DETAILED DESCRIPTION

[0033] The present invention is a physically robust light emitting diode that offers high reliability in standard packaging and will withstand high temperature and high humidity conditions.

[0034] As noted in the background, ohmic contacts must be protected from physical, mechanical, environmental and packaging stresses to prevent degradation of Group III nitride LEDs.

[0035] In this regard, FIG. 1 is a photograph of an entire LED (“die”). In the device of FIG. 1 the passivation layer of silicon dioxide (glass) has been removed except around the outside edge of the die. The portions where glass is still present are generally indicated by the spotted or stained-appearing portions around the perimeter of the generally square die. This mottled appearance results from a varying gap of air under the glass as it delaminates from the die. In the die illustrated in FIG. 1 the delamination begins at about the three o'clock (moving clockwise) and reaches approximately the 11:00 o'clock position. The passivation layer is absent from the center of the die and the wire ball bond can be seen at the very center of the die still attached to the bond pad. In this particular example, the center portion of the passivation layer was removed while the die was being de-encapsulated after testing.

[0036] The passivation layer of the die illustrated in FIG. 1 had delaminated in the package during testing, and allowed moisture to penetrate beneath the passivation layer. The resulting delamination reduced the initial light output of this particular device by about 20%. Subsequently the moisture, which tends to permeate through the epoxy lens of an LED lamp and around the leads coming out of the bottom of the lamp package, causes the thin semi-transparent ohmic contact to degrade and eventually fail completely. This failure in turn causes the light output to continue to fall and eventually increase the forward voltage of the device. In the device photographed in FIG. 1 , the failure of the contact appears as the dark or rough areas just to the right of the center of the die.

[0037] FIG. 2 is a magnified view of the die photographed in FIG. 1 . FIG. 2 illustrates that the glass remaining on the perimeter has broken off of the inner mesa of the device and that the p-contact has failed. The dark, rough appearing areas are positions where the ohmic contact (titanium and gold in this example) has balled up. As best understood, as the contact becomes less compatible with the p-type layer it tends to bead up rather than wet the p-type layer. In turn, as the Ti/Au balls up around the bond pad, the device slowly becomes disconnected. Furthermore, light is no longer generated in areas where the contact becomes discontinuous. Because a p-type gallium nitride surface is not a good conductor, and generally exhibits high resistivity, the poor current spreading in the void areas fail to provide a current path which would help generate light.

[0038] FIG. 3 illustrates a first embodiment of the diode of the invention that will withstand high temperature and high humidity conditions. The diode is generally designated at 10 and includes a silicon carbide substrate 11 , the production and nature of which are clearly set forth in other U.S. patents assigned to assignee of this invention, including for example No. RE 34,861 (formerly No. 4,866,005). In preferred embodiments, the silicon carbide substrate is a single crystal selected from the group consisting of the 3C, 4H, 6H, and 15R polytypes of silicon carbide, and is n-type.

[0039] In preferred embodiments, the LED of the present invention further comprises a buffer structure 12 on the silicon carbide substrate 11 . The buffer structure helps provide a crystalline and mechanical transition from the silicon carbide substrate 11 to the remaining Group III nitride portions of the device. Appropriate buffer structures are set forth for example in U.S. Pat. Nos. 5,393,993; 5,523,589; 5,592,501; and 5,739,554, all of which are commonly assigned with the present invention, and each of which is incorporated entirely herein by reference. The diode 10 further comprises an active region 13 of Group III nitride heterojunction diode structure formed on the buffer structure 12 . The active region 13 may comprise a single heterostructure, double heterostructure, single quantum well or multi-quantum well structure. Examples of such structures are disclosed in co-pending and commonly assigned U.S. application Ser. No. 09/154,363 filed Sep. 16, 1998 for “Vertical Geometry InGaN Light Emitting Diode” which is incorporated entirely herein by reference.

[0040] A p-type Group III nitride contact layer 14 is formed on the active region 13 . A metal contact 15 is made to the substrate 11 and another metal contact 16 is made to the p-type gallium nitride epitaxial layer. Preferably, metal contacts 15 and 16 form ohmic (i.e. non-rectifying) contacts to substrate 11 and contact layer 14 respectively. The ohmic contact 16 is selected from the group consisting of platinum, palladium, gold, a combination of titanium and gold, a combination of platinum and gold, a combination of titanium, platinum and gold, or a combination of platinum and indium tin oxide, and is most preferably formed of platinum or palladium. Nickel (Ni) is a preferred ohmic contact metal to the n-type substrate. The device is completed with a passivation layer 17 on the ohmic contact 16 , of which appropriate candidate materials are recited above, but that is most preferably formed of silicon nitride.

