RU2240632C2 - Photon integrated circuit manufacturing process - Google Patents

Abstract

FIELD: integrated circuit manufacture.

SUBSTANCE: proposed manufacturing process for photon integrated circuit that has complicated semiconductor structure with quantum well region includes irradiation of structure using photon source for generating defects, photon energy E being not lower than displacement energy E _d of one element of compound semiconductor. Then structure is subjected to annealing to encourage mixture of quantum wells. Preferable radiation source is plasma generated with aid of electronic cyclotron resonance system. Structure can be masked in any way to ensure its selective mixing by controlling exposure of structure sections.

EFFECT: ability of mixing quantum wells by structure irradiation with photon radiation source.

16 cl, 14 dwg

Description

BACKGROUND OF THE INVENTION

The monolithic integration of several optoelectronic devices in optoelectronic integrated circuits (OEIS) and photon integrated circuits (FIS) is of significant interest in connection with the development of communication systems.

In OEIS, optical devices, such as lasers and electronic devices, such as transistors, are integrated in a single crystal (chip) to achieve high performance, since the tight packing of the devices minimizes the stray reactance of electrical connections.

FIS are a separate type of OEIS, characterized by the absence of electrical components, as well as the fact that the connection or connection between optoelectronic and / or photonic devices is carried out only by means of photons. The driving force behind the development of FIS is the desire to simplify (improve) next-generation optical communication lines, network architectures and switching systems, for example, multi-channel wavelength division multiplexing (MRD) and high-speed time division multiplexing (MRI) systems. In addition to reducing cost, reducing size and increasing complex stability, the main advantage of FIS is that all internal connections between individual optical fiber optic devices are precisely and constantly located (aligned) with respect to each other, since the optical fibers are formed by lithographic method.

In the process of integration, complex devices are built from components of various functional purposes, for example, light sources, optical fibers, modulators and detectors. To create each component with optimal characteristics, different material structures are required. Therefore, to create OEIS and FIS, you must be able to change the width (energy) of the forbidden zone and the refractive index of materials. Several methods have been developed for this, including growing and recrystallizing, selective epitaxy or growing on a patterned substrate, and mixing quantum wells (CQW).

The method of growing and recrystallization is complex and expensive and involves the growth, etching and recrystallization of the layers of a quantum well (QW) in selected areas of bulk material. The same upper shell is built up on these multilayer structures, but they have different active regions. The disadvantage of this approach is the mismatch of the optical propagation coefficient and the mismatch of the dimensions of the fiber at the recrystallization interface. In addition, this process provides low yield (Efficiency) and productivity, which leads to an increase in the cost of the final product.

The method of selective growth is based on differences in the composition and thickness of the epitaxial layer grown through the mask to achieve spatial selectivity when changing the width of the energy gap. Before embarking on epitaxial growth, the substrate is patterned with a dielectric mask, for example of SiO ₂ , in which slits of different widths are formed. The growth rate in open areas depends on the width of the hole and the patterning of the mask. In places where there is a dielectric coating, cultivation is impossible. However, within a certain distance, surface migration of particles to the nearest hole across the mask can occur. The advantage of this approach is the reduction in the total number of processing steps, which allows one to form essentially optimal sections of multiple quantum wells (MQWs) of the laser and modulator in one epitaxial growth step. This process gives good results provided that the set of parameters is precisely controlled, but in general it is difficult to control. In addition, this method provides a low spatial resolution of about 100 μm, which is why the passive section usually has relatively high losses.

The QW method is based on the fact that the QW is by its nature a metastable system due to the high gradient of atomic concentration across the interface between the QW and the barrier. Nevertheless, this allows one to change the energy gap of the QW structures in the selected regions by mixing the QW with barriers to form “fused” semiconductors. This method allows, after growing, to effectively carry out horizontal integration of sections with different widths of the energy forbidden zone, refractive index, and optical absorption within the same epitaxial layers.

The CQW method has been widely recognized and widely used in the field of construction of optoelectronic integrated circuits, which include, for example, electroabsorption modulators with an adjustable energy gap, lasers with an adjustable energy gap, low-loss fibers for the internal connection of OEIS or FIS components, large integrated resonators volume for lasers with reduced line width, single-frequency lasers with distributed Bragg reflector (RBO), lasers with synchronization modes, nonabsorbing mirrors, amplification or phase diffraction gratings for distributed feedback lasers (ROS), superluminescent diodes (SLDs), polarization insensitive QW modulators and amplifiers, and multifrequency lasers.

