Silicon has historically dominated IC technology development as the primary semiconductor material. The principal focus in recent years on a silicon alternative has concentrated on III-V compounds, such as gallium arsenide (GaAs), as a substrate material. As a semiconductor material, GaAs exhibits increased capabilities over silicon, such as electron mobility 5 to 6 times that of silicon. This characteristic, coupled with the potential semi- insulating properties of GaAs, leads to increased performance in both speed and power consumption.

GaAs has a zinc blende-structure consisting of two interpenetrating face-centred cubic sublattices which relate to the growth of high quality ingot material. The technology involved in the growth of GaAs is considerably more complicated than that employed for silicon, as a more complicated two-phase equilibrium and a highly volatile component, arsenic (As), is involved. Precise control of the As vapour pressure in the ingot growth system is required to maintain exact stoichiometry of the GaAs compound during the growth process. Two primary categories of III-V semiconductor display and device production have economically feasible processing procedures—LED displays and microwave IC devices.

LEDs are fabricated from single-crystal GaAs in which p-n junctions are formed by the addition of suitable doping agents—typically tellurium, zinc or silicon. Epitaxial layers of ternary and quaternary III-V materials such as gallium arsenide phosphide (GaAsP) are grown on the substrate and result in an emission band of specific wavelengths in the visible spectrum for displays or in the infrared spectrum for emitters or detectors. For example, red light with a peak at about 650 nm comes from the direct recombination of the p-n electrons and holes. Green-emitting diodes are generally composed of gallium phosphide (GaP). The generalized LED processing steps are covered in this article.

Microwave IC devices are a specialized form of integrated circuit; they are used as high-frequency amplifiers (2 to 18 GHz) for radar, telecommunications and telemetry, as well as for octave and multi-octave amplifiers for use in electronic warfare systems. Microwave IC device manufacturers typically purchase single-crystal GaAs substrate, either with or without an epitaxial layer, from outside vendors (as do silicon device manufacturers). The major processing steps include liquid-phase epitaxial deposition, fabrication and non-fabrication processing similar to silicon device manufacturing. Processing steps which warrant description additional to that for LED processing are also discussed in this article.

Wafer Manufacturing

Similar to the silicon ingot growth process, elemental forms of gallium and arsenic, plus small quantities of dopant material—silicon, tellurium or zinc—are reacted at elevated temperatures to form ingots of doped single-crystal GaAs. Three generalized methods of ingot production are utilized:

• horizontal or vertical Bridgeman

• horizontal or vertical gradient freeze

• high- or low-pressure liquid encapsulated Czochralski (LEC).

The bulk polycrystalline GaAs compound is normally formed by the reaction of As vapour with Ga metal at elevated temperatures in sealed quartz ampoules. Typically, an As reservoir located at one end of the ampoule is heated to 618°C. This generates approximately 1 atmosphere of As vapour pressure in the ampoule, a prerequisite for obtaining stoichiometric GaAs. The As vapour reacts with the Ga metal maintained at 1,238°C and located at the other end of the ampoule in a quartz or pyrolytic boron nitride (PBN) boat. After the arsenic has been completely reacted, a polycrystalline charge is formed. This is used for single-crystal growth by programmed cooling (gradient freeze) or by physically moving either the ampoule or furnace to provide proper temperature gradients for growth (Bridgeman). This indirect approach (arsenic transport) for compounding and growth of GaAs is used because of the high vapour pressure of arsenic at the melting point of GaAs, about 20 atmospheres at 812°C and 60 atmospheres at 1,238°C, respectively.

Another approach to the commercial production of bulk single-crystal GaAs is the LEC technique. A Czochralski crystal puller is loaded with chunk GaAs in a crucible with an outer graphite susceptor. The bulk GaAs is then melted at temperatures close to 1,238°C, and the crystal is pulled in a pressurized atmosphere which could vary by manufacturer typically from a few atmospheres up to 100 atmospheres. The melt is completely encapsulated by a viscous glass, B₂O₃, which prevents melt dissociation as the As vapour pressure is matched or exceeded by the pressure of an inert gas (typically argon, or nitrogen) applied in the puller chamber. Alternatively, monocrystalline GaAs can be synthesized in situ by injecting the As into the molten Ga or combining As and Ga directly at high pressure.

