JP7604311B2 - Ultraviolet semiconductor light emitting element - Google Patents

Description

The present invention relates to an ultraviolet semiconductor light-emitting device, in particular a nitride semiconductor light-emitting device that emits ultraviolet light.

In recent years, semiconductor light-emitting devices that emit light in the deep ultraviolet region have been attracting attention as light sources that have the effect of inactivating and sterilizing bacteria and viruses.

For example, Patent Document 1 discloses an AlGaN-based semiconductor light-emitting device in which the p-type semiconductor layer on the electron blocking layer is a compositionally graded layer.

Patent Documents 2 and 3 disclose group III nitride semiconductor light-emitting devices in which the electron blocking layer is doped with Si and Mg.

JP 2019-160974 A JP 2020-057783 A JP 2019-083221 A

In conventional ultraviolet semiconductor light-emitting devices, it was difficult to realize devices that were not only highly efficient and high-output, but also highly reliable. In particular, when driven with a large current to obtain high optical output, there was a problem in that the device deteriorated early. In other words, it was difficult to achieve both high output characteristics and high reliability (long life).

The inventors of the present application have discovered the trade-off relationship and degradation mechanism that the higher the efficiency and output of an element, the greater the drop in output after current is passed through it, and the shorter the element lifespan. The present invention was made based on this knowledge, and aims to provide an ultraviolet semiconductor light-emitting element that is highly efficient and high output, has minimal degradation, and has an excellent element lifespan.

An ultraviolet semiconductor light emitting device according to one embodiment of the present invention comprises:
a semiconductor structure layer in which an n-type AlGaN layer, an active layer, an _{AlY1Ga1 -Y1N} layer (0.8≦Y1≦1.0) and a p-type _{AlY2Ga1 -Y2N} layer (0.5≦Y2≦1.0, Y2≦Y1) are epitaxially grown in this order;
the p-type _{AlY2Ga1 -Y2N} layer is co-doped with a p-type impurity serving as an acceptor and an n-type impurity serving as a donor, and a concentration ratio (Nd/Na) of a concentration (Nd) of the n-type impurity to a concentration (Na) of the p-type impurity satisfies 0.009≦(Nd/Na)<0.185;
the p-type impurity concentration of the p-type Al _Y2 Ga _1-Y2 N layer is 1×10 ¹⁷ to 1.2×10 ²⁰ cm ⁻³ ,
The _{Al.sub.Y1Ga.sub.1 -Y1N} layer does not contain the n-type impurity, or the concentration of the n-type impurity is less than the concentration of the n-type impurity in the p-type _{Al.sub.Y2Ga.sub.1 -Y2N} layer.

1 is a cross-sectional view showing a schematic structure of an ultraviolet semiconductor light-emitting element 10 according to an embodiment of the present invention. FIG. 2 is a diagram illustrating a band diagram of an ultraviolet LED 10. 1 is a table showing the characteristics of ultraviolet LEDs 10 obtained from two types of wafers having different Mg concentrations in the p-type Al _{Y 2} Ga _{1-Y 2} N layer 15 . 1 is a graph plotting the relationship between the Si concentration in the p-type Al _Y2 Ga _1-Y2 N layer 15 and the luminous efficiency (arbitrary unit) at the beginning of current application. 1 is a graph plotting the relationship between the Si/Mg ratio (Si concentration/Mg concentration) of the p-type Al _{Y 2} Ga _{1-Y 2} N layer 15 and the luminous efficiency (arbitrary unit) at the beginning of current application. 1 is a graph showing the output maintenance rate (@100h) after 100 hours have elapsed from the initial time versus the Si concentration in the p-type Al _Y2 Ga _1-Y2 N layer 15 . 1 is a graph showing the Si/Mg ratio (Si concentration/Mg concentration) of a p-type Al _{Y 2} Ga _{1-Y 2} N layer 15 and the output maintenance ratio. 1 is a graph showing the output maintenance rate (@100 h) versus the initial light output (arbitrary units) at the start of energization for elements obtained from structures 1 and 2 (EX1 and EX2) and comparative examples 1 and 2 (CM1 and CM2).

In the following, preferred embodiments of the present invention will be described, but these may be modified and combined as appropriate. In the following description and accompanying drawings, substantially the same or equivalent parts are denoted by the same reference numerals.

[Structure of ultraviolet semiconductor light-emitting element]
1 is a cross-sectional view showing a schematic structure of an ultraviolet semiconductor light-emitting element 10 according to one embodiment of the present invention. The ultraviolet semiconductor light-emitting element 10 is an ultraviolet light-emitting diode (hereinafter referred to as an ultraviolet LED 10), and is formed by epitaxially growing an n-type AlGaN layer 12, an active layer 13, an AlGaN layer 14, a p-type AlGaN layer 15, and a p-type GaN layer 16 on a substrate 11, in that order.

Figure 2 shows a schematic band diagram of the ultraviolet LED 10. A more detailed explanation will be given with reference to Figures 1 and 2.

The substrate 11 is preferably a substrate having a low dislocation density so that the dislocation density in the active layer is low, but is not particularly limited to this. A material that can reduce the dislocation density in the active layer to 10 ⁹ cm ⁻² or less, preferably 10 ⁸ cm ⁻² or less, is preferable. For example, an AlN template substrate in which an AlN film is laminated on a sapphire substrate as described in Patent Document 1 or a single crystal AlN substrate as described in Patent Document 2 can be used.

When using an AlN template substrate, it is preferable to select one in which the dislocation density of the outermost surface of the AlN layer is 10 ⁹ cm ⁻² or less, more preferably 10 ⁸ cm ⁻² or less. The method for growing the AlN layer on the template substrate is not particularly limited as long as the above-mentioned dislocation density is satisfied, and physical vapor deposition methods such as metalorganic chemical vapor deposition (MOCVD), hydride vapor phase epitaxy (HVPE), and sputtering can be used.