Recent studies have focused on improving the CQW method using approaches such as pure vacancy-stimulated disordering (BWRM), laser-stimulated disordering (LSR), and impurity-stimulated disordering (RPS). Each of these methods of SCW has its own advantages and disadvantages.

The BPVR method involves the deposition of a dielectric coating material on materials with QW and subsequent high-temperature annealing to stimulate the generation of vacancies from the dielectric coating in materials with QW and, therefore, to enhance mixing in selected areas.

For example, it is known that the presence of SiO ₂ in GaAs-AlGaAs-based QW materials stimulates the back diffusion (yield diffusion) of Ga atoms during annealing, which leads to the formation of group III vacancies in the QW material. The thermal stress at the interface between the GaAs and SiO ₂ layers plays an important role. The thermal expansion coefficient of GaAs is ten times greater than that of SiO ₂ . During high-temperature annealing, bonds in a highly porous SiO ₂ layer deposited by plasma-stimulated chemical vapor deposition (PSCOF) can be destroyed due to the presence of a voltage gradient between GaAs and the SiO ₂ film. Thus, the reverse diffusion of Ga contributes to the weakening of the tensile stress in GaAs. These Ga vacancies propagate further into the QW structure and increase the interdiffusion rate of Ga and Al, and thus lead to QW. At the end of the mixing process, the energy forbidden zone in the material with QW expands, and the refractive index decreases.

To achieve the selectivity of this method, an SrF ₂ layer can be used, which prevents the back diffusion of Ga, which makes it possible to suppress the QW process. This method has been successfully applied in the formation of such devices as a multi-frequency laser with an adjustable energy gap and a multi-channel light guide photodetector.

Although the BWRV method has been successfully applied in the GaAs / AlGaAs system, in the InGaAs / InGaAsP systems, it provides low reproducibility. In addition, it was found that, due to the low heat resistance of InGaAs / InGaAsP materials, the BHWR process, which requires high-temperature annealing, provides low energy band gap selectivity in InGaAs / InGaAsP based QW structures.

Laser-stimulated disordering (LSR) is a promising QW process to achieve disordering in materials with InGaAs / InGaAsP based QWs due to the low heat resistance of these materials. According to the method of photoabsorption-induced disordering (FASR), the radiation of a laser generating a continuous wave (HB) is absorbed in the QW regions, causing heat release and thermally stimulated mixing. Although the resulting material has high optical and electrical characteristics, the spatial selectivity of this method is limited to about 100 μm due to horizontal overflow. The modified FASR method, called pulsed FASR (I-FASR), uses high-energy Q-switched laser pulses with AID (yttrium aluminum garnet) with Nd to irradiate InP-based material. The absorption of pulses leads to damage to the crystal lattice and an increase in the density of point defects. During high-temperature annealing, these point defects diffuse in the QW, which leads to an increase in the mixing rate of the QW. Although the I-FASR method can provide a spatial resolution of over 1.25 microns and the possibility of direct recording, “mixed” materials are of poor quality due to the formation of extended defects.

Of all the methods of QW, impurity-stimulated disordering (PSR) is the only process that requires the introduction of impurities in materials with QW to implement the mixing process. These impurities can be introduced using a focused ion beam, diffusion of the impurity in the furnace, and ion implantation.

RPS is a relatively simple and highly reproducible mixing process. It allows you to provide high spatial resolution necessary for the integration of small-sized devices, and to control the displacement of the energy gap due to implantation parameters. This method is usually used to provide horizontal electrical and optical confinement in semiconductors, which allows one to achieve a low threshold current and operate on the same transverse mode. In addition, the RPS process is of significant interest for the integration of MRL systems, for example, for the formation of multi-frequency laser sources, low-loss optical fibers, modulators, and even detectors.

It is well known that the effect of RPS consists of two stages. At the first stage, impurities are implanted in the material with QW. Then, the material is annealed to stimulate the diffusion of impurities and point defects into QWs and barriers, and, therefore, the interdiffusion of the main elements between QWs and barriers. It is believed that in the QW system based on InGaAs / InGaAsP, the interdiffusion of Group V elements from the barrier to the well, leading to a blue shift of the energy band gap, is caused by the diffusion of point defects that arise during implantation, spontaneous interdiffusion at elevated temperatures (thermal displacement), and diffusion of implanted particles.

In the process of implantation, impurities, as well as point defects, such as group III vacancies and interstitial atoms, are introduced into the material in selected areas. Diffusion of these point defects and impurities at elevated temperature increases the rate of interdiffusion between the QW and barriers and, therefore, stimulates mixing after annealing. Under the influence of injected impurities, the QW composition profile changes from square to parabolic. As a result, at the end of the interdiffusion process, the local energy band gap increases, and the corresponding refractive index decreases.