[0041] Silicon nitride is preferred over silicon dioxide in particular because it forms a better seal over the device, preventing contaminants such as water from reaching the epitaxial layers of the device and causing degradation such as is described above.

[0042] In a most preferred embodiment, the silicon nitride is deposited by means of sputtering. Sputtering is a well known technique for depositing thin layers of material in a vacuum or near-vacuum environment. A technique for sputtering SiN on microwave transistor structures is described in U.S. patent application Ser. No. 09/771,800 entitled “Group III Nitride Based FETs and HEMTs with Reduced Trapping and Method for Producing the Same” filed Jan. 29, 2001, which is hereby incorporated herein by reference.

[0043] FIG. 5 shows a simplified sputtering chamber 100 that can be used to deposit material on a substrate or a device or a device precursor. In operation, a semiconductor device 101 is placed on an anode 102 . The chamber 103 is then evacuated and an inert gas 104 such as argon is bled through the valve 105 to maintain a background pressure. The cathode 106 is made of the material (or a component of the material) to be deposited on the substrate or device. With the application of a high voltage between electrodes 107 , the inert gas is ionized and the positive ions 110 accelerate to the cathode 106 . Upon striking the cathode 106 , they collide with the cathode atoms 112 , giving them sufficient energy to be ejected. The sputtered cathode atoms 112 travel through space, eventually coating the anode 102 and the semiconductor device 101 on it. Other sputtering units can be more complex and detailed, but they work on much the same basic physical mechanisms. Using the more complex sputtering systems, it is possible to sputter and deposit a range of metals and dielectric layers.

[0044] In a preferred embodiment, cathode 106 is a pure silicon target. Nitrogen is provided for silicon nitride formation by flowing nitrogen gas through the sputter chamber 103 along with the inert gas. Because the sputtered target material (in this case, silicon) reacts with a reaction gas (nitrogen) to form SiN, this form of sputtering is known as “reactive sputtering.”

[0045] In a first embodiment, the sputter deposition may be performed at a temperature in excess of 200° C., and most preferably at a temperature of about 440° C. to maximize encapsulation and produce a more hermetic seal on the device. The pressure of the chamber 103 should be maintained at less than 20 millitorr (mTorr) and preferably between about 10-20 mTorr in a mixed atmosphere of argon and nitrogen gas. The sputter rate should be maintained at about 45 Å/min, and a total film thickness of greater than about 1000 Å should be deposited for optimum encapsulation. In this embodiment, sputtering of silicon nitride may be accomplished using an Endeavor sputtering system manufactured by Sputtered Films, Inc. Although the inventors do not wish to be bound by any particular theory, at this pressure (20 mTorr or less), it is presently believed that the sputtering process causes substantial ion bombardment damage to the device. Nevertheless, it is also presently believed that the increased sputter temperature acts to anneal the ion bombardment damage out of the device.

[0046] In this embodiment, the preferred sputtering process includes the specific steps of pumping down the chamber 103 to a low pressure of less than about 20 mTorr, flowing argon gas at a rate of about 40 standard cubic centimeters per minute (sccm), and flowing nitrogen gas flow at a rate of about 25 sccm. The temperature of the chamber 103 is raised above 200° C. and preferably to about 440° C. An RF power of about 100W and a DC power of about 700-800 W is applied to the terminals 107 to create an ionized plasma. This condition is maintained for about 40 minutes to sputter the Si cathode 106 . The sputtered silicon reacts with the nitrogen resulting in deposition of silicon nitride on the wafer.

[0047] In an alternative embodiment, the sputter deposition may be performed at room temperature but at a higher pressure, e.g. between about 80-100 mTorr in a mixed Ar/N ₂ atmosphere. A pulsed DC power supply should be used to reduce “spitting” and arcing between the sputter electrodes. Using a pulsed DC power supply increases the amount of ion bombardment on the sputter target, but it has been found that annealing the device to remove ion damage is not needed if the sputter is performed at the higher pressure. It is presently believed that in this embodiment, the peak ion energy of the sputtered ions is reduced while the ion flux is increased, resulting in negligible ion bombardment damage while retaining an acceptable sputter rate. In this embodiment, the sputter rate is preferably maintained at about 50 Å/min, and sputtering may be performed using a CVC 2800 sputtering system.