In addition, since a large surface roughness of the template substrate can cause abnormal growth in the AlGaN layer grown on the substrate, the surface roughness (RMS) is preferably 1.0 nm or less, and more preferably 0.5 nm or less. In order to obtain this surface roughness or to remove a damaged layer formed on the surface after growth, the surface of the template substrate may be polished by a known polishing method such as chemical mechanical polishing (CMP). In addition, a superlattice layer can be further provided on the AlN template substrate to interfere with the strain with the n-type AlGaN layer.

From the viewpoint of reducing the dislocation density in the active layer, it is preferable to use a single crystal AlN substrate as the substrate 11. The dislocation density of the single crystal AlN substrate is preferably 10 ⁸ cm ⁻² or less, more preferably 10 6 cm ⁻² or less, and most preferably 10 ⁴ cm ⁻² or less. By using an AlN substrate with a lower dislocation density of 10 ⁶ cm ⁻² or less, or even 10 ⁴ cm ⁻² , it is possible to prevent a decrease in the light emission efficiency in the ^active layer due to dislocations, and further to prevent problems such as the diffusion of impurities through dislocations and an increase in leakage current that occur when a current is passed through the ultraviolet light-emitting device.

Furthermore, for the same reasons as for the AlN template substrate described above, the surface roughness (RMS) of the single crystal AlN substrate 11 is preferably 1.0 nm or less, and more preferably 0.5 nm or less. Naturally, the surface of the AlN substrate may also be polished by a known polishing method such as chemical mechanical polishing (CMP).

Furthermore, if the absorption coefficient of the substrate for the ultraviolet light emitted from the active layer is large, there is a concern that the total amount of ultraviolet light that can be extracted to the outside will decrease, leading to a decrease in light emission efficiency. Therefore, the absorption coefficient of the AlN substrate and the AlN layer of the AlN template is preferably 20 cm ^-1 or less, and more preferably 10 cm ^-1 or less. By making it 10 cm ^-1 or less, it is possible to ensure a linear transmittance of 90% or more even if the plate thickness of the AlN substrate 11 is 100 μm, for example.

The n-type AlGaN layer 12 is an n-type conductive layer doped with Si (silicon). The Al composition of the n-type AlGaN layer can be appropriately determined so as to obtain sufficient transparency for the desired emission wavelength of ultraviolet light. In the ultraviolet semiconductor light-emitting element, the ultraviolet light emitted from the light-emitting layer passes through the n-type AlGaN layer 12 and the substrate 11 and is emitted to the outside. In addition, as the Al composition of the n-type AlGaN layer increases, the band gap of the n-type AlGaN layer increases, and accordingly, ultraviolet light of a shorter wavelength can be transmitted.

The n-type AlGaN layer 12 may be formed of a plurality of layers having different Al compositions, and may be a compositionally graded layer in which the Al composition is graded in the stacking direction. For example, the n-type _AlGaN layer 12 may be composed of a first n-type Al _X1 Ga _1-X1 N layer 12A and a second n-type Al X2 Ga _1-X2 N layer 12B. The first n-type Al _X1 Ga _1-X1 N layer 12A may be a compositionally graded layer in which the Al composition X1 decreases from 1.0 to 0.75 in the stacking direction, and the second n-type Al _X2 Ga _1-X2 N layer 12B may be a compositionally graded layer in which the Al composition X2 decreases from 0.75 to 0.70.

The thickness of the n-type AlGaN layer 12 is not particularly limited and is determined appropriately. When a single crystal AlN substrate is used for the substrate 11, the thickness of the n-type AlGaN layer 12 is preferably 0.5 μm or more and 2 μm or less. From the viewpoint of reducing the resistance value of the n-type AlGaN layer, it is preferable that the n-type AlGaN layer is thicker. However, when an AlN substrate is used for the substrate 11, if the n-type AlGaN layer is too thick, the n-type AlGaN layer will undergo lattice relaxation and dislocations will be more likely to occur.

For example, when the n-type AlGaN layer 12 is formed as a laminated structure consisting of the first n-type _{Al.sub.X1Ga.sub.1 -X1N} layer 12A and the second n-type _{Al.sub.X2Ga.sub.1 -X2N} layer 12B, the first n-type _{Al.sub.X1Ga.sub.1 -X1N} layer 12A can have a thickness of 200 nm, and the second n-type _{Al.sub.X2Ga.sub.1 -X2N} layer 12B can have a thickness of 1000 nm. The thicknesses of the first n-type _{Al.sub.X1Ga.sub.1 -X1N} layer 12A and the second n-type _{Al.sub.X2Ga.sub.1 -X2N} layer 12B are not limited to the exemplified values, and can be appropriately determined so that the total thickness is 2.0 μm or less.

On the other hand, when an AlN template substrate is used, the thickness of the n-type AlGaN layer 12 is preferably 1.5 μm or more and 10 μm or less. This is because a thicker n-type AlGaN layer 12 is expected to have the effect of reducing dislocations in addition to reducing the resistance value, so a thickness of 1.5 μm or more is considered preferable. Also, when an AlN template substrate is used, the upper limit of the thickness of the n-type AlGaN layer 12 is considered to be about 10.0 μm, taking into account industrial viewpoints such as manufacturing time.

The Si concentration doped into the n-type AlGaN layer 12 may be appropriately determined so as to obtain the desired n-type conductivity, but is preferably 1×10 ¹⁸ to 1×10 ²⁰ cm ^-3 , and more preferably 5×10 ¹⁸ to 5×10 ¹⁹ cm ^-3 , from the viewpoint of lowering the resistance value of the n-type AlGaN layer 12. The Si doping concentration may be constant in the film thickness direction in the n-type AlGaN layer, or it may be modulated doping in which the Si concentration varies in the film thickness direction.