Using the PSR method, it is possible to achieve mixing in a selected area of the plate using an implantation mask of variable thickness of SiO ₂ . However, this method involves numerous stages of lithography and etching, which complicates the production process.

An article entitled “Integration process for photonic integrated circuits using plasma damage induced layer intermixing”. Electronics Letters, Vol. 31, 449, 1995, BS Ooi, A.S. Wuss and JH Marsh describe the mixing of quantum wells based on damage caused by bombardment by active ions. According to this method, H ₂ plasma treatment with a high RF power, and therefore with a high damage ability, was used to introduce point defects on the surface of the samples, which were then annealed to ensure diffusion of point defects in the QW region. Plasma exposure was performed using a reactive ion etching apparatus (RIT) with parallel plates. Similarly, in the article “Disordering Stimulated by Implantation of Ar ⁺⁴ 'Ions by Plasma Immersion in the Structure of Multiple Voltage InGaAsP Based Quantum Wells” (Plasma Immersion Ar ⁺ Ion Implantation Induced Disorder in Strained InGaAsP Multiple Quantum Wells), LM Lam et al ., Electronic Letters, t. 34, No. 8, April 16, 1998, disclosed is the process of ion implantation by plasma immersion, which uses the RIT apparatus. According to each of these methods, the QW process is based on damage caused by ion bombardment and requires several cycles to achieve any noticeable displacement of the energy band gap.

The ability to control the energy gap within the III-V semiconductor wafer is a key requirement for the manufacture of monolithic FIS. For the manufacture of such elements of integrated circuits as lasers, modulators, and low-loss fibers, spatial adjustment of the edge of the absorption band of QW structures along the plate is necessary. Although CJW methods have great advantages over growth and recrystallization methods and selective epitaxial growth methods from the point of view of the formation of the energy forbidden zone, the spatial control provided by these methods is indirect and complex.

The tremendous growth of Internet graphics, multimedia services and high-speed data services requires communication owners to quickly and economically increase the capacity of their networks. It is well known that there are three ways to increase capacity, namely, laying new fibers, increasing the bit rate of communication systems, and using wavelength division multiplexing (WDM). Due to the fact that the first method is fraught with problems of high costs and the need for an exclusion band, and the second method has limited growth potential due to internal systemic limitations, the third method is the most attractive because it makes it possible to multiply the network capacity at moderate costs.

SUMMARY OF THE INVENTION

The present invention provides a method for manufacturing a photon integrated circuit containing a complex semiconductor structure having a quantum well region, comprising the steps of irradiating said structure with a photon radiation source to generate defects, the photons having an energy (E) of at least equal to the energy (Its ) the displacement of at least one element of a complex semiconductor, and subsequent annealing of the above structure to stimulate the mixing of quantum wells.

The preferred source of photon radiation is plasma, although other sources of high-energy photons can be used. Plasma sources such as electron cyclotron resonance (ECR) systems, inductively coupled plasma (ICP) systems, a plasma disk excited by a low-energy electron beam in a vacuum, or soft X-ray plasma devices (MRI) are suitable. Other suitable sources of high-energy photon radiation include electric gas discharge devices, excimer lasers, synchrotron devices, pulsed x-ray devices, and gamma radiation sources.

The method may include the step of masking part of the structure to control the degree of radiation damage. So, the mask can be adapted to completely prevent mixing. However, it is preferable to mask the structure for selective, spatially controlled mixing by controlling the exposure of sections of the structure in a predetermined manner.

There are several suitable types of exposure masks, including binary masks, phase masks, grayscale masks, dielectric or metal masks, and photoresist masks. Spatial mixing control is preferably performed using a variable profile mask template. Our internationally filed application PCT / GB01 / 00904, published under No. WO 01/67497, describes a method for patterning a structure by exposing a layer of photoresistive material through a halftone mask (i.e. a mask with different gray levels). The degree of mixing of the quantum wells is controlled with the possibility of spatial selectivity depending on the optical transmission characteristics of the grayscale mask. This method is particularly suitable for use in the present invention, because it allows you to design such a mask that can control the exposure of the structure to radiation with high energy. The photo-resistive mask template can be used on its own to control exposure, or can instead be used to translate the mask template into underlying material, such as a dielectric layer, through an etching process.