GaAs wafer manufacturing represents the semiconductor manufacturing process with the greatest potential for significant, routine chemical exposures. While GaAs wafer manufacturing is done only by a small percentage of semiconductor manufacturers, particular emphasis is needed in this area. The large amounts of As used in the process, the numerous steps in the process and the low airborne exposure limit for arsenic make it difficult to control exposures. Articles by Harrison (1986); Lenihan, Sheehy and Jones (1989); McIntyre and Sherin (1989) and Sheehy and Jones (1993) provide additional information on the hazards and controls for this process.

Polycrystalline ingot synthesis

Ampoule load and seal

Elemental As (99.9999%) in chunk form is weighed and loaded into a quartz boat in an exhausted glove box. Pure liquid Ga (99.9999%) and the dopant material are also weighed and loaded into a quartz or pyrolytic boron nitride (PBN) boat(s) in the same manner. The boats are loaded into a long cylindrical quartz ampoule. (In the Bridgman and gradient freeze techniques, a seed crystal with the desired crystallographic orientation is also introduced, whereas in the two-stage LEC technique, where only poly GaAs is needed at this stage, a polycrystalline GaAs is synthesized without the seed crystal.)

The quartz ampoules are placed in a low-temperature furnace and heated while the ampoule is purged with hydrogen (H₂), in a process known as hydrogen reduction reaction, to remove oxides. After purging with an inert gas such as argon, the quartz ampoules are attached to a vacuum pump assembly, evacuated, and the ampoule ends are heated and sealed with a hydrogen/oxygen torch. This creates a charged and sealed quartz ampoule ready for furnace growth. Hydrogen purging and the hydrogen/oxygen torch system is a potential fire/explosion hazard if proper safety devices and equipment are not in use (Wade et al. 1981).

Because the arsenic is being heated, this assembly is maintained under exhaust ventilation. Arsenic oxide deposits can form in the exhaust duct supporting this assembly. Care must be taken to prevent exposure and contamination should the ducts be disturbed in any way.

Storage and handling of arsenic chunks is a concern. For security, often the arsenic is kept under locked storage and with a tight inventory control. Typically the arsenic is also kept in a fire-rated storage cabinet to prevent its involvement in event of a fire.

Furnace growth

The Bridgeman and the gradient freeze methods of single-crystal ingot growth both utilize charged and sealed quartz ampoules in a high-temperature furnace enclosure which is vented to a wet scrubber system. The primary exposure hazards during furnace growth relate to the potential for the quartz ampoule to implode or explode during ingot growth. This situation occurs on a rather sporadic and infrequent basis, and is the result of one of the following:

• the partial pressure of the As vapour which results from the high temperatures used in the growth process

• devitrification of the quartz ampoule glass, which creates hairline cracks and the attendant potential for de-pressurization of the ampoule

• lack of accurate high-temperature control devices on the heating source—usually resistance type—with the resultant over-pressurization of the quartz ampoule

• thermocouple malfunction or failure, resulting in over-pressurization of the quartz ampoule

• excess As or too little Ga in the ampoule tube, resulting in extremely high As pressure, which can cause catastrophic depressurization of the ampoule.

The horizontal Bridgeman system consists of a multizone furnace in which the sealed quartz ampoule has separate temperature zones—the arsenic “cold” finger end at 618°C and the quartz gallium/dopant/seed crystal boat containing the melt at 1,238°C. The basic principle in the horizontal Bridgeman system involves traversing two heated zones (one above the melting point of GaAs, and one below the melting point) over a boat of GaAs to provide the precisely controlled freezing of molten GaAs. The seed crystal, maintained at all times in the freeze zone, provides the initial crystal starting structure, defining the direction and orientation of the crystalline structure within the boat. The quartz boat and ampoule of Ga and As are suspended within the heater chamber by a set of silicon carbide liners called support tubes, which are positioned within the resistance heater assembly to mechanically move the full distance of the ampoule. Additionally, the furnace assembly rests on a table which must be tilted during growth to provide the proper interface of the synthesized GaAs melt with the seed crystal.