[0048] In this embodiment, the preferred sputtering process includes the specific steps of pumping down the chamber 103 to a pressure of about 80-100 mTorr, flowing argon gas at a rate of about 80 sccm, and flowing nitrogen gas flow at a rate of about 10 sccm. The temperature of the chamber 103 is kept at room temperature. A pulsed DC voltage having a power of about 1000W with a pulse period of about 5 μs and a duty cycle of about 40% is applied to the terminals 107 to create an ionized plasma. This condition is maintained for about 75 minutes to sputter the Si cathode 106 . The sputtered silicon reacts with the nitrogen resulting in deposition of silicon nitride on the wafer.

[0049] In either of the foregoing embodiments, the silicon nitride is preferably deposited as a silicon nitride composition that is slightly silicon-poor with respect to the stoichiometry of silicon nitride (Si ₃ N ₄ ). That is, the silicon nitride is preferably deposited in a manner that results in a non-stoichiometric composition. Rather, the proportion of silicon in the film is reduced to enhance light extraction. The proportion of silicon in the film may be adjusted by increasing or decreasing the flow of nitrogen gas into the chamber.

[0050] Stated differently, as used herein, the term “silicon nitride composition” refers to a composition that includes both silicon and nitride, including silicon and nitrogen chemically bonded to one another, and potentially including some bonded in the stoichiometric relationship of Si ₃ N ₄ . The composition can also include non-stoichiometric combinations in which the relationship of some or all of the composition is other than Si ₃ N ₄ .

[0051] In the present invention, the sputtered silicon nitride composition is preferred to the conventional plasma enhanced chemical vapor deposition (PECVD) method because the sputtering technique avoids introducing undesirable levels of hydrogen into the SiN film. As is known to those skilled in the art, hydrogen can passivate Mg-acceptors in a GaN-based semiconductor. Although the precise mechanism is not completely understood and the inventors do not wish to be bound by any particular theory of operation, it is currently understood that when silicon nitride is deposited by means of PECVD at a deposition temperature in excess of 200° C., hydrogen in the film can diffuse through the thin ohmic contact and into the p-type Group III nitride contact layer 14 , causing layer 14 to become passivated in a region close to the surface thereof. That is, in a region near the surface, a substantial number of acceptor ions are rendered neutral by the introduction of hydrogen in the film. Accordingly, the interface between the ohmic contact and the nitride material is degraded, and the contact metal does not exhibit ideal ohmic characteristics. This can result in an increase in forward voltage (V _F degradation) in the device. Essentially, the device behaves as though the interface between metal 16 and contact layer 14 forms a schottky contact instead of an ohmic contact.

[0052] In contrast, because it is deposited in a vacuum or near-vacuum, sputtered the silicon nitride composition is believed to be substantially free of hydrogen impurities. Accordingly, it is also preferable to ensure that all parts of the sputter system are clean and dry to avoid any hydrogen contamination. This may require bake-out of parts prior to sputtering.

[0053] In addition, once the LED chip has been manufactured and diced, it is necessary to mount the chip in a lamp package, as described in more detail below. The process of packaging a chip often results in the chip being exposed to high temperatures for a period of time. The chip can also be exposed to high temperatures during subsequent operation. In a chip on which silicon nitride has been deposited using the PECVD method, such exposure can result in an increase in forward voltage over time (V _F degradation). It is presently understood that this voltage increase results from the diffusion of hydrogen from the silicon nitride passivation layer 17 into the Mg-doped contact layer 14 . By depositing the silicon nitride composition layer using a sputtering technique (resulting in sputter-deposited silicon nitride composition), the resulting degradation is substantially reduced or eliminated.

[0054] If PECVD deposition is unavoidable, it is possible to compensate somewhat for the V _F degradation by doping the Mg-doped contact layer 14 at a higher doping level in order to offset the passivation caused by hydrogen diffusion. However, increasing the doping level of Mg in contact layer 14 can have a detrimental effect on the device by impairing crystal quality and surface morphology in the Mg-doped layer 14 .

