Mixing and melting the composites of QDs precursor with the glass batch together, then a glass without QDs will be obtained. By certain process of nucleating, QDs will subsequently grow inside the glass. Borrelli and Smith used this method to fabricate QDs-doped glass as early as 1994 (Borrelli and Smith, 1994). Silva et al. successfully doped S into SiO₂-Na₂CO₃-Al₂O₃-PbO₂-B₂O₃-based glass to grew PbS QDs after certain thermal treatment (Silva et al., 2003, 2006a). The absorption and PL spectra of PbS QD-doped glass heat-treated for 3 and 5 h are shown in Figure 3, respectively (Silva et al., 2003). It is clear that a red-shift occurred when the QD-doped glass was heat-treated for a longer period, which indicates that the size of QDs grows larger.

In our group, we used SiO₂-B₂O₃-K₂O-BaO-ZnO glass system as matrix to mix with 1–1.5 mol% PbS for the formation of PbS QDs (P1, P1.5). In contrast, another system using ZnS-PbO to indirectly produce PbS as the starting materials (Z2; Guoping et al., 2011). The as-prepared glasses were subsequently heat-treated at several temperatures to precipitate PbS QDs. Figure 4 shows the transmission electron microscope (TEM) images and size distribution of PbS QDs in representative P1 and Z2 glass samples, which suggest that using ZnS-PbO to replace PbS as precursor of PbS QDs, the size distribution of PbS QDs in glasses becomes more uniform.

In Figure 5, broadband near-infrared (NIR) absorption and emission peaks are observed in the spectra, which is due to the recombination of electron and hole pairs created by the excitation source. With the enhanced heat-treatment temperature, PbS QDs grow larger, which will result in the red-shift of PL emission band. The size distribution of PbS QDs in glasses also becomes wider.

Although the method is convenient to prepare QDs-doped glasses, it requires particular high temperature to melt raw materials. Several other groups also developed other low temperature routines to fabricate QD-doped glasses. Several glass systems like tellurite and phosphate glasses are developed to lower the melting temperature, but they usually have a lower mechanical property than silica glasses (Kolobkova et al., 2002; Savitski et al., 2002; Jacob et al., 2005).

Colloidal QDs systems including CdS, ZnS, PbS, and ZnO, have been successfully reported in many previous investigations (Bahnemann et al., 1984; Henglein, 1988; Kormann et al., 1988). Since Rajh et al. utilized a sol-gel method to incorporate QDs in transparent silicate glasses conveniently at low temperature (Rajh et al., 1988), sol-gel method is now considered as a reliable way to produce QD-doped glass (Planelles-Arago et al., 2008; Mashford et al., 2011). Planelles-Aragó’s group reported highly transparent and colorless silica embedded with Eu³⁺-doped ZnS QDs (Planelles-Arago et al., 2008). In their works, the solution of zinc nitrate and europium nitrate were first dissolved in DMSO [dimethyl sulfoxide, (CH₃)₂SO] and then stirring for 30 min. The resulting solution subsequently added to the mixture of TEOS [tetraethoxysilane, Si(OC₂H₅)₄], DMSO, and water. After stirring and oven dry, the transparent and colorless xerogels were fired in a furnace at 500°C, and finally formed a QD-doped glass. The TEM images in Figure 6 show that typical spherical ZnS QDs were embedded in the amorphous silica medium. Emission spectra in Figure 7 illustrate a broad blue emission of ZnS QDs and red ⁵D₀ → ⁷F₂ transition (615 nm) emission of Eu³⁺ ions, respectively. Both the TEM and emission results confirmed that QDs were successfully doped into the glass matrix by sol-gel method.

Compared to melt-quenching, the sol-gel route is helpful to lower the melting temperature and disperse the precursors homogeneous easily. However, fabrication in large bulk material is still a challenge because the residual stress is hard to eliminate. Therefore, the sol-gel method is only suitable for QDs-doped thin films. In addition, the difficult removal of organic ligands introduced by synthesis procedure and the restricted silica glass matrix may limit the development of sol-gel route.