In the gradient freeze method, a multizone high temperature furnace utilizing resistance heating is kept at 1,200 to 1,300 °C (1,237°C is the melt/freeze point of GaAs). The total ingot growth process duration is typically 3 days and comprises the following steps:

• furnace firing to temperature

• GaAs synthesis

• seeding the melt

• cool down/crystal growth.

The quartz ampoule is also tilted during the growth process by the use of a scissors-type manual jack.

Ampoule breakout

After the single-crystal GaAs ingot is grown within the sealed quartz ampoule, the ampoule must be opened and the quartz boat containing the ingot plus seed crystal removed. This is accomplished by one of the following methods:

• cutting off the sealed end of the ampoules with a wet circular saw

• heating and cracking the ampoule with a hydrogen/oxygen torch

• breaking the bagged ampoule with a hammer while under exhaust to control the airborne arsenic.

The quartz ampoules are recycled by wet etching the condensed arsenic on the interior surface with aqua regia (HCl,HNO₃) or sulphuric acid/hydrogen peroxide (H₂SO₄/H₂O₂).

Ingot beadblasting/cleaning

In order to see polycrystalline defects and remove exterior oxides and contaminants, the single-crystal GaAs ingot must be beadblasted. The beadblasting is done in an exhausted glove-box unit utilizing either silicon carbide or calcined alumina blasting media. Wet cleaning is done in chemical baths provided with local exhaust ventilation and utilizing aqua regia or alcohol rinses (isopropyl alcohol and/or methanol).

Monocrystalline ingot growth

The polycrystalline GaAs ingot retrieved from the ampoule is broken into chunks, weighed and placed into a quartz or PBN crucible, and a boron oxide disc is placed on top of it. The crucible is then placed into a crystal grower (puller) pressurized in an inert gas, and heated to 1,238°C. At this temperature, the GaAs melts, with the lighter boron oxide becoming a liquid encapsulant to prevent the arsenic from dissociating from the melt. A seed crystal is introduced into the melt below the liquid cap and while counter-rotating, is slowly withdrawn from the melt, thereby solidifying as it leaves the “hot-zone”. This process takes approximately 24 hours, depending on the charge size and crystal diameter.

Once the growth cycle is completed, the grower is opened to retrieve the monocrystalline ingot and for cleaning. Some amount of arsenic escapes from the melt even with the liquid cap in place. There can be significant exposure to airborne arsenic during this step of the process. To control this exposure, the grower is cooled to below 100°C, which results in the deposition of fine arsenic particulate on the interior surface of the grower. This cooling helps minimize the amount of arsenic that becomes airborne.

Heavy deposits of arsenic-containing residues are left on the inside of the crystal grower. Removal of the residues during routine preventive maintenance can result in significant airborne concentrations of arsenic (Lenihan, Sheehy and Jones 1989; Baldwin and Stewart 1989; McIntyre and Sherin 1989). Controls used during this maintenance operation often include scavenger exhaust ventilation, disposable clothing and respirators.

When the ingot is removed, the grower is dismantled. A HEPA vacuum is utilized to pick up arsenic particulates on all parts of the grower. After vacuuming, the stainless steel parts are wiped with an ammonium hydroxide/hydrogen peroxide mixture to remove any residual arsenic, and the grower is assembled.

Wafer processing

X-ray diffraction

The crystalline orientation of the GaAs ingot is determined by the use of an x-ray diffraction unit, as in silicon ingot processing. A low-powered laser can be used to determine the crystalline orientation in a production setting; however, x-ray diffraction is more accurate and is the preferred method.

When x-ray diffraction is used, often the x-ray beam is totally enclosed in a protective cabinet that is periodically checked for radiation leakage. Under certain circumstances, it is not practical to fully contain the x-ray beam in an interlocked enclosure. In this instance operators may be required to wear radiation finger badges, and controls similar to those used for high-powered lasers are used (e.g., enclosed room with limited access, operator training, enclosing the beam as much as practical, etc.) (Baldwin and Williams 1996).

