Silicate- and borosilicate-based PbS:glass material and borosilicate-glass-based fibers are fabricated and analyzed. Optical properties including absorption and emission are characterized and related to growth and annealing conditions. In silicate glasses PbS volume fractions of exceeding 0.4 percent and almost octave-spanning emission spectra with a halfwidth of 940 nm are achieved. Fiber bundles with a core being surrounded by three PbS:Glass fibers are pulled. A confinement factor of Γ = 0.00406 is determined. Emission properties, in particular emission bandwidth, are subsequently tuned and spectrally widened by annealing fibers in a gradient furnace. The results pave the way towards optically pumped broad-bandwidth light emitters based either on ‘bulk’ PbS:glass or PbS:glass-based fibers.
©2013 Optical Society of America
Glass being doped with lead sulfide nanocrystals (PbS:Glass) represents an attractive option as photonic material for a number of applications. This involves the use as gain material , as saturable absorber , and as source for infrared lighting. For the latter application, a broad emission band can be tailored by employing the small band gap energy of bulk PbS of Eg = 0.41 eV (wavelength λ~3 µm) on the one hand, and the wide tuneability of the fundamental electronic resonance by quantum-size effects (even below λ = 1 µm) on the other hand. This allows either to tailor a special wavelength for a given application, or to arrange broad-bandwidth light sources with emission bandwidths exceeding one octave. A part of these applications involve the areas of fiber and integrated optics. Therefore it is a vital option to have the infrared emission available in a fiber  or within a waveguide . Consequently not only coupling into, but also light generation within fibers or waveguides are crucial approaches. This latter topic is alternatively addressed by strategies that circumvent the growth of bulk PbS:Glass by ex post processes. Along these lines, e.g., microstructured optical fibers have been filled with PbS nanocrystal colloidal solutions . Such approaches, where the nanocrystals are exposed to the ambient, however, often do not achieve the long-term stability of PbS:Glass.
We report results on fabrication and spectroscopic characterization of genuine PbS:Glasses, which are designed as optically pumped broad-bandwidth near-infrared light source, either as bulk material, e.g., as cap atop of LEDs, or as a part of a fiber bundle. This report is organized as follows. After addressing the samples and the details about the formation of PbS nanocrystals, we point to the luminescence properties of pristine and annealed bulk material. Then we come to the properties of fibers pulled from this material. Finally, we present the properties of fibers with size-graded PbS nanocrystal distributions along the fiber. We find both bulk PbS:glass as well as PbS:glass-based fibers to be suited as optically pumped broad-bandwidth light emitters.
2. Experimental details
The PbS:Glass samples are grown according to the methods reported by Joshi  (borosilicate glass, sample A) and Borrelli  (silicate glass, sample B1 and B2). The mixture of the reagent-grade materials was molten for 90 min in an alumina crucible. Temperatures and compositions are listed in Table 1. Subsequently, the material was casted and re-molten for additional 30 min in order to improve homogeneity. About 3.5 mm thick glass plates were obtained from each batch. Samples were prepared from this ‘pristine’ material as well as from those parts of the same batch, which subsequently became annealed. For transmission measurements samples have been optically polished on both sides.
The motivation for involving here two kinds of PbS:glass is the following: The temperature, where PbS nanocrystals are formed (crystallization temperature: Tc) depends crucially on thereabouts chemical ambient. In borosilicate glasses Tc is higher than in silicate glasses. The difference in Tc depends of course on the particular glasses, however, for the material compositions relevant for this study pretests delivered a value of ~90 K. Taking into account the by only 20-30 K different melting temperatures (Tm) for the two kinds of glasses,one expects pristine borosilicate glasses to contain substantially less PbS and smaller nanocrystals than silicate glasses do. This offers for borosilicate glasses a wider range for additional successive thermal treatments, which all eventually cause the nanocrystals to grow in size. Thermal treatments with relevance to applications are
- • annealing in order to tailor the fundamental resonance towards the infrared spectral range, and
- • fiber pulling.
