JPS6244835B2 - - Google Patents

Description

[Detailed Description of the Invention] GaP light emitting diodes have a red color and a wavelength of 570 nm.
A light-emitting diode of a certain degree of yellowish-green (usually called green, but here called yellowish-green according to the color) was known. The former is a deep Donna-type impurity O
A composite luminescent center formed by the combination of a p-type (acceptor) shallow impurity Zn is used, while a yellow-green light emitting diode uses N (nitrogen) as the luminescent center, but N is electrically neutral. , a phenomenon such as the coupling between the shallow acceptor Zn and the deep impurity O as in red light emitting diodes does not occur. The appropriate impurity density for providing high brightness naturally differs depending on the difference in these luminescent centers and the impurity that provides the conductivity type.For example, for red light emitting diodes, ``Gallium Phosphide Motoko”
Appropriate values for impurity densities on the n-side and p-side of the p-n junction are listed. According to this, the complex luminescent center is p
side, that is, the Zn-doped side, the donor density on the n-side is generally higher than the acceptor density on the p-side. This meets the conditions for increasing the injection efficiency to the p-side. On the other hand, in the case of a yellow-green light emitting diode, N, which is a neutral impurity, can be doped on either the p or n side, so the conditions are clearly different from those for a red light emitting diode. In order to reduce the density of defects that occur on the n-side, a p ⁺ n junction with a lower impurity density on the n-side is used, but defects due to shallow impurities are more likely to occur on the n-side than on the p-side. We have not reached that conclusion. Furthermore, as a method for manufacturing a yellow-green light-emitting diode doped with N, a method of diffusing Zn into an n-type crystal is known in Japanese Patent Application Laid-open No. 51-38887 entitled "Method for Manufacturing a - Group Semiconductor Light-Emitting Device Containing Phosphorus". ,
In this case, circumstances arise that are different from the optimal conditions for the impurity distribution of the pn junction by the liquid phase epitaxial method. In other words, since a p-n junction is formed by diffusing p-type Zn impurities into an n-type crystal, it is inevitable that
This results in a p ⁺ n junction. In other words, the optimal impurity density distribution for the brightness of a light-emitting diode differs depending on the shallow impurity that determines the conductivity type and the difference in the emission center, and it also has a correlation with the manufacturing method of the p-n junction and the crystal itself. By the way, in the GaP light emitting diode, the present inventor applied the vapor pressure controlled temperature difference method liquid phase growth (abbreviated as TDM/CVP method) method to fabricate the pn junction by the liquid phase epitaxial method. Even without doping, a pure green light-emitting diode with a clear wavelength of about 550 nm can be obtained by simply doping the p-n junction with shallow impurities that determine the conductivity type, that is, doping the p-side with Zn and doping the n-side with Te, S, Se, etc. (Japanese Journal of Applied Physics, Vol. 19, No. 1, pp. 377-382 (1980)
(Jpn.J.Appl.Phys.19 Suppl.19-1 pp.377~
382 (1980))).

The reason why pure green color can be obtained without adding N is as follows.
Crystals produced by the TDM/CVP method have a low density of defects due to non-stoichiometric composition, so there is a low density of non-emissive centers, and this type of crystal is produced by recombination through shallow impurities or free recombination to provide a conductive type. It is thought that light with a short wavelength can be obtained. The present invention relates to a light-emitting diode (hereinafter referred to as a pure green light-emitting diode to distinguish it from yellow-green) that emits green color without adding N. It is clear from the above that impurity density conditions must be set differently from those for diodes and yellow-green light emitting diodes. Already reported simple p-n junctions (including p ⁺ np junctions and n ⁺ p junctions)
In pure green light emitting diodes, the impurity density had the disadvantage that it was not possible to achieve sufficiently high brightness because the conditions for providing a conductive type and providing sufficient luminescent center density were not controlled separately. . The present invention aims to provide a high brightness pure green light emitting diode free from such drawbacks.