Ingot cropping, grinding and slicing

The ends or tails of the single-crystal ingot are removed, using a water-lubricated single-bladed diamond saw, with various coolants added to the water. The monocrystalline ingot is then placed on a lathe which shapes it into a cylindrical ingot of uniform diameter. This is the grinding process, which is also a wet process.

After cropping and grinding, GaAs ingots are epoxy or wax mounted to a graphite beam and sawed into individual wafers through the use of automatically operated inside diameter (ID) diamond-blade saws. This wet operation is done with the use of lubricants and generates a GaAs slurry, which is collected, centrifuged and treated with calcium fluoride to precipitate out the arsenic. The supernatant is tested to ensure that it does not contain excess arsenic, and the sludge is pressed into a cake and disposed of as hazardous waste. Some manufacturers send the collected slurry from the ingot cropping, grinding and slicing processes for Ga reclaim.

Arsine and phosphine may be formed from the reaction of GaAs and indium phosphide with moisture in the air, other arsenides and phosphides or when mixed with acids during the processing of gallium arsenide and indium phosphide; 92 ppb arsine and 176 ppb phosphine have been measured 2 inches away from the slicing blades used to cut GaAs and indium phosphide ingots (Mosovsky et al. 1992, Rainer et al. 1993).

Wafer washing

After GaAs wafers are dismounted from the graphite beam, they are cleaned by sequential dipping in wet chemical baths containing solutions of sulphuric acid/hydrogen peroxide or acetic acid and alcohols.

Edge profiling

Edge profiling is also a wet process performed on sliced wafers to form an edge around the wafer, which makes it less prone to breakage. Because only a thin cut is made on the surface of the wafer, only a small amount of slurry is generated.

Lapping and polishing

Wafers are wax mounted on a lapping or grinding plate, using a hotplate, and are lapped on a machine exerting a set rotational speed and pressure. A lapping solution is fed onto the lapping surface (a slurry of aluminium oxide, glycerine and water). After a brief lapping period, when the desired thickness is achieved, the wafers are rinsed and mounted on a mechanical polishing machine. Polishing is performed using a sodium bicarbonate, 5% chlorine, water (or sodium hypochlorite) and colloidal silica slurry. The wafers are then dismounted on a hotplate, the wax is removed using solvents and the wafers are cleaned.

Epitaxy

The single-crystal GaAs wafers are used as substrates for the growth of very thin layers of the same or other III-V compounds having the desired electronic or optical properties. This must be done in such a way as to continue, in the grown layer, the crystal structure of the substrate. Such crystal growth, in which the substrate determines the crystallinity and orientation of the grown layer, is called epitaxy, and a variety of epitaxial growth techniques are used in III-V display and device production. The most common techniques are:

• liquid-phase epitaxy (LPE)

• molecular-beam epitaxy (MBE)

• vapour-phase epitaxy (VPE)

• metallorganic chemical-vapour deposition (MOCVD)—also known as organometallic vapour-phase epitaxy (OMVPE).

Liquid-phase epitaxy

In LPE a layer of doped III-V material is grown directly on the surface of the GaAs substrate using a graphite holder that contains separate chambers for the material to be deposited on the wafers. Weighed quantities of deposition materials are added to the upper chamber of the holder, while the wafers are placed in a lower chamber. The assembly is placed within a quartz reaction tube under a hydrogen atmosphere. The tube is heated to melt the deposition materials, and when the melt equilibrates, the upper section of the holder is slid so that the melt is positioned over the wafer. The furnace temperature is then lowered to form the epitaxial layer.

LPE is primarily used in microwave IC epitaxy and for manufacturing LEDs of certain wavelengths. The major concern with this LPE process is the use of highly flammable hydrogen gas in the system, which is mitigated by good engineering controls and early warning systems.

Molecular-beam epitaxy

Vacuum epitaxy in the form of MBE has developed as a particularly versatile technique. MBE of GaAs consists of an ultrahigh-vacuum system containing sources for atomic or molecular beams of Ga and As and a heated substrate wafer. The molecular-beam sources are usually containers for liquid Ga or solid As. The sources have an orifice that faces the substrate wafer. When the effusion oven (or container) is heated, atoms of Ga or molecules of As effuse from the orifice. For GaAs, growth usually takes place with a substrate temperature above 450°C.