The main feature of the present invention is the use of a source of such radiation, which leads to radiation damage to the crystal structure. This requires a strictly defined minimum energy transfer. This energy is called the displacement energy and is denoted by E _C. The transfer of energy exceeding E _C leads to the displacement of the atom, which is either the primary displacement when the ion of the main element is knocked out by one of the bombarding particles, or the secondary displacement when the energy is transferred from the previously knocked out atom of the main element. Below are the values of energy E _C in electron volts (eV) for a number of semiconductor materials from elements of groups III-V:

Although it is known that ultraviolet radiation in vacuum (VUV) can cause damage to semiconductor structures, this phenomenon was previously investigated for how to fix or at least fix this damage by annealing so that these defects do not affect the operation of the device.

This new, economical and simple method can be used for the manufacture of FIS in general, as well as sources for WDM in particular. Using the QW method in accordance with the present invention, it is possible to adjust the energy gap of the QW material so that it assumes different values along the plate. This allows not only integrating monolithic multi-frequency lasers, but also integrating them with modulators and connectors in a single chip (chip). This method can also be used to simplify the manufacturing and design processes of superluminescent diodes (SLD) by expanding the gain spectrum to a maximum after epitaxial growth.

Researchers involved in the integration of photonic devices are currently considering the CQW method as a promising approach only for two-section photonic devices, because otherwise traditional CQW processes become cumbersome and complex. For multisectional integration, researchers prefer to use selective epitaxy, despite its complexity and inefficiency. The present invention shows that the use of SCW is not limited to two-section devices. In addition, this method is more economical and provides higher productivity and profitability compared to selective epitaxy.

Brief Description of the Drawings

We proceed to a detailed description of embodiments of the present invention with reference to the accompanying drawings, in which:

figure 1 is an illustrative diagram of an ECR system;

figure 2 - schematic representation of the multilayer structure of the NQF based on InGaAs / InGaAsP and zone diagram of the structure;

figa and 3B are graphs illustrating the PL spectra (photoluminescence) of samples irradiated with Ar plasma;

4 is a graph illustrating the relationship between the exposure time Ar and the relative displacement of the energy gap for various microwave power values;

5 is a graph illustrating the relationship between the operating temperature and the relative displacement of the energy gap;

6 is a graph illustrating the relationship between the working pressure and the relative displacement of the energy gap;

7 is a schematic view of a sample partially masked by a layer of photoresistive material;

Fig.8 is a graph illustrating the PL spectra obtained on the sample shown in Fig.7, after irradiation with plasma Ar;

Fig.9 is a graph illustrating the relationship between the RF power and the relative displacement of the energy gap;

10 is a graph illustrating the relationship between microwave power and relative displacement of the energy gap;

11 is a graph illustrating the relative displacement of the energy gap for samples irradiated with Ar plasma at different thicknesses of the SiO ₂ mask;

12 is a schematic view of a large area laser with a light guide formed by gain distribution; and

Fig.13 is a graph illustrating the spectra of the device depicted in Fig.12.

DETAILED DESCRIPTION OF THE INVENTION

The present invention is based on the discovery that a more efficient version of a plasma-stimulated QW can be obtained using high-energy radiation, for example, VUV radiation generated in a plasma formed by ECR. This plasma process operates in a completely different mode compared to the above described QW method using plasma-stimulated disordering. By controlling the microwave power in the ECR process, it is possible to generate high-energy radiation, which cannot be obtained using the traditional RIT apparatus. As a result, the blue shift of the CQW obtained using high-energy radiation is much stronger.

In an ECR system, a magnetic field acts in conjunction with an exciting electromagnetic wave. Under the influence of these fields, the electrons make a circular or orbital motion, the radius of which depends on the intensity of the fields, and the frequency of their rotation is called the electron cyclotron frequency. When the frequency of the electromagnetic wave is equal to the cyclotron frequency, phase coherence occurs, leading to a continuous increase in electron energy. Under such conditions, radiative energy transfer of the exciting electromagnetic field to the electrons, which is known as the resonant process, occurs. In this resonant process, electrons in the plasma volume increase their energy due to microwave excitation and transfer energy to the molecules through collisions leading to electron-impact ionization and high-density plasma generation. Highly ionized ions emit photons in a narrow VUV band.

The ECR method of plasma formation is becoming increasingly popular in microelectronic processing, for example, during etching and deposition of thin films, due to its ability to maintain highly dissociated and highly ionized plasmas at relatively low pressures and temperatures. This makes it possible to use such a plasma at a lower pressure (usually from 10 ^-3 to 10 ^-2 Torr) than the traditional HF RIT plasma, and in some cases its degree of ionization reaches about 10%.