Silicate glasses with their lower Tc have larger lead (PbO) and sulfur (ZnS) concentrations; see Table 1. More lead and sulfur promise eventually larger PbS volume fractions (f). Their stability against thermal processing, however, is expected to be poorer. Thus both kinds of PbS:glass address different application prospects.
PbS:Glass fibers are pulled from borosilicate glass. Figure 1(a) shows the interior part of a preform. It contains three nominally identical PbS:glass primary rods arranged like a cloverleaf around an undoped core. For applications the core is expected to serve as waveguide for pump radiation, while evanescent light leaking from there into the three PbS:glass fibers will accomplish the excitation. Figures 1(b) and 1(c) show the pulled structure including the claddings that maintain mechanical stability as well as the inner part with the three PbS:glass fibers attached to the core, respectively. Numbers refer to the materials involved into the fiber bundle architecture; see Table 2 for details. A typical thermal load put on the material during a standard pulling process corresponds roughly to 19 min annealing at 670°C.
The transmittance T is measured using a Perkin Elmer Lambda 900 spectrophotometer (200-1800 nm) and a Bruker IFS66v Fourier-transform-spectrometer (FT, 700-5000 nm). Typically from each sample a set of 2-3 homogeneous plates with thicknesses 25 µm<d<1.5 mm are analyzed. Spectra of the absorption coefficient α(λ) are compiled by using contributions from the spectra from those plates, where α × d = 1 is best matched. Errors and noise of the α(λ)-spectra are minimized in this way.
The FT-spectrometer has been also employed for photoluminescence (PL) spectroscopy in step-scan mode by either employing liquid nitrogen cooled Ge- or HgCdTe-photodiodes. Typically, we present those spectra being taken with the Ge-photodiode because of better signal-to-noise-ratios. If emission beyond 1700 nm has is observed, we give the spectra taken with the MCT detector. Excitation is provided by HeCd-, Ti:sapphire-, and Nd:YAG-lasers, at 442, 800, and 1064 nm, respectively. PL has been performed in two different geometries: First, there is the standard ‘backscattering’ or ‘reflection’ type one. Second, we also measured ‘transmission’ type PL, where a sample is excited from one side, while the PL is detected from the other side after the luminescence passed through the sample. We will refer to this special type of measurement as ‘transmission-PL’.
Since we not only use spectral shapes and spectral positions of PL bands, but also spectrally integrated PL intensities, we checked for the reproducibility of this parameter for standard PL and found an error not exceeding 20 percent. Note that intensities of standard and ‘transmission-PL’ among themselves are not comparable.
3. Results and discussion
3.1 Bulk borosilicate and silicate glass
We start with borosilicate glass; samples A. Pristine PbS:Glass samples from this batch appear almost transparent to the naked eye. Transmission measurements yield α(400 nm) = 3-4 cm−1. According to Moreels et al. [8,9], who established a linear relation between α(400 nm)-value and the PbS volume fraction (f), this low absorption coefficient points to f = 1-2 × 10−5. Figure 2 compares PL data obtained from pristine and annealed A-samples. The annealing conditions have been chosen in order to model the thermal process that is involved in the fiber pulling technology. An estimate of the f-values of the annealed samples can be made on the base of the PL intensities and provides a factor of ~102. Multiplying this value with the above determined f-value, we end up with values of f = 0.1-0.2 percent, which match well f-values for our standard borosilicate glasses after annealing. During the two annealing processes both nucleation and nanocrystal growth have taken place simultaneously, and the slightly from 670 to 690°C increased annealing temperature has obviously a strong impact on the nanocrystal size resulting in an increasing radius (R) from R = 2.9 to R = 4.3 nm; estimate made according to Nanda et al. .
We now focus on bulk silicate glasses; samples B. Figure 3 shows PL and α(λ)-spectra of sample B1. We start with the source material. The comparison of the α(400 nm)-values of pristine borosilicate (A) and silicate glass samples (B1) yields ~3-4 cm−1and ~180 cm−1, respectively. The very different α(400 nm)-values of the silicate glass indicate a by virtually two orders of magnitude higher f-value compared to pristine borosilicate glass. This finding shows that spontaneous nucleation in silicate glass has already taken place even in the source material. This is also consistent with PL data, which show a significantly higher PL magnitude from sample B1 than that from sample A. Comparing pristine borosilicate (A) and silicate (B1) samples, we find
- • the full width at half maximum (FWHM) doubled from 290 nm to 580 nm, and
- • the infrared cutoff wavelength shifts from 1400 to 2100 nm.