Ions accelerated by electric field are injected into surface of the material. Ions implanted will react with the material, creating defect or forming composite. This method was originally used for semiconductor doping, then developed into a wider range of applications in the preparation of integrated circuits and modification of metals. Shortly after, there were numerous researches about the effects of implantation on silica matrix (Primak and Kampwith, 1968; Arnold, 1973; EerNisse, 1974). Glasses implanted with various ions were begun to study afterward (Arnold, 1975; Mattern et al., 1976). Ion-implantation is mainly used to nanocomposites like metallic nanocrystals, semiconductor nanocrystals, and ferromagnetic nanoparticles (Meldrum et al., 2001). After decades of investigation, ion-implantation still draws high attention (Lee et al., 2005; Carder et al., 2013). In the work conducted by Carder’s et al., slides of SiO₂ was used as a substrates, which were ion-implanted with Pb (29 keV) and Se (25 keV) ions under vacuum at 5 × 10⁻⁸ mbar (Carder et al., 2013). The samples were then transferred to an electron beam annealing (EBA) system and maintained at a temperature around 760–915°C.

The TEM images confirm that PbSe QDs have been successfully implanted into the silica matrix (Figures 8A,B), and there are a band of nanoclusters at about 50–70 nm beneath the surface (Figure 8A). Figure 9 shows the PL spectra of samples applied to different annealing process, indicating that QDs embedded inside the glass can be further tuned by EBA temperature.

Although ion-implantation can induce the formation of tunable QDs beneath the surface of glass matrix, the ions implanted are limited around the surface, and barely reach the inner part. Therefore, it is a challenge to produce scaling QDs-doped bulk glasses.

Sol-gel route is the most convenient protocol to synthesis QDs with excellent performance. However, for further application, QDs must be incorporated into various bulk matrixes. Ion-implantation is able to synthesize all kinds of QDs-doped glasses, the shallow implantation depth of QDs beneath the matrix surface limits its further applications. In spite of high temperature and relatively complex procedure, melting-quenching is a high prior to synthesize a QDs-doped glass due to the controllable size and distribution of QDs in various glass matrixes.

There are also some other methods like atomic layer epitaxy (ALE; Mukai et al., 1997), molecular beam epitaxy (MBE; Grundmann et al., 1995; Mirin et al., 1995), metal-organic chemical vapor deposition (MOCVD; Heinrichsdorff et al., 1997; Tan et al., 2005), and aqueous synthesis (Rogach et al., 2007), which are relatively complex and less used. Therefore, they are not discussed in detail here.

Since the quantum confinement of QDs first observed by Efros and Ekimov (Efros and Efros, 1982; Ekimov et al., 1985), Brus modeled the excited electronic state theoretically in order to understand the redox potential that depends upon size (Brus, 1984). Later on, Henglein et al. successfully synthesized QDs with different sizes (Weller et al., 1984; Spanhel et al., 1987), which create a precedent in the research field.

Energy spectrum of electrons in a bulk material is almost continuous, therefore, electrons in fact is free (Figure 1A). However, when the material gets smaller, electrons will generally be restricted and exhibit a discrete energy spectrum in macro (Figures 1B–D). Especially when a particle’s size tend to the Bohr radii of the bound state of an electron and a hole, the movement of electrons is strongly confined, lending to a unique tunable optical property to QDs.

Due to the quantum confinement effect, the band gap of electrons depends upon their sizes. In other words, QDs get smaller, their band gaps become wider. It offers a convenient way to tune the QDs absorption and PL wavelength by controlling their size.