Resonance, i.e. peak absorption of energy occurs when the frequency of an alternating electric field coincides with the cyclotron frequency. In this case, the spiral motion of the electrons coincides in phase with the alternating electric field, which provides resonant acceleration of electrons with each polarity reversal. At an industrial microwave frequency of 2.45 GHz, resonance occurs in the presence of a constant magnetic field of 873 G. For resonant energy absorption to occur efficiently, electrons must move in their cyclotron orbits without collisions with neutral particles. Collisions prevent the absorption of energy due to the transfer of energy to neutral particles and randomization (i.e. randomization) of direction. In the general case, collisions lead to inefficient electron cyclotron heating at pressures above 20 mTorr. With an effective discharge in the ECR process, it is possible to achieve a concentration of ions and electrons up to 10 ¹² cm ^-3 . This is approximately 100-1000 times the concentration achievable in the plasma generated by traditional RIT systems.

For processing the samples described below, the Plasma Quest Series II PQM-9187-A type ECR system was used. Its circuit is shown in figure 1. System 10 consists of a microwave generator 11 with a radiation frequency of 2.45 GHz, which enters the resonator 12 of the ECR system through a quartz window. Microwave power varies between 0-1500 watts. The microwave generator is connected to a three-loop regulator, consisting of three impedance matching loops installed in a 13-inch waveguide 13 9 inches (22.86 cm) long. This is done to reduce the reflected power when applying microwave energy to an easily adjustable plasma source or user load. Additional permanent magnets 14 of the Nd-Fe-B type are installed around the perimeter of the reactor and embedded in a grounded upper electrode. The magnetic field created by such a design holds plasma better. Under its action, plasma ions are concentrated in the center of the chamber, moving away from the walls of the chamber, which reduces charge leakage through the walls.

The ECR reactor also contains a sample holder 15 connected to an RF radiation source 16 operating at a frequency of 13.56 MHz. The maximum power provided by the RF generator is 500 watts. The microwave power determines the degree of dissociation and generation of active particles. On the other hand, the RF source biases the substrate and thus controls the flow of ions to the substrate, increasing the directivity of the process.

The In _z Ga _1-z As / In _x Ga _1-x AS _y P _1-y structures used in the examples described below were grown by chemical vapor deposition of organometallic compounds (MOHOPF) on an InP substrate. The region of a single quantum well (NQW) is undoped and is formed by the structure of a QW based on In _z Ga _1-z As 5.5 nm wide, limited by barriers based on In _x Ga _1-x AS _y P _1-y 12 nm wide (λ _g = 1.26 μm). The active region is bounded by retention layers of In _x Ga _1-n AS _y P _1-y with a stepped refractive index profile (SPP). The thickness and composition of these layers were 50 nm at λ _g = 1.18 μm and 89 nm at λ _g = 1.05 μm, respectively. The structure, which was completely (everywhere) coordinated with InP in the crystal lattice, was supplemented by a layer of InP upper shell of 1.4 μm thick and a layer of In _x Ga _1-x AS _y P _1-y with a thickness of 0.65 μm, followed by layer In _z Ga _1-z As, acting as a contact layer. The layer of the lower shell was doped with sulfur at a concentration of 2.5 × 10 ¹⁸ cm ^-3 . The first layer of the upper shell (InP) was doped with Zn at a concentration of 7.4 × 10 ¹⁷ cm ^-3 , and the following layers were doped with Zn at a concentration of 2 × 10 ¹⁸ cm ^-3 1 and 1.3 × 10 ¹⁹ cm ^-3, respectively. The main data on the multilayer structure and its graphical representation are shown in table 1 and figure 2, respectively.

The structure of the SPPP is used to obtain better optical confinement due to the difference in the refractive index, i.e. higher refractive index of the QW structure compared to barriers. The lower region of the SPPP is doped with S (n-type), while the upper region of the SPPP (layers 7-8) is not doped with an p-type impurity, i.e. Zn, in order to prevent its diffusion into the QW region at the QW stage, which would lead to a deterioration in the quality of the active layer. The upper InGaAs layer is used as the contact layer, and the InGaAsP layer lies between the InP and InGaAs layers, smoothing the sharp transition from the InP structure to the InGaAs structure.

First, samples 17 were cleaned and cut to 2 × 2 mm ² . Then they were irradiated with Ar plasma in the ECR device 10 shown in FIG. 1 under various process conditions. For the first series of samples subjected to plasma treatment, the RF and microwave power was fixed at 450 W (proper DC bias of about -35 V) and 1400 W, respectively, at a flow rate of Ar 50 cm ³ / min (sсcm) and an operating pressure of 30 mTorr. The exposure time varied from 1 to 15 minutes. Another series of samples was irradiated with Ar plasma under the same process conditions, except that the microwave power was reduced to 800 W (proper DC bias of about -60 V). The exposure time varied from 1 to 9 minutes. After plasma irradiation, the samples were annealed at 600 ° С for 2 min using a fast heat treatment apparatus (BTO). At the annealing stage, a tight-fitting GaAs cap was used to ensure that As exerts pressure on the samples.

3A and 3B respectively show the PL spectra of samples irradiated with Ar plasma at different exposure times and microwave powers of 1400 W and 800 W. Figure 4 shows the relative displacement of the energy band gap with respect to the just grown sample as a function of exposure time for Ar plasma generated at an RF power of 450 W and a microwave power of 800 W and 1400 W, respectively.

From figure 4 it is clear that in the samples irradiated with Ar plasma, the QW effect is observed, which causes an expansion of the energy gap and a blue shift of the luminescence wavelength. The degree of displacement gradually increases with increasing exposure time for samples exposed at 1400 W. The energy band gap displacement reaches saturation by approximately 106 nm (72 meV) after 10 min of plasma treatment. Saturation of the energy gap displacement means that after the exposure time of 10 min, the number of point defects caused by ion bombardment and radiation damage reaches a maximum. The results obtained for samples exposed at a power of 800 W showed approximately the same trend as at 1400 W, but with a lower blue shift. This can be explained by the fact that a lower microwave power provides a lower degree of plasma ionization Ar. Under these exposure conditions, the highest achieved blue shift was about 66 nm (42 meV) for the sample treated for 9 min.

From figure 5 it follows that the displacement of the energy gap and the operating temperature are not connected by any linear relationship. The maximum energy gap of 32 nm was obtained at an operating temperature of 100 ° C. It is generally expected that, at a higher temperature, the degree of SCW will be higher due to ion impact damage. However, in this case, this phenomenon was not observed. From this we can conclude that the concentration of defects generated in this process is below a certain threshold for the activation of SCW.

Figure 6 shows the displacement of the energy forbidden zone depending on the operating pressure. The energy band gap reaches a maximum value of 49 nm at an operating pressure of 30 mTorr and gradually decreases with increasing pressure. The results can be explained by the fact that with an increase in operating pressure from 10 to 30 mTorr, the concentration of neutral and ionized plasma particles increases. Thus, the more defects occur, the higher the degree of displacement. However, as the pressure increases further, the mean free path of ions begins to decrease. This leads to a significant decrease in the number of ions and neutral particles colliding with the surface of the sample, and thereby to a decrease in the number of generated defects. An increase in the degree of ionization due to an increase in pressure should cause an increase in radiation damage. However, the results indicate a minimal change in the radiation intensity and show that its effect on the QW within the pressure range remains almost constant.

The process of SCW is advisable to apply only in the case when it can be localized in the desired sections of the semiconductor, i.e. if possible selective mixing. Selectivity is an important aspect of the process because it provides the ability to integrate. With respect to SCJ, the sharpness of the interface between the mixed and unmixed regions is referred to as spatial resolution. A high spatial resolution of the mixing process should be sought, since it provides a compact instrument integration.

To study the selectivity of plasma processes, samples were prepared with 20 sizes of 2 × 4 mm ² (Fig. 7). Half of the samples were patterned with photoresistive material or photoresist 21. These samples 20 were irradiated with Ar plasma at an RF power of 450 W and a microwave power of 1400 W for 5 min. The area masked by photoresist 21 is protected from damage caused by exposure to Ar plasma, and thus is not exposed or minimally affected by the QW after the BTO process.

On Fig shows the PL spectra obtained on samples 20 after irradiation of Ar and subsequent thermal annealing. It can be seen from the graph that the region masked by the photoresist layer 21 received a slight displacement of the energy gap (-10 nm), while the region 22 exposed to the plasma received a much larger energy gap of 64 nm; Thus, between the masked and unmasked regions, a relative energy gap of 54 nm was formed. This result clearly indicates that in InGaAs-InGaAsP samples when using only a photoresist as a masking layer, high selectivity can be achieved. A slight shift in the energy gap in the masked region can be explained by a change in the energy gap due to thermal effects.

It is assumed that plasma generated using only RF radiation provides mainly ion impact damage. The main reason is the high potential difference between the plasma and the semiconductor, which can reach 130 eV. By exposing the sample to the influence of such a plasma, it is possible to study the QW mechanism in a plasma medium with a predominance of ion bombardment.

A group of samples was irradiated with Ar plasma generated under different RF conditions, while other operating parameters remained constant. The exposure time in all cases was 5 minutes. Figure 9 shows the relative displacement of the energy gap as a function of RF power. According to Fig.9, samples treated with plasma in the presence of only RF exhibit a slight displacement of the energy gap at a maximum displacement of 22 nm (10 meV).

The displacements of the energy band gap at other RF power values are also quite small.

Another group of samples was irradiated with plasma generated under various microwave conditions, while other operating parameters remained constant. The exposure time in all cases was 5 minutes. At the end of the exposure, the samples were annealed at 600 ° С for 2 min. Figure 10 shows the relative displacement of the energy gap as a function of microwave power.

According to figure 10, samples treated with plasma in the presence of only microwave, show a shift of the energy band gap to 66 nm (42 meV). The magnitude of the energy gap is also increasing with increasing microwave power. This result suggests that the high-energy VUV radiation generated by high-density ECR plasma has a stronger effect on the QW effect than ion bombardment. Thus, it plays an important role in the QW in InGaAs / InGaAsP structures when using this process.

Table 2 shows the operating parameters discussed above, their possible ranges of values and preferred ranges of values.

In the following example, the SiO ₂ layer is used as an exposure mask for Ar plasma to study the dependence of the mixing rate on the thickness of SiO ₂ deposited on the InGaAs / InGaAsP based MQW structure. The ability to control the degree of mixing by changing the thickness of SiO ₂ would make it possible to horizontally change the energy gap in the sample. This would allow the creation of devices requiring the distribution of various working wavelengths across the sample, for example, multi-frequency lasers.

Samples with MQW structures based on InGaAs / InGaAsP 2 × 2 mm ² in size were cut, and SiO ₂ layers _of different thicknesses were deposited onto the samples using the PSKHFF system. The thickness of SiO ₂ ranged from 100 to 1200 nm. Four samples were used for each thickness of SiO ₂ ; This was done to study the repeatability of the process.

All samples were irradiated with Ar plasma at an RF power of 450 W and a microwave power of 1400 W for 10 min. At the end of the exposure, two samples for each thickness of SiO _{2 were} placed in a solution of HF: H ₂ O with a ratio of 2: 1. This was done to remove the SiO ₂ layer from the samples before proceeding to the annealing process. Thus, the annealing effect was studied in the presence and absence of a SiO ₂ coating. Then, the samples were annealed in BTO at a temperature of 590 ° С for 2 min. After that, PL measurements were carried out in order to analyze the degree of CQW.

11 shows the relative energy shift of the band gap for samples irradiated with Ar plasma at different values of the thickness of SiO ₂ . 11, the degree of mixing gradually decreases as the thickness of SiO ₂ increases. However, with a SiO ₂ thickness _of less than 500 nm, the degree of mixing remains fairly constant, with the energy gap shifting within 40–50 meV. With a SiO ₂ coating thickness _of more than 800 nm, no significant energy gap was observed. In the range of SiO ₂ thicknesses _of 500-800 nm, a significant decrease in the degree of mixing with increasing thickness was found.

So, we found out that the QW process in the QW structures based on InGaAs / InGaAsP irradiated with Ar plasma can be controlled by changing the thickness of the SiO ₂ deposited on the sample before exposure. The ability to control the degree of mixing allows the manufacture of devices that require the distribution of the width of the energy gap over the sample. By controlling the thickness of SiO ₂ within the wafer prior to Ar betrothal, it is possible to create devices such as multi-frequency lasers used in MRI systems. Using the new halftone mask lithography method described in our jointly filed international application No.PCT / GB 01/00904, manufacturing can be further simplified since it only requires a one-step processing of the RIT to transfer the variable thickness SiO ₂ template to the samples. Alternatively, the mask may consist only of a photoresist pattern of variable thickness superimposed by the same halftone mask method.

To study the laser wavelength in materials subjected to the QW process, large-area lasers with a light guide formed by the distribution of amplification were made from a newly grown sample (which had not undergone plasma treatment and annealing), a control sample (which had not undergone plasma processing, but was annealed) and a sample subjected to mixing under the action of Ar plasma.

Samples of 6 × 6 mm ^{2 were} cut along the crystallographic axis from a plate with a MQW structure based on InGaAs / InGaAsP. These samples were irradiated with Ar plasma at an RF power of 450 W and a microwave power of 800 W for 5 min. To stimulate UQW, an annealing step was carried out at 590 ° С for 120 s. Then, a 200-nm-thick SiO ₂ dielectric coating was applied to the samples by the PSHOPF method. After that, using photolithography, strip windows of 50 μm wide were mapped and dry and wet etching was used to open these windows. To minimize RIT damage using CF ₄ and O ₂ , dry etching was first carried out for 5 min, and then wet etching was carried out for 10 s to remove the remaining 75 nm SiO ₂ . The resulting lasers are lasers with a fiber formed by gain distribution, since the injected current generated by the inverse population and the thin waveguide act only in strip regions with a width of 50 μm. After this, the metallization of the front contact surface (p-type: Ti / Au, 50 nm / 200 nm) was carried out using an electron beam evaporator. Then the samples were thinned to a thickness of about 180 microns. The back contact surface was metallized (Au / Ge / Au / Ni / Au, 14 nm / 14 nm / 14 nm / 11 nm / 200 nm) by evaporation and the samples were annealed using BTO at 360 ° C for 60 seconds. To characterize the treated samples, they were scribed (i.e. marked or engraved) on individual lasers with different resonator lengths. 12 is a diagram of an oxide strip laser 30 with a biased energy band gap.

13 shows the spectra of a laser made from a just-grown structure, a control structure, and a structure subjected to mixing under the influence of Ar plasma. According to the figure, the laser lines for the control sample and the just grown sample have close wavelengths of about 1.55 μm, and the laser line for the laser that has been mixed in the Ar plasma has a wavelength of 1.517 μm, i.e. shifted by 38 nm.

Claims (16)

1. A method of manufacturing a photon integrated circuit containing a complex semiconductor structure having a quantum well region, comprising the steps of irradiating said structure with a photon radiation source to generate defects, the photons having an energy (E) of at least equal to energy (E _s ) the displacement of at least one element of a complex semiconductor, and subsequent annealing of the above structure to stimulate the mixing of quantum wells. 2. The method according to claim 1, characterized in that the source of photon radiation is plasma. 3. The method according to claim 2, characterized in that the plasma is generated using either an electron cyclotron resonance (ECR) system, or an inductively coupled plasma (ICP) system, or a plasma disk excited by a low-energy electron beam in vacuum, or soft x-ray plasma devices radiation (MRI). 4. The method according to claim 1, characterized in that the photon radiation source is selected from the group consisting of electric gas-discharge devices, excimer lasers, synchrotron devices, pulsed x-ray devices and gamma radiation sources. 5. The method according to claim 1, characterized in that the method before the irradiation step further includes the step of masking part of said structure to control the degree of radiation damage. 6. The method according to claim 5, characterized in that the masking step is performed using a mask adapted to completely prevent mixing. 7. The method according to claim 5, characterized in that the structure is masked in various ways to provide selective spatially controlled mixing of said structure by controlling the exposure of sections of the structure in a predetermined manner. 8. The method according to any one of claims 5 to 7, characterized in that the mask is selected from the group consisting of binary masks, phase masks, halftone masks, dielectric or metal masks and photoresist masks. 9. The method according to claim 1, characterized in that the method further includes the step of spatial control of the blending using a variable profile mask template. 10. The method according to claim 1, characterized in that the method before the irradiation step further includes the steps of forming a photoresist on said structure, exposing the regions of the photoresist in various ways, providing spatial selectivity depending on the required degree of mixing of the quantum wells and the subsequent manifestation of the photoresist. 11. The method according to claim 10, characterized in that the method further includes the step of applying an optical mask to said photoresist and exposing the photoresist through an optical mask, the optical transmission of which is different to provide spatial selectivity. 12. The method according to claim 11, characterized in that the said optical mask is a halftone mask. 13. The method according to any one of paragraphs.10-12, characterized in that the said photoresist is formed on a masking layer. 14. The method according to item 13, wherein the said masking layer is dielectric. 15. The method according to item 13, wherein the method further includes the step of etching the structure with the developed photoresist in situ to provide a differently etched mask layer. 16. The method according to claim 2, characterized in that for the generation of plasma using an electronic cyclotron resonance system (ECR), and the microwave power of the ECR system is from 300 to 3000 W, more preferably from 1000 to 2000 W, the operating temperature is from 25 to 500 ° C, more preferably 25 to 200 ° C, the operating pressure is 0.1 to 100 mTorr, more preferably 20 to 50 mTorr, and the exposure time is 30 s to 1 h, more preferably 4 to 14 min .

Images