1 Introduction

In recent years, the need for small-size and cost-efficient integrated optical devices has drawn much attention and importance in communication industries and telecom networks [1]. Integrating many components on a single optical chip significantly enhances the energy consumption and efficiency of the telecommunication network. Integrated optical devices can also be used as high-speed interconnects for processors and networks-on-chip [2] and also for sensing applications such as microfluidics [3, 4]. Integrated optical sections are among the most challenging parts of optical telecommunication technology. Today, making passive optical components, such as planar optical waveguides, has become quite efficient. Optical amplifiers need to be integrated with other components to establish such technologies. However, the main challenge is to provide optical amplifiers and lasers with monolithic gain materials in such integrated optical circuits [5]. Amplifiers were conventionally implemented using III–V materials, which are hard to integrate for large-scale production. This is due to the packaging process and bonding in the integration process [6]. Another common approach for on-chip signal amplification is attained using rare-earth dopants [7]. Specifically, Er-dopants are exciting materials for active waveguide fabrication [8].

Er–Yb-doped waveguide amplifiers (EYDWA) were first introduced in 1987 [7]. They deliver a valuable gain at telecom wavelengths such as L-band (1565–1610 nm) and C-band (1525–1565 nm) [9, 10]. Its considerable gain and low noise, nonlinearity, and compatibility with optical networks [11] makes it an outstanding amplification module for telecommunication systems. Due to these properties, Er-doped waveguides have been broadly researched in the past three decades, and multiple glass-based materials have been utilized to develop the EYDWAs [11, 12]. Nevertheless, it should be noted that desirable EYDWAs as a module must have the ability to integrate as an element of a robust and low-cost integrated circuit and be compatible with other Si-based integrated platforms such as Silicon Nitride (SiN). However, this integration will bring many engineering issues and challenges. However, such issues can be addressed by introducing various engineered devices, materials, and diverse integration platforms. Because of compatibility and similarity with silica-based optical fibers, earlier investigations were done on phosphosilicate glass as a host [13]. Er-doped phosphate glasses are exciting materials for high-gain amplifiers and active waveguide fabrication [14]. Throughout many explorations over the years, a diverse number of Er–Yb-doped waveguide amplifiers (EYDWAs) and lasers were also studied using crystals and polymers.

This work presents the fabrication of an Er-Yb-doped waveguide amplifier (EYDWA) with low loss for integrated optics and optical devices. Firstly, Sect. (2) explains this study’s methodology. Section (3) will discuss the fabrication process. Er-doped phosphate glass waveguides are deposited using the sol–gel method, which is convenient for preparing active films on several substrates. First, spin-coating is used to make sol–gel films ready. Working with a SiO2–Er(NO3)3 5H2O–Yb (NO3)3 9H2O–Al (NO3)3 9H2O system, the optimum composition for the required concentration will be attained. The waveguide fabrication method utilizes the Ag-Na ion exchange process. In such a method, ions in the core material and molten salt are exchanged, and the waveguide is immersed. Section (4) will consider characterization methods, results, and discussions. A high-power laser at 1.48 μm is used as the pump source, and an adjustable laser at 1.54 μm is used as the signal laser. Finally, The net gain of around 2dB at a wavelength of approximately 1.54 μm in Er:Yb:Al: PSG will be achieved. This section shows that by reducing the width of the waveguide to 0.5 μm, the confinement and the gain can be increased. Without more pump power, >1dB net gain is achieved in that case. Lastly, Sect. (5) will conclude and summarize the achievements of the work.

2 Methodology

The Er-Yb-doped waveguide in this work is doped with Er3+ and Yb3+ using the sol–gel method [15]. The selected molar composition in this method is 4.8gr TEOS–0.5gr HCL–0.5gr P2O5–0.2gr AL2O3–0.4gr Er–0.2gr Yb [16]. The films were spin-coated on Si/SiO2 substrate, then annealed at 750 °C, presenting a homogeneous and continuous surface. Er3+ is stimulated to one of its upper energy layers using an external laser. The Er–Yb-doped waveguide amplifiers can be pumped at 0.98 or 1.48 μm. When the 1.54 μm signal passes through a waveguide with an Er3+ dopant, it causes a stimulated emission, and the excited Er3+ returns to the base state from the first excitation layer, which will amplify the signal. If the stimulated Er3+ returns to the base state in the absence of the main signal, it will cause a spontaneous transmission, and this will not be followed by any gain increase [17]. The Er3+-ion doped Al2O3 has also received considerable attention as an assuring candidate for the development of 1.5 μm optical waveguide amplifiers. To attain a full background knowledge of the process used in this work, different physical considerations and definitions are explained in more detail in the following subsections.