The Si concentration and the Mg concentration described later can be measured by known secondary ion mass spectrometry ( _SIMS ) analysis. The Si concentration and Mg concentration measurements in this application are quantitative values obtained by using standard samples of AlN, _{Al0.65Ga0.35N} , and GaN for the AlN layer, AlGaN layer, and GaN layer, respectively.

The active layer (ACT) 13 is a quantum well structure composed of a barrier layer made of an _{AlAlGa1 -A1N} layer and a well layer made of an _{AlAlGa1 -A2N} layer. The emission peak wavelength of the active layer 13 is in the range of 210 to 300 nm. Since the wavelength of the light emitted from the active layer 13 is determined by the Al composition and film thickness of the well layer, the Al composition and film thickness can be appropriately determined so as to obtain a desired emission wavelength within the above wavelength range.

For example, the thickness of the well layer can be set in the range of 2 to 10 nm, and the Al composition can be determined so as to obtain the desired emission wavelength. The Al composition and thickness of the barrier layer are not particularly limited, but for example, the Al composition can be set in the range of A2 < A1 ≦ 1.0, and the thickness can be set in the range of 2 to 15 nm.

The well layer and the barrier layer may be n-type layers doped with Si. Both the well layer and the barrier layer may be Si-doped layers, or only the well layer or only the barrier layer may be doped with Si. The doping Si concentration is not particularly limited, but is preferably in the range of 1×10 ¹⁷ to 5×10 ¹⁸ cm ^-3 .

The number of quantum well layers is not particularly limited, and may be a multi-quantum well (MQW: Multi Quantum Well) structure in which multiple well layers are formed, or a single quantum well (SQW: Single Quantum Well). It is preferable to determine the number of well layers appropriately in the range of 1 to 5.

The Al _Y1 Ga _1-Y1 N layer 14 is a layer provided adjacent to the active layer 13. The Al _Y1 Ga _1-Y1 N layer 14 functions as an electron blocking layer (EBL) for suppressing the overflow of electrons injected into the active layer 13 to the p-type Al _Y2 Ga _1-Y2 N layer 15. Therefore, the Al _Y1 Ga _1-Y1 N layer 14 has a larger band gap than the active layer 13 and the p-type Al _Y2 Ga _1-Y2 N layer 15 described later, and the Al composition Y1 of the Al _Y1 Ga _1-Y1 N layer 14 is determined in the range of 0.8<Y1≦1.0.

As the emission wavelength becomes shorter, the Al composition of the AlGaN layer epitaxially grown on the substrate 11 becomes higher, and when the emission wavelength is shorter than 270 nm, in order to fully exhibit the function as an electron blocking layer, it is preferable that the Al composition Y1 is 0.9≦Y1≦1.0. In this embodiment, AlN (Y1=1) is used as the Al _Y1 Ga _1-Y1 N layer 14.

In addition, the Al _Y1 Ga _1-Y1 N layer 14 may be an undoped layer or may be doped with a p-type dopant as long as it can function as an electron blocking layer. Examples of p-type dopant materials that can be used in the Al _Y1 Ga _1-Y1 N layer 14 include Mg (magnesium), Zn (zinc), Be (beryllium), and C (carbon). In particular, it is preferable to use Mg, which is generally used as a p-type dopant material for AlGaN layers, and Mg is also used in the examples of the present invention described below.

The p-type dopant material may be uniformly doped in the stacking direction of the Al _Y1 Ga _1-Y1 N layer 14, or the concentration of the dopant material may be changed in the stacking direction. For example, a stacked structure may be formed of an undoped AlN layer 14A (Y1=1) and a Mg (magnesium)-doped p-type AlN layer 14B from the side in contact with the active layer.

The p-type dopant concentration in the Al _Y1 Ga _1-Y1 N layer 14 is not particularly limited, but in order to obtain the function as an electron blocking layer, it is preferably 5×10 ¹⁸ to 1×10 ²⁰ cm ⁻³ , and from the viewpoint of increasing the efficiency of carrier injection into the light-emitting layer, it is particularly preferably 1×10 ¹⁹ to 8×10 ¹⁹ cm ⁻³ .

The Al _Y1 Ga _1-Y1 N layer 14 of the present invention does not contain an n-type dopant, or may contain an n-type dopant at a concentration lower than the n-type dopant contained in the p-type Al _Y2 Ga _1-Y2 N layer 15 described below. Specifically, the n-type impurity concentration in the Al _Y1 Ga _1-Y1 N layer 14 is preferably 1×10 ¹⁸ cm ^-3 or less. According to the knowledge of the present inventors, it has been found that during the growth of the p-type Al _Y2 Ga _1-Y2 N layer 15, diffusion of dopant occurs between adjacent Al _Y1 Ga _1-Y1 N layers 14. Therefore, when the n-type dopant concentration in the Al _Y1 Ga _1-Y1 N layer 14 is higher than that in the p-type Al _Y2 Ga _1-Y2 N layer 15, the n-type dopant diffuses from the Al _Y1 Ga _1-Y1 N layer 14 to the p-type Al _Y2 Ga _1-Y2 N layer 15, which may make it difficult to precisely control the n-type dopant concentration in the p-type Al _Y2 Ga _{1 -Y2} N layer 15. In order to prevent a change in the n-type dopant concentration in the p-type Al _Y2 Ga 1-Y2 N layer 15 due to this diffusion, the n-type dopant in the Al _Y1 Ga _1-Y1 N layer 14 must have a concentration lower than that of the n-type dopant contained in the p-type Al _Y2 Ga _1-Y2 N layer 15.

