1255053 九、發明說明： 【發明所屬之技術領域】 本發明係有關具多量子井結構發光層的pn接合型m族 氮化物半導體發光裝置，其中含有I[[族氮化物半導體的并層 與阻障層係週期性交錯堆疊於在結晶基板上形成並含有瓜 族氮化物半導體的η型覆層與p型覆層之間。 【先前技術】 m族氮化物半導體通常係作爲組成諸如發光二極體 # (LED)之PN異質接面結構的m族氮化物半導體發光裝置的 功能材料，其可發出短波長可見光（見JP-A 2 00 0-332364)。 例如，當要形成發射近紫外光區、藍光區或綠光區之光的 LED時，其係使用n型或p型氮化鋁鎵（AlxGaYN : 0 $ X， YS1’ X+Y=l)形成覆層（見 jp_A 2003-229645)。氮化 鎵銦（GaYInzN: OS Y，ZS 1，Υ+Ζ=1)用於形成活化層（發 光層）（見 JP-B SHO 55-3 834 )。 用於由單一 GaYInzN(〇$Y，ZS1，Y+Z=l)形成發 ^ 光層的技術先前業已揭示（見JP-B SHO 55-3834)。由被稱 爲量子井結構的超晶格結構形成發光層的實施例亦爲所熟 知（見JP-A 200 1 - 102 6 29 )。量子井結構爲阻障層與井層週 期性交錯堆疊的結構（見J p _ A 2 0 0 0 - 1 3 3 8 8 3 )。例如，所揭 示的實施例係將GaG.7InQ 3N層作爲井層且將氮化鎵（GaN)作 爲阻障層，以及形成多量子井結構發光層（見jp_A 2000- 1 02629中的實施例〇 。 然而’在前揭習用技術中，發光層的二個端面不僅限於 1255053 ^阻障層，且當其一端面爲井層時，載體（電子）易於經由井 層而向覆層散佈，而具有發光輸出會以等量縮減的問題。 縱使接觸P型覆層之發光層的另一端面形成有阻障層， 倘阻障層厚度不足，來自η型覆層的載體便會流向p型覆 層，而具有發光輸出會受此影響而縮減的問題。 【發明內容】 本發明已鑑於前揭狀況而完成，且本發明目的在於提供 ρν接合型m族氮化物半導體發光裝置，縱使發光層爲多量 • 子井結構時，其仍可抑制載體向覆層散佈，而大幅提高發光 輸出。 爲達前揭目的，本發明提供一種具有多量子井結構發光 層的PN接合型瓜族氮化物半導體發光裝置，其中含有皿族 氮化物半導體的井層與阻障層係週期性交錯堆疊於形成在 結晶基板上並含有H[族氮化物半導體的η型覆層與p型覆層 之間，其中發光層的一端面層係最靠近並正對於η型覆層， 而發光層的另一端面層係最靠近並正對於ρ型覆層，該一端 Φ 面層與另一端面層爲阻障層，以及該另一端面層較該一端面 層的阻障層爲厚。 在ΡΝ接合型ΙΠ族氮化物半導體發光裝置中，各阻障層 的厚度由該一端面層向該另一端面層漸增。 在第一及第二次提及的ρν接合型m族氮化物半導體 中，該另一端面層的雜質濃度在其與井層的接面部位爲低， 在其中間部位爲最高，且由中間部位向P型覆層漸減。 在第一至第三次提及之PN接合型ΠΙ族氮化物半導體的 1255053 . 任一個中’該另一端面層已接合有一未刻意摻有雜質的井 層。 根據本發明，在多量子井結構發光層中，形成發光層之 二個端面的薄層爲阻障層。由於載體被封在井層中，因此得 以抑制載體向覆層散佈，而提高發光輸出。 與P型覆層接觸之形成多量子井結構發光層二端面的該 阻障層的該另一端面層較該一端面層爲厚。因此，來自η型 覆層的載體不會流向ρ型覆層，而可提高發光輸出。 # 根據本發明，發光效率可提高約1.5倍。因此，LED燈 的發光輸出與發光轉換效率可提高約1.5倍，而可節約能源。 根據本發明，各阻障層厚度係由該一端面層向該另一端 面層漸增。因此，可將載體更緊密地封在井層中。在該狀況 中’各阻障層的厚度最好設定在使來自井層的發光波長不會 改變的數値。 此外，該另一端面層的雜質濃度在其與井層的接面部位 低’在其中間部位濃度爲最高，且由中間部位向P型覆層漸 # 減。因此，影響與其他層接觸之井層的環境及對多量子井層 的電洞充電效率皆得到滿足。 此外，未刻意摻入雜質的井層係接合於該另一個端面 層。因此，可提高井層結晶度與發光效率。 實行本發明的最佳態樣 本發明的實施例將詳細說明如下。 根據本發明，PN接合型瓜族氮化物半導體發光裝置係 由形成於結晶基板上的η型或p型瓜族氮化物半導體層所形 1255053 成，特別是形成於單晶基板上者。 亦即，本發明的pn接合型m族氮化物半導體發光裝置 具有多量子井結構發光層，其中含有m族氮化物半導體的井 層與阻障層係週期性交錯堆疊於形成在結晶基板上並含有 瓜族氮化物半導體的η型覆層與p型覆層之間’發光層的一 端面層係最靠近並正對於η型覆層，而發光層的另一端面層 係最靠近並正對於Ρ型覆層，該一端面層與另一端面層皆爲 阻障層，以及該另一端面層較該一端面層的阻障層爲厚。1255053 IX. Description of the Invention: [Technical Field] The present invention relates to a pn junction type m-nitride semiconductor light-emitting device having a multi-quantum well structure light-emitting layer, which contains I [[substrate and resistance of a family of nitride semiconductors] The barrier layer is periodically interleaved between the n-type cladding layer and the p-type cladding layer formed on the crystalline substrate and containing the quaternary nitride semiconductor. [Prior Art] A group m nitride semiconductor is generally used as a functional material of a group m nitride semiconductor light-emitting device constituting a PN heterojunction structure such as a light-emitting diode # (LED), which emits short-wavelength visible light (see JP- A 2 00 0-332364). For example, when forming an LED that emits light in a near-ultraviolet, blue, or green region, it uses n-type or p-type aluminum gallium nitride (AlxGaYN: 0 $ X, YS1' X+Y=l) Form a coating (see jp_A 2003-229645). Indium gallium nitride (GaYInzN: OS Y, ZS 1, Υ + Ζ = 1) is used to form an active layer (light-emitting layer) (see JP-B SHO 55-3 834). A technique for forming a light-emitting layer from a single GaYInzN (〇$Y, ZS1, Y+Z=l) has been previously disclosed (see JP-B SHO 55-3834). An embodiment in which a light-emitting layer is formed of a superlattice structure called a quantum well structure is also known (see JP-A 200 1 - 102 6 29 ). The quantum well structure is a structure in which the barrier layer and the well layer are periodically staggered and stacked (see J p _ A 2 0 0 0 - 1 3 3 8 8 3 ). For example, the disclosed embodiment uses a GaG.7 InQ 3N layer as a well layer and gallium nitride (GaN) as a barrier layer, and a multi-quantum well structure light-emitting layer (see the example in jp_A 2000-01829) However, in the prior art, the two end faces of the luminescent layer are not limited to the 1255053 ^ barrier layer, and when one end face is a well layer, the carrier (electron) is easily spread to the cladding via the well layer, and has The light output will be reduced by the same amount. Even if the other end surface of the light-emitting layer contacting the P-type cladding layer is formed with a barrier layer, if the thickness of the barrier layer is insufficient, the carrier from the n-type cladding layer flows to the p-type cladding layer. The present invention has been made in view of the foregoing, and an object of the present invention is to provide a ρν junction type m-nitride semiconductor light-emitting device, even though the luminescent layer is a large amount. • When the sub-well structure is used, it can still suppress the dispersion of the carrier to the cladding layer, and greatly increase the light-emitting output. To achieve the foregoing, the present invention provides a PN-bonded melon-type nitride having a multi-quantum well structure light-emitting layer. a semiconductor light-emitting device in which a well layer and a barrier layer containing a nitride semiconductor are periodically interleaved and stacked between an n-type cladding layer and a p-type cladding layer formed on a crystalline substrate and containing an H-type nitride semiconductor. Wherein one end layer of the luminescent layer is closest to and facing the n-type cladding layer, and the other end surface layer of the luminescent layer is closest to and facing the p-type cladding layer, the one end Φ surface layer and the other end surface layer being barrier The layer, and the other end face layer is thicker than the barrier layer of the one end face layer. In the meandering-type bismuth nitride semiconductor light-emitting device, the thickness of each barrier layer is from the one end face layer to the other end face layer In the first and second mentioned ρν junction type m-type nitride semiconductors, the impurity concentration of the other end face layer is lower at the junction portion with the well layer, and the middle portion is the highest. And gradually decreasing from the intermediate portion to the P-type cladding layer. In the first to third mentioned PN junction type lanthanum nitride semiconductor 1255053. In any one of the 'end surface layer has been bonded with an unintentionally doped impurity Well layer. According to the invention, in a multi-quantum well In the light-emitting layer, the thin layer forming the two end faces of the light-emitting layer is a barrier layer. Since the carrier is sealed in the well layer, the carrier is prevented from being scattered to the cladding layer, and the light-emitting output is improved. Contact with the P-type cladding layer The other end face layer of the barrier layer forming the two end faces of the multi-quantum well structure light-emitting layer is thicker than the one end face layer. Therefore, the carrier from the n-type cladding layer does not flow to the p-type cladding layer, and the light-emitting output can be improved. According to the invention, the luminous efficiency can be increased by about 1.5 times. Therefore, the luminous output and the luminous conversion efficiency of the LED lamp can be improved by about 1.5 times, and energy can be saved. According to the invention, the thickness of each barrier layer is formed by the one end face. The layer is gradually increased toward the other end face layer. Therefore, the carrier can be more closely packed in the well layer. In this case, the thickness of each barrier layer is preferably set so that the wavelength of the light emitted from the well layer does not change. Counting. Further, the impurity concentration of the other end face layer is lower at the junction portion with the well layer, and the concentration is the highest in the middle portion thereof, and is gradually decreased from the intermediate portion to the P-type cladding layer. Therefore, the environment affecting the well layers in contact with other layers and the charging efficiency of the holes in the multi-quantum well layer are all satisfied. Further, a well layer not intentionally doped with impurities is bonded to the other end face layer. Therefore, the well layer crystallinity and luminous efficiency can be improved. BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below. According to the invention, the PN junction type cerium nitride semiconductor light-emitting device is formed of 1255053 in an n-type or p-type quaternary nitride semiconductor layer formed on a crystal substrate, in particular, formed on a single crystal substrate. That is, the pn junction type m-nitride semiconductor light-emitting device of the present invention has a multi-quantum well structure light-emitting layer in which a well layer and a barrier layer containing a group m nitride semiconductor are periodically stacked alternately on a crystalline substrate and The one end layer of the 'light-emitting layer' between the n-type cladding layer and the p-type cladding layer containing the cuban nitride semiconductor is closest to the n-type cladding layer, and the other end surface layer of the light-emitting layer is closest and positive In the Ρ-type cladding layer, the one end surface layer and the other end surface layer are both barrier layers, and the other end surface layer is thicker than the barrier layer of the one end surface layer.
較靠近結晶基板的覆層可爲η型或ρ型。 當形成ΙΠ族氮化物半導體層時，係使用單晶作爲基板材 料。得以使用下列材料作爲基板材料：諸如藍寶石（cx-A1203 單晶）、氧化鋅（ZnO )與氧化鋰鎵（LiGa02 )之具有相當 高熔點與耐熱性的氧化物單晶材料，諸如矽（Si )單晶、立 方晶體與六方晶碳化矽之IV族半導體單晶，以及諸如磷化鎵 (GaP)之ΙΠ-V族化合物半導體卓晶材料。來自發光層之光 可穿透的透光單晶材料可有效地作爲基板使用。The coating layer closer to the crystalline substrate may be n-type or p-type. When a bismuth nitride semiconductor layer is formed, a single crystal is used as a base material. The following materials can be used as substrate materials: oxide single crystal materials such as samarium (cx-A1203 single crystal), zinc oxide (ZnO) and lithium gallium oxide (LiGaO 2 ) having a relatively high melting point and heat resistance, such as germanium (Si) a group IV semiconductor single crystal of a single crystal, a cubic crystal and a hexagonal tantalum carbide, and a bismuth-V compound semiconductor crystal material such as gallium phosphide (GaP). Light permeable, single crystal material that is permeable to light from the luminescent layer can be effectively used as a substrate.
設於單晶基板上的η型與ρ型Π族氮化物半導體層（覆 層）具有組成物化學式爲 AlxGaYlnzNuMa的組成物 (OgXgl，0SYS1，0SZS1，Χ+γ+Ζ=1，符號 μ 代表 異於氮的V族元素，以及0 S a < 1 )。當於單晶基板與形成 在基板上的瓜族氮化物半導體層（覆層）之間有晶格失配存 在時，最好將調適該失配並成爲具有極佳結晶度上層的低或 高溫緩衝層置於這些層之間。緩衝層可由諸如氮化鋁鎵 (AlxGaYN: 0SXS1，0‘YS1，Χ+Υ=ι)所形成。 1255053 η型與p型覆層不一定要由具相同組成物比例的1Π族氮 化物半導體材料所形成。例如，η型覆層可由氮化鎵（G aN ) 形成，而P型覆層可由氮化鋁鎵（AlxGaYN )形成。η型與p 型覆層最好是由能帶間隙較組成發光層量子井結構之阻障 層爲大的m族氮化物半導體材料所形成。 配置於η型覆層與p型覆層間的發光層爲多量子井結 構，其中由Π族氮化物半導體所形成的井層與阻障層係週期 性交錯堆疊，且二個端面層（一端面層與另一端面層）爲阻 φ 障層。此乃因更有效避免載體向η型與p型覆層散佈。在組 成本發明發光層的薄層中，最靠近並正對於η型覆層的該一 端面層及最靠近並正對於Ρ型覆層的該另一端面層爲阻障 層。 井層由能帶間隙小於阻障層的HI族氮化物半導體所形 成。例如，井層由氮化鎵銦所形成（GaYInzN : 0‘ YS 1， 〇SZgl，Y+Z=l) 。GaYInzN的銦組成物比例（鎵組成物 比例）係經適當選擇，以獲得希冀波長的光。發射近紫外光 # 區、藍光區或綠光區短波長光之發光裝置的發光層可由銦組 成物比例約0.40 (=40%)以下（最好爲0.25)的GaYInzN (0.25SZS0.40，Y+Z=l)所形成。 組成多量子井結構發光層的井層可由含氮及氮以外之 V族元素的瓜族氮化物半導體所形成，諸如磷化鎵氮 (GaNi_aPa : 〇$ a< 1 )。在GaNbaPa中，能帶間隙依磷組 成物比例（=a )(換言之，依氮化物（N)組成物比例（=；l - a )) 而急遽變化。能夠射出由紫外光帶至紅光帶之光的井層可由 1255053 I 使用該能帶弓曲的GaN^Pa所形成。 爲獲得具低正向電壓（Vf )或閾電壓（vth ) 強度的發光裝置，形成多量子井結構之井層數最好 上且爲/、個以下。要接合於形成量子井結構該另一* 阻障層的井層係由未摻雜m族氮化物半導體層所形 定來自1多量子井結構發光層的發光波長。其他井層 接合於該一端面層之井層）可由摻有雜質的瓜族氮 體層所形成。接合於該另一端面層之層以外的井層 Φ 發光波長。倘若接合於該另一端面層之層以外的井 有雜質的m族氮化物半導體層所形成，則會提高降 V t h的效果。 相較於組成井層之m族氮化物半導體的能帶間 井結構的阻障層係由能帶間隙爲室溫時之電子熱 0.026 eV )之約10倍以上的m族氮化物半導體層 阻障層最好由含複數種V族元素的m族氮化物半 成，諸如八1乂0&丫叫0^乂^1’0^丫^1，\+丫=1)或 •所形成。 在本發明中，該另一端面層的阻障層係由厚度 一阻障層的m族氮化物半導體層所形成。此乃因爲 層穿經該一端面層之阻障層而進入發光層的電子 合於該另一端面層的井層內。量子井結構之該另一 外之阻障層的適當厚度爲15nm以上且50 nm以下 端面層之阻障層的厚度爲其他阻障層厚度之1.2 2.5倍以下。 及高發光 爲三個以 端面層之 “成，以穩 (諸如要 化物半導 不會影響 層係由摻 :低V f或 丨隙，量子 動能（= 所形成。 導體所形 GaN 1 -aPa 大於該另 由η型覆 係封在接 端面層以 。該另一 倍以上且 -10- 1255053 & 縱使量子井結構係使用厚度由該一端面層向該另一端 面層逐漸增加的阻障層所形成，其仍有效於抑制來自η型覆 層的電子過度流入（電子溢流）ρ型覆層。例如，倘若該一 端面層之阻障層的厚度定義爲1.0做爲參考値，則較靠近該 另一端面層而配置之阻障層的厚度爲1 .5，次靠近該另一端 面層配置之阻障層的厚度爲2.0，且該另一端面層之阻障層 的厚度爲2.5。或者，該一端面層與靠近該一端面層之阻障 層厚度相同，在量子井結構中間部位的數個阻障層厚度更爲 # 增加，靠近該另一端面層的阻障層厚度更爲增加，該另一端 面層厚度最大，這些層的厚度係由該一端面層向該另一端面 層增加。 形成該另一端面層的阻障層係由HI族氮化物半導體層 所形成，該m族氮化物半導體層的雜質濃度在其接合於井層 的端面爲低且在其中間部位爲最高，且由中間部位向ρ型覆 層逐漸降低。藉此結構，可穩定來自發光層的發光波長。亦 即，該另一端面層係由摻雜物濃度由該層中間部位向ρ型覆 ^ 層降低的m族氮化物半導體層所形成。組成量子井結構的另 一阻障層可由未摻雜或有摻雜雜質之層所形成。倘若該另一 阻障層及該另一端面層由摻有雜質的π族氮化物半導體所 形成，則可降低Vf或Vth。在該阻障層的狀況中，摻雜雜 質之層中的雜質濃度分佈方式並未受限。該分佈可同於該另 一端面層的阻障層，或可在厚度方向上均勻分佈。 組成本發明量子井結構的阻障層及井層可使用金屬有 機化學氣相沈積（MOCVD、MOVPE或OMVPE)、分子束磊 1255053 晶（MBE )、鹵化物氣相沈積及氫化物氣相沈積而形成。 Μ Ο C V D特別有效，因爲m族氮化物半導體包含高揮發性元 素，諸如磷（P )、砷（As )及類似物。組成m族氮化物半 導體發光裝置的組成層可使用不同的蒸氣相成長法形成。例 如，形成發光層的阻障層或井層可藉由常壓（實質大氣壓） 或減壓MOCVD形成，而η型或p型覆層可藉由MBE形成， 惟其易於使用相同蒸氣相成長法形成該組成層。 適用於獲得摻雜有雜質之阻障層（諸如η型阻障層）之 # 雜質的實施例爲諸如矽（Si)、鍺（Ge)和錫（Sn)之IV族元素及 諸如硒（Se)、碲（Te)之VI族元素。適用於獲得p型阻障層之 雜質的實施例爲諸如鎂（mg)和鈣（C a)之Π族元素。摻雜在阻 障層中的雜質濃度可使用裝置分析方法量測，諸如二次離子 質譜（SIMS)或歐傑電子分析（Auger electronanolysis)。 該另一端面層的阻障層得以未摻雜狀態成長，亦即未供 應雜質於成長反應系統中。當阻障層厚度達約希冀厚度一半 時，急遽添加大量雜質，而於阻障層中間部位形成含高雜質 ^ 濃度的區域。在形成具希冀厚度且於中間部位形成高濃度雜 質的區域後，要添加的雜質量會隨著時間而減少，並持續成 長直至獲得希冀厚度。倘若要添加至成長反應系統的雜質量 在此方法中隨著時間而變化，則另一端面層的雜質濃度在其 與井層的接面部位爲低，在其中間部位濃度最高，且由中間 部位向P型覆層漸減。 較佳方式爲形成該另一端面層之阻障層中的高雜質濃 度區厚度爲2.5 nm以上且40nm以下。該高雜質濃度區中的 1255053 較佳雜質濃度範圍爲lxl〇18cm_3以上且lxl〇19cm·3以下。倘 若該區域的雜質濃度隨著厚度變薄而增加，則有助於避免正 向電壓增加。例如，當高雜質濃度區厚度爲2.5nm至5 〇nm 時，倘雜質濃度爲5xl〇18cm·3至ixi〇19cm·3，則可獲低正向 電壓。 形成接合於井層之該另一端面層的阻障層的內部區域 可不爲未摻雜區域。必要條件僅爲相較於該另一端面層的中 間區，所要添加的雜質濃度低。該低雜質濃度係小於 # lxl〇16cm·3 以上或小於 ixi〇18cm·3，以及最好爲 5xl〇ucm-3 以上且5xl017cm_3以下。 較佳方式爲將未刻意摻入雜質的井層接合於形成該另 一端面層的阻障層。倘接合具微雜質濃度的高純度井層，則 可避免雜質所形成之部位引起的二次發光，並可獲具極佳單 色的光。可降低雜質造成扭曲生成的機率，並具可穩定地獲 得希冀波長光的優點。 接合於發光層該一端面層的η型覆層可由能帶間隙大於 _ 井層的 η 型 GaYInzN(〇^X，YS1，χ+γ=ι)形成，或可 由 η 型 AlxGaYN(〇SX，YS1，χ+γ=ι)形成。倘 η 型覆 層由GaYInzN形成，則可形成晶體扭曲減弱的發光層於覆層 上。此外’可獲得具極佳發光效率的瓜族氮化物半導體發光 裝置。得使用氮化鎵銦層製成的 GaY1InzlN/GaY2Inz2N (OSY!，Y2，Ζ2^ 1，Υ1^γ2，ζ1^Ζ2，Υ1+Ζ1 = 1，Υ2+Ζ2=1) 超晶格結構作爲n型覆層。倘若使用該超晶格結構作爲基 層’則可形成晶格扭曲減弱的發光層，並可獲得發光效率提 1255053 高的瓜族氮化物半導體發光裝置。 本發明的m族氮化物半導體發光裝置設有位 構上的歐姆電極，該堆疊結構具有含本發明結構之 多量子井結構發光層。當結晶基板由具極佳導電率 成時（諸如矽或立方晶3 C晶體型、六方晶體4 Η每 型碳化矽（SiC))，會設置與形成基板之結晶材料 類型的η型或p型歐姆電極之一。適用於另一導電 姆電極設在與基板晶體導電類型相反的覆層上，或 Φ 具本發明結構之多量子井結構上之覆層上方的接觸 當基板由高電阻或電絕緣藍寶石形成時，η型 姆電極無法直接接觸基板。因此，其各設於具堆· 型或Ρ型層上。例如，η型（ρ型）歐姆電極設於 型）覆層上、或設於相同覆層上的接觸層上。或者 含 GaYInzN(0SY，Zgl，Y+Z=l)的 η 型（ρ 型 上，以和緩地降低發光層能帶間隙差或提高結晶度 中間層配置於η型（ρ型）覆層與η型（ρ型）發为 • 絕緣或高電阻率結晶基板不適於設有導電 極，或者，若是光學透明基板，則將一反射膜設在 疊結構之表面的正對基板表面上，其中該反射膜具 方向上將穿經基板之光反射的功能、且由含單一金 重金屬膜的多重金屬膜所形成。倘若使用具發射光 的透明高電阻率基板，則可獲得具高發光能力的m 半導體LED。反射藍光帶、綠光帶之光的反射層可 屬、鎳（Ni)或其合金所形成。 於堆疊結 阻障層的 的材料形 $ 6H晶體 相同導電 類型的歐 設在位於 3層上。 或P型歐 ^結構的η ^ η 型（ρ ，其設於 )中間層 :.，其中該 i層之間。 性歐姆電 形成有堆 有在射出 屬膜或多 反射功能 族氮化物 爲稀有金 1255053 用於形成低接觸電阻率η型或p型歐姆電極的η型或p 型接觸層係由具低電阻率的η型或p型瓜_V族氮化物半導 體材料所形成。發光層的光得以穿透至外部之方向上的接觸 層係由具有較高能帶間隙的材料所形成，且光可穿過對應於 發射光波長的能帶間隙。例如’其由能帶間隙大於形成本發 明多量子井結構之井層材料的瓜族氮化物半導體材料所形 成。例如，其可由磷化硼（BP )或以磷化硼爲底的高能帶間 隙混合晶體所形成，諸如磷化硼鎵（BQGaRP : 0< Q，R< 1， _ Q+R=l)或磷化硼銦（BQInRP:0<Q，R<l，Q+R=i)。 在單體BP的狀況中，藉由調整成長溫度便可輕易獲得η型 與Ρ型電阻層（縱使在未摻雜狀態下亦然），因而爲適於形 成接觸層的材料。 倘若設於本發明多量子井結構發光層之光透至LED外 部的方向上的η型或ρ型歐姆電極爲具開孔的篩狀電極或電 極爲矩陣布置的格狀電極，則可獲高發射強度LED。倘若電 極的形狀會使設有歐姆電極之覆層或接觸層的表面被部分 • 覆蓋’則光吸收率會因電極材料而降低，且未爲開孔吸收並 穿透的光量會增加。因此，可獲具高發光強度的LED。 如前所述，根據本發明實施例，在多量子井結構發光層 中’因爲形成二個端面的薄層爲阻障層，所以可抑制載體由 發光層散佈至覆層。亦即，因爲載體封在井層中’所以可提 高發光輸出。 在形成多量子井結構發光層之二個端面的阻障層中，接 觸於ρ型覆層的該一端面層較該另一端面層爲厚。因此’可 -15- 1255053 抑制來自η型覆層的載體（電子）流向P型覆層’並可提高 發光輸出。 倘若井層同習用技術般存在於發光層的一端面與另一 端面，則因該井層直接接觸於覆層，所以僅有該井層接合於 不同結晶度與不同能帶間隙的薄層’且相較於其他井層’其 有結晶度、扭曲及發光波長不同的問題。另一方面，根據本 發明實施例，因爲形成發光層二個端面的薄層爲阻障層’所 以接觸於井層的薄層皆爲阻障層，而使井層周圍的環境皆相 Φ 同。因此，井層的發光效應爲固定的，而使發光波長穩定且 變化微小。如前所述，倘若存在接觸於覆層的井層，則該井 層和一與夾合於井層內部形成的阻障層之間之井層接觸之 層是不同的，因而影響發光效果，波長變長，且發光峰値的 半寬增加。 根據本發明實施例，接觸於該井層的薄層皆爲阻障層， 成長順序爲阻障層—井層—阻障層，因此所有井層的成長條 件基本上相同，井層可更穩定地形成，可抑制點缺陷形成， ® 且其性質極佳。基於此，可改善發光輸出。倘若接觸於覆層 的井層係以習用技術的方式存在，則該井層和一與夾合於阻 障層之間並於井層內部形成之井層接觸之層是不同的，且覆 層與阻障層的成長條件不同。因此，於接觸於覆層的井層中 會形成點缺陷，而使發光輸出劣化。 根據本發明，發光效率可提高約1 · 5倍。因此，LED燈 的發光輸出與電光轉化效率可提高約1 · 5倍，而可節約能源。 實施例1 -16- 1255053 本發明將根據使用形成於藍寶石基板上之本發明多量 子井結構發光層所製造的m族氮化物半導體發光二極體的 例子而做說明。 第1圖爲用於製造實施例1之LED的堆疊結構示意剖 面圖。第2圖爲第1圖所示堆疊結構之發光層結構的示意 圖。第3圖爲由第1圖所示堆疊結構製造之LED的示意平 面圖。 爲形成用於製造LED 10的堆疊結構11，藍寶石基板101 • 係置於周圍配置有感應加熱射頻（RF)線圈的石英MOCVD 反應爐中。氮氣得以流入反應爐中1 0秒，以沖洗反應爐內 部，再於10分鐘內將基板101溫度由室溫增加至1 150°c。 當基板101溫度維持在1 150°c時，得將氫氣與氮氣流入其 中，並使爐體在該狀態下維持1 0分鐘，以熱清洗基板1 0 1 表面。 第一個步驟爲將藍寶石基板1 0 1進行基板處理，其中含 以莫耳比例1 : 2混合之三甲基鋁（TMA1 )蒸氣與三甲基鎵 ® (TMGa)蒸氣混合氣體與氨氣（NH3)得以對基板101起作 用。在第一個步驟中的希冀 v族/m族比例（nh3/ (TMAl + TMGa )濃度比例）設在約85。表面處理在ι ι 5 0 °C 進行約6分鐘，其次終止供應含TMA1與TMGa蒸氣的氣體 至反應爐。 其次，在第二個步驟中，在流入TMGa與氨氣時，以將 未摻雜η型氮化鎵（GaN)層1〇2形成於（0001)藍寶石基 板101上一小時，直至0.8微米厚爲止。 1255053 . 摻Si之η型GaN層103形成於未摻雜之η型GaN層ι〇2 上。當形成層103時，其係使用矽烷氣體（SiH4)作爲摻雜源， 而慘雜電子濃度爲5xl018cnT3的砂。摻Si之η型GaN層103 的厚度爲2微米。 其次’使用SiH4作爲搶雜源並使用TMGa，TMA1及NH3 作爲原料，而形成摻Si之η型氮化鋁鎵（Al〇.Q9Ga().9lN)層 104於摻Si之η型GaN層103上。Alo.o9Gao.MN層的載體 (電洞）濃度爲7xl017cm·3，且厚度爲8nm。 • 當氨氣輸入反應爐時，基板101溫度由11501降低至 8 30°C。其次，當使用三甲基銦（TMIn)作爲銦原料時，膜 厚爲50nm之摻Si之氮化銦鎵（Ino.^Gao.^N)的n型覆層 105 係形成於 Al〇.Q9Ga〇.9iN 層 104 上。 由GaN阻障層21與In〇.〇4Ga〇.96N井層22組成的多量 子井結構發光層2係形成於η型覆層105上。在形成多量子 井結構當中’摻S i之G a Ν阻障層2 1 m係首先形成於摻§ i 之 I η 〇. 〇 i G a 〇. 9 9 N 的 η 型覆層 1 〇 5 上，且 I η 〇 . 〇 4 G a 〇 · 9 6 N 井層 • 22係形成於摻Si之GaN阻障層21m上。各井層22厚度設 爲2nm。之後將摻Si之GaN阻障層21與Ino.o4Gao.96N井層 2 2係重複堆疊五次，摻s丨之g aN阻障層2 1 η接合於第五個 In0.G4Ga().96N井層22，且多量子井結構的一端面層21m與ρ 型覆層107邊上的另一端面層21η係形成爲摻Si之GaN阻 障層2 1。 摻Si之GaN阻障層21的總數爲6，且各該五個摻si 之GaN阻障層21除另一端面層21η外，厚度均設爲i5nm。 1255053 -形成另一端面層21η的摻Si之GaN阻障層厚度設爲20nm， 其較其他阻障層爲厚。 由摻Mg之Alo.07Gao.93N製成的P型覆層107係形成於 多量子井結構發光層2上。p型覆層1〇7的載體（電洞）濃 度爲5xl017cnT3，且膜厚爲10nm。 此外，摻Mg之GaN層得於P型覆層107上長成P型接 觸層108。卩型接觸層108的載體（電洞）濃度爲8xl〇17cm-3, 且膜厚爲nm。 Φ 在p型接觸層1 08的成長完成後，終止供應電源至感應 加熱器，且基板101溫度在20分鐘內降低至室溫。當溫度 由成長溫度（1 100 °C )降低至300 °C時，反應爐內係使用氮 氣作爲載體氣體，且使1體積％的NH3得以流入。其次，當 基板101溫度達300°c時，終止nh3流量，且大氣之氣體僅 爲氮氣。當基板101溫度降低至室溫時，由反應爐取出堆疊 結構1 1。 . 根據前揭溫度降低與冷卻作業，縱使未進行將摻雜作爲 • P型雜質之Mg電活化的退火處理，其仍可獲得p型接觸層 108的p型導電率。 當摻Si之GaN層103表面預定要形成^型歐姆電極1〇9 的區域暴露出時，以前揭方式形成的堆疊結構1 1係使用一 般選擇性圖樣化技術進行選擇性蝕刻。其次，η型歐姆電極 109設於摻Si之GaN層103上，其中η型歐姆電極1〇9之 Νι、銘（Α1)、鈦（Τι)、金（Au)四層依序堆疊且鎳（Ni)層接觸於 暴露的摻Si之GaN層103表面。如第3圖所示，透明的金 1255053 .(Au)p型歐姆電極1 10係形成於留置在堆疊結構i i表面上之 P型接觸層108的整個表面上方。鈦（Ti)、鋁（A1)、金（Au) 依序堆疊的接合墊（基座電極）1 1 1係接合於p型歐姆電極 110° 其次，將厚度3 50微米的藍寶石基板101背面拋光，以 形成厚度1 〇〇微米的薄板，並進一步拋光與修整至平坦鏡 面。其次，將之切割成邊長350微米（上視）的方形LED 晶片。LED晶片1 0接合於引腳架，以使裝置驅動電流得經 # 由η型歐姆電極109與p型基座電極111而穿經LED晶片 10 ° 正向裝置驅動電流得以在η型歐姆電極109與p型歐姆 電極1 1 0間流動，以使LED晶片1 0發光。當正向電流設爲 20mA時，發光波長爲3 95nm。使用一般積分球量測得的發 光輸出高達8.7 mW。藉此，可獲高發光輸出的近紫外光LED。 比較例 . 在比較例中，其係使用具相同厚度阻障層之多量子井結 I 構的發光層而形成LED。比較其發光特性。 亦即，在實施例1所描述的多量子井結構中，該另一端 面層（第2圖中的21η )係由厚度同於其他阻障層的摻Si 之GaN阻障層所形成。除多量子井結構發光層以外的，因素 皆同實施例1中所描述的堆疊結構與電極結構。 20mA的正向電流通過引腳架而流經以實施例1相同方 式安裝的LED晶片，並發射具395nm波長之近紫外區的光。 雖該LED晶片的發光波長同實施例1的LED晶片10的發光 1255053 . 波長，但正向電壓高達3.5V。然而，發光輸出降低至5.9 mW， 且不具實施例1之LED 10的特性。 實施例2 在實施例2中，本發明的內容將根據使用具阻障層之多 量子井結構發光層所形成的m族氮化物LED的例子而做具 體說明。 在實施例2中，在形成實施例1所描述之組成發光層的 多量子井結構當中，該一端面層（第2圖的2lm)的阻障層 # 厚設爲15nm，第二阻障層厚設爲16nm，第三阻障層厚設爲 17nm，第四阻障層厚設爲i8nm，以及第五阻障層厚設爲 1 9 n m 〇 亦即，所形成多量子井結構之摻Si之GaN阻障層厚度 係由η型覆層（第1與2圖的105)向p型覆層（第1與2 圖的107)縮減。形成該另一端面層21η的摻Si之GaN阻 障層厚度設爲20nm。 . 20mA的正向裝置驅動電流得以流經以實施例1相同方 Φ 式安裝的LED晶片，並評估其特性。發光波長爲395 nm。雖 該正向電壓高達3.3V，但發光輸出同實施例1高達8.9 mW， 並獲得高輸出與近紫外光區LED。倘若將該LED比較例中 的習用LED比較，則實施例2的正向電壓與發光輸出更佳。 實施例3 本發明將根據使用有雜質散布之阻障層作爲該另一端 面層之多量子井結構發光層所形成的瓜族氮化物半導體層 的例子而做具體說明。 1255053 _ 在實施例3中，形成實施例1所描述之多量子井結構的 另一端面層21η係由摻Si之GaN層所形成，且該摻Si之 GaN層之雜質濃度在與井層之接面區的端面上爲低，在該層 中間部位爲最高，且在厚度增加方向上由中間部位向P型覆 層107漸減。 在另一端面層中，在厚度增加方向上由井層接面區至 5nm內部區的矽雜質濃度設定在3xl017Cm·3。在阻障層中間 區（在厚度方向上由井層接面區延伸5nm至lOnm)的矽雜 φ 質濃度設爲4xl018cnT3。在由中間區至p型覆層接合區的區 域中，矽雜質濃度係由4xl018cnT3線性降低至7xl〇17cm_3。 當製造形成該另一端面層的阻障層時，藉由隨時間改變供應 至反應系統的SiH4氣體流速便可調整各區域中的矽雜質濃 度。 以實施例1中的相同方式形成LED晶片，且正向電流 得以流經晶片。當疋向電流設爲20mA時，發光波長爲 3 95 nm。當正向電壓低達3.2V，發光輸出爲8.6mW ;且倘若 φ 與比較例中的習用LED比較，則實施例3之LED的正向電 壓與發光輸出更佳。 產業利用性 根據本發明的m族氮化物半導體發光裝置具有高結晶 度的井結構，以使發光輸出與電光轉化效率大幅提高，而節 省能源。 【圖式簡單說明】 第1圖爲用於製造第一個實施例LED之堆疊結構的示 1255053 意剖面圖。 第2圖爲第1圖所示堆疊結構之發光層的示意圖。 第3圖爲由第1圖所示堆疊結構製造之LED的示意平 面圖。 【元件符號說明】 2 發光層 10 LED晶片 11 堆疊結構 2 1 阻障層 2 1η 摻Si之GaN阻障層 2 1m 摻Si之GaN阻障層 22 井層 101 基板 102 未摻雜之η型GaN層 103 摻Si之η型GaN層 104 摻S i之η型氮化鋁鎵 105 η型覆層 107 Ρ型覆層 108 Ρ型接觸層 109 η型歐姆電極 1 10 ρ型歐姆電極 111 Ρ型基座電極 -23 -The n-type and p-type lanthanum nitride semiconductor layers (cladding) provided on the single crystal substrate have a composition formula of AlxGaYlnzNuMa (OgXgl, 0SYS1, 0SZS1, Χ+γ+Ζ=1, and the symbol μ represents a different a group V element of nitrogen, and 0 S a < 1 ). When there is a lattice mismatch between the single crystal substrate and the quaternary nitride semiconductor layer (cladding) formed on the substrate, it is preferable to adjust the mismatch and become a low or high temperature having an excellent crystallinity upper layer. A buffer layer is placed between the layers. The buffer layer may be formed of, for example, aluminum gallium nitride (AlxGaYN: 0SXS1, 0'YS1, Χ+Υ=ι). 1255053 The n-type and p-type cladding layers do not have to be formed of a 1 lanthanide nitride semiconductor material having the same composition ratio. For example, the n-type cladding layer may be formed of gallium nitride (G aN ), and the p-type cladding layer may be formed of aluminum gallium nitride (AlxGaYN ). Preferably, the n-type and p-type cladding layers are formed of a group m nitride semiconductor material having a band gap which is larger than a barrier layer constituting the quantum well structure of the light-emitting layer. The light-emitting layer disposed between the n-type cladding layer and the p-type cladding layer is a multi-quantum well structure, wherein the well layer and the barrier layer formed by the lanthanum nitride semiconductor are periodically staggered and stacked, and the two end face layers (one end face) The layer and the other end layer) are barrier layers. This is because it is more effective to prevent the carrier from spreading to the n-type and p-type cladding layers. In the thin layer of the light-emitting layer of the invention, the one end face layer closest to and facing the n-type clad layer and the other end face layer closest to the n-type clad layer are barrier layers. The well layer is formed of a HI-based nitride semiconductor having a band gap smaller than that of the barrier layer. For example, the well layer is formed of indium gallium nitride (GaYInzN: 0' YS 1, 〇SZgl, Y+Z=l). The indium composition ratio (gallium composition ratio) of GaYInzN is appropriately selected to obtain light of a desired wavelength. The light-emitting layer of the light-emitting device emitting short-wavelength light in the near-ultraviolet light # region, the blue region or the green region may be GaYInzN (0.25SZS0.40, Y) having an indium composition ratio of about 0.40 (=40%) or less (preferably 0.25). +Z=l) is formed. The well layer constituting the light-emitting layer of the multi-quantum well structure may be formed of a cuban nitride semiconductor containing a group V element other than nitrogen and nitrogen, such as gallium phosphide nitride (GaNi_aPa: 〇$ a < 1 ). In GaNbaPa, the energy band gap is proportional to the phosphorus composition ratio (=a) (in other words, depending on the ratio of nitride (N) composition (=; l - a )). A well layer capable of emitting light from the ultraviolet light to the red light band can be formed by 1255053 I using the band-shaped GaN^Pa. In order to obtain a light-emitting device having a low forward voltage (Vf) or threshold voltage (vth) intensity, the number of well layers forming the multi-quantum well structure is preferably at least /. The well layer to be bonded to the other * barrier layer forming the quantum well structure is characterized by an illuminating wavelength from the undoped m-type nitride semiconductor layer from the luminescent layer of the 1-multi-quantum well structure. Other well layers joined to the well layer of the one end layer may be formed of a melon-based nitrogen layer doped with impurities. a well layer Φ emitting wavelength outside the layer of the other end face layer. If the m-type nitride semiconductor layer having impurities in the well other than the layer of the other end face layer is formed, the effect of lowering V t h is enhanced. The barrier layer of the band-in-well structure of the m-type nitride semiconductor constituting the well layer is an element of the m-group nitride semiconductor layer having a band gap of about 10 times the electron heat of 0.026 eV at room temperature. The barrier layer is preferably formed by a semi-nitride containing a plurality of group V elements, such as 八乂0, 丫; 0^乂^1'0^丫^1, \+丫=1) or •. In the present invention, the barrier layer of the other end face layer is formed of a m-type nitride semiconductor layer having a thickness of a barrier layer. This is because the layer passes through the barrier layer of the one end layer and the electrons entering the luminescent layer merge into the well layer of the other end layer. The outer barrier layer of the quantum well structure has a suitable thickness of 15 nm or more and 50 nm or less. The thickness of the barrier layer of the end face layer is 1.2 2.5 times or less of the thickness of the other barrier layer. And high luminescence for the three end face layers to form, to stabilize (such as the semiconductor semi-conducting does not affect the layer system by doping: low V f or crevice, quantum kinetic energy (= formed. Conductor shaped GaN 1 - aPa Further than the other, the n-type cladding is sealed on the end face layer. The other time is more than -10- 1255053 & even if the quantum well structure uses a barrier whose thickness is gradually increased from the one end face layer to the other end face layer a layer formed which is still effective for suppressing excessive electron inflow (electron overflow) p-type cladding from the n-type cladding layer. For example, if the thickness of the barrier layer of the one end layer is defined as 1.0 as a reference, The barrier layer disposed closer to the other end face layer has a thickness of 1.5, the barrier layer disposed next to the other end face layer has a thickness of 2.0, and the barrier layer of the other end face layer has a thickness of 2.5. Alternatively, the thickness of the one end face layer is the same as the thickness of the barrier layer adjacent to the one end face layer, and the thickness of the plurality of barrier layers in the middle portion of the quantum well structure is increased by #, and the thickness of the barrier layer near the other end face layer is increased. Increasingly, the thickness of the other end face layer is the largest, and the layers are The thickness is increased from the one end face layer to the other end face layer. The barrier layer forming the other end face layer is formed of a HI group nitride semiconductor layer, and the impurity concentration of the m group nitride semiconductor layer is bonded thereto The end face of the well layer is low and is the highest at the middle portion thereof, and gradually decreases from the intermediate portion to the p-type cladding layer. With this structure, the wavelength of light emitted from the light-emitting layer can be stabilized. That is, the other end face layer is doped The impurity concentration is formed by the m-type nitride semiconductor layer in which the intermediate portion of the layer is lowered toward the p-type cladding layer. The other barrier layer constituting the quantum well structure may be formed of a layer doped or doped with impurities. The other barrier layer and the other end face layer are formed of a π-nitride semiconductor doped with impurities, thereby reducing Vf or Vth. In the state of the barrier layer, the impurity concentration in the layer doped with impurities The distribution pattern is not limited. The distribution may be the same as the barrier layer of the other end face layer, or may be evenly distributed in the thickness direction. The metal-organic chemical gas may be used for the barrier layer and the well layer of the quantum well structure. Phase deposition (MOCVD, MOVPE OMVPE), molecular beam ray 1255053 crystal (MBE), halide vapor deposition and hydride vapor deposition. Μ CVD CVD is particularly effective because the m-type nitride semiconductor contains highly volatile elements such as phosphorus (P), Arsenic (As) and the like. The constituent layers constituting the group-m nitride semiconductor light-emitting device can be formed by using different vapor phase growth methods. For example, the barrier layer or the well layer forming the light-emitting layer can be formed by atmospheric pressure (substantial atmospheric pressure). Or formed by decompression MOCVD, and the n-type or p-type cladding layer can be formed by MBE, but it is easy to form the composition layer by the same vapor phase growth method. It is suitable for obtaining a barrier layer doped with impurities (such as an n-type barrier layer). Examples of the # impurity of the layer are a group IV element such as germanium (Si), germanium (Ge), and tin (Sn), and a group VI element such as selenium (Se) or tellurium (Te). Examples of impurities suitable for obtaining a p-type barrier layer are lanthanum elements such as magnesium (mg) and calcium (C a). The impurity concentration doped in the barrier layer can be measured using a device analysis method such as secondary ion mass spectrometry (SIMS) or Auger electronanolysis. The barrier layer of the other end face layer is grown in an undoped state, i.e., no impurities are supplied to the growth reaction system. When the thickness of the barrier layer is about half of the thickness of the barrier layer, a large amount of impurities are rapidly added, and a region containing a high impurity concentration is formed in the middle portion of the barrier layer. After forming a region of a desired thickness and forming a high concentration of impurities in the middle portion, the amount of impurities to be added decreases with time and continues to grow until the desired thickness is obtained. If the amount of impurities to be added to the growth reaction system changes over time in this method, the impurity concentration of the other end layer is lower at the junction with the well layer, and the concentration is highest in the middle portion, and is intermediate The part is gradually reduced to the P-type cladding layer. Preferably, the thickness of the high impurity concentration region in the barrier layer forming the other end face layer is 2.5 nm or more and 40 nm or less. The preferred impurity concentration of 1255053 in the high impurity concentration region is lxl 〇 18 cm _ 3 or more and l x l 〇 19 cm · 3 or less. If the impurity concentration of the region increases as the thickness becomes thinner, it helps to avoid an increase in the forward voltage. For example, when the high impurity concentration region has a thickness of 2.5 nm to 5 〇 nm, a low forward voltage can be obtained if the impurity concentration is 5 x 1 〇 18 cm · 3 to ixi 〇 19 cm · 3. The inner region of the barrier layer forming the other end face layer bonded to the well layer may not be an undoped region. The necessary condition is that the concentration of impurities to be added is low as compared with the intermediate portion of the other end face layer. The low impurity concentration is less than # lxl 〇 16 cm · 3 or less or less than ixi 〇 18 cm · 3, and preferably 5 x l 〇 ucm -3 or more and 5 x 10 17 cm -3 or less. Preferably, the well layer not intentionally doped with impurities is bonded to the barrier layer forming the other end face layer. When a high-purity well layer having a fine impurity concentration is joined, secondary light emission due to a portion formed by impurities can be avoided, and an excellent single-color light can be obtained. It can reduce the probability of distortion caused by impurities, and has the advantage of stably obtaining the light of the wavelength. The n-type cladding layer bonded to the one end surface layer of the light-emitting layer may be formed of n-type GaYInzN (〇^X, YS1, χ+γ=ι) having a band gap larger than _ well layer, or may be n-type AlxGaYN (〇SX, YS1) , χ + γ = ι) formed. If the n-type cladding layer is formed of GaYInzN, a light-emitting layer with reduced crystal distortion can be formed on the cladding layer. Further, a cuban nitride semiconductor light-emitting device having excellent luminous efficiency can be obtained. GaY1InzlN/GaY2Inz2N (OSY!, Y2, Ζ2^ 1, Υ1^γ2, ζ1^Ζ2, Υ1+Ζ1 = 1, Υ2+Ζ2=1) made of indium gallium nitride layer is used as the n-type Cladding. If the superlattice structure is used as the base layer, a light-emitting layer having a reduced lattice distortion can be formed, and a cuban nitride semiconductor light-emitting device having a high luminous efficiency of 1255053 can be obtained. The group-m nitride semiconductor light-emitting device of the present invention is provided with a ohmic electrode in a structure having a multi-quantum well structure light-emitting layer comprising the structure of the present invention. When the crystalline substrate is formed with excellent electrical conductivity (such as yttrium or cubic 3 C crystal form, hexagonal crystal 4 Η each type of tantalum carbide (SiC)), an n-type or p-type of the type of crystalline material forming the substrate is set. One of the ohmic electrodes. Applicable to another conductive electrode disposed on a coating opposite to the conductivity type of the substrate crystal, or Φ contact above the cladding on the multi-quantum well structure having the structure of the present invention when the substrate is formed of high resistance or electrically insulating sapphire The n-type electrode cannot directly contact the substrate. Therefore, each is provided on a stack type or a ruthenium type layer. For example, an n-type (p-type) ohmic electrode is provided on the type of cladding layer or on a contact layer provided on the same cladding layer. Or an η-type containing GaYInzN (0SY, Zgl, Y+Z=l) (on the p-type, to gently reduce the band gap of the luminescent layer or to increase the crystallinity, the intermediate layer is disposed in the η-type (p-type) cladding layer and η Type (p type) is: • The insulating or high resistivity crystalline substrate is not suitable for providing a conductive electrode, or, in the case of an optically transparent substrate, a reflective film is disposed on the surface of the substrate opposite the surface of the stacked structure, wherein the reflection The film has a function of reflecting light passing through the substrate and is formed of a multiple metal film containing a single gold heavy metal film. If a transparent high-resistivity substrate with emitted light is used, an m semiconductor having high light-emitting capability can be obtained. LED. The reflective layer that reflects the blue light and the green light can be formed by nickel (Ni) or its alloy. The material shape of the stacked junction barrier is 6 6H crystal. The same conductivity type is set on the 3rd layer. Or a P-type ohmic structure of η ^ η type (ρ , which is set in the middle layer: . , between the i layers. Sexual ohmic electricity is formed with a stack of films in the exiting film or multi-reflective functional group For rare gold 1255053 used to form low contact electricity The n-type or p-type contact layer of the resistive n-type or p-type ohmic electrode is formed of a low-resistivity n-type or p-type melon-V nitride semiconductor material. The light of the light-emitting layer is penetrated to the outside. The contact layer in the direction is formed by a material having a higher energy band gap, and the light can pass through an energy band gap corresponding to the wavelength of the emitted light. For example, 'the energy band gap is larger than the well layer forming the multi-quantum well structure of the present invention. The material is formed of a cuban nitride semiconductor material. For example, it may be formed of boron phosphide (BP) or a high energy band gap mixed crystal based on boron phosphide, such as borophosphide gallium (BQGaRP: 0<Q, R<lt; 1, _ Q+R = l) or boron indium phosphide (BQInRP: 0, R < l, Q + R = i). In the state of monomer BP, it can be easily obtained by adjusting the growth temperature An n-type and a Ρ-type resistive layer (also in the undoped state), and thus a material suitable for forming a contact layer. If the light provided in the luminescent layer of the multi-quantum well structure of the present invention penetrates into the direction of the outside of the LED The n-type or p-type ohmic electrode is a sieve electrode with an opening or a grid electrode with electrodes arranged in a matrix A high emission intensity LED can be obtained. If the shape of the electrode is such that the surface of the coating or contact layer provided with the ohmic electrode is partially covered, the light absorption rate is lowered by the electrode material and is not absorbed and worn by the opening. The amount of light transmitted will increase. Therefore, an LED having high luminous intensity can be obtained. As described above, in the multi-quantum well structure light-emitting layer, 'because the thin layer forming the two end faces is a barrier layer, according to an embodiment of the present invention, Therefore, it is possible to suppress the carrier from being dispersed from the light-emitting layer to the cladding layer, that is, because the carrier is encapsulated in the well layer, so that the light-emitting output can be improved. In the barrier layer forming the two end faces of the multi-quantum well structure light-emitting layer, contact with ρ The one end layer of the cladding is thicker than the other end layer. Therefore, ' -15 - 1255053 suppresses the flow of the carrier (electron) from the n-type cladding layer to the p-type cladding layer' and can improve the light-emitting output. If the well layer exists on the one end surface and the other end surface of the light-emitting layer as in the conventional technology, since the well layer is in direct contact with the coating layer, only the well layer is bonded to the thin layer with different crystallinity and different energy band gaps. And compared to other well layers, it has the problem of different crystallinity, distortion and wavelength of light emission. On the other hand, according to the embodiment of the present invention, since the thin layers forming the two end faces of the light-emitting layer are the barrier layers, the thin layers contacting the well layers are all barrier layers, and the environment around the well layers is the same. . Therefore, the luminescence effect of the well layer is fixed, and the illuminating wavelength is stabilized and the change is small. As mentioned above, if there is a well layer in contact with the coating, the layer of the well layer and the layer in contact with the barrier layer formed inside the well layer are different, thereby affecting the luminous effect. The wavelength becomes longer and the half width of the luminescence peak 增加 increases. According to an embodiment of the invention, the thin layers contacting the well layer are all barrier layers, and the growth order is a barrier layer-well layer-barrier layer, so the growth conditions of all the well layers are basically the same, and the well layer can be more stable. The formation of the ground suppresses the formation of point defects, and its properties are excellent. Based on this, the light output can be improved. If the well layer in contact with the cladding is present in the manner of conventional techniques, the well layer and the layer that is in contact with the well layer sandwiched between the barrier layer and formed inside the well layer are different, and the cladding layer Different from the growth conditions of the barrier layer. Therefore, point defects are formed in the well layer which is in contact with the coating, and the light-emitting output is deteriorated. According to the present invention, the luminous efficiency can be improved by about 1.5 times. Therefore, the luminous output and electro-optic conversion efficiency of the LED lamp can be improved by about 1.5 times, and energy can be saved. Embodiment 1 - 16 - 1255053 The present invention will be described based on an example of a group m nitride semiconductor light-emitting diode manufactured using the multi-well structured light-emitting layer of the present invention formed on a sapphire substrate. Fig. 1 is a schematic cross-sectional view showing a stack structure for manufacturing the LED of Example 1. Fig. 2 is a schematic view showing the structure of the light-emitting layer of the stacked structure shown in Fig. 1. Fig. 3 is a schematic plan view of an LED manufactured by the stacked structure shown in Fig. 1. To form the stacked structure 11 for fabricating the LED 10, the sapphire substrate 101 is placed in a quartz MOCVD reactor equipped with an induction heating radio frequency (RF) coil. Nitrogen gas was allowed to flow into the reactor for 10 seconds to rinse the inside of the reactor, and the temperature of the substrate 101 was increased from room temperature to 1 150 ° C in 10 minutes. When the temperature of the substrate 101 was maintained at 1 150 ° C, hydrogen gas and nitrogen gas were flowed thereinto, and the furnace body was maintained in this state for 10 minutes to thermally clean the surface of the substrate 10 1 . The first step is to perform substrate processing on the sapphire substrate 101, which contains a mixture of trimethylaluminum (TMA1) vapor and trimethylgallium® (TMGa) vapor in a molar ratio of 1:2 (a mixture of ammonia and ammonia). NH3) is allowed to act on the substrate 101. The ratio of the group v/m group (nh3/(TMAl + TMGa) concentration ratio) in the first step is set at about 85. The surface treatment was carried out at ι 5 0 ° C for about 6 minutes, and then the gas containing TMA1 and TMGa vapor was supplied to the reaction furnace. Next, in the second step, when the TMGa and the ammonia gas are flowed, the undoped n-type gallium nitride (GaN) layer 1〇2 is formed on the (0001) sapphire substrate 101 for one hour until 0.8 μm thick. until. 1255053. A Si-doped n-type GaN layer 103 is formed on the undoped n-type GaN layer ι 2 . When the layer 103 is formed, it uses a decane gas (SiH4) as a doping source, and a miscellaneous electron concentration of 5x1018cnT3. The Si-doped n-type GaN layer 103 has a thickness of 2 μm. Secondly, using SiH4 as a source of impurity and using TMGa, TMA1 and NH3 as raw materials, a Si-doped n-type aluminum gallium nitride (Al〇.Q9Ga().9lN) layer 104 is formed on the Si-doped n-type GaN layer 103. on. The carrier (hole) of the Alo.o9Gao.MN layer has a concentration of 7xl017cm·3 and a thickness of 8 nm. • When ammonia is fed into the reactor, the substrate 101 temperature is reduced from 11501 to 8 30 °C. Secondly, when trimethylindium (TMIn) is used as the indium raw material, an n-type cladding layer 105 of Si-doped indium gallium nitride (Ino.^Gao.^N) having a film thickness of 50 nm is formed on Al〇.Q9Ga. 〇.9iN layer 104. A multi-well structure light-emitting layer 2 composed of a GaN barrier layer 21 and an In〇.〇4Ga〇.96N well layer 22 is formed on the n-type cladding layer 105. In the formation of a multi-quantum well structure, the S 1 -doped G a Ν barrier layer 2 1 m is first formed in the η-type cladding layer 1 〇 5 of I η 〇 〇i G a 〇. 9 9 N Upper, and I η 〇. 〇4 G a 〇· 9 6 N Well layer • 22 is formed on the Si-doped GaN barrier layer 21m. The thickness of each well layer 22 is set to 2 nm. Then, the Si-doped GaN barrier layer 21 and the Ino.o4Gao.96N well layer 2 2 series are repeatedly stacked five times, and the s-doped g aN barrier layer 2 1 η is bonded to the fifth In0.G4Ga().96N. The well layer 22, and the end face layer 21m of the multi-quantum well structure and the other end face layer 21n on the side of the p-type clad layer 107 are formed as a Si-doped GaN barrier layer 2 1 . The total number of Si-doped GaN barrier layers 21 is 6, and each of the five Si-doped GaN barrier layers 21 has a thickness of i5 nm except for the other end face layer 21n. 1255053 - The thickness of the Si-doped GaN barrier layer forming the other end face layer 21n is set to 20 nm, which is thicker than the other barrier layers. A P-type cladding layer 107 made of Mg-doped Alo.07Gao.93N is formed on the multi-quantum well structure light-emitting layer 2. The carrier (hole) concentration of the p-type cladding layer 1〇7 was 5xl017cnT3, and the film thickness was 10 nm. Further, the Mg-doped GaN layer is grown on the P-type cladding layer 107 to form a P-type contact layer 108. The carrier (hole) concentration of the ruthenium-type contact layer 108 is 8 x 1 〇 17 cm -3 and the film thickness is nm. Φ After the growth of the p-type contact layer 108 is completed, the supply of power is terminated to the induction heater, and the temperature of the substrate 101 is lowered to room temperature within 20 minutes. When the temperature was lowered from the growth temperature (1 100 °C) to 300 °C, nitrogen gas was used as a carrier gas in the reactor, and 1% by volume of NH3 was allowed to flow. Secondly, when the temperature of the substrate 101 reaches 300 ° C, the flow rate of nh 3 is terminated, and the atmosphere gas is only nitrogen. When the temperature of the substrate 101 is lowered to room temperature, the stacked structure 11 is taken out from the reaction furnace. According to the foregoing temperature reduction and cooling operation, the p-type conductivity of the p-type contact layer 108 can be obtained even if the annealing treatment for electrically energizing Mg as a p-type impurity is not performed. When the region of the surface of the Si-doped GaN layer 103 where the ohmic electrode 1〇9 is to be formed is exposed, the stacked structure 11 formed by the prior art is selectively etched using a general selective patterning technique. Next, the n-type ohmic electrode 109 is disposed on the Si-doped GaN layer 103, wherein the n-type ohmic electrode 1〇9 is 四ι, 铭(Α1), titanium (Τι), and gold (Au), and the four layers are sequentially stacked and nickel ( The Ni) layer is in contact with the surface of the exposed Si-doped GaN layer 103. As shown in Fig. 3, a transparent gold 1255053. (Au) p-type ohmic electrode 1 10 is formed over the entire surface of the P-type contact layer 108 remaining on the surface of the stacked structure i i . Titanium (Ti), aluminum (A1), gold (Au) sequentially stacked bonding pads (base electrode) 1 1 1 is bonded to the p-type ohmic electrode 110° Next, the back surface of the sapphire substrate 101 having a thickness of 3 50 μm is polished To form a thin plate with a thickness of 1 μm and further polish and trim to a flat mirror. Second, it was cut into square LED chips with a side length of 350 microns (top view). The LED chip 10 is bonded to the lead frame so that the device drive current is passed through the n-type ohmic electrode 109 and the p-type pedestal electrode 111 through the LED chip 10°. The forward device drive current is applied to the n-type ohmic electrode 109. Flows between the p-type ohmic electrode 1 10 to cause the LED chip 10 to emit light. When the forward current is set to 20 mA, the emission wavelength is 3 95 nm. The illuminating output measured with a general integrating sphere is as high as 8.7 mW. Thereby, a near-ultraviolet LED with high illumination output can be obtained. Comparative Example In the comparative example, an LED was formed using a light-emitting layer of a multi-quantum well structure having the same thickness barrier layer. Compare its luminescent properties. That is, in the multi-quantum well structure described in Embodiment 1, the other end face layer (21n in Fig. 2) is formed of a Si-doped GaN barrier layer having a thickness similar to that of the other barrier layers. The factors other than the multi-quantum well structure light-emitting layer are the stack structure and electrode structure described in Embodiment 1. A forward current of 20 mA was passed through the lead frame through the LED chip mounted in the same manner as in Example 1 and emitted light having a near ultraviolet region having a wavelength of 395 nm. Although the light emission wavelength of the LED chip is the same as that of the LED wafer 10 of the first embodiment, the forward voltage is as high as 3.5V. However, the light output was lowered to 5.9 mW, and the characteristics of the LED 10 of Example 1 were not obtained. [Embodiment 2] In Embodiment 2, the content of the present invention will be specifically described based on an example of a group m nitride LED formed using a multi-quantum well structure light-emitting layer having a barrier layer. In the second embodiment, among the multi-quantum well structures forming the constituent light-emitting layer described in Embodiment 1, the barrier layer # of the one end face layer (2 lm of FIG. 2) is set to 15 nm thick, and the second barrier layer is formed. The thickness is set to 16 nm, the thickness of the third barrier layer is set to 17 nm, the thickness of the fourth barrier layer is set to i8 nm, and the thickness of the fifth barrier layer is set to 19 nm, that is, the Si doping of the multi-quantum well structure is formed. The thickness of the GaN barrier layer is reduced by the n-type cladding layer (105 of Figs. 1 and 2) to the p-type cladding layer (107 of Figs. 1 and 2). The thickness of the Si-doped GaN barrier layer forming the other end face layer 21n was set to 20 nm. A 20 mA forward device drive current was passed through the LED chip mounted in the same manner as in Example 1, and its characteristics were evaluated. The emission wavelength is 395 nm. Although the forward voltage is as high as 3.3V, the luminous output is as high as 8.9 mW as in the first embodiment, and a high output and near ultraviolet region LED is obtained. The forward voltage and the light output of Example 2 are better if the conventional LEDs in the LED comparative example are compared. Embodiment 3 The present invention will be specifically described based on an example in which a barrier layer having impurity dispersion is used as a melon nitride semiconductor layer formed of a multi-quantum well structure light-emitting layer of the other end layer. 1255053 _ In Embodiment 3, another end face layer 21n forming the multi-quantum well structure described in Embodiment 1 is formed of a Si-doped GaN layer, and the impurity concentration of the Si-doped GaN layer is in the well layer The end face of the junction region is low, the highest portion in the middle portion of the layer, and gradually decreases from the intermediate portion toward the P-type cladding layer 107 in the direction of increasing thickness. In the other end face layer, the erbium impurity concentration from the well layer junction region to the 5 nm inner region in the thickness increase direction was set at 3xl017Cm·3. The doping φ mass concentration in the intermediate portion of the barrier layer (5 nm to lOnm extending from the well layer junction region in the thickness direction) was set to 4x1018cnT3. In the region from the intermediate portion to the p-type cladding junction region, the germanium impurity concentration was linearly reduced from 4xl018cnT3 to 7xl〇17cm_3. When the barrier layer forming the other end face layer is formed, the concentration of germanium in each region can be adjusted by changing the flow rate of the SiH4 gas supplied to the reaction system over time. The LED wafer was formed in the same manner as in Example 1, and a forward current was allowed to flow through the wafer. When the 疋 current is set to 20 mA, the illuminating wavelength is 3 95 nm. When the forward voltage is as low as 3.2 V, the light output is 8.6 mW; and if φ is compared with the conventional LED in the comparative example, the forward voltage and the light output of the LED of Example 3 are better. Industrial Applicability The m-group nitride semiconductor light-emitting device according to the present invention has a well crystal structure having a high crystallinity, so that the light-emitting output and electro-optic conversion efficiency are greatly improved, and energy is saved. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a cross-sectional view showing the structure of a stack of LEDs of the first embodiment. Fig. 2 is a schematic view showing the light-emitting layer of the stacked structure shown in Fig. 1. Fig. 3 is a schematic plan view of an LED manufactured by the stacked structure shown in Fig. 1. [Component Symbol Description] 2 Light Emitting Layer 10 LED Wafer 11 Stack Structure 2 1 Barrier Layer 2 1η Si-Doped GaN Barrier Layer 2 1m Si-Doped GaN Barrier Layer 22 Well Layer 101 Substrate 102 Undoped n-type GaN Layer 103 Si-doped n-type GaN layer 104 S1-doped n-type aluminum nitride gallium 105 n-type cladding layer 107 Ρ-type cladding layer 108 Ρ-type contact layer 109 η-type ohmic electrode 1 10 ρ-type ohmic electrode 111 Ρ-type base Seat electrode -23 -