Work of Murray et al. can show what above more directly: they synthesize CdE (E = S, Se, Te) from ~12 to ~115 Å in diameter by sol-gel method (Murray et al., 1993). The series spans a range of sizes from nearly molecular species to bulk lattice. An obvious red-shift can be observed in Figure 10 with the size of CdSe QDs getting larger, which matches well with the theory mentioned above. As Figure 11 illustrated, transition energy of CdSe crystallites as a function of size is compared with the prediction of the effective mass approximation. Experiment and theory fit well when the size of QDs is larger enough, but it does not in the case of small size because of the non-parabolicity of the bands at higher wave vectors and the finite potential barrier at the surface of the particles.

During the last decade, synthesis of QDs have been improved and further expanded to control the QDs’ morphology and structure (Peng et al., 1998; Manna et al., 2000; Peng and Peng, 2001a). Figure 12 shows the influences of CdSe QDs’ shape and structure on their optical properties: (1) A wider splitting between absorbing and emitting states was shown in rods than that in dots (Figures 12A,B); (2) quantum yields rises at first and goes down latter with the increasing thickness of a shell (CdS or ZnS) grown on the CdSe core (Figure 12C); (3) polarization anisotropy can be observed from quantum rod while QD cannot (Figure 12D).

All examples above only focused on the tunable optical properties of QDs themselves. However, for better application, controlling QDs in glasses is more important. Heo and his coworkers have successfully controlled the size of PbS QDs using rare earth ions and Ag⁺ ions (Heo and Liu, 2007; Chao and Jong, 2014; Xu and Heo, 2014). Take RE ion for example: the size of QDs decreased with RE ion concentration increases under the same heat-treatment conditions and thereby provided another pathway for the controlled formation of QDs (Chao and Jong, 2014). Their group also synthesized glasses doped with different percents of Nd₂O₃, and their absorption and PL spectra are shown in Figure 13. The peak wavelength of the first absorption bands shifted from 1415 to 1274 nm as Nd₂O₃ content increased from 0 to 0.6 mol% under the same heat-treatment (Figure 13A). Emission peak wavelength of the PL bands also shifted from 1516 to 1370 nm at the same time (Figure 13B).

In addition to the quantum confinement, some external field factors, such as temperature, pump intensity, can also influence QDs’ optical properties (Olkhovets et al., 1998; Schaller et al., 2003). Olkhovets et al. observed that the temperature coefficients of electron-hole pair energies (dE/dT) in PbS and PbSe QDs depend strongly on their size (Olkhovets et al., 1998), resulting in the special electronic and vibrational properties in QDs (Wise, 2000). The experimental data are shown in Figure 14. It is clear induced that the absorption spectra of smaller QDs are almost independent of temperature while the larger ones strongly depend on the ambient temperature.

Schaller and his coworkers also demonstrated that the amplified spontaneous emission (ASE) from PbSe QDs with emission energies tunable in the NIR (Schaller et al., 2003). They synthesized PbSe QDs by sol-gel method to ensure a good surface passivation and high filling factor. An emission band at 1555 nm is observed when the pump intensity increased (Figure 15), showing that PbSe QDs have a strong dependence on the pump intensity.

By achieving tunable optical properties, people can obtain QDs in glass with multi-wavelength NIR emission. And it is the first step to obtain tunable NIR fiber amplifiers and lasers. In the next section, we will discuss about the optical amplification of QDs-doped glasses. Combing their tunable NIR emission with the optical amplification, the NIR broad amplifiers and tunable lasers will have a bright future.

Since the CdSe QDs were first demonstrated to be available in the visible optical amplification (Klimov et al., 2000), nowadays, the wavelength of optical amplification have explored from visible to infrared. For infrared broadband PL semiconductor QDs, the research system is mainly about PbS, InAs, PbSe, HgTe, and so forth (Steiner et al., 2004; O’Connor et al., 2005; Sukhovatkin et al., 2005). Among them, PbS QDs with a large Bohr exciton radii (~18 μm) and a moderate band gap at near/mid-infrared wavelength have gained great attention as NIR broadband amplified medium (Sargent, 2005; Heo and Liu, 2007; Chao et al., 2008).