The thickness of the Al _Y1 Ga _1-Y1 N layer 14 may be appropriately determined so as to function as an electron blocking layer and to efficiently inject holes from the p-type Al _Y2 Ga _1-Y2 N layer 15 into the active layer, but is preferably in the range of 1 to 30 nm. If the thickness is less than 1 nm, electrons tunnel, decreasing the function as an electron blocking layer, while if the thickness exceeds 30 nm, holes are less likely to be injected from the p-type Al _Y2 Ga ₁ -Y2 N layer 15 into the active layer. Taking these into consideration, the thickness of the Al _Y1 Ga _1-Y1 N layer 14 is preferably 2 to 20 nm, and more preferably 5 to 15 nm.

As described above, the Mg doped in the Al _Y1 Ga _1-Y1 N layer 14 can have a concentration difference in the stacking direction. For example, an undoped AlN layer 14A with a thickness of 1 to 5 nm can be stacked on the side in contact with the active layer 13, and an Mg-doped p-type AlN layer 14B with a thickness of 5 to 15 nm can be stacked on top of that. In this case, the Mg doping concentration is preferably 5×10 ¹⁸ to 1×10 ²⁰ cm ^-3 , as described above, and particularly preferably 1×10 ¹⁹ to 8×10 ¹⁹ cm ^-3 .

The p-type Al _Y2 Ga _1-Y2 N layer 15 of the present invention is formed on the Al _Y1 Ga _1-Y1 N layer 14 and functions as a p-type cladding layer. The p-type Al _Y2 Ga _1-Y2 N layer 15 is co-doped with a p-type impurity that serves as an acceptor and an n-type impurity that serves as a donor.

As the p-type impurity to be doped into the p-type Al _{Y 2} Ga _{1-Y 2} N layer 15, Mg (magnesium), Zn (zinc), Be (beryllium), C (carbon), etc. can be used. Among them, it is preferable to use Mg, which is generally used as a p-type dopant material for AlGaN semiconductors. As the n-type impurity, Si, Ge (germanium), Se (selenium), S (sulfur), O (oxygen), etc. can be used. Among them, it is preferable to use Si, which is generally used as an n-type dopant material.

In the p-type _{Al.sub.Y2Ga.sub.1 -Y2N} layer 15 of the present invention, the ratio (Nd/Na) of the n-type impurity concentration (Nd) to the p-type impurity concentration in the p-type _{Al.sub.Y2Ga.sub.1 -Y2N} layer 15 satisfies formula (1).

0.009≦(Nd/Na)<0.185...Formula (1)

Furthermore, it is even more preferable that Nd/Na satisfies one of the following formulas:

0.009≦(Nd/Na)<0.135...Formula (2)
0.038≦(Nd/Na)<0.185...Formula (3)

By co-doping the p-type Al _Y2 Ga _1-Y2 N layer 15 with Nd/Na so as to satisfy any one of the above formulas (1) to (3), an ultraviolet LED having high luminous efficiency and good output maintenance rate (device life) is realized.

The amount of p-type impurities doped into the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably 1×10 ¹⁷ to 1.2×10 ²⁰ cm ⁻³ . As theoretically shown in J. Applmaru Physmaru, Volmaru 95, No. 8, 15 April (2004), the amount of nitrogen defects, which are considered to be a cause of deterioration, is thought to increase with the amount of p-type impurities in the p-type Al _Y2 Ga _1-Y2 N layer 15. Therefore, when the amount of p-type impurities exceeds 1.2×10 ²⁰ cm ⁻³ , the amount of nitrogen defects formed in the initial stage becomes too large, making it difficult to obtain a high output maintenance rate.

Furthermore, when the p-type impurity concentration is low, particularly when the Al composition Y2 is constant, the hole concentration decreases and the minority carrier (electron) mobility increases, and when the Al composition Y2 is inclined, the minority carrier (electron) mobility increases, causing a decrease in output and making it difficult to obtain high luminous efficiency. Therefore, the p-type impurity concentration can be appropriately determined within the above range, taking such a trade-off into consideration, but in order to obtain a higher output maintenance rate and high output, it is preferably 1×10 ¹⁹ to 5×10 ¹⁹ cm ^-3 , and more preferably 1×10 ¹⁹ to 4×10 ¹⁹ cm ^-3 .

The amount of n-type impurity doped into the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably 1.1×10 ¹⁸ or more and 9.0×10 ¹⁸ cm ^-3 or less, and more preferably 1.8×10 ¹⁸ or more and 8.0×10 ¹⁸ cm ^-3 or less. With these amounts of n-type impurities, a light emitting device 10 with high luminous efficiency can be obtained.

The p-type impurity and n-type impurity doped in the p-type Al _Y2 Ga _1-Y2 N layer 15 may have a constant concentration within the layer, or may have a concentration difference in the stacking direction. For example, the side in contact with the Al _Y1 Ga _1-Y1 N layer 14 may be a co-doped layer, and the remaining p-type Al _Y2 Ga _1-Y2 N layer 15 may be a layer not doped with an n-type impurity.

The Al composition Y2 of the p-type Al _Y2 Ga _1-Y2 N layer 15 is equal to or greater than 0.5 and equal to or less than 1.0, and is equal to or less than the Al composition Y1 of the Al _Y1 Ga _1-Y1 N layer 14 .