These findings indicate that nucleation goes another way under the condition of rich PbO and ZnS supply; see Table 1: While in borosilicate glasses indeed small nuclei are expected to dominate the ensemble, in case of silicate glasses larger PbS nanocrystals are additionally created at the same time.
In order to check whether the nucleation process is fully completed already in the pristine material, we applied long-term annealing at lower temperatures (475°C for 1320 min + 525°C for 75 min) to sample B1; see Fig. 3. Such type of thermal treatment is known to promote nucleation . At first glance the result shown in Fig. 3 is surprising and contradicts the intuition, namely a blue-shifted PL is observed as a result of a thermal treatment. This, however, can be explained by presuming that the nucleation process in the pristine material was not fully completed. The newly generated PL band at 1170 nm points to nucleation of small nanocrystals (R = 2.3 nm, f = 0.19 percent) from Pb and S, while the one at 1600 nm is almost unchanged. The creation of additional nanocrystals is independently confirmed by the nearly doubling α(400 nm)-values that determine the PbS volume fraction. Except for the answer about the completeness of the nucleation, this experiment provides a guideline for the production of PbS:Glass with a bi- or multimodal nanocrystal size distribution, what is essential for assembling a broad-bandwidth light source: If different nucleation processes become activated, one can expect from each of them a distinct nanocrystal size. In our experiment a bimodal nanocrystal size distribution is formed by spontaneous and induced nucleation acting in concert.
The application of a ‘regular’ annealing process (590°C for 20 min) to the pristine material leads to a comparable result as we have seen for annealed samples A (PL band at ~1600 nm; see Fig. 2 annealed for 12 min at 690°C); see Fig. 3. The by 100 K lowered annealing temperature was chosen to account for the lower Tc in silicate glass. The observed increase in f from the pristine to the annealed material from 0.08 to 0.41 percent [see α(400 nm)-values in Fig. 3)] is by one order of magnitude smaller than the corresponding increase found for the borosilicate glasses. As elucidated, this is an effect of spontaneous nucleation that has taken place in pristine silicate glasses already.
Sample B2 has been synthesized in order to check, to which degree f can be enhanced by increasing the PbS-content; see Table 1. Figure 4 shows absorption and PL spectra of the pristine sample. The α(400 nm)-value amounts to ~900 cm−1 pinpointing f = 0.41 percent without any annealing. This huge value is accompanied by an ultrabroad PL spectrum spanning from the excitation wavelength (1064 nm) to about 2500 nm. Having said that we must state that in the pristine sample B2 precipitates are observed, which were not dissolvable by any further thermal treatment. Therefore sample B2 marks probably a border case of the present PbS:Glass technology.
3.2 Borosilicate glass-based fibers
PL is also used for the optical characterization of fibers. Figures 5(a), 5(b) and 5(c) show results as obtained from polished cross-sectional areas of a fiber section with a length of l = 21 cm. The PL spectra taken in ‘regular’ reflectance geometry from the entire slice plane at excitation wavelengths of λexc = 442, 800, and 1064 nm show a single peak at about 1300 nm and remarkably matching spectral shapes. The shapes of the transmission-PL spectra, however, systematically depend on the excitation wavelength. Transmission-PL pickup is possible only if the excitation spot (∅<10 µm) has been focused to the undoped fiber core; cf. Figure 1(c). Thus it is rather straightforward to interpret the altered shape of the transmission PL compared to the regular ones as a result of re-absorption. Therefore we can relate transmission-PL spectrum (ITPL) and ‘regular’ PL spectrum (IPL) by :Fig. 5.
