulse pumping source was turned on, a huge jump in the transmitted seed power demonstrated that the seed power was amplified nearly 5 folds. Interestingly, when the pumping beam was turned off, a huge spike occurred, and then followed by a slowly trailing–off process. To one’s surprise, after the pumping beam was off completely, the seed beam had still appreciable gain until 1.5 s. It is worthwhile to point out that this very trailing-off time interval is 4000 times longer than the lifetime corresponding to the transition from ⁴F_3/2 manifold to ⁴I_11/2 manifold, which is 250 µs. When the seed power increased, the tailing-off time decreased. Then a question arises: what is the source for this extra energy after the pumping beam was off for seconds? Indeed, there must be a certain optoenergy stored in the gain medium and released slowly by certain relevant physical stimuli after the pumping source was completely off.

In order to get an in-depth understanding of the aforementioned special feature of the slow response and relevant optoenergy storage in Nd³⁺:PLZT ceramics, it is better to take a step backwards from the ceramic and proceed downward to the individual building blocks or unit cells that make up the crystallites. In PLZT family, a typical unit cell is known as the ABO₃ perovskite structure (see more details in [10]. It consists of A sites occupied with Pb²⁺ and La³⁺ ions and linked network of oxygen octahedral B sites occupied with Zr⁴⁺ and Ti⁴⁺ ions within the octahedral cage. Lanthanum cations replaces some lead cations in the A site of the perovskite ABO₃ ionic structure. Since La³⁺ (added as La₂O₃) substitutes for Pb²⁺, charge balance is maintained by creation of lattice site vacancies. Similar vacancies can also be formed at B sites. Additionally, as widely accepted, oxygen vacancies can also be formed in a complex oxide like PLZT ceramic [10]. Furthermore, because of the intrinsic disordered nature of the PLZT ceramics, carrier traps can also form on boundaries of grains [18], due to slight variation of the composition from grain to grain. Typical grain sizes range from 1.0 to 10 µm. All these carrier traps make PLZT host material relatively rich in charge carrier traps. As illustrated by the model proposed by Lewandowski and McKeever in dealing with thermally stimulated processes [19,20] (see our modification drawing in Fig. 3(b)), the shallow traps (ST) are almost empty at room temperature. The thermally disconnected traps (TDTs) are occupied by electrons undisturbedly without optical stimulus from photons. In the case when thermal stimulus dominates, only the thermally active traps (TAT), with activation energy between E_a and E_h are the most active, the electrons trapped in TDTs are hardly released thermally [19,20]. However, when the large photon flux of the pumping beam bombards the specimen, the electrons at the TDTs can be excited to the conduction band and partly transported to the thermally active traps. As the result, the trap distribution, and hence occupancy of the traps, are reshaped temporarily. In other words, the occupancy of TATs is much fuller than usual (see Fig. 3 (b)). Right after this, even though the pumping beam is shut off, the thermally stimulated process can still liberate swarms of electrons from the TATs. This might be one of the reasons behind the slowly trailing-off in the optical amplification. The detrapping rate is fast enough to provide a great number of electrons to the upper manifold energy level to feed up weak seed amplification. However, since the thermally released electrons are limited in numbers, for a strong seed beam, the amplification is capped by the relatively small numbers of upper level electrons. In short, this picture also provides an explanation to the near exponential dependence of the single-pass gain on the seed power in Fig. 3 (a). In general, a typical photoluminescence life time was measured at relatively low intensity and very short time. So the samples were not heated up to a high temperature to release many carrier traps. When we heated up the samples with relatively strong cw diode laser beam, the spontaneous luminescence decay curve had two portions. One process was extremely fast, in the order of hundreds of µs, followed by a slow decay in order of hundreds milliseconds to seconds. However, the magnitude of the latter was much lower than that of the former.

After achieving single-pass gains in the four-level-system, Nd³⁺:PLZT ceramic, it is natural to step forward towards lasing action in the same system. For this purpose, the specimens were also prepared by lapping, polishing, and then AR coating treatments (reflectivity R<0.2% at 1064 nm, and R<0.5% at 805 nm). The gain medium used for the lasing action was also a 1.0 mol % Nd³⁺:PLZT plate, 4.0 mm × 4.0 mm × 4.0 mm in full dimension. The input coupler was a flat dielectric mirror, with 99.98% reflectivity at 1064.4 nm. The output coupler was a concave dielectric mirror, with 98.2% reflectivity at 1064.4 nm, and 25 cm in radius of curvature. The pumping source was the same one used in the optical amplification experiments described above. The pulse width was 100 ms, with a time interval of 5.0 s. A relatively weak lasing action was seen in this gain material family. The laser output was 10.0 mW for a 3 W pumping power and the wavelength of the laser output was centered at 1064.4 nm, and FWHM was 0.4 nm. The laser power depended on the input pump power, but it was not a linear one for some unknown reasons. The laser output was unstable and the lasing action could stop occasionally. Regarding this unstable output, there are several possible reasons behind it. The thermal conductivity of Nd:PLZT was (19 ± 3) x 10⁻³ W·cm⁻¹K⁻¹. This relatively low thermal conductivity might cause this unstable laser output. In addition, since the relative high scattering in this disordered gain medium, there was possibility of random lasing involvement in the process, especially when the gain medium used in lasing action has pyroelectric effect and thermo-electro-induced scattering in turn might contribute to the unstable lasing output. It should be pointed out that this lasing action, though is relatively weak in terms of the output power currently, is a very good start. Intensive research activities to improve the power and stability of the laser output are still undergoing. Exhibited in Fig. 4(a) is the photograph taken from the laser setup with Nd³⁺:PLZT ceramic gain medium. Shown in Fig. 4(b) are the laser pulses taken with the HP Infinium oscilloscope. It should be noted that observed was a similar intriguing slowly trailing-off feature in the laser dynamics. After the pumping pulse was completely stopped, the lasing action was still on for a while, though the power level was much lower than that when the pumping was on. Once again, these observations suggested that a certain long persistent optoenergy storage mechanism does exist in the Nd³⁺:PLZT gain medium, which is equivalent to a long lifetime of the upper manifold. This property is highly desirable for designing high-power lasers. If the energy thus released could be accumulated and used, the peak power of ceramics laser can be increased tremendously. Since the pulse mode lasing action had more repeatability, the pulse mode laser output was given in this paper. In addition, lasing action in cw mode had been performed and the output was pretty unstable. Detailed work on lasing action will be reported in the future publications.

 

Fig. 4 (a) A photograph of the setup of lasing action in Nd³⁺:PLZT ceramics: LD: a fiber-pigtailed laser diode (at 805 nm); L1 and L2: lenses; IC and OC:the input and output couplers, respectively; SL: a seed laser for alignment; (b) A typical dynamic curve of the laser output.

Download Full Size | PDF

3. Conclusion

In conclusion, strong optical amplification and initial lasing action were both seen with Nd³⁺ doped PLZT EO ceramic media, which have good electrooptical coefficients in their wide optical transmission window from visible to mid-wave IR regime, with a pulsed diode laser pumping source. This work has opened a door leading to implementation of self Q-switched lasers without introducing additional foreign components into the laser cavities and zero loss optical devices. Our investigations pertinent to the single-pass gains and the lasing action in Nd³⁺:PLZT plates has identified the optoenergy source for slowly trailing-off phenomenon in the unique material family. The rich carrier traps in the ceramics might be responsible for slowly trailing-off process and hence the resultant large gains.