[0055] FIG. 6 is a plot of V _F versus anneal temperature for an LED die processed with PECVD silicon nitride deposition on the one hand and the sputtered silicon nitride composition on the other. The LED die were manufactured by depositing epitaxial layers on a silicon carbide substrate. The substrate was then sawed in half. A silicon nitride passivation layer was deposited on each half of the wafer. PECVD deposition was used to deposit the silicon nitride passivation layer on one half of the wafer, and the high-temperature sputtering process described above was used to deposit the silicon nitride composition passivation layer on the other half of the wafer. The remainder of the LED fabrication process was conventional, and LED die were fabricated from each half of the wafer. Five die having sputtered silicon nitride composition and three die having PECVD-deposited silicon nitride passivation layers were subjected to annealing in a rapid thermal anneal chamber for five minutes at an anneal temperature of 250° C. The forward voltage of each die was measured before and after the anneal. The die were then subjected to a subsequent anneal for five minutes at an anneal temperature of 290° C. The forward voltage of each die was measured after the subsequent anneal. The average results from these tests are plotted in FIG. 6 . As can be seen in FIG. 6 , LEDs on which silicon nitride was deposited using PECVD exhibited an average V _F increase of just over 0.1V after being annealed for five minutes at 250° C., while LEDs on which the silicon nitride composition was sputter deposited showed a slight reduction (i.e. improvement) in V _F . LEDs on which silicon nitride was deposited using PECVD exhibited an average V _F increase of over 0.7V after being annealed for five minutes at 290° C., while LEDs on which the silicon nitride composition was sputter deposited showed a reduction in V _F of almost 0.1V after being annealed for five minutes at 290° C.

[0056] Accordingly, in one aspect, the present invention includes a method of manufacturing a light emitting diode comprising the steps of: forming a buffer layer on a substrate, forming an active region on the buffer layer, forming a p-type contact layer on the active region, forming a metal contact on the contact layer, and sputter-depositing a silicon nitride composition passivation layer on the metal contact. Preferably, the substrate is a conductive, single crystal silicon carbide substrate, the contact layer comprises Mg-doped GaN and the metal contact comprises platinum.

[0057] In the most preferred embodiment, the heterostructure diode is single heterostructure, double heterostructure, single quantum well or multi-quantum well structure such as described in the previously incorporated U.S. application Ser. No. 09/154,363 filed Sep. 16, 1998 for “Vertical Geometry InGaN Light Emitting Diode.”

[0058] Table 1 summarizes these ohmic contact materials in terms of their suitability for devices according to the claimed invention. In the rating scale used in Table 1, “A” refers to superior characteristics, while “C” refers to generally weak characteristics. 1

[0059] As illustrated in FIG. 3 , in preferred embodiments the ohmic contact 16 covers a substantial portion of the p-type gallium nitride layer to encourage current spreading across the p-type gallium nitride layer. Because it covers the light emitting portions of the device, the ohmic contact 16 is preferably thin enough to be semi-transparent.

[0060] The diodes illustrated in FIG. 3 can be used in a number of specific applications. One useful application is as a display, typically referred to as “numeric” or “alphanumeric” displays, although certainly not limited to such, that incorporate a plurality of the light emitting diodes according to the invention. In certain embodiments, blue emitting diodes according to the present invention are incorporated with red and green LEDs to form red-green-blue (“RGB”) pixels. Because such pixels individually produce the three primary colors, they have the capability to produce almost all colors visible to the human eye.

[0061] In other applications, diodes such as the diode 10 illustrated in FIG. 3 are incorporated into LED lamps. FIG. 4 accordingly illustrates one version of such a typical lamp. It will understood, of course, that FIG. 4 is simply exemplary of the type of lamp structure that can be used to incorporate a diode according to the present and is in no sense limiting of the type of lamp with which the diode of the invention can be used.

[0062] In FIG. 4 , the lamp 20 includes the diode 10 according to the invention encapsulated in a plastic (i.e., polymeric) lens 21 . The plastic material for the lens can be selected from a wide variety of polymeric materials that are well known to those of ordinary skill in this art and without undue experimentation. In many circumstances, the lens 21 is formed of an epoxy resin. The lamp 20 further comprises a metal lead frame 22 for electrically connecting the lamp to other electronic circuit elements. As illustrated in FIG. 4 , the metal lead frame 22 incorporates the anode 23 and the cathode 24 .

[0063] As in the diode embodiment of the invention, a plurality of the lamps 20 can be incorporated to form an appropriate display device. In particular, because gallium nitride devices of this type emit in the blue portion of the visible spectrum, lamps such as those according to the present invention can be advantageously incorporated along with red and green LED lamps to form a full color display. Examples of such displays are set forth in for example, co-pending and commonly assigned applications Ser. No. 09/057,838, which is a divisional of 08/580,771, filed Dec. 29, 1995, for “True Color Flat Panel Display Module;” and U.S. Pat. No. 5,812,105 issued on Sep. 22, 1998 for “LED Dot Matrix Drive Method and Apparatus.”

[0064] In the drawings and specification, there have been disclosed typical embodiments of the invention, and, although specific terms have been employed, they have been used in a generic and descriptive sense only and not for purposes of limitation, scope of the invention being set forth in the following claims.