Enhancement of the Electroluminescence from Amorphous Er-Doped Al2O3 Nanolaminate Films by Y2O3 Cladding Layers Using Atomic Layer Deposition

Amorphous Al2O3-Y2O3:Er nanolaminate films are fabricated on silicon by atomic layer deposition, and ~1530 nm electroluminescence (EL) is obtained from the metal-oxide-semiconductor light-emitting devices based on these nanofilms. The introduction of Y2O3 into Al2O3 reduces the electric field for Er excitation and the EL performance is significantly enhanced, while the electron injection of devices and the radiative recombination of doped Er3+ ions are not impacted. The 0.2 nm Y2O3 cladding layers for Er3+ ions increase the external quantum efficiency from ~3% to 8.7% and the power efficiency is increased by nearly one order of magnitude to 0.12%. The EL is ascribed to the impact excitation of Er3+ ions by hot electrons, which stem from Poole-Frenkel conduction mechanism under sufficient voltage within the Al2O3-Y2O3 matrix.


Introduction
Rare earth (RE) ions are generally efficient luminescence centers in various matrices. Nowadays diverse RE-doped insulating materials have been developed for the applications in solid state lasers and phosphors [1,2]. Erbium (Er) ions are one of the most researched luminescence centers due to their near-infrared (NIR) 1.53 µm emission which coincides with the window of optical telecommunication [3,4] Aiming for the realization of Si-integrated optoelectronics, the 1.53 µm electroluminescence (EL) from Er 3+ ion has been researched extensively in many materials, including SiO x , SiN x , TiO 2 and ZnO [5][6][7][8]. However, the efficiencies of the devices based on these aforementioned materials are still far from practical application, due to the limitations in doping tolerance and excitation efficiency. Y 2 O 3 is one of the attractive doping hosts for RE ions as the substitution of other RE 3+ ions in Y 2 O 3 is quite easy without charge compensation and severe lattice distortion. In addition, Y 3+ ions are not luminescent and Y 2 O 3 has a large bandgap (5.8 eV) and high stability [9,10]. In our previous study, Al 2 O 3 has been proved to be a suitable matrix for the excitation of RE 3+ ion to realize the EL emissions but the doping concentration is still limited [11][12][13]. Therefore, using Y 2 O 3 as a cladding layer in Er-doped Al 2 O 3 could utilize the merits of both oxides, the Er-clustering and resultant concentration quenching could be reduced while the optical-active Er 3+ ions can be excited more effectively [14].
In this work, we fabricate the metal-oxide-semiconductor light-emitting devices (MOSLEDs) based on the amorphous Al 2 O 3 -Y 2 O 3 :Er nanolaminate films, which are deposited using atomic layer deposition (ALD). Due to the unique growth mechanism based on the successive self-limiting gas-surface reactions, ALD realizes the precise control of the thickness of different compositions with excellent homogeneity [15,16]. By alternating deposition sequence of Al 2 O 3 and Y 2 O 3 , nanolaminate Al 2 O 3 -Y 2 O 3 :Er films with interlayers of different thicknesses are fabricated. Under sufficient forward bias, such devices exhibit 1530 nm emissions originating from the infra-4f transitions of Er 3+ ions. Inserting of the Y 2 O 3 cladding layers increases the external quantum efficiency (EQE) from 3% to 8.7% and almost upgrades the power efficiency (PE) by one order of magnitude, while the excitation and recombination of the Er 3+ ions are not affected. We believe that this work contributes to the development of silicon-based light sources for integrated optoelectronic applications.