In the case of a structure in which Y2 is a constant value in the stacking direction, the Al composition Y2 of the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably greater than the Al composition of the barrier layer of the active layer and less than or equal to the Al composition Y1 of the Al _Y1 Ga _1-Y1 N layer 14. By setting the Al composition Y2 of the p-type Al _Y2 Ga _1-Y2 N layer 15 within the above range, a high carrier overflow suppression effect can be obtained even when the injection current amount of the ultraviolet light-emitting device is large. In order to obtain a higher effect, it is preferable that the difference between the Al composition of the barrier layer of the active layer and the Al composition Y2 of the p-type Al _Y2 Ga _{1-Y2 N layer 15 is 0.5 or more. In addition, it is preferable that the Al composition Y2 of the p-type Al Y2 Ga 1-Y2 N layer} 15 is greater than the Al composition of the n-type AlGaN layer 12, which enhances the effect of suppressing carrier overflow to the p-type layer and increases the luminous efficiency of the ultraviolet light-emitting device. From the above, in the case of a structure in which Y2 is a constant value in the stacking direction, the Al composition Y2 of the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably 0.6 or more and 0.9 or less.

Moreover, the p-type Al _Y2 Ga _1-Y2 N layer 15 may be a composition gradient layer in which the Al composition Y2 changes in the stacking direction. In particular, it is preferable that the structure has an Al composition Y2 that decreases in the stacking direction from the side in contact with the Al _Y1 Ga _1-Y1 N layer 14. This allows a polarization doping effect to be obtained in the p-type Al _Y2 Ga _1-Y2 N layer 15, making it easier to obtain a higher hole concentration, and as a result, the efficiency of hole injection into the active layer is increased. For example, when the emission wavelength is 270 nm or less, the Al composition on the side in contact with the Al _Y1 Ga _1-Y1 N layer 14 is preferably 0.95 to 1.0, and the Al composition in the surface layer of the p-type Al _Y2 Ga _1-Y2 N layer 15 on the opposite side is preferably 0.60 to 0.85. By adopting such a structure, the above-mentioned polarization doping effect can be enhanced and transparency to the emission wavelength can be maintained, making it easier to obtain high emission efficiency.

The thickness of the p-type Al _Y2 Ga _1-Y2 N layer 15 is not particularly limited, but may be appropriately determined within the range of 10 to 150 nm. If the thickness of the p-type Al _Y2 Ga _1-Y2 N layer 15 is less than 10 nm, it becomes difficult to obtain the above-mentioned effect of suppressing carrier overflow, while if the thickness is thick and exceeds 150 nm, the resistance value of the p-type Al _Y2 Ga _1-Y2 N layer 15 increases, resulting in an increase in the operating voltage of the ultraviolet light-emitting device. From this viewpoint, the thickness of the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably 40 to 120 nm, and particularly preferably 50 to 100 nm.

On the p-type Al _Y2 Ga _1-Y2 N layer 15, a p-type GaN layer 16 doped with a p-type dopant may be formed in order to reduce the contact resistance with the electrode. The above-mentioned known p-type dopant materials can be used as the p-type dopant material, but for the same reason, it is preferable to use Mg. The Mg doping concentration in the p-type GaN layer 16 is not particularly limited, but is preferably 1×10 ¹⁸ to 2×10 ²⁰ cm ⁻³ in order to reduce the resistance value in the p-type GaN layer and the contact resistance with the electrode. The film thickness of the p-type GaN layer 16 is also not particularly limited, and may be appropriately determined in the range of 5 to 500 nm.

When AlN is used as the substrate 11, all of the AlGaN layers 12, 13, 14, and 15, except for the p-type GaN layer 16, are crystal-grown in a state of being lattice-bonded to the AlN substrate 11, and therefore have a low dislocation density equivalent to that of the AlN substrate 11. Specifically, the dislocation density is 10 ⁵ cm ⁻² or less.

Although the ultraviolet semiconductor light-emitting element 10 has been described as being a light-emitting diode (LED), it may also be configured as a semiconductor laser element (LD: Laser Diode).

As the ultraviolet semiconductor light-emitting element 10, a semiconductor light-emitting element having a substrate 11 has been described, but the substrate 11 can be removed as necessary or can be provided as an optional element, depending on the properties of the substrate used, etc.

Although the ultraviolet semiconductor light emitting element 10 has been described as a semiconductor light emitting element having the substrate 11 on the n-type AlGaN layer 12 side, a support substrate other than the substrate 11 used for epitaxial growth may be provided on the p-type GaN layer 16 side. In this case, the material of the support substrate is not particularly limited, and any known material used as a support substrate member for a light emitting element, such as _{polycrystalline} AlN, Si, _Al2O3Cu , or CuW, may be used without limitation.

Below, we will explain the ultraviolet semiconductor light-emitting element 10 as a semiconductor light-emitting element having a substrate 11 as a growth substrate of a semiconductor laminate structure, but as mentioned above, it is also possible to use a semiconductor light-emitting element having a substrate 11 that is different from the growth substrate such as a support substrate.

Next, a method for manufacturing an ultraviolet LED having the structure described above will be described. The ultraviolet LED 10 of the present invention can be manufactured by known crystal growth methods such as MOCVD and molecular beam epitaxy (MBE). Among these, the MOCVD method is preferred because it has high productivity and is widely used industrially. For the Group III (Al, Ga) source gas and Group V (N) source gas used in the present invention, any known source gas can be used without particular restrictions.

For example, gases such as trimethylaluminum, triethylaluminum, trimethylgallium, and triethylgallium can be used as Group III source gases. Ammonia is usually used as Group V source gas.

In addition, known materials can be used without restrictions as dopant source gases for Mg and Si, such as biscyclopentadienyl magnesium, monosilane, and tetraethylsilane.

The above source gases are supplied onto the substrate 11 together with a carrier gas such as hydrogen and/or nitrogen to grow the element layer of the UV LED 10.

The supply ratio of the Group III source gas to the Group V source gas (V/III ratio) may be appropriately determined so as to obtain the desired characteristics, but is preferably set within the range of 500 to 10,000.

The growth temperature of the element layers constituting the UV-LED 10 is not particularly limited and may be appropriately determined so as to obtain the desired characteristics of each layer and the characteristics of the UV-LED 10. However, growth at 1000 to 1200°C is preferable, and 1000 to 1150°C is more preferable.