First, let's compare p ⁺ n type and n ⁺ p type pure green light emitting diodes. In the p ⁺ n type, hole injection is mainly to the n-side, so the n-side must be the light-emitting region, but the defect density that appears to exist on the n-side increases the amount of shallow donor impurities Te and S. A phenomenon can be seen that increases with age. This has already been clarified by the present inventors in GaAs, and for this reason, n-type crystals have low luminous efficiency, but GaAs
It is also shown that similar defects occur in the n-type region. Therefore, in order to increase the luminous efficiency of the p ⁺ n junction, the n-side impurity density must be further lowered. However, by doing so, n
Naturally, the density of the luminescent centers on the side also decreases, so the brightness does not become very high after all. Currently, in pure green light emitting diodes made using the vapor pressure controlled temperature difference method without adding nitrogen, the efficiency of the p ⁺ n type is about 0.16%, and the majority of the contribution is still from non-radiative recombination due to the search level. It is.

By the way, on the other hand, if you lower the density of zinc in the p-type region,
When the n ⁺ p type, that is, p〓10 ¹⁷ to 10 ¹⁶ /cm ³ is used, the brightness becomes extremely low and does not reach the brightness of the p ⁺ n junction. This fact is also observed in GaP light emitting diodes made by the slow cooling method. Therefore, it has been argued that defects are more likely to occur in the n-type region of GaP.
There is no proof of that. Conventionally, the existence of a compensation effect has not been pointed out as the reason why the p ⁺ n type has higher luminance than the n ⁺ p type, but this will be pointed out below.

When forming a GaPp ⁺ n junction by liquid phase growth, the diffusion coefficient of Zn, which usually takes about 0.5 to 2 hours to grow the p-layer and n-layer, is higher than that of donors such as Te, S, and Se. Since it is extremely large and diffuses to a considerable extent during growth, an actual pn boundary is formed within the n-type growth layer as shown in FIG. Moreover, it can be seen from FIG. 1 that when the diffusion length of minority carriers is about 2 to 3 μm, the slope of the Zn distribution is relatively gentle.

FIG. 1 shows an example of impurity distribution when a first layer with a donor density N _D1 is grown on an n-type substrate and a second layer with a zinc density N _A is grown thereon. This figure shows an example of the impurity distribution and carrier density distribution on the first layer side when the diffusion coefficient D _Zo of Zn is assumed to be D _Zo ≒ 1.0 × exp (-2.1 eV/kT) according to the literature. It is assumed that the growth temperature of the second layer, that is, the diffusion temperature of Zz into the first layer is 800° C., the growth time, that is, the diffusion time t is 1 hour, and that the growth rate is sufficiently faster than the diffusion rate. Diffusion effects were ignored because the diffusion coefficient of donor impurities was small. In the figure, the sizes of the diffusion lengths Lp and Ln of minority carriers are shown for comparison. Lp, Ln are μn〓μP〓100cm ³ /V・S τn
From the typical value of 〓τp〓50ns,
Approximately Lp〓Ln〓3μm.

In the figure, a case is shown in which the Zn density of the second layer N _A =2×10 ¹⁸ /cm ^{3 and} the donor density N _D of the first layer is 1×10 ¹⁷ /cm ³ . It is said that the impurity distribution of Zn is not a unitary error function when the density is extremely low, but it is an exponential function that is close to the error function in areas not far from the second layer. The approximate impurity distribution can be found by approximating . As shown in the figure, the results show that the impurity distribution of Zn cannot be said to be steep compared to the diffusion length of minority carriers. The actual growth temperature ranges from 700°C to 900°C and the growth time ranges from about 30 minutes to 2 hours, but in either case, the distribution of Zn is not steep enough compared to the diffusion length of the minority carriers, as shown in Figure 1. Therefore p ⁺ n
If you look at the junction near the pn boundary, that is, at a distance of about the diffusion length of minority carriers, it is essentially close to a p ^- n ^- junction with strong compensation on both sides. As long as the impurity distribution of Zn is close to an exponential function, even if you intend to further lower the donor density on the n side to make it stronger p ⁺ n type, the N _D line in Figure 1 will just go down, so it will still be essentially p ^- It is an n ^- junction. The formation of a luminescent center is advantageous because both sides are strongly compensated, but the fact that it is essentially a p ^- n ^- junction near the p-n boundary is disadvantageous for injection because there are fewer majority carriers. . Furthermore, since the impurity density N _D in the n region must be lowered as much as possible in order to reduce defects, there is a drawback that both acceptor and donor densities are low at the pn boundary, that is, the luminescent center density is low. In particular, since GaP is an indirect transition type semiconductor, the lifetime of the luminescent center is long, and therefore the contribution of the search level is relatively significant. Therefore, it is a major drawback that the density of luminescent centers must be reduced. An object of the present invention is to provide a pure green light-emitting diode with high brightness, which fully exhibits the compensation effect as explained above, and which does not have a drop in injection amount due to a drop in majority carrier density.

First, the pn junction of the pure green light emitting diode of the present invention is the opposite of the p ⁺ n type, that is, the n ⁺ p type or np type junction,
Light is emitted mainly in the p region. Furthermore, in order to produce the same compensation effect that is clearly produced in the p ⁺ n type, donor type impurities with an appropriate shallow density are added to the p region, which is the light emitting region, and the implantation amount is not reduced. do. In other words, the p-layer forming the p-n junction is an acceptor that provides p-type conduction type.
Contains shallow donor impurities for compensation in addition to Zn. If the acceptor density is N _A and the donor density is N _D2 , the hole density is p ₂ =N _A -N _D2 , so N _A >N _D2 (2). Also, if the shallow donor impurity density on the n side of the pn junction is N _D , then N _D1 > N _A ......(3) Otherwise, the actual pn boundary will be buried in the n layer due to the diffusion of Zn, and the conventional p ⁺ It becomes the same as n junction. Naturally, the impurity density on the n side is higher than the impurity density on the p side, that is, in order to form an n ⁺ p junction, N _A must be selected so that N _A −N _D2 < N _D2 (4). −N _D2 〓N _D1
may be possible, but it will be difficult to satisfy conditions (1) and (2). The shallow donor N _D2 doped on the p side is implanted as shown in the band diagram of the pn junction in
The probability that electrons, which are minority carriers diffusing to the side, will be captured by the donor D increases compared to the probability that they will be captured by the deep level T, which is a non-luminous center. Since the electrons once captured by the donor emit light by recombining with the holes, high luminous efficiency can be obtained, unlike a normal n ⁺ p junction. The density _N Since defects are likely to occur, it is desirable that the density be between 1×10 ¹⁶ and 1×10 ¹⁷ /cm ³ . Therefore, it is comparable to the donor density in the n region in a conventional p ⁺ n junction, but in the case of the present invention, the aim is to trap the minority carriers and not to supply majority carriers, so the donor density is at this level. It can be said that the density is quite large.

The example of the impurity density distribution of the present invention described above is shown in the second example.
As shown in the figure. That is, pn on the n-type GaP substrate (S)
An n-type GaP first layer 1, which is the n-side region of the junction, is grown. Impurities include S, Se, Te, etc., but S is most desirable because it causes fewer defects. The impurity density of the first layer is 1 to 5 × 10 ¹⁷ /cm ³ , and even though it is an n ⁺ layer, if the impurity density is too high, defects will increase, so the p
The above range is appropriate because it is better for the impurity density to be as low as possible within a range that does not reduce the efficiency of implantation into the side. The second region, which is the p-side region of the p-n junction,
Layer 2 is doped with the Zn mentioned earlier and a shallow donor, and if the relationships shown in equations (2), (3), and (4) are met, then
It becomes an n ⁺ p junction, and carriers are mainly injected to the p side. At this time, the donor density for compensation is preferably between 1×10 ¹⁶ and 1×10 ¹⁷ /cm ² as described above.

The easiest way to form such a second p-layer is to add impurities to the second layer melt.
For example, add the necessary amounts of S and Zn. Depending on the growth temperature, this method may cause bonding of Zn and S to occur in the melt, forming precipitates in the crystal and deteriorating crystallinity. Therefore, as shown in Fig. 4, the first layer is the same as before, but the second layer is doped with only donors with an impurity density N _D2 , and on top of that is a p layer 3 with an impurity density N _A3 .
grow. This impurity is Zn, which diffuses into the second layer during growth and inverts the second layer to p ^- type.
Therefore, the pn boundary is close to the boundary between the first layer and the second layer as shown in the figure. For this purpose, the thickness of the second layer is determined by the impurity diffusion distance L, which is determined by the Zn diffusion time during the growth of the third layer _.
=√ Should not be much thicker than _Zo . In other words, the impurity density can be designed to satisfy the conditions (2), (3), and (4) using equation (1), which preferably has a thickness of _d√Zo . Of course, since it diffuses to the first layer, N _D1 must be large enough that the carrier density in this region does not drop too much. For example, N _D1 is 1×10 ¹⁷ ~5×10 ¹⁷ /cm ³ N _D2 1×10 ¹⁶ ~3×10 ¹⁷ /cm ³ N _A3 1×10 ¹⁸ ~5×10 ¹⁸ /cm ³ 1d ₂ 10μm Third layer growth temperature: 750 DEG C. to 850 DEG C. Third layer growth time: By selecting a value from about 30 minutes to 4 hours, it is possible to appropriately design a distribution as shown in FIG. 4.

Even in the case of Fig. 2, growing a p-type third layer with a high carrier density reduces the series resistance and allows a uniform current to flow through the junction surface. It is particularly useful for diodes.

The above is an example using an n-type substrate crystal, but since the design of the impurity density is a problem in the present invention, the substrate crystal is a p-type substrate crystal.
A similar distribution can be achieved even if the shape is different. However, it must be taken into consideration that diffusion during growth is utilized. For example, as shown in FIG. 5, a p-type substrate (S) is used, and a second layer 2 having a donor density N _D2 is grown thereon, and a first layer 1 having a higher donor density N _D1 is grown thereon. Since Zn diffuses from the substrate into the first layer and is distributed as shown in the figure, the second layer becomes p-type and has an impurity distribution similar to that shown in Figure 4. Moreover, in this case, a layer corresponding to the third layer is not necessarily required. do not have.

However, as mentioned before, the first layer cannot be made very thick. When the dislocation density or defect density of the substrate crystal is high, the crystallinity of the second layer will be affected and the characteristics will deteriorate, so a p-type third layer is first grown on the p-type substrate as shown in Figure 6. After that, it is necessary to grow the second layer and the first layer in sequence. If the donor density is 10 ¹⁸ /cm ³ or more, the defect density will increase and the crystallinity will deteriorate significantly, so considering that the defects will diffuse into the light emitting region, N _D1 should be 5 × 10 ¹⁷ /cm ³ or less. is desirable. Furthermore, if the donor density N _D2 of the second layer, which is the light emitting region, is less than 1×10 ¹⁶ /cm ³ , it is not easy to control the doping level from the conventional embodiment, and it is too large as long as the defect density is not increased. Since the luminescent center density is higher, it is practical to set the density to 1×10 ¹⁶ /cm ³ or more.
Also, considering that N _D1 is less than 5×10 ¹⁷ /cm ³ , N _D2 must be less than 3×10 ¹⁷ /cm ³ or N _A −N _D2
<N It becomes difficult to choose _D1 . That is, N
_D1 〓5×10 ¹⁷ /cm ³ then N _D2 〓3×10 ¹⁷ then N _A
Unless it is chosen to be 3×10 ¹⁷ /cm ³ <N _A <5×10 ¹⁷ /cm ³ on the pn boundary side, conditions (2), (3), and (4) will not be satisfied. Although it is possible to control N _A within this range with good yield, if N _D2 is increased beyond this range, N _A must be controlled using two significant figures, which is not practical.