Experimental
The luminescent Al 2 O 3 -Y 2 O 3 :Er nanolaminates were grown on <100>-oriented ntype silicon (2-5 Ω·cm) using the thermal ALD system (NanoTech Savannah 100, Cambridge, MA, USA). The growth chamber was first evacuated to a base pressure of 0.3 Torr. Trimethylaluminum [TMA, Al(CH 3 )], Y(THD) 3 and Er(THD) 3 (THD = 2,2,6,6-tetramethyl-3,5-heptanedionate) were used as the precursors for Al 2 O 3 , Y 2 O 3 and Er 2 O 3 , respectively, with ozone acting as the oxidant. During the ALD process, the Al precursor was maintained at room temperature (RT), while Y and Er precursors were maintained at 180 • C and 190 • C, respectively. The precursor delivery lines were heated at 190 • C. N 2 was used as the carrier and purge gas with a flow rate of 20 sccm. The pulse time for Al and RE precursors are 0.015 s and 2 s, respectively. One growth cycle consists of one precursor pulse, the 5 s N 2 purge, a 1.8 s ozone pulse, and the 9 s N 2 purge. Based on the former research, the Er dopant cycles are fixed at 2, which are preferable concerning both the efficient doping and the absence of RE clustering [11,13,17,18] are used to calculate the corresponding dopant amount of Er 3+ ions. After the deposition, the films were annealed at 800 • C in N 2 atmosphere for 1 h to enable activation of the dopants. Subsequent device procedures were as previously mentioned [12,13,17,18], resulting in the multilayer-structured MOSLEDs of ZnO:Al/TiO 2 -Al 2 O 3 /Al 2 O 3 -Y 2 O 3 :Er/Si/Al. The top ZnO:Al electrodes were lithographically patterned into 0.5 mm circular dots, while the TiO 2 -Al 2 O 3 nanolaminates were used to enhance the operation stability of the devices.
The film thickness was measured by an ellipsometer with a 632.8 nm He-Ne laser at an incident angle of 69.8 • . The phase and the crystal structure of the films were identified by an X-ray diffractometer (XRD, D/max 2500/pc, Rigaku) using the Cu Kα radiation. To activate EL from the MOSLEDs, appropriate forward bias was applied with the negative voltage connecting to the n-Si substrates. EL and Current-Voltage (I-V) characteristics were recorded by a Keithley 2410 SourceMeter. The EL signal was collected by a 0.5 m monochromator and detected by an InGaAs detector connected to a Keithley 2010 multimeter. The absolute EL power from the device surface was measured using a calibrated Newport 1830-C optical power-meter with an 818-IR Sensor. All measurements were performed at RT.

Results and Discussion
The XRD patterns of all the Al 2 O 3 -Y 2 O 3 :Er films annealed at 800 • C confirm that the nanolaminates are amorphous, one representative XRD pattern from the nanolaminate using 3 nm Al 2 O 3 interlayers and 0.2 nm Y 2 O 3 cladding layers is shown in Figure 1. The Al 2 O 3 layers are not crystalized at such a relatively low temperature of 800 • C that beneficial for the EL performance from RE-doped Al 2 O 3 films, while the crystallization of the sub-nanometer Y 2 O 3 layers is restricted [17,20]. The amorphous nanolaminate films are quite smooth under the observation of scanning electron microscope, with a root-square roughness of only 0.56-0.7 nm scanned by the atomic force microscopy (AFM, Dimension Icon, Bruker) [21]. monochromator and detected by an InGaAs detector connected to a Keithley 2010 multimeter. The absolute EL power from the device surface was measured using a calibrated Newport 1830−C optical power-meter with an 818−IR Sensor. All measurements were performed at RT.