The present invention will be specifically explained below using an example in which an ultraviolet LED with an emission wavelength of 265 nm was fabricated, but the present invention is not limited to this example.

[Fabrication of element]
The substrate on which the UV-LED element layer was grown was an AlN single crystal substrate fabricated by the method described in Applied Physics Express 5 (2012) 122101. Specifically, it was a laminated substrate comprising a C-plane AlN seed substrate fabricated by physical vapor transport (PVT) and an AlN thick film grown by HVPE. The dislocation density of this AlN substrate was 10 ⁵ cm ^-2 or less, and the surface roughness (RMS) in an area of 5 × 5 μm ² was 0.1 nm.

On this AlN substrate, an AlN layer (100 nm), a first n-type AlGaN layer 12A (200 nm), and a second n-type AlGaN layer 12B (1000 nm) were grown by MOCVD. The first n-type AlGaN layer 12A and the second n-type AlGaN layer 12B are both composition-graded layers, and the Al composition of the first n-type AlGaN layer 12A decreases from 1.0 to 0.75 from the side in contact with the AlN layer, and the Al composition of the second n-type AlGaN layer 12B decreases from 0.75 to 0.70 from the side in contact with the first n-type AlGaN layer 12A. The Si concentration in the n-type AlGaN layer 12 was controlled to be 1×10 ¹⁹ cm ⁻³ .

Next, an active layer 13 was grown, which was a triple quantum well layer consisting of a barrier layer made of n-type Al _0.6 Ga _0.4 N (7 nm) and a well layer made of Al _0.5 Ga _0.5 N (4 nm). The Si concentration of the barrier layer was controlled to be 1×10 ¹⁸ cm ^-3 .

Next, an electron blocking layer 14 made of AlN (10 nm) was grown. The electron blocking layer on the side in contact with the barrier layer was an undoped layer 14A (2 nm), and the remaining electron blocking layer 14B was a layer doped with Mg at 4×10 ¹⁹ cm ^-3 . Note that the undoped AlN layer 14A and the Mg-doped p-type AlN layer 14B were not doped with Si.

Next, a p-type AlGaN layer 15 (60 nm) was grown. The p-type AlGaN layer 15 is a composition gradient layer in which the Al composition decreases from 1.0 to 0.8 from the side in contact with the electron blocking layer. A wafer (hereinafter referred to as structure 1 (EX1)) in which the Mg concentration of the p-type AlGaN layer 15 is 1.0×10 ²⁰ cm ⁻³ (hereinafter sometimes written as 1.0E20 cm ⁻³ ) and a wafer (hereinafter referred to as structure 2 (EX2)) in which the Mg concentration of the p-type AlGaN layer 15 is 4.3E19 cm ⁻³ were fabricated. In each structure, a plurality of wafers were fabricated in which the co-doped Si concentration was adjusted in the range of 1×10 ¹⁸ to 9×10 ¹⁸ cm ⁻³ .

Next, a p-type GaN layer 16 (270 nm) was grown to complete the deep ultraviolet LED element layer. The Mg concentration in the p-type GaN layer 16 was set to 5×10 ¹⁹ cm ^-3 .

Next, the second n-type AlGaN layer was exposed by ICP (inductively coupled plasma) dry etching, after which an n-type electrode was formed by stacking a Ti layer and a Au layer, and heat-treated in a nitrogen atmosphere at 900°C. Next, a p-electrode was formed by stacking a Ni layer and a Au layer on the p-type GaN layer, and heat-treated in an oxygen atmosphere at 500°C.

Next, the back surface of the growth substrate (the side opposite to the UV LED element layer) was mechanically polished until the AlN thick film formed by the HVPE method was exposed, completing the UV LED. Next, the substrate was cut into chip shapes of 0.75 mm x 0.95 mm by dicing, and then flip-chip bonded onto a ceramic submount to complete the UV LED element.

Figure 3 is a table showing the Mg concentration, Si concentration, and the ratio of Si to Mg concentration (Si/Mg) in the p-type AlN layer 14 and the p-type AlGaN layer 15.

In addition, wafers not co-doped with Si were also produced as Comparative Example 1 and Comparative Example 2 (hereinafter referred to as CM1 and CM2) for Structure 1 (EX1) and Structure 2 (EX2). The Mg concentrations of the p-type AlGaN layer 15 of the CM1 and CM2 wafers were 1.0E20 cm ^-3 and 4.3E19 cm ^-3 , respectively, which are the same as those of the wafers of Structure 1 (EX1) and Structure 2 (EX2). Si was also incorporated in the structure of the comparative example without co-doping, and about 7 x 10 ¹⁶ to 1 x 10 ¹⁷ cm ^-3 of Si was detected by SIMS (Secondary Ion Mass Spectrometry) measurement, but this shows the lower limit of detection in SIMS analysis, and the actual concentration is lower. In other words, the measured concentration of about 7×10 ¹⁶ to 1×10 ¹⁷ cm ⁻³ indicates that the concentration is no higher than this.

[Element Evaluation]
The Mg concentration, Si concentration, and light emission characteristics of the various ultraviolet LEDs 10 described above were evaluated. The Mg concentration and Si concentration were measured by SIMS measurement for each element. For the quantification of the AlN layer and the AlGaN layer, AlN and Al _0.65 Ga _0.35 N were used as standard samples, respectively.

(1) Initial Light Emission Efficiency Figure 4A is a graph plotting the relationship between the Si concentration of the p-type Al _Y2 Ga _1-Y2 N layer 15 (hereinafter also referred to as the p-type cladding layer 15) and the light emission efficiency (arbitrary unit) at the beginning of current flow. In both Structure 1 and Structure 2, the initial light emission efficiency hardly changed or only showed a slight decreasing tendency by co-doping with Si. However, in Structure 1, when the Si concentration increased to 8E18 cm ^-3 , the light emission efficiency rapidly decreased, and in Structure 2, when the Si concentration increased to 9E18 cm ^-3 , the light emission efficiency rapidly decreased.

4B is a graph plotting the relationship between the Si/Mg ratio (Si concentration/Mg concentration) of the p-type Al _Y2 Ga _1-Y2 N layer 15 and the luminous efficiency (arbitrary unit) at the beginning of current flow. In Structure 1, the luminous efficiency decreased rapidly when the Si/Mg ratio was 0.135 or more, and in Structure 2, the luminous efficiency decreased rapidly when the Si/Mg ratio was 0.185 or more.

That is, it was found that the Si concentration of the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably less than 9E18 cm ^-3 , and more preferably less than 8E18 cm ^-3 . Also, it was found that the Si/Mg ratio of the p-type Al _Y2 Ga _1-Y2 N layer 15 is preferably less than 0.185, and more preferably less than 0.135.

(2) Output Maintenance Rate Fig. 4C is a graph showing the output maintenance rate (@100h) after 100 hours from the initial time versus the Si concentration of the p-type cladding layer 15. More specifically, the output maintenance rate, which is the rate of change in element output after 100 hours (hr) of current flow to the element from the start of current flow to the element, is plotted versus the Si concentration of the p-type cladding layer 15. Note that current was passed through the element in a constant current (ACC) drive mode in which a drive current of 450 mA was passed through the element.

Figure 4D is a graph showing the Si/Mg ratio (Si concentration/Mg concentration) of the p-type cladding layer 15 and the output maintenance rate. In both the case of structure 1 (EX1) and structure 2 (EX2), the output maintenance rate shows a tendency to improve as the Si concentration and Si/Mg ratio of the p-type cladding layer 15 increase.

(Structure 1: EX1)
4C and 3, in the case of Structure 1 (EX1: 1.0E20 cm ⁻³ ) in which the Mg concentration of the p-type cladding layer 15 is high, the output maintenance rate improved when the Si concentration was 1.1E18 cm ⁻³ or more. Also, when the Si concentration was 9.0E18 cm ⁻³ or less, no output reduction occurred.

As for the Si/Mg ratio, referring to FIG. 4D and FIG. 3, the output maintenance rate improved when the Si/Mg ratio was 0.009 or more. Also, when the Si/Mg ratio was less than 0.135, there was no decrease in output.

(Structure 2: EX2)
4C and 3, in the case of structure 2 (EX2: 4.3E19 cm ⁻³ ) in which the Mg concentration of the p-type cladding layer 15 is low, the output maintenance rate improved when the Si concentration was 1.8E18 cm ⁻³ or more. Also, when the Si concentration was 8.0E18 cm ⁻³ or less, no output reduction occurred.

As for the Si/Mg ratio, referring to FIG. 4D and FIG. 3, the output maintenance rate improved when the Si/Mg ratio was 0.038 or more. Also, when the Si/Mg ratio was less than 0.185, there was no decrease in output.

(Si/Mg ratio)
4D, in both the case of Structure 1 and Structure 2, the power retention ratio increases almost linearly with an increase in the Si/Mg ratio, and it is considered that the Si/Mg ratio is a dominant physical parameter that influences the power retention ratio. Note that the Si/Mg ratio is estimated to be inversely proportional to the concentration of nitrogen defects formed in the undoped state.

Specifically, when the Si/Mg ratio is 0.009 or more, 0.038 or more, and 0.064 or more, the output maintenance ratio is expected to be 0.7 or more, 0.8 or more, and 0.9 or more, respectively.

Even when Si is not co-doped (CM1 (Comparative Example 1) and CM2 (Comparative Example 2)), Comparative Example 2 (CM2), which has a low Mg concentration, shows a higher output maintenance rate. As will be described later (Figure 5), this is thought to be because Structure 1 and Comparative Example 1, which have a high Mg concentration, have a large initial value for the nitrogen vacancy concentration, which is thought to be the true cause of deterioration.

Therefore, by co-doping the p-type cladding layer 15 so that the Si/Mg ratio satisfies the following formula, an ultraviolet LED with high luminous efficiency and good output maintenance rate (device life) is realized.

0.009≦(Si/Mg)<0.185...Formula (1)

It is further preferable that the Si/Mg ratio satisfies one of the following formulas:

0.009≦(Si/Mg)<0.135...Formula (2)
0.038≦(Si/Mg)<0.185...Formula (3)

(Initial output and output maintenance rate)
FIG. 5 is a graph showing the output maintenance rate (@100 h) versus the initial light output (arbitrary units) at the start of energization for elements obtained from Structure 1 (EX1), Structure 2 (EX2), Comparative Example 1 (CM1), and Comparative Example 2 (CM2).

In the elements of Comparative Examples 1 and 2 (CM1, CM2), reliability tests show that the output maintenance rate is small at the beginning of current application (up to 100 hours), and the drop in light output is large. In addition, there is a trade-off in that the higher the efficiency (i.e., the higher the output), the greater the drop in output, making it difficult to produce an element that is both high output and highly reliable.

On the other hand, in the elements of structures 1 and 2 (EX1, EX2) of the present invention, the output maintenance rate is large at the beginning of charging (up to 100 hours), and it can be seen that the decrease in optical output is suppressed. Also, structure 2 (EX2), which has a low Mg concentration in the p-type cladding layer 15, has a higher output maintenance rate than structure 1 (EX1).

It was confirmed that there is a correlation between the output maintenance rate (@100h) and the long-term life (1000h). Therefore, it was confirmed that the elements of structures 1 and 2 (EX1, EX2) are also superior in terms of long-term life, and according to the present invention, it is possible to realize an element that is highly efficient, has high output, and is highly reliable.

[Estimation of Deterioration Causes]
It has been reported that a common cause of output reduction in ultraviolet LEDs is degradation of the active layer, that is, an increase in point defects in the active layer, which reduces the recombination probability (radiative transition probability).
Based on the above findings and the results of verification experiments carried out in the process of completing the present invention, the inventors have come to deduce the degradation mechanism of UV LEDs as follows. That is, it is presumed that doping the p-type Al _Y2 Ga _1-Y2 N layer 15 with Si, which is a donor, reduces the nitrogen defects in the p-type Al _Y2 Ga _1-Y2 N layer 15. As a result, it is presumed that the amount of nitrogen defects diffusing into the active layer is reduced, suppressing the degradation of the active layer and achieving a high output maintenance rate.

(Co-doped impurities)
In the above embodiment, Mg is used as the p-type impurity in the p-type Al _Y2 Ga _1-Y2 N layer 15 that is the p-type cladding layer, and Si is used as the co-doped impurity, but the present invention is not limited to this.

As described above, the mechanism of the present invention is _believed to be that the effect of _reducing nitrogen defects is exerted by the interaction between Mg, which is an acceptor impurity, and Si, which is a donor impurity, doped into the p-type Al _Y2 Ga _{1-Y2 N layer 15. Therefore, any material that functions as an acceptor and a donor for the p-type Al Y2 Ga 1-Y2} N layer 15 can be used without any restrictions.

Specifically, in addition to Mg, Zn (zinc), Be (beryllium), C (carbon), etc. can be used as p-type impurities that act as acceptors. In addition to Si, Ge (germanium), Se (selenium), S (sulfur), O (oxygen), etc. can be used as n-type impurities (co-doped impurities) that act as donors.

In general, therefore, when the p-type impurity concentration of the p-type Al _{Y 2} Ga _{1-Y 2} N layer 15 is Na and the n-type impurity concentration is Nd, it is preferable that Nd/Na satisfies the following formula:

0.009≦(Nd/Na)<0.185...Formula (4)

It is even more preferable that Nd/Na satisfies one of the following formulas:

0.009≦(Nd/Na)<0.135...Formula (5)
0.038≦(Nd/Na)<0.185...Formula (6)

Therefore, by co-doping the p-type cladding layer 15 with Nd/Na satisfying the above formula, an ultraviolet LED with high luminous efficiency and good output maintenance rate (device life) is realized.

As described above in detail, this embodiment can provide an ultraviolet LED that is highly efficient, has high output, and has minimal degradation and an excellent element lifespan.

10: ultraviolet semiconductor light-emitting element, 11: substrate, 12: n-type AlGaN layer, 12A: first n-type _{Al.sub.X1Ga.sub.1 -X1N} layer, 12B: second n-type _{Al.sub.X2Ga.sub.1 -X2N} layer, 13: active layer, 14: _{Al.sub.Y1Ga.sub.1 -Y1N} layer, 14A: AlN layer, 14B: p-type AlN layer, 15: p-type AlGaN layer, 16: p-type GaN layer

Claims (9)

a semiconductor structure layer in which an n-type AlGaN layer, an active layer, an _{AlY1Ga1 -Y1N} layer (0.8≦Y1≦1.0) and a p-type _{AlY2Ga1 -Y2N} layer (0.5≦Y2≦1.0, Y2≦Y1) are epitaxially grown in this order;
the p-type _{AlY2Ga1 -Y2N} layer is co-doped with a p-type impurity serving as an acceptor and an n-type impurity serving as a donor, and a concentration ratio (Nd/Na) of a concentration (Nd) of the n-type impurity to a concentration (Na) of the p-type impurity satisfies 0.009≦(Nd/Na)<0.185;
the p-type impurity concentration of the p-type Al _Y2 Ga _1-Y2 N layer is 1×10 ¹⁷ to 1.2×10 ²⁰ cm ⁻³ ,
The _{Al.sub.Y1 Ga.sub.1-Y1} N layer does not contain the n-type impurity or the concentration of the n-type impurity is less than the concentration of the n-type impurity in the p-type _Al.sub.Y2 Ga.sub.1 _-Y2 N layer. The ultraviolet semiconductor light-emitting element according to claim 1, wherein the p-type impurity is Mg (magnesium) and the n-type impurity is Si (silicon). The ultraviolet semiconductor light-emitting element according to claim 2, wherein the ratio of the concentration of the n-type impurity to the concentration of the p-type impurity, Si/Mg, satisfies 0.009≦(Si/Mg)<0.135. The ultraviolet semiconductor light-emitting element according to claim 2, wherein the concentration ratio satisfies 0.038≦(Si/Mg)<0.185. 5. The ultraviolet semiconductor light emitting element according to claim 1, wherein said _{Al.sub.Y1Ga.sub.1 -Y1N} layer contains Mg as an impurity. 6. The ultraviolet semiconductor light emitting element according to claim 1, wherein the p-type _{Al.sub.Y2 Ga.sub.1-Y2} N layer is a composition gradient layer in which the Al composition decreases with increasing distance from the _Al.sub.Y1 Ga.sub.1 _-Y1 N layer. 7. The ultraviolet semiconductor light-emitting element according to claim 1, wherein the _{Al.sub.Y1Ga.sub.1 -Y1N} layer includes an undoped layer formed on the active layer, and an Mg-doped layer formed on the undoped layer. 8. The ultraviolet semiconductor light-emitting element according to claim 1, wherein the _{Al.sub.Y1Ga.sub.1 -Y1N} layer is an AlN layer. The semiconductor structure layer is formed on a substrate,
9. The ultraviolet semiconductor light-emitting device according to claim 1, wherein the substrate is a single crystal AlN substrate.

Images