When a p-type substrate is used as shown in FIGS. 5 and 6, the diffusion time t is the sum of the growth times of the second layer and the first layer. In the case of slow cooling, the growth temperature decreases over time, resulting in a change in the diffusion coefficient, which has the disadvantage that precise design is not possible. Therefore, it is necessary to determine the parameters experimentally, but even then, the desired n ⁺ p junction is often not achieved due to subtle changes in temperature. In the temperature difference liquid phase growth method or the vapor pressure controlled temperature difference method, since the growth temperature remains unchanged, the impurity distribution can be manufactured very precisely as designed, and a high yield can be obtained. Furthermore, the slow cooling method involves a drop in temperature, which is undesirable because many defects occur at layer boundaries due to overcooling during sliding from the first layer to the second layer and remelting, and this boundary almost coincides with the pn boundary. However, the temperature difference method and the vapor pressure controlled temperature difference method eliminate this drawback.
The diffusion of Zn is not necessarily an error function, but since we are not concerned with low impurity densities, it can be approximated as an exponential function. It is sufficient to design it as Note that if impurities diffuse during growth, computer analysis will be required if the diffusion rate is close to or faster than the growth rate, but growth should be done so that the diffusion rate of Zn is sufficiently slow compared to the growth rate. If the growth rate is selected, there is no need to consider the effect of the growth rate, and in that case, N _A (o) is given by N _A /2.

However, N _A is the Zn density inside the layer on the diffusion source side. On the other hand, although the design is more complicated, when the growth rate is close to or smaller than the diffusion rate, there is an advantage that the defect density of the second layer, which is the light emitting layer, can be reduced. One reason is simply that the growth rate is slow, but Zn at the growth interface
When Te and S are introduced into the growth layer, the free energy is lower when these donors combine with Zn than when they generate defects. Since it has the effect of suppressing generation, it becomes possible to increase the donor density and increase the density of luminescent centers.

As described above, the present invention provides a light emitting diode that exhibits green color by adding shallow acceptors and shallow donors without adding N, in which the main light emitting region is set on the p side, shallow donor impurities are added to the p side, and This is a high-brightness light emitting diode with a high carrier injection effect.

In the case of a green light emitting diode, Zn is used as the acceptor impurity, and S, Se, Te, etc. are used as the donor impurity, but S and Se are preferable because Te easily creates defects.

[Brief explanation of the drawing]

Figure 1 shows an example of impurity distribution when a first layer with a donor density is grown on an n-type substrate, and a second layer with a zinc density is grown on top of that. Figure 2 shows an example of impurity distribution in an example of the present invention. Distribution, Figure 3 is an example of a band diagram of a pn junction,
FIG. 4, FIG. 5, and FIG. 6 show impurity distributions of other examples of the present invention.

Claims (1)

[Claims] 1. Forming one of the pn junctions in a GaP green light emitting diode that has a pn junction manufactured by liquid phase epitaxial method, does not contain nitrogen, and has a shallow acceptor and a shallow donor added. When the carrier density p of the p region is smaller than the carrier density n of the n region forming the other side of the p-n junction, and a shallow donor impurity is added to the p region, and the shallow donor impurity density of the p region is N _D2 . 1×10 ¹⁶ /cm ³ N _D2
3×10 ¹⁷ /cm ³ , and the shallow acceptor in the p region is
A GaP green light emitting diode characterized by being Zn. 2 The density of shallow acceptors in the p region is
A GaP green light emitting diode according to claim 1, characterized in that the density is lower than the shallow donor impurity density doped in the region. 3. The GaP green light emitting diode according to claim 1 or 2, wherein the donor impurity is S or Se. 4 has a p-type third region containing Zn in contact with the p region, and is an acceptor impurity for the p region.
4. A GaP green light emitting diode according to any one of claims 1 to 3, characterized in that Zn is added from the third region by impurity diffusion during growth. 5. The GaP green light emitting diode according to any one of claims 1 to 4, wherein the third region containing Zn is a GaP substrate. 6. The GaP according to any one of claims 1 to 5, wherein the third region containing Zn is an epitaxial layer formed on a p-type GaP substrate. green light emitting diode.