In fact, PbS QDs were precipitated in glasses and a gain of 80⁻¹ cm at 1300 nm was obtained at room temperature (Wundke et al., 1999). For further research, Huang et al. reported the tunable absorption and PL of PbS QDs in the glasses precipitated through thermal treatment (Huang et al., 2008). As shown in Figure 16, optical amplification at 1300 nm is observed, and the ASE spectrum is also measured to confirm the optical gain from PbS QDs.

Recently our group first demonstrated that PbS QDs-doped glasses exhibit two optical amplifications at the traditional ~1330 and ~1550 nm optical communication windows, which is helpful to expand the applications of QDs in optical communication (Guoping et al., 2011). In order to further evaluate the potential of PbS QDs-doped glass applied in a broadband fiber amplifier, it is necessary to investigate the broadband optical amplification at several optical communication windows. We successfully obtained the optical amplification at both 1330 and 1550 nm simultaneously in single PbS QDs-doped glass (as shown in Figure 17) by tuning the QDs’ size, distribution, and PL properties (Guoping et al., 2014). Our studies have confirmed that PbS QDs-doped glasses are promising to be applied as the gain medium of broadband fiber amplifiers and tunable fiber laser at optical communication windows.

Our studies also demonstrated that utilizing ZnS + PbO instead of PbS as the precursor of QDs can effectively inhibit sulfur volatilization during glass melting, leading to the uniform and controllable growth of PbS QDs in glass matrix. From the optical gain spectra in Figure 18, it can be deduced that using ZnS + PbO as precursor in PbS QD-doped glasses, the optical gain and its slope value (optical gain vs. pumping power) improve substantially (Huang et al., in press).

Herein, QD-embedded glass with multi-wavelength optical amplification is promising as the gain medium of broadband fiber amplifiers and tunable fiber lasers.

When a system (molecule or atom) absorbs a photon, it gains energy and activates into an excited state. One way for the system to relax is to emit a photon, thus releasing its energy (another way would be the transformation of heat energy). When the emitted photon has less energy than the absorbed photon, this energy difference is the Stokes shift (Figure 25). With the decreasing of Stokes shift, the overlap of absorption and fluorescence spectrum increases, leading to the self-absorption which reduces the fluorescence efficiency. In general, a big Stokes shift is desirable to elevate emission efficiency within the interesting waveband (Cheng et al., 2011). Blue- or red-shift will occur by changing the size and distribution of QDs, with which the Stokes shift can be adjusted (Silva et al., 2006b). As the size of QDs increases, the Stokes shift will decrease (Cheng et al., 2011). This phenomenon was also reported in literature (Silva et al., 2003), which can be explained by the Kubo theory (Yu and Peng, 2002). The level spacing as a function of particle diameter can be expressed as follows (Kubo, 1962): δ=4EF3N∝V−1 where δ is the level spacing, N is the total number of conduction electrons of a nanoparticle, V is the volume of a nanoparticle, and E_F is the Fermi level.

Through the above analysis, it can be deduced that the key to solve the problem of overlapping is to control the size and distribution of QDs. One can improve synthetic method of QDs, continuously explore new synthetic techniques to realize the controllable growth of QDs. By this, people can control the size and distribution of QD as required. In recent years, organic synthesis have been improved, and successfully prepared various bivariate and multivariate semiconductor QDs (such as: CdSe, PbS, InP, ZnSe, ZnxCd_1-xSe, CdTe_xSe_1-x, and so forth), realizing the controllable synthesis in size, morphology, composition, and properties of QD basically (Murray et al., 1993; Peng and Peng, 2001b; Yu and Peng, 2002).