High exposures to arsine can occur during the maintenance of solid-source MBE systems. Room air concentrations of 0.08 ppm were detected in one study when the chamber of the MBE unit was opened for maintenance. The authors hypothesized that transient arsine generation may be caused by a reaction of very fine particulate arsenic with water vapour, with aluminium acting as a catalyst (Asom et al. 1991).

Vapour phase epitaxy

Degreased and polished wafers undergo an etch and clean step prior to epitaxy. This involves a sequential wet-chemical dipping operation utilizing sulphuric acid, hydrogen peroxide and water in a 5:1:1 ratio; a de-ionized water rinse; and an isopropyl alcohol clean/dry. A visual inspection is also performed.

Two major techniques of VPE are in use, based on two different chemistries:

• the III-halogens (GaCl₃) and V-halogens (AsCl₃) or V-hydrogen (AsH₃ and PH₃)

• the III metal-organics and V-hydrogen, such as Ga(CH₃)₃ and AsH₃—OMVPE.

The thermochemistries of these techniques are very different. The halogen reactions are usually “hot” to “cold” ones, in which the III-halogen is generated in a hot zone by reaction of the III element with HCl, and then diffuses to the cold zone, where it reacts with the V species to form III-V material.The metal-organic chemistry is a “hot wall” process in which the III metal-organic compound “cracks” or pyrolyzes away the organic group and the remaining III and hydride V react to form III-V.

In VPE, GaAs substrate is placed in a heated chamber under a hydrogen atmosphere. The chamber is heated by either RF or resistance heating. HCl is bubbled through a Ga boat, forming gallium chloride, which then reacts with the AsH₃ and PH₃ near the surface of the wafers to form GaAsP, which is deposited as the epitaxial layer on the substrate. There are a number of dopants that can be added (depending on the product and the recipe). These include low concentrations of tellurides, selenides and sulphides.

A common technique used for VPE in LED processing is the III-halogen and V-hydrogen (hydride) system. It involves a two-cycle process—initially growing the epitaxial layer of GaAsP on the GaAs substrate and, lastly, an etch cycle to clean the graphite/quartz reactor chamber of impurities. During the epitaxial growth cycle, the pre-cleaned GaAs wafers are loaded onto a carousel located inside a quartz reactor chamber containing a reservoir of elemental liquid gallium through which anhydrous HCl gas is metered, forming GaCl₃. The hydride/hydrogen gas mixtures (e.g., 7% AsH₃/H₂ and 10% PH₃/H₂) are also metered into the reactor chamber with the addition of ppm concentrations of organometallic dopants of tellurium and selenium. The chemical species in the hot zone, the upper part of the reaction chamber, react, and, in the cold zone, the lower part of the chamber, form the desired layer of GaAsP on the wafer substrate as well as on the interior of the reactor chamber.

Effluents from the reactor are routed to a hydrogen torch system (combustion chamber or burnbox) for pyrolysis and are vented to a wet scrubber system. Alternatively, the reactor effluents can be bubbled through a liquid medium to trap most of the particulates. The safety challenge is reliance on the reactors themselves to “crack” the gases. The efficiency of these reactors is approximately 98 to 99.5%; therefore, some unreacted gases may be coming off of the bubbler when they are taken out by the operators. There is off-gassing of various arsenic- and phosphorus-containing compounds from these bubblers, requiring that they be quickly transported to a vented sink for maintenance, where they are purged and cleaned, in order to keep personnel exposure low. The occupational hygiene challenge of this process is profiling the exhaust effluent, since most of the out-gassed compounds from various parts of the reactor, especially the bubbler, are unstable in air and the available conventional collection media and analytical techniques are not discriminatory towards the different species.

Another concern is prescrubbers for VPE reactors. They can contain high concentrations of arsine and phosphine. Exposures above occupational exposure limits can occur if these prescrubbers are indiscriminately opened (Baldwin and Stewart 1989).

The etch cycle is performed at the end of the growth cycle and on new reactor parts to clean the interior surface of impurities. Undiluted HCl gas is metered into the chamber for periods of approximately 30 minutes, and the reactor is heