JPS6158992B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a semiconductor heterojunction light emitting diode whose main components are Ga, Al, and As.

An example of the structure of a GaAlAs heterojunction light _emitting diode and changes in the forbidden band width is shown in Figure _1a , as shown in Fig _.
layer 2, n-type Ga ₁ - _y Al _y As layer 3, electrodes 4 and 5, and the substrate 1 is made of p-type GaAs, for example.
use. Conventionally, such a heterojunction light emitting diode is known in Japanese Patent Application Laid-Open No. 157084/1984 entitled "Semiconductor Light Emitting Diode and Manufacturing Method," which describes both a structure in which light is extracted from the surface and a structure in which light is extracted from the side. Furthermore, Japanese Patent Application Laid-Open No. 53-6591 ``Light-emitting diode for pattern display and method for manufacturing the same'' describes a structure for extracting light from the surface. However, the drawback is that the luminance is low because no special consideration is given to the carrier density value of each layer. The present invention provides a heterojunction light emitting diode with extremely high luminance, and as will be described in detail later, extremely high luminance can be achieved by setting the carrier density of each layer within a certain range. In particular, of the two structures mentioned above, the structure that extracts light from the surface has a difference in characteristics from the latter in JP-A-50-157084, although the difference in characteristics from the latter is not mentioned. By taking this into consideration, it is possible to create a high-brightness light-emitting diode that would not be possible with a structure that extracts light from the side of a Pn junction. That is, the light emitting diode according to the present invention has a forbidden band width of the first
As shown in Figure b, the forbidden band width is widened on the side opposite to the substrate via the Pn heterojunction 6, and light is extracted from the surface of the part without electrodes through the side with the large forbidden band width. involved.

First, we will explain the principle of how the carrier density of each layer should be selected.

Usually, light is emitted mainly in the p region 2 due to recombination of electrons injected into the p region 2 via the heterojunction 6,
The emission wavelength corresponds to the forbidden band of this p region 2. Generally, in a light emitting diode made of a - group compound semiconductor, the luminous efficiency in the p region is higher than that in the n region. However, there is also a report that in GaP, the luminous efficiency due to recombination of holes injected into the n region is higher than the efficiency in the p region.

In the case of semiconductor lasers, the carrier lifetime of stimulated emission is extremely short, so recombination occurs mainly through stimulated emission rather than through non-emissive centers.
The effect of the non-emissive center is not so large, but in the case of light-emitting diodes, the carrier lifetime of spontaneous luminescent recombination is long, so the minority carriers injected from the p-n junction are captured by the non-emissive center, and do not emit light there. The probability of recombination is extremely high. Therefore, in light emitting diodes, the efficiency decrease due to non-light emitting centers is decisively large.

Whether the p-region or the n-region serves as the main light-emitting region is related to the likelihood of defects occurring. In other words, a region in which fewer defects, which serve as non-light emitting centers, occur becomes the main light emitting region. GaAs and a mixed crystal containing GaAs
In GaAlAs, defects are likely to occur in the n region, so the p
The area serves as the main light emitting area. As an important factor that tends to cause defects in the n-region, vacancies that electrically compensate for donor impurities such as Te, Se, and S doped in the n-region are formed depending on the doping amount, and these This is thought to be because it combines with other molecules or acts as a very effective non-luminescent center on its own.

Therefore, in the GaAlAs system, 2 in Figure 1a is p.
Layer 3 is an n layer, and this p layer is the main light emitting region, and the light emitted in the p layer is transmitted through the n layer, which has a large forbidden band width, without being absorbed, and is extracted from the surface.

By the way, in the case of a light emitting diode having a pn homojunction with no change in the forbidden band, especially when manufactured using impurity diffusion, it is usually made by diffusing Zn as an n-type crystal and p-type impurity, so the carrier in the p-type region is The density must be greater than the carrier density of the n-type region. Since there is no suitable n-type impurity with a large diffusion coefficient that rivals Zn,
Conversely, it is not possible to diffuse n-type impurities into a p-type substrate to make the carrier density in the p-region lower than the carrier density in the n-region.

It is well known that light emitting diodes manufactured by liquid phase growth have fewer defects and exhibit much higher luminous efficiency than light emitting diodes manufactured by impurity diffusion. Of course, the purpose of the light emitting diode used in the present invention is to obtain high brightness, and since it is a heterojunction, it is manufactured by a liquid phase growth method.
On the other hand, in a pn homojunction made by a liquid phase growth method, unlike a pn homojunction made by a diffusion method, there is no restriction on carrier density, so the carrier density is selected to increase the injection efficiency into the light emitting region. For example, when the p region is the main light emitting region, the amount of atoms injected into the p region is made sufficiently larger than the amount of holes injected into the n region. That is, impurity doping is performed so that the carrier density in the p region, which is the light emitting region, is smaller than that in the n region. Naturally, it is clear that this will be the opposite if the injection of holes is to be mainly performed in the n-region. In any case, the main light emitting region side is doped with impurities so as to have a low impurity density.

However, in the case of a light emitting diode using a pn heterojunction as shown in the present application, the injection efficiency is determined by the potential barrier due to the difference in forbidden band width between the p region and the n region, and the impurity density is no longer the main factor.

Therefore, in order to increase the number of luminescent centers, it is better to increase the impurity density, and hence the carrier density, in the main luminescent region as much as possible. However, when the doping impurity density is increased, various defects including point defects occur, forming non-emissive centers. This is the first factor that determines the upper limit of the preferred impurity density. GaAs and
In GaAlAs, defects are more likely to occur in the n-type, so the appropriate carrier density, and therefore the impurity density, is higher in the p-type region, which is the main light emitting region, than in the n-type region, which is the optimal condition for a homo p-n junction. becomes something else.

In the n-type region, when the carrier density exceeds 10 ¹⁸ /cm ³ , the luminous efficiency is significantly reduced due to the occurrence of defects. However, if the carrier density is too low, the resistance will increase, which is not desirable, so 10 ¹⁷ /cm ³
The above is desirable. On the other hand, p of the main light emitting region
Since the layer has fewer defects than the n-layer, the carrier density can exceed 10 ¹⁸ /cm ³ .

The second factor that determines the upper limit of impurity density is the diffusion of impurities during growth. P-type impurities include Zn, Ge, etc., but the one with the highest luminous efficiency is
It is a Zn impurity. Since the diffusion coefficient of Zn is large,
Diffuses into the n-region during growth. If the diffusion of Zn is strong, part of the n-type region will be inverted to p-type, and the position of the p-n junction will shift. However, due to diffusion, n of the p region
Zn
Even if the impurity density becomes larger than the impurity density in the n region, the pn boundary does not immediately go deep into the region with a large forbidden band.

Taking the configuration of FIG. 1 as an example, the impurity density distribution near the pn junction 6 between the p-type region 2 and the n-type region 3 will be explained using FIG. 2. Figure 2 a shows pn junction 6
Nearby impurity density distribution, FIG. 2b shows the forbidden band distribution near the same pn junction 6. The horizontal axis represents the distance perpendicular to the joint surface, and corresponds to both figures a and b. In FIG. 2a, the vertical axis represents the difference N _D -N _A between the donor density N _D and the acceptor density N _A , or the donor density N _D in the positive direction and the acceptor density N _A in the negative direction. It is assumed that the acceptor density doped into the p-type region 2 is N _AO and the donor density doped into the n-type region 3 is N _DO . Acceptor density N _A is pn
In the p-type region 2 other than the vicinity of the junction 6, the distribution is the same as the _solid line 10 indicating _N
It is distributed as shown by dotted line 12 in the vicinity of . Similarly, the donor density N _D is the same as the distribution 10 of N _D -N _A in the n-type region 3 other than the vicinity of the p-n junction 6, but
In the vicinity of the pn junction 6, the distribution is as shown by the dotted line 13. Distribution of donor density N _D near p-n junction 6 13
and the distribution 12 of the acceptor density N _A compensate for each other, and N _A -N _D becomes like the solid line 10. Even if the doped Zn acceptor density N _AO is larger than the doped donor density N _DO , the Zn density near the pn interface decreases as Zn diffuses, so conductivity type reversal is unlikely to occur, and there is also a region compensated by Zn diffusion. n
It remains in shape. If the doping Zn density is too large compared to the donor density, the pn boundary will enter a region with a large forbidden band width, but the limit to which such encroachment occurs can be determined by observing the emission spectrum. That is, if the Zn density becomes too high and the pn interface moves into a region with a large forbidden band, the emission wavelength will correspondingly become shorter. In addition, in the case of GaAlAs, indirect transition occurs even with a composition with a large forbidden band width, but even if it is a direct transition, the efficiency decreases because the electron density in the indirect transition band increases relatively.

Based on the above considerations, p-type
A suitable range of impurity density when epitaxially growing the GaAlAs layer 2 and the n-type GaAlAs layer 3 was determined. The luminous intensity of the fabricated light emitting diode is shown in FIGS. 3 and 4 as a function of impurity density.

As shown in FIGS. 3 and 4, the carrier density in the p region is 4.5×10 ¹⁷ /cm ³ <p<2.5×10 ¹⁸ /cm ^{3 and} the carrier density in the n region is 2×10 ¹⁷ /cm ³ < Luminous efficiency is extremely high when n<1.0×10 ¹⁸ /cm ³ .

As is clear from Figures 3 and 4, by selecting each carrier density within this range,
High brightness of 100mcd or more can be obtained. Furthermore, the third
To explain the figure, in this experiment in which the carrier density of the p-layer was varied, the carrier density of the n-layer was 2×10 ¹⁷ cm ^-3 <n<8×10 ¹⁷ cm ^-3 , that is, as the range of the carrier density of the n-layer. These are the results obtained by selecting from the range listed above, and the same can be said for Figure 4. Furthermore, as is clear from Figures 3 and 4, within the carrier density range that provides this high brightness, all p values except for the special case of p = 6 × 10 ¹⁷ cm ^-3 in Figure 3
>n, and the length of the arrow indicates that in the special case described above, a diode with low brightness can also be obtained. Therefore, in order to reliably obtain a high-brightness light emitting diode, the third point is that it is desirable that p>n.
Figures and Figure 4 tell a story.

To manufacture this light emitting diode, a patent number is required.
It is preferable to use the temperature difference method liquid phase growth described in No. 85754. In other words, the conventional manufacturing of light emitting diodes by liquid phase growth used a slow cooling method, that is, a method of epitaxial growth by gradually lowering the temperature of the solution. The degree of deviation from the growth layer changes as the temperature decreases, and therefore it is impossible to avoid changes in these amounts in the thickness direction of the grown layer. On the other hand, according to the temperature difference method, the material is transported by diffusion from a high temperature area kept at a constant temperature to a low temperature area due to the temperature difference in the solution, and is epitaxially grown on a substrate in contact with the solution in the low temperature area. Therefore, the crystal growth temperature is constant over time and, in principle, does not fluctuate as described above. To grow multiple layers, the substrates placed on a slider are brought into contact with different melt baths one after another. Therefore, except for variations in the impurity density and mixed crystal composition near the hetero boundary due to solid diffusion during growth, it is possible to obtain a flat distribution for each layer that is incomparable to the slow cooling method. For example, p-type Ga ₁ - _x Al _x As layer 2 and n-type
Ga ₁ - _y Al _y The fluctuations in x and y of As layer 3 are △x except for the boundary region where solid diffusion is a problem as mentioned above.
Keeping it at 0.01, △y0.01 means that the growth layer thickness is 10
Even if it is more than μm, it is easy, △x＜0.002, △
In contrast, in the slow cooling method, the Al composition decreases as the growth progresses, and the thickness decreases to 10
A reduction of △x = 0.02 per μm is normal.

In light-emitting diodes, in order to increase luminous efficiency, the lower the density of non-emissive centers, the longer the diffusion length of minority carriers becomes, so the growth layer, especially p, becomes the main emissive region.
The thickness of the layer must be relatively thick. In a high-quality GaAlAs light-emitting diode, the diffusion length is several μm or more, so the thickness of the light-emitting layer is 1 μm as in the past.
It is insufficient to have a thickness of 2 to 3 μm or less, and it is necessary to make it at least 5 μm or thicker.
It is desirable that the thickness be 10 μm or more. At the same time, the temperature difference method allows extremely good control of impurity density, and as a result of the slow cooling method, there is less variation in the actual doping level from the designed value, making it possible to control the impurity doping amount within the range mentioned above. Can be done accurately. Therefore, the yield of highly efficient diodes is very high, for example, at 20 mA, the luminous intensity is 80 mcd or more (value when coated with epoxy resin).
We were able to increase the yield of diodes exhibiting 70% or more.

An example of a pn heterojunction light emitting diode manufactured in this way has the structure _shown in _{FIG .} First layer 2 and n-type epitaxially grown on the first layer
A Ga ₁ - _y Al _y As second layer 3 is formed, and x<y so that light can be extracted efficiently. Light emitted from the p-layer passes through the n-layer, which has a larger forbidden band, and is extracted in a direction perpendicular to the junction surface. Such a junction has the characteristic that light emitted closer to the pn interface has less absorption loss.

Therefore, the brightness is extremely sensitive to defects near the pn interface.

On the other hand, in a light emitting diode having a structure different from the structure to which the present invention relates, in which light is extracted in the side direction of the junction, the luminance cannot be increased because the self-absorption loss of the light emitting layer itself is large. In addition, in the case of such a structure, since the emission is primarily observed from the p-side part, which is far from the p-n junction boundary, the effect of self-absorption is far greater than the effect of defects occurring on the n-side. Therefore, even if a carrier density range within the range of the present invention is selected, the brightness cannot be significantly increased.

At this time, the composition x of the p layer is selected from a value within the range of 0.30 to 0.37, while the value of y is selected from a value within the range of 0.4 to 0.7. . When the value of x exceeds 0.4, the indirect transition mechanism becomes dominant and the luminous efficiency decreases;
When the wavelength is less than 0.30, the visibility is significantly reduced at a wavelength, and the brightness is also reduced.

The composition values were measured using an X-ray microanalyzer, but there may be slight numerical deviations due to differences in calibration methods. Rather, the optimum composition can be determined by focusing on the emission wavelength peak. In other words, the relationship between the photon energy hν of light emission in the direct transition region and the composition x is almost linear unless the range of x is widened, and it is given as hν (ev) = 1.371 + 1.429x, and the relationship between hν and wavelength λ (Å) The relationship is given by λ=12400/hν(Å). However, since the optimal wavelength range is at most a narrow range from 6550 Å to 6900 Å, it can be assumed that the wavelength λ and the composition x have a close linear relationship in that range.In that case, instead of using the above formula, calculate as λ (Å) = 8468−5260x. Good too.

Note that the carrier density becomes smaller than the doping impurity density in the vicinity of a junction, due to growing impurities, interdiffusion, and compensation effects. The carrier density referred to here is not the value in the immediate vicinity of the junction obtained from CV characteristics, but is the average carrier density in each region determined from the amount of doping impurities. As a measurement method, for example, a shot joint may be attached to the surface of each region or the side surface of the angle-wrapped surface, and each region may be measured independently.

In addition, in the above example, we talked about a heterojunction made of Ga, Al, and As, but if a small amount of P is added to this as a component of a semiconductor, the lattice constant can be well matched without significantly changing other physical properties of the crystal. Since it is possible to fabricate a heterojunction, defects such as misfit dislocations at the hetero boundary can be reduced, and luminous efficiency can be further improved.

[Brief explanation of the drawing]

Figure 1a shows an example of the configuration of a GaAlAs heterojunction light emitting diode, Figure 1b shows an example of the change in the forbidden band in Figure 1a, Figure 2a shows the impurity distribution near the pn junction, Figure 2b Figure 3 shows the distribution of prohibited band width.
4 and 4 are examples of characteristics for explaining the present invention.

Claims (1)

[Claims] 1. A p-type GaAs substrate, a p-type GaAs substrate formed on the substrate
Ga ₁ - _x Al _x As epitaxial layer and the p-type
n formed on Ga ₁ - _x Al _x As epitaxial layer
Ga ₁ - _y Al _y As epitaxial layer, x<
y relationship, and the light generated in the p-type Ga ₁ - _x Al _x As epitaxial layer is transferred to the n-type Ga ₁ - _y Al _y As.
In an incoherent light emitting semiconductor device that passes through a layer and is extracted from its surface, the n-type
It suppresses the occurrence of defects due to the increase in carrier density in the Ga ₁ - _y Al _y As epitaxial layer, and
_The carrier density of _the p-type Ga ₁ - _{x Al x As} epitaxial layer is In order to make it big,
The carrier density p of the p-type Ga ₁ - _x Al _x As epitaxial layer and the carrier density n of the n-type Ga ₁ - _y Al _y As epitaxial layer are 4.5×10 ¹⁷ cm ^-3 ＜p＜2.5×10 ¹⁸ cm ^{- 3} and 2×10 ¹⁷ cm ⁻³ <n<1×10 ¹⁸ cm ⁻³ to increase luminance of light emitted light. 2. The light emitting semiconductor device according to claim 1, wherein the impurity that substantially determines the carrier density of the p-layer is zinc. 3. The light emitting semiconductor device according to claim 1 or 2, wherein the composition x of Al in the p layer is selected from the range of 0.3<x<0.4.