Results and Discussion
The XRD patterns of all the Al2O3-Y2O3:Er films annealed at 800 °C confirm that the nanolaminates are amorphous, one representative XRD pattern from the nanolaminate using 3 nm Al2O3 interlayers and 0.2 nm Y2O3 cladding layers is shown in Figure 1. The Al2O3 layers are not crystalized at such a relatively low temperature of 800 °C that beneficial for the EL performance from RE-doped Al2O3 films, while the crystallization of the sub-nanometer Y2O3 layers is restricted [17,20]. The amorphous nanolaminate films are quite smooth under the observation of scanning electron microscope, with a root-square roughness of only 0.56-0.7 nm scanned by the atomic force microscopy (AFM, Dimension Icon, Bruker) [21].   Figure 2b shows the NIR EL spectra of MOSLEDs based on the Al2O3−Y2O3:Er films of different Y2O3 cladding layers (with the thickness of x nm). The EL peaks centered at ~1530 nm correspond to the infra-4f 4 I13/2→ 4 I15/2 transitions of the Er 3+ ions. The presence of other shoulder peaks is ascribed to the splitting levels associated with the Stark effect [22]. These EL peaks are similar in positions and sharps in the Al2O3−Y2O3:Er nanofilms with different Y2O3 cladding layers, thus the incorporation of Y2O3 cladding layers imposes no apparent effect on the Er 3+ intra-4f transitions. In comparison with the Er-emissions from different matrices, the spectra also confirm that the Al2O3−Y2O3:Er films are amorphous due to the absence of companion peaks [20,23].   Figure 2b shows the NIR EL spectra of MOSLEDs based on the Al 2 O 3 -Y 2 O 3 :Er films of different Y 2 O 3 cladding layers (with the thickness of x nm). The EL peaks centered at~1530 nm correspond to the infra-4f 4 I 13/2 → 4 I 15/2 transitions of the Er 3+ ions. The presence of other shoulder peaks is ascribed to the splitting levels associated with the Stark effect [22]. These EL peaks are similar in positions and sharps in the Al 2 O 3 -Y 2 O 3 :Er nanofilms with different Y 2 O 3 cladding layers, thus the incorporation of Y 2 O 3 cladding layers imposes no apparent effect on the Er 3+ intra-4f transitions. In comparison with the Er-emissions from different matrices, the spectra also confirm that the Al 2 O 3 -Y 2 O 3 :Er films are amorphous due to the absence of companion peaks [20,23].   [13,18,24,25]. Beneath the threshold electric field, the defect states contribute to the low background currents. In the working voltage region, the currents increase exponentially until breakdown. The difference on the current injection will be discussed afterwards concerning the conduction mechanism. All the EL intensities also present an exponential relationship with the applied voltages until reaching saturation. The MOSLED with 0.2 nm Y2O3 cladding layers presents the highest EL intensity, with the lowest threshold voltage and the highest injection current. The devices with thicker Y2O3 layers underperform in EL intensities and the injection currents are restricted. Despite the uncertainty brought about by the device preparation, Y2O3 cladding layers with suitable thickness can effectively enhance the current injection and promote the EL emissions from these Al2O3−Y2O3:Er MOSLEDs. As previously reported, the incorporation of Y 3+ ion makes the crystal field around Er 3+ ions less symmetric and introduces distortion in the crystal field, moreover the Er 3+ ions are dispersed to suppress the concentration quenching, resulting in the enhanced radiation probability [26]. Therefore, the 0.2 nm Y2O3 layers act as cladding layers that inhibit the Er-clustering, while the thicker Y2O3 layers inhibit the electron injection, which is ascribed to the higher dielectric index of Y2O3 and the disruptive interfaces among Al2O3 and Y2O3 interlayers.   [13,18,24,25]. Beneath the threshold electric field, the defect states contribute to the low background currents. In the working voltage region, the currents increase exponentially until breakdown. The difference on the current injection will be discussed afterwards concerning the conduction mechanism. All the EL intensities also present an exponential relationship with the applied voltages until reaching saturation. The MOSLED with 0.2 nm Y 2 O 3 cladding layers presents the highest EL intensity, with the lowest threshold voltage and the highest injection current. The devices with thicker Y 2 O 3 layers underperform in EL intensities and the injection currents are restricted. Despite the uncertainty brought about by the device preparation, Y 2 O 3 cladding layers with suitable thickness can effectively enhance the current injection and promote the EL emissions from these Al 2 O 3 -Y 2 O 3 :Er MOSLEDs. As previously reported, the incorporation of Y 3+ ion makes the crystal field around Er 3+ ions less symmetric and introduces distortion in the crystal field, moreover the Er 3+ ions are dispersed to suppress the concentration quenching, resulting in the enhanced radiation probability [26]. Therefore, the 0.2 nm Y 2 O 3 layers act as cladding layers that inhibit the Er-clustering, while the thicker Y 2 O 3 layers inhibit the electron injection, which is ascribed to the higher dielectric index of Y 2 O 3 and the disruptive interfaces among Al 2 O 3 and Y 2 O 3 interlayers.  Figure 3b shows the dependence of EL intensities on the injection currents for the Al2O3−Y2O3:Er MOSLEDs with different Y2O3 cladding layers. The threshold currents for all the devices are ~0.2 μA, the EL intensities and the injection currents present linear relationship. In comparison, the devices with different Y2O3 cladding layers exhibit similar EL, which increases more prominently than that based on the Al2O3:Er film. Y2O3 also lessens the saturation of EL intensities at higher injection currents. Considering the thick interlayers among RE layers (the Al2O3 interlayers with the thickness of at least 3 nm), the acceleration distance for hot electrons is sufficient; therefore the enhanced EL should result from more optical-active Er dopants as the Er 3+ ions disperse into the Y2O3 layers and the Er-clustering is suppressed.