2.1 Dopants used in active waveguides

Erbium is one of the rare earth elements in the lanthanide group. When this element is placed in the solid, it is usually assumed to be in the trivalent oxidation state. The Er3+ has a semi-filled 4f layer. Therefore, due to the interaction of spin-spin and spin-orbit, they can have different electronic arrangements with different energies. Er3+ can be easily pumped directly to its first excited layer (4I11/2) by a 1.48 μm diode laser and, with the help of a 0.98 μm laser diode, go to the higher absorption layer (4I13 / 2). In the second case, Er3+ quickly returns to the first layer of stimulation; usually, the lifetime is about 1nS–100 µS. Figure 1 exhibits the energy transfer diagram for the Er3+–Er3+ energy transfer. Another substance used in optical amplifiers as a dopant is Neodymium (Nd). The performance mechanism of Nd is like Er. However, unlike Er3+, which makes amplification at 1.5 μm, Nd3+ makes amplification at 1.3 μm. Yb3+ is also used to increase the pump absorption and amplify at 0.98 μm wavelength.

Fig. 1
figure 1

Energy diagram for Er3+–Er3+ energy transfer

Full size image

The main reason limiting the gain in systems with large Er3+ dopants is the interaction between ions. By giving its energy to its neighboring ion and raising it to the 4I9/2 level, excited Er3+ can return to the ground state by this mechanism. To achieve a specific amount of inversion, more pump power is therefore required. The resonance level exists at double the energy of the initial excited level, making co-operative upconversion conceivable. The placements of the energy levels, cross sections, dielectric constant, and characteristic phonon energy of the host material all affect the co-operative upconversion coefficient, which depends on the host material. Co-operative upconversion is a significant gain limiting factor in real-world applications. The co-operative upconversion impact depends on the microscopic distribution of Er3+ in the host material as well as the average quantity of Er dopant. The upconversion coefficient might vary up to three times in a host material. The fabrication process should also be tuned to have a consistent distribution of Er3+ to provide the best amplifier possible. Modifications should be made to the host material’s network to increase Er3+’s solubility (such as silica-based glasses). For instance, phosphorus can be added to the glass matrix to boost Er3+’s solubility. Adding high refractive index ceramic materials like Al2O3 to glasses improves significant Er3+’s solubility. In addition, fabrication techniques also have a significant impact on it. The density needed to achieve the desired gain in a few centimeters is roughly 1019–1020 Er/cm3.

2.2 The required density for Er dopings

The density of Er3+ is the determining parameter for amplification, which can be optically activated. By multiplying the stimulated emission cross-section and activated Er3+ density, one can obtain the maximum gain per unit length emission cross-sections (σem). Since the usual value for σem is about 1020–1021/ cm3, the required density to reach the acceptable net gain to increase the device’s sensitivity and reduce loss is about 1019–1020 Er / cm3. However, changes must be made to the host material (such as silica-based glass) to improve the solubility of Er3+ [11].

2.3 Pump wavelength

Er-based amplifiers are pumped with a 0.98 μm wavelength (4I11 / 2 level) or 1.48 μm wavelength (4I13/2 level). In order to obtain a three-level quasi-surface system, the first stimulated layer of Er3+ must be pumped with a 1.48 μm laser. Due to the similar mode sizes, the 1.54 μm signal mode overlaps well with the pump mode. Besides, wave scattering losses usually increase for lower wavelengths, so a laser with a longer wavelength is more desirable. The stimulated emission decreases when the pump wavelength is 0.98 μm. In this case, the stimulated Er3+ quickly returns to the first non-irradiated stimulus layer. The cross-sectional absorption for the 0.98 μm level is lower than the 1.48 μm level.

2.4 Pump absorption

To reach the inversion, a relatively large pump intensity is needed because of the small absorption cross-section for Er3+ energy levels. This issue can be solved by inserting a high absorption cross-section to transfer the energy to Er3+. The Yb3+ with a high absorption cross-section at 0.98 μm is a proper candidate and can transfer its energy to a non-radiant Er3+ to be stimulated to 4I11 / 2 level [18]. Figure 2 exhibits the energy diagram for the Er–Yb energy transfer.