Besides temperature, new fields including femtosecond laser, electrochemical synthesis, and magnetic field induction may also be utilized to tuning the distribution and growth of QDs in glass substrate. Stone et al. pointed out that the ferroelectric single crystal LaBGeO₅ can controllably grow directly in 25La₂O₅-25B₂O₃-50GeO₂ (mole percentage) glass by femtosecond laser irradiation (Stone et al., 2009). Later, Fan et al. also use femtosecond laser to control the micron-region formation of PbS QDs inside the glass. These fully prove that ultrafast laser radiation is one of the most effective ways to induce the controllable QDs formation in glass matrix (Fan et al., 2012). Now many scholars focus on finding out the effective laser radiation parameters to control crystallization behavior.

In summary, the size and distribution of the QDs can be designed to get the requiring Stokes shift, thereby weakening the overlap and self-absorption to further improve the QDs’ quantum yield.

The key to solve this problem is to control the size of QDs, and they will not grow up in the role of light (Lagatsky et al., 2004). As is mentioned above, QD-doped glass and fiber can be prepared by composite approach of the QD nanocrystals with glass. Based on this method, QDs nanocrystals can be further cladding, their surface were passivation. And then they were compounded with glass, no any excessive ions (such as Pb and S elements for PbS QDs) existed in glass matrix. This makes the resistance of QDs growth much bigger, harder to grow when influenced by external field conditions.

A stable and widely covered way is growing inorganic material shell on the surface of QDs, forming core/shell structure. Inorganic material shell with the same clan system can grow on the surface of QD’s nuclear epitaxial, it will be more insulation than semiconductor, effectively protect the carrier at the inside of the light. Among them, the ZnS shell can coat in the nuclear of CdSe (both are II–VI), which is the most common core/shell structure. The research about them is the thorough currently, and they have been widely used in various applications such as subsequent biological, photoelectric fields (Colvin et al., 1994; Bruchez et al., 1998).

When drawing PbS QD-doped optical fiber from the preform, due to the excessive Pb and S elements in glass matrix, the secondary growth of QDs will occur via ion-migration mechanism under the driving force for the rise of temperature, so that the size of QDs is hard to control. Therefore, confinement effect is difficult to play since QDs easily grow larger than the exciton Bohr radius.

As is mentioned above, we can obtain fiber by melt-extraction method, and its advantage is easy to control the size of QD. But it is only used to prepare bare fiber. To obtain optical fiber that can be used for practical application, we must modify its fiber-drawing process, making direct access to prepare whole QD-doped glass fiber containing outer cladding layer.

Based on the successful fabrication of luminescent PbS QD-doped bare fibers by melt-extraction method mentioned above, a traditional type of fiber-drawing technology - double crucible method is expected to directly fabricate optical fiber containing outer cladding layer from glass melt. Sanghera et al. have fabricated single mode optical fibers using the double crucible technique, and that double crucible process enables adjustment of the core/clad diameter ratio during fiber drawing by independent pressure control above each melt (Sanghera et al., 2008). The mechanical property of glass fibers fabricated by double crucible process is better (Dianov et al., 1990). These all confirmed the feasibility of double crucible technique. After drawing by double crucible process, and then heat-treatment, moderate size of QDs can be obtained in fiber by controlling the heat-treatment schedule.

By composite approach of the QDs and glass, it is also expected to prepare QD-doped glass and fiber. The QDs should be controllably synthesized firstly, and then composited with the substrate glass together through co-melting technique (Chai et al., 2012), which can be further drawn into optical fiber by controlling the reasonable temperature and melting state. During the composite approach, there are no residual Pb and S elements in PbS QD-doped glass, and the QDs can only grow through the Oswald ripening process. Since the resistance of QDs growth via Oswald ripening process is much bigger than the above ion-migration force of secondary growth, it is much more difficult for the secondary growth of QDs by composite approach during the fiber-drawing process. In our previous work, we have successfully synthesized YAG nanocrystals, and making them recombined with substrate tellurate glass by novel co-melting technology. Intense 2.7 μm emission was obtained in the YAG:Er³⁺,Tm³⁺ nanocrystals-tellurate glass composites (Chai et al., 2012). Those works have opened a bright future work to fabricate high-quality QDs-doped glass fibers and devices.