In our previously reported MOSLEDs based on RE-doped Al2O3, the RE-related EL is triggered by the direct impact excitation of the RE ions by the hot electrons accelerated under sufficient bias voltages [12,25]. As the I-V characterization are accordingly comparable, it is rational to ascribe the NIR EL from these Al2O3−Y2O3:Er MOSLEDs to the same mechanism. Considering the high bandgap of the matrix materials and the barrier for electrons to be injected from the Si substrates into the conduction band of the oxides, the current conduction of these MOSLEDs has been ascribe to the Poole-Frenkel (P-F) mechanism, in which the electrons hop via the defect-related trap states under sufficient electrical field [11,26,27]. In simplicity, the plot of the ln(J/E) versus E 1/2 presents linear relationship in P-F conduction mechanism, where J and E are the current density and the electric field, respectively [28,29]. Figure 3c shows the plots of the I-V characteristics derived from Figure 3a, the electrical fields across the luminescent films are roughly calculated in terms of electrostatics [30], and the well-defined linearity is established for all the MOSLEDs in the EL-enabling region. Thus the electron transport through the Al2O3−Y2O3:Er nanofilms is governed by the P-F mechanism. These electrons tunnel into the conduction band of  Figure 3b shows the dependence of EL intensities on the injection currents for the Al 2 O 3 -Y 2 O 3 :Er MOSLEDs with different Y 2 O 3 cladding layers. The threshold currents for all the devices are~0.2 µA, the EL intensities and the injection currents present linear relationship. In comparison, the devices with different Y 2 O 3 cladding layers exhibit similar EL, which increases more prominently than that based on the Al 2 O 3 :Er film. Y 2 O 3 also lessens the saturation of EL intensities at higher injection currents. Considering the thick interlayers among RE layers (the Al 2 O 3 interlayers with the thickness of at least 3 nm), the acceleration distance for hot electrons is sufficient; therefore the enhanced EL should result from more optical-active Er dopants as the Er 3+ ions disperse into the Y 2 O 3 layers and the Er-clustering is suppressed.
In our previously reported MOSLEDs based on RE-doped Al 2 O 3 , the RE-related EL is triggered by the direct impact excitation of the RE ions by the hot electrons accelerated under sufficient bias voltages [12,25]. As the I-V characterization are accordingly comparable, it is rational to ascribe the NIR EL from these Al 2 O 3 -Y 2 O 3 :Er MOSLEDs to the same mechanism. Considering the high bandgap of the matrix materials and the barrier for electrons to be injected from the Si substrates into the conduction band of the oxides, the current conduction of these MOSLEDs has been ascribe to the Poole-Frenkel (P-F) mechanism, in which the electrons hop via the defect-related trap states under sufficient electrical field [11,26,27]. In simplicity, the plot of the ln(J/E) versus E 1/2 presents linear relationship in P-F conduction mechanism, where J and E are the current density and the electric field, respectively [28,29]. Figure 3c shows the plots of the I-V characteristics derived from Figure 3a, the electrical fields across the luminescent films are roughly calculated in terms of electrostatics [30], and the well-defined linearity is established for all the MOSLEDs in the EL-enabling region. Thus the electron transport through the Al 2 O 3 -Y 2 O 3 :Er nanofilms is governed by the P-F mechanism. These electrons tunnel into the conduction band of oxides and transport by hopping among trap states in the Al 2 O 3 -Y 2 O 3 nanolaminates under sufficient electric field. Certain parts of the electrons are accelerated therein and become hot electrons that excite the Er 3+ ions by inelastic impact, the subsequent recombination gives rise to the characteristic EL emissions. Apparently, the Y 2 O 3 cladding layers decrease the working electric field prominently. As mentioned in the discussion on the I-V characteristics, the Y 2 O 3 cladding layers increase both the injection currents and EL intensities, we conclude that ultrathin Y 2 O 3 layers introduce defect sites within Al 2 O 3 , via which electrons transport by the P-F hopping mechanism; therefore the injection currents are enhanced. Since the accelerated electrons collide with the doped Er 3+ ions and contribute to the NIR EL, adding the aforementioned crystal field distortion and cluster dispersion effects of the Y 2 O 3 on Er 3+ ions, the EL performance are greatly enhanced by the Y 2 O 3 cladding layers in Al 2 O 3 films. However, the Y 2 O 3 cladding layers should be thin enough to not impact the carrier transport which could be ascribed to the formation of distinct Al 2 O 3 -Y 2 O 3 interfaces when using thicker Y 2 O 3 cladding layers.
In evaluation of the thickness of Al 2 O 3 interlayers on the EL performance, the dependence of the EL intensities from each dopant cycle on the injection currents for the Al 2 O 3 -Y 2 O 3 :Er MOSLEDs with different Al 2 O 3 interlayers (with the thickness of y nm) are shown in Figure 4a, the thickness of Y 2 O 3 cladding layers is the optimal 0.2 nm. Again, the EL intensities increase almost linearly with the injection currents. The difference on the EL-I-V characteristics among these MOSLEDs with different Al 2 O 3 layers are small (not shown herein), and the increase in EL intensity with the Al 2 O 3 thickness could be ascribed to the less concentration quenching of doped Er 3+ ions together with the longer acceleration distance. When the thickness of Al 2 O 3 declines, the EL intensity decreases greatly due to the cross-relaxation of Er 3+ between adjacent dopant layers when the inter-distance (the Al 2 O 3 thickness) is smaller enough, and the limited acceleration length for the hot electrons to gain energy to excite the Er 3+ ions [18,21,[31][32][33]. Cross-relaxation is a common phenomenon that occurs among the same ions or different ions of similar energy intervals. One ion in the excited state ( 4 I 13/2 in the case of Er 3+ ion) transfer the energy to another one (in the ground state of 4 I 15/2 in this case of Er 3+ ions), excite the latter to higher energy levels ( 4 I 13/2 ) while relaxing itself to lower energy levels ( 4 I 15/2 ) without radiation. The interaction of energy transfer by cross-relaxation could finally disperse the excitation energy through phonons instead of luminescent emissions. Nanomaterials 2023, 13, x FOR PEER REVIEW 6 of 9 oxides and transport by hopping among trap states in the Al2O3−Y2O3 nanolaminates under sufficient electric field. Certain parts of the electrons are accelerated therein and become hot electrons that excite the Er 3+ ions by inelastic impact, the subsequent recombination gives rise to the characteristic EL emissions. Apparently, the Y2O3 cladding layers decrease the working electric field prominently. As mentioned in the discussion on the I−V characteristics, the Y2O3 cladding layers increase both the injection currents and EL intensities, we conclude that ultrathin Y2O3 layers introduce defect sites within Al2O3, via which electrons transport by the P-F hopping mechanism; therefore the injection currents are enhanced. Since the accelerated electrons collide with the doped Er 3+ ions and contribute to the NIR EL, adding the aforementioned crystal field distortion and cluster dispersion effects of the Y2O3 on Er 3+ ions, the EL performance are greatly enhanced by the Y2O3 cladding layers in Al2O3 films. However, the Y2O3 cladding layers should be thin enough to not impact the carrier transport which could be ascribed to the formation of distinct Al2O3-Y2O3 interfaces when using thicker Y2O3 cladding layers.
In evaluation of the thickness of Al2O3 interlayers on the EL performance, the dependence of the EL intensities from each dopant cycle on the injection currents for the Al2O3−Y2O3:Er MOSLEDs with different Al2O3 interlayers (with the thickness of y nm) are shown in Figure 4a, the thickness of Y2O3 cladding layers is the optimal 0.2 nm. Again, the EL intensities increase almost linearly with the injection currents. The difference on the EL−I−V characteristics among these MOSLEDs with different Al2O3 layers are small (not shown herein), and the increase in EL intensity with the Al2O3 thickness could be ascribed to the less concentration quenching of doped Er 3+ ions together with the longer acceleration distance. When the thickness of Al2O3 declines, the EL intensity decreases greatly due to the cross-relaxation of Er 3+ between adjacent dopant layers when the inter-distance (the Al2O3 thickness) is smaller enough, and the limited acceleration length for the hot electrons to gain energy to excite the Er 3+ ions [18,21,[31][32][33]. Cross-relaxation is a common phenomenon that occurs among the same ions or different ions of similar energy intervals. One ion in the excited state ( 4 I13/2 in the case of Er 3+ ion) transfer the energy to another one (in the ground state of 4 I15/2 in this case of Er 3+ ions), excite the latter to higher energy levels ( 4 I13/2) while relaxing itself to lower energy levels ( 4 I15/2) without radiation. The interaction of energy transfer by cross-relaxation could finally disperse the excitation energy through phonons instead of luminescent emissions. In the RE-doped Al2O3 MOSLEDs, the Al2O3 sublayer thickness affects the cross relaxation between excited RE ions, and the acceleration distance for injected electrons.  In the RE-doped Al 2 O 3 MOSLEDs, the Al 2 O 3 sublayer thickness affects the cross relaxation between excited RE ions, and the acceleration distance for injected electrons. Figure 4b shows the dependence of the integrated 1530 nm EL intensity per Er cycle on the Nanomaterials 2023, 13, 849 7 of 9 thickness of Al 2 O 3 interlayers under different injection currents. Under all these injection currents, with the increase in the thickness of Al 2 O 3 interlayers, the contribution of single Er cycle to the EL intensity firstly increases and then saturates as the Al 2 O 3 interlayer thickness reaches 3 nm. This is still in consistency with the common characteristic for the luminescent RE 3+ ions in Al 2 O 3 matrix that the distance for the presence of non-radiative interaction and adequate electron acceleration is around 3 nm [11,12,21,33].
Considering the total EL intensity from the MOSLEDs with Al 2 O 3 interlayers of different thicknesses (marked as y nm here) shown in Figure 5a, the device using 3 nm Al 2 O 3 interlayers presents the optimal emission intensity in the operation range, with the highest power density of 4.6 mW/cm 2 . External efficiency is widely used to evaluate LED performance. Figure 5b shows the EQE and PE of these MOSLEDs based on different Al 2 O 3 -Y 2 O 3 :Er nanolaminate films. These EL efficiencies sustain a broad maximum, and fall down at higher currents. Generally, the EQE of the devices with 2-3 nm Al 2 O 3 interlayers are the highest. As aforementioned, this phenomenon could be ascribed to the sufficient distance for electron acceleration and suppression of the cross-relaxation among adjacent Er 2 O 3 dopant layers. The Y 2 O 3 cladding layers somewhat decrease this critical distance which is beneficial for higher doping concentrations. The optimal device with 3/0.2 nm Al 2 O 3 /Y 2 O 3 interlayers achieves the maximum EQE of 8.7% and a corresponding PE of 0.12%. These values are comparable to our Yb 2 O 3 :Er MOSLEDs but with lowered working voltages. In comparison, the control Al 2 O 3 :Er MOSLED presents only an EQE of 3% and a PE of 0.014%, much lower than the Al 2 O 3 -Y 2 O 3 :Er MOSLEDs. The Y 2 O 3 cladding layers with suitable thickness enhance the efficiencies from the MOSLEDs to a great extent. We have found that by using a thicker luminescent layer, the efficiency of the Al 2 O 3 :RE MOSLED might be further increased to higher than 10% [25,34]. These efficiencies are superior to that from Si-based EL devises in literature, thus further optimization of the luminescent Al 2 O 3 -Y 2 O 3 :Er nanolaminates would supply potential light source for the applications in Si-based optoelectronics. Nanomaterials 2023, 13, x FOR PEER REVIEW 7 of 9 thickness of Al2O3 interlayers under different injection currents. Under all these injection currents, with the increase in the thickness of Al2O3 interlayers, the contribution of single Er cycle to the EL intensity firstly increases and then saturates as the Al2O3 interlayer thickness reaches 3 nm. This is still in consistency with the common characteristic for the luminescent RE 3+ ions in Al2O3 matrix that the distance for the presence of non-radiative interaction and adequate electron acceleration is around 3 nm [11,12,21,33]. Considering the total EL intensity from the MOSLEDs with Al2O3 interlayers of different thicknesses (marked as y nm here) shown in Figure 5a, the device using 3 nm Al2O3 interlayers presents the optimal emission intensity in the operation range, with the highest power density of 4.6 mW/cm 2 . External efficiency is widely used to evaluate LED performance. Figure 5b shows the EQE and PE of these MOSLEDs based on different Al2O3−Y2O3:Er nanolaminate films. These EL efficiencies sustain a broad maximum, and fall down at higher currents. Generally, the EQE of the devices with 2-3 nm Al2O3 interlayers are the highest. As aforementioned, this phenomenon could be ascribed to the sufficient distance for electron acceleration and suppression of the cross-relaxation among adjacent Er2O3 dopant layers. The Y2O3 cladding layers somewhat decrease this critical distance which is beneficial for higher doping concentrations. The optimal device with 3/0.2 nm Al2O3/Y2O3 interlayers achieves the maximum EQE of 8.7% and a corresponding PE of 0.12%. These values are comparable to our Yb2O3:Er MOSLEDs but with lowered working voltages. In comparison, the control Al2O3:Er MOSLED presents only an EQE of 3% and a PE of 0.014%, much lower than the Al2O3−Y2O3:Er MOSLEDs. The Y2O3 cladding layers with suitable thickness enhance the efficiencies from the MOSLEDs to a great extent. We have found that by using a thicker luminescent layer, the efficiency of the Al2O3:RE MOSLED might be further increased to higher than 10% [25,34]. These efficiencies are superior to that from Si-based EL devises in literature, thus further optimization of the luminescent Al2O3−Y2O3:Er nanolaminates would supply potential light source for the applications in Si-based optoelectronics.

Conclusions
In summary, significantly enhanced~1530 nm NIR EL emissions are achieved from the MOSLEDs based on the amorphous Al 2 O 3 :Er nanolaminate films by the insertion of cladding Y 2 O 3 sub-nanolayers, which are fabricated by ALD on Si substrates. The Y 2 O 3 cladding layers reduce the threshold electric field for excitation and increase the radiative possibility of doped Er 3+ ions, resulting in improved EL performance. The Al 2 O 3 -Y 2 O 3 :Er MOSLEDs with 0.2 nm Y 2 O 3 and 3 nm Al 2 O 3 interlayers present an EQE of 8.7% and a corresponding PE of 0.12%, which are much higher than that of the counterpart without Y 2 O 3 cladding layers. The incorporation of Y 2 O 3 does not change the electron injection mode under sufficient electric field that conforms to P-F mechanism, the resultant energetic electrons trigger the impact-excitation of Er 3+ ions and subsequent EL emissions. The strategy of Y 2 O 3 -cladding by ALD can be employed to improve the EL performance from LEDs based on RE-doped oxides.