Fig. 2
figure 2

Energy diagram for Er–Yb energy transfer

Full size image

The most critical parameter for amplifying the waveguide is the concentration of Er3+, which are optically activated and have a high quantum efficiency. Multiplying the cross section for σem and the density of activated Er3+ yields the maximal gain per unit length. Since σem typically has a value of around 1021–1020 cm2, the density needed to achieve the desired gain in a few centimeters is roughly 1019–1020 Er/cm3. Pump absorption is increased by using Yb3+. Amplification occurs in this rare-earth ion at a wavelength of 0.98 μm. Er3+ energy levels have a small absorption cross-section. The parameter value of Er3+ absorption cross-section at 1.55 μm is around 6.6 × 10−25 m2. Therefore inversion requires a significantly greater pump intensity. By introducing a substance with a large absorption cross-section that may transfer energy to Er3+, this issue is resolved. Yb3+-ion is an illustration of this class of ions. At 0.98 μm, the Yb3+ cross-section has strong absorption and may non-radiatively transmit its energy to Er3+ ions, which are then stimulated to the 4I11/2 level. The material will need to be co-doped with Yb3+-ions for greater resonant pump absorption near 0.98 μm and subsequent energy transfer of the excitation energy to Er3+ to achieve noticeably shorter device lengths and, consequently, higher integration densities.

3 Fabrication

At the beginning, Si must be separated from the other deposited layers on it. As depicted in Fig. 3, SiO2, as a total insulator layer, has been grown and established the essential separation. Moreover, the HF acid takes off all the contaminants from the Si surface. Then, Si is oxidized in the furnace at about 1050 °C. Thermal oxidation is done in two processes: dry and wet oxidation. In dry oxidation, the sample takes place under a pure oxygen atmosphere. On the other hand, in the wet thermal oxidation method, the oxygen is led through a bubbler vessel filled with heated water. This process is done by 800–1050 °C. The oxide layer grows to a thickness of about 0.5 μm as shown in Fig. 3.

Fig. 3
figure 3

Thermal oxidation of Si substrate (at 1050 °C) with 0.5 μm thickness

Full size image

The next step is to apply Er to the oxide layer using the sol–gel method. In the sol–gel method, the molecular precursor must be prepared at the beginning. The sol–gel solution, as shown in Fig. 4, is made using a combination of TEOS, Ethanol, H2O, HCL, P2O5, Al (NO3)3 9H2O, Yb (NO3)3 9H2O, and Er (NO3)3 5H2O with specific ratios. Tetraethyl orthosilicate (TEOS) (Merck) and ethanol were mixed together in a 4:1 volume ratio. HCL and deionized water were mixed together in a 1:1 volume ratio. Then left for 6 min under magnetic stirrer to be homogenized. For the Al (NO3)3 9H2O (Merck) as Al2O3 source and P2O5 (Merck) was first dissolved in absolute ethanol in a 1:2 volume ratio, then this solution was added to the previous solution. Er (NO3)3 5H2O (Sigma Aldrich) as Er source and Yb (NO3)3 9H2O (Sigma Aldrich) as Yb source dissolved in absolute ethanol in a 1:4 volume ratio. The final solution was left for 30 min under a magnetic stirrer. This solution homogenizes in almost 15 h, and it turns into a gel after 48 h. Afterward, the 3500 rpm speed rotation of the spin coating will cause the deposition of the homogeneous solution on Si/SiO2 substrate.

Fig. 4
figure 4

The sol–gel solution: a Heterogeneous solution at the beginning, b Homogeneous solution after 15 h and c The gel after 48 h

Full size image

For examining the morphology of the samples, a high-resolution scanning electron microscope (SEM) is required. More specifically, the TESCAN VEGA model, at a speed of 30 kV was used in our experiment. The Samples in this method have cracks on the surface (See inset in Fig. 5a, b). To repair the surface cracks and impurities, the samples must be spun again by the sol–gel solution (See inset in Fig. 5c).

Fig. 5
figure 5

SEM images of the Er: Yb doped SiO2–Al2O3 films with one deposited layer by spin coating in scales: a 200 μm, b 20 μm. c with two deposited layers by spin coating and annealing in scale 2 μm

Full size image

The thermal treatment of the samples by vacuum annealing is done in 1 Pa at temperature 750 °C for 1 h. Then, a 200 nm Titanium film is sputtered on the sample layer as a hard mask. By placing metals such as Chromium on the semiconductor, a depletion region is created due to the difference in the work functions of the metal and the semiconductor. Therefore, an electric field is created due to the presence of this depletion area.

These fields attract ions to themselves and destroy the uniform distribution of ions. Ions accumulate in the lateral regions of the waveguide, and their density decreases in the center. For this reason, it is better to use masks such as Titanium and Aluminum oxide. In the next step, using lithography, a layer of resist is spun on the entire surface of the sample with a rotation speed of 5000 rpm and is baked at a temperature of 150 °C. Then, the waveguide pattern is created using electron beam lithography. At this stage, the challenge is to create a pattern with smooth and continuous edges; if in EBL programming the dtset is set to more than 750, the waveguide pattern is created intermittently with rough edges, as shown in Fig. 6a. Where dtset is defined as the time that the electron beam is stopped until the energy hits the sample. During multiple tests, the dtset must be set between 500 and 750, in which case a continuous pattern can be created, as depicted in Fig. 6b.

Using the chemical etching methods, the titanium film can remain in the desired parts. This is done using the etching process with 100 ml of water + 10 ml of hydrogen peroxide + 10 ml of HF, as shown in Fig. 7a. Various methods are used to create waveguides, including the reactive ion etching in both dry and wet methods and the ion-exchange method [19]. However, the advantage of the ion-exchange method over the RIE method is that it prevents the formation of shark fin structures on the waveguide surface and.

Fig. 6
figure 6

SEM images of the waveguide in sol–gel Er: Yb doped SiO2–Al2O3 coatings patterned by Electron Beam Lithography: a intermittent waveguide pattern, b continuous waveguide pattern

Full size image

creates a completely smooth and polished structure. In the method of making ridge waveguides, due to surface and edge imperfections, waveguide losses such as propagation and scattering losses are increased, while the method of making buried waveguides does not result in such losses. Ag–Na ion exchange is used because of the large change of refractive index that can be induced. Initially, the sample is cleaned with Propanol with the formula CH3CH2CH2OH and Acetone and washed in DI water. Then completely dried under Nitrogen gas. The AgNO3 salt (Merck) is placed in a ceramic boot inside the furnace. Firstly, the sample is plunged in a molten solution of AgNO3 at a temperature of about 300 °C in the presence of argon gas for 10 min. Additionally, conservation of charge neutrality happens when we let the N+ into the melt. This operation needs to release Ag+ into the substrate using a chemical potential gradient. Accordingly, thermal diffusion causes Ag+ to be distributed in the substrate after the execution of the ion-exchange process. Figure 7b is the thermal ion exchange through a patterned mask from a molten salt source. After the sample has cooled down to room temperature, they are washed with DI water to remove the excessive salt. Finally, chemical etching process is used to remove the Titanium film totally.

Fig. 7
figure 7

Simulation of a removing titanium film by wet etching, and b ion exchange process in molten salt (AgNO3)

Full size image

The penetration constant for silver impurities in the process of impurity penetration into the Erbium and the ion exchange method is expressed in Eq. (1). Here, Cs is surface concentration and D is Diffusion coefficient, which has different amounts for different impurities obtained by Eq. (2). Moreover, D0 is the internal diffusion coefficient and EA is the Activation energy. According to [20], these values are obtained.

$${C}_{x}={C}_{s}\times erfc\left(\frac{x}{2\sqrt{Dt}}\right)$$
(1)
$$D={D}_{0}\times \text{e}\text{x}\text{p}\left(\frac{-{E}_{A}}{{K}_{B}T}\right)$$
(2)

4 Characterization, results and discussions

When the pump power is sufficient, a population inversion occurs between the first excitation layer (4I11 / 2) and the ground state (4I15 / 2). The gain will not increase if the excited Er3+ returns to the ground state without the main signal (1550 nm), as it will cause spontaneous emission. However, if the excited Er returns to the ground state with the main signal (1550 nm) present, it will cause excited emission and return to the ground state from the first excitation layer. Figure 8a shows the signal changes and the spontaneous emission with different pump powers.

By pumping an Er:Yb:Al: PSG waveguide amplifier with a dopant concentration of 1 × 1020 cm3 at each wavelength, the net gain at 1.55 μm can be compared. In order to obtain a quasi-three-level pumping system, which overlaps with the pump mode due to similar mode sizes, one must pump the first Er-stimulated layer with a 1.48 μm laser. Moreover, longer wavelength lasers are also more beneficial since shorter wavelengths often result in increased waveguide scattering losses. However, the pump also produces excited emission at 1.48 μm [21] due to the overlap of the pump wavelength of 1.48 μm with emission 4I13 / 2 -> 4I15 / 2. Such non-ideal effects can limit the rate of population inversion. The excited emission decreases when the pump wavelength of 0.98 μm is used. In this case, the excited Er quickly returns to the first non-radially excited layer, preventing pump-stimulated emissions from the 0.98 μm level [21]. However, internal relaxations also reduce the pump’s efficiency due to converting relaxation energy into heat. The threshold power is higher for the 0.98 m level due to a lower absorption cross-section than for the 1.48 μm level. Still, the 1.48 μm pump signal may travel farther in an Er amplifier without experiencing a drop as significant of a drop compared to the 0.98 μm signal. Therefore, a 1.48 μm pump is used for pumping. The internal net gain with 1.48 μm and 0.98 μm is shown in Fig. 8b.

Fig. 8
figure 8

a Comparison of signal power and spontaneous emission power. b An Er:Yb:Al:PSG waveguide with an Er3+ concentration of 1 × 1020 cm3 and pump wavelengths of 0.98 and 1.48 μm. c Amplifier spectra with 180 mW pump power and without pump power. d Simulated internal relative gain with different amounts of Er concentration. e Difference of relative gain with pump power for EYDWA with Er:Yb:PSG, Er:Al:PSG and Er:Yb:Al:PSG, respectively from bottom to top. f Absorption cross-section of an Er:Yb:Al:PSG waveguide

Full size image

The amplifier’s insertion losses are the most important number that must be established. The mode mismatch between the circular fiber and the rectangular waveguide causes the coupling losses. On the other hand, propagation losses mostly result from Rayleigh scattering brought on by inhomogeneity or impurities inside the waveguide, which typically have a size considerably lower than the operating wavelength of 1.55 μm. The roughness of the sidewall can also cause scattering losses. However, because the buried waveguide has been discussed in this work, a significant amount of the waveguide walls’ roughness has been eliminated. The essential idea is that given an amplifier of a specific length, the background losses are first measured at a wavelength far beyond the erbium absorption bands. The exact measurements are then carried out on the same amplifier. By comparing the two readings, we can determine the propagation losses caused by our amplifier per unit length. The same procedure may be performed to get an average value of the propagation losses and lower the experimental error. As previously noted, these measurements may be carried out at a wavelength outside the erbium band. The inaccuracy of our results is significantly increased by the fact that the erbium band may sometimes go beyond 1600 nm and that laser power levels at such wavelengths are relatively low. We also employed a 1.48 μm laser source to determine the background losses more precisely. We have absorption (negative gain) at all wavelengths for the ground state (lowest curve in Fig. 8c). While there is still net absorption, for example, at 1500 nm with 50% excitation of the erbium ions (middle curve in Fig. 8c), that absorption is now much weaker than for the ground state. Instead, we see some gain in the 1550 nm region. With 80% excitation, the net gain increases significantly around 1535 nm and in the 1550 nm region. There is now some gain even at 1500 nm. This suggests that a shorter pump wavelength is necessary to limit stimulated emission by the pump wave and make it possible to achieve such a high excitation level. Stronger excitation levels alter the gain spectrum’s form and offer additional gain. Figure 8c shows the waveguide spectrum for various pump powers. At a wavelength of roughly 1535 nm, the waveguide’s maximum net gain is obtained, which is around 1.7 dB. Inversion is visible even at very long wavelengths above 1600 nm. The range of about 1450 nm appears to be the safest wavelength for calculating background loss. The overall insertion losses in this scenario are in the range of 0.8 dB. The waveguide exhibits insertion losses of 1.3 dB, which suggests that coupling losses are between 0.15 and 0.2 dB per facet, and background propagation losses at 1450 nm are roughly 0.25 dB/cm. As a result, it can be concluded that Rayleigh scattering is the primary source of loss for propagation losses at this wavelength, which are calculated to be 0.35  dB/cm. This indicates that Loss is (dB /cm) \(\lambda = {\text{a}}{ / }\lambda ^{4}\), where a is a constant that relies on the core’s material qualities [22]. Using the relationship, we can determine that the propagation losses are 0.18 dB/cm at the maximum gain wavelength (1535 nm) and 1.1 dB/cm elsewhere. For an Er3+ concentration of 1 × 1020, a larger gain per unit length and overall gain may be obtained by increasing the pump power to 180 mW. Therefore the optimum Er3+ concentration will depend on the available pump power and the application, which determines the gain required and the amplifier length.

Moreover, Fig. 8c shows the transmission spectrum with pump power and when no pump power is applied for EYDWA. The net gain is about 2 dB for the maximum signal power of 180mW at a wavelength of approximately 1.54 μm. To increase Er3+-Ions, modifications must be made to the host material network (such as glass formed of silica) [23]. The required density to reach the acceptable gain is about 1019–1020 Er /cm3. In case of high Er dopants, the interaction between Er-ions is the most critical limiting factor of gain. In this process, an excited Er3+-ion can return to its base by transferring its energy to its side ion and raise its side ion to the 4I9 / 2 level. Therefore, the highest amount of gain at a density of 1020 Er/cm3 is obtained (Fig. 8d). The variation of net gain with pump power for EYDWAs in the different compositions is shown in Fig. 8e. As it is evident from the results, the Er:Yb:Al:PSG EYDWA performance is significantly better than samples without Al and Yb.

Since Er energy levels have a low absorption cross-section, higher pump intensities are needed to obtain inversion. By using a high absorption cross-section material, such Yb, to transfer its energy to Er3+, this problem may be solved. Yb3+ may be stimulated to the 4I11/2 level non-radiatively and has a high absorption cross-section at 0.98 μm [24]. As it is evident in Fig. 8f, the peak emission cross-sections in EYDWAs are lower compared to crystalline hosts. Nevertheless, the broad smooth emission spectrum gives a relatively flat gain over a wide wavelength.

The main goal of the fabrication of Er:Yb integrated optical amplifier is optimizing the amplifier concerning the threshold power at a pump wavelength of 1.48 μm and 0.98 μm. The amplifier gain is probed as a function of amplifier length, background loss, propagation loss, scattering loss, Erbium concentration, and pump power for a given waveguide geometry and Erbium distribution. With the sol–gel method, Erbium clustering can be prevented, leading to the upconversion phenomenon and reducing the gain. Therefore, it causes uniform Er doping in the substrate. Moreover, P2O5 is added to the solvents due to the homogeneous distribution without clustering. The concentration of 1 × 1020/cm3 was achieved for Er. However, efficient pumping needs a relatively high Erbium concentration, possibly even for optimized amplifiers. Nevertheless, ion-ion interactions significantly reduce the gain at such high Erbium concentrations. This method is the most common way to improve pump efficiency and achieve efficient high-gain Erbium-doped integrated optical amplifiers.

5 Conclusion

This work presented the fabrication of an Erbium-ytterbium-doped waveguide amplifier (EYDWA) as an integrated optical amplifier device that boosts the intensity of optical signals at communication wavelengths. Furthermore, spin-coating in the fabrication development resulted in the sol–gel films. Using the SiO2–Er (NO3)3 5H2O–Yb (NO3)3 9H2O–Al (NO3)3 9H2O system, it was found that the optimum composition for 1 × 1020/cm3 (1.2 wt%) concentration was 4.8gr TEOS–0.5gr HCL–0.5gr P2O5–0.2gr Al2O3–0.4gr Er–0.2gr Yb. The waveguide fabrication was done using the Ag–Na ion exchange process. In such a process, ions trade between the core material and the molten salt. Then, the waveguide is immersed. A high-power laser at 1.48 μm was used as the pump source, and an adjustable laser at 1.54 μm was used as the signal laser. The final net gain of around 2 dB at a wavelength of approximately 1.54 μm in Er:Yb:Al: PSG was achieved. Obtaining the best time for spin coating on the sample, appropriate temperature and humidity control are the main challenges to acquire high-quality sol–gel films. Moreover, the pumped light passing through the waveguide can be reflected and reduce the pumping efficiency. Disturbances in the refractive index due to waveguide roughness can cause such reflections. Thus, it is mandatory to create a pattern with continuous smooth edges. Optimizing material composition and device geometries are among future research directions to increase the net gain in Er-doped optical amplifiers. Also, the fabrication of coupled resonator optical waveguide (CROW) structures in the active substrate, doped with erbium ions, can be utilized for integrated optics applications.