Figure 5(d) shows the results obtained for the absorbance α(λ)·d(λexc) as determined from the data given in Figs. 5(a-c) using Eq. (1). Within the limits set by noise, all three α(λ)·d(λexc)-spectra are represented by (almost) straight lines with slopes of (−6.8 ± 0.2) × 10−3, (−3.3 ± 0.2) × 10−3, and (−0.53 ± 0.01) × 10−3 nm−1, for λexc = 442, 800, and 1064 nm, respectively (determined by linear fits in the 1250-1550 nm range). In other words: Except for an additive constant, which is not known here (since we can’t compare absolute IPL- and ITPL-values), the 3 spectra can be described by the same spectral shape, namely α(λ), multiplied by a constant being represented by d(λexc). The assumption of such a separation ansatz for α(λ)·d(λexc) also reflects well the physics involved: The α(λ)-spectrum of the fiber should be doubtlessly fully independent on λexc, since the PL created by λexc serves here as a tool only to generate inside the fiber the ‘test radiation’ for the subsequent transmission-like measurement. The slope, however, is expected to be proportional to d and in this way dependent on λexc; see inset in Fig. 5.
In view of these results it is worthwhile seeking quantitative knowledge about absorption of the materials. Therefore the same pristine material as being used in the fibers is annealed under thermal conditions emulating the fiber pulling process (19 min at 670°C). Transmission measurements delivered α = 250, 20, and 12 cm−1 for 442, 800, and 1064 nm, respectively. Even in the PL emission range of 1100-1600 nm (cf. Fig. 5), α never falls below 7 cm−1. Given these values, the ‘penetration depth’ 1/α at any of these wavelengths is small against l = 21 cm. Therefore, we can conclude that the pure high-absorptive PbS:Glass is definitely not the material, in which excitation and re-absorption take place. In concert with the fiber length of 21 cm, only a by orders of magnitude lower effective α(λ)-values of the bundle could explain our findings. This requisition exactly matches what one expects from a non-absorbing fiber core, being surrounded by absorptive material, into which an evanescent component of the core-guided light is leaking. Such a system can be described by a confinement factorFig. 5(d) to be proportional to d and using
For implementing fiber-based broad-bandwidth light sources it is potentially desirable to achieve a further spectral broadening of the emission band. Here fibers offer an additional unique degree of freedom, namely the option of annealing them in a temperature gradient along l. More specifically, annealing of a PbS:Glass-fiber within a gradient furnace is expected to allow implementing a longitudinally size-graded nanocrystal distribution. For the corresponding feasibility test, we used a piece of the fiber, whose properties have been probed before. The furnace profile was adjusted in a way that a 9 cm long fiber experiences at its ends temperatures of 400 and 610°C (with an almost uniform temperature gradient of ~23 K/cm in between) for 630 min.
Figure 6 shows PL data demonstrating the success of the inhomogeneous annealing approach. The spectra have been taken in standard PL geometry from both fiber ends (442 nm excitation). Assuming the same confinement factor as determined for the homogeneous fiber, the PL information depth 1/α(442 nm) in this measurement is about 1 cm. Thus an independent evaluation of the two fiber ends is possible. While the fiber end at 400°C maintains the original PL peak position at 1300 nm, the other end having been at 610°C shows an infrared-shifted PL pointing to increased nanocrystal sizes. At the same time, the FWHM increases from 190 to 320 nm. This increased FWHM-value of 320 nm, more precisely even 321 nm, is monitored also for transmission-PL measurements at this graded fiber bundle with an excitation wavelength of 1064 nm. This combination emulates already a LED-pumped broad-bandwidth light source, representing an important motivation of this study.
Borosilicate- and silicate-based PbS:glass bulk material as well as borosilicate-glass-based fibers are fabricated and analyzed. Optical properties including the emission of pristine and annealed samples are characterized. In silicate glasses PbS volume fractions of exceeding 0.4 percent and almost octave-spanning emission spectra (FWHM = 940 nm) are achieved.
Despite the concerns emphasized by Auxier et al. , we successfully demonstrate that fiber pulling of semiconductor doped glass can be accomplished using borosilicate glasses, and even tailoring of optical properties is still feasible. Setting up absorption is made employing of a fiber bundle with a core being surrounded by three PbS:Glass fibers. A confinement factor of Γ = 0.00406 is determined. Emission properties, in particular emission bandwidth, are subsequently tuned and spectrally widened by annealing fibers in a gradient furnace. As a result, optically pumped broad-bandwidth light emitters can be realized based on both ‘bulk’ PbS:Glass as well as PbS:Glass-based fibers.
The authors thank Sandy Schwirzke-Schaaf for expert technical assistance. This work is supported by the BMWi of Germany (MF090215), the NSFC (60906043), the PCSIRT and the SRF for the Doctoral Program of Higher Education of China (20090076120010).
References and links
1. K. Wundke, J. Auxier, A. Schulzgen, N. Peyghambarian, and N. F. Borrelli, “Room-temperature gain at 1.3 µm in PbS-doped glasses,” Appl. Phys. Lett. 75(20), 3060–3062 (1999). [CrossRef]
2. P. T. Guerreiro, S. Ten, N. F. Borrelli, J. Butty, G. E. Jabbour, and N. Peyghambarian, “PbS quantum-dot doped glasses as saturable absorbers for mode locking of a Cr:forsterite laser,” Appl. Phys. Lett. 71(12), 1595–1597 (1997). [CrossRef]
3. C. Cheng, H. L. Jiang, D. W. Ma, and X. Y. Cheng, “An optical fiber glass containing PbSe quantum dots,” Opt. Commun. 284(19), 4491–4495 (2011). [CrossRef]
4. J. M. Auxier, S. Honkanen, A. Schulzgen, M. M. Morrell, M. A. Leigh, S. Sen, N. E. Borrelli, and N. Peyghambarian, “Silver and potassium ion-exchanged waveguides in glasses doped with PbS semiconductor quantum dots,” J. Opt. Soc. Am. B 23(6), 1037–1045 (2006). [CrossRef]
5. E. F. Chillcce, R. E. Ramos-Gonzales, C. M. B. Cordeiro, L. Gutierrez-Rivera, H. L. Fragnito, C. H. D. Cruz, A. C. Bordonalli, H. E. Hernandez-Figueroa, R. L. Braga, and L. C. Barbosa, “Luminescence of PbS quantum dots spread on the core surface of a silica microstructured optical fiber,” J. Non-Cryst. Solids 356(44-49), 2397–2401 (2010). [CrossRef]
6. S. Joshi, S. Sen, and P. C. Ocampo, “Nucleation and growth kinetics of PbS quantum dots in oxide glass: spectroscopic and microscopic studies in the dilute range,” J. Phys. Chem. C 111(11), 4105–4110 (2007). [CrossRef]
7. N. F. Borrelli and D. W. Smith, “Quantum confinement of PbS microcrystals in glass,” J. Non-Cryst. Solids 180(1), 25–31 (1994). [CrossRef]
8. I. Moreels, K. Lambert, D. Smeets, D. De Muynck, T. Nollet, J. C. Martins, F. Vanhaecke, A. Vantomme, C. Delerue, G. Allan, and Z. Hens, “Size-dependent optical properties of colloidal PbS quantum dots,” ACS Nano 3(10), 3023–3030 (2009). [CrossRef] [PubMed]
9. I. Moreels, D. Kruschke, P. Glas, and J. W. Tomm, “The dielectric function of PbS quantum dots in a glass matrix,” Opt. Mater. Express 2(5), 496–500 (2012). [CrossRef]
10. K. K. Nanda, F. E. Kruis, and H. Fissan, “Energy levels in embedded semiconductor nanoparticles and nanowires,” Nano Lett. 1(11), 605–611 (2001). [CrossRef]
11. W. Vogel, Glass Chemistry (Springer-Verlag Berlin and Heidelberg GmbH & Co. KG, 1994).
12. B. Ullrich, R. Schroeder, and H. Sakai, “Intrinsic gap emission and its geometry dependence of thin-film CdS excited by two-photon absorption,” Semicond. Sci. Technol. 16(12), L89–L92 (2001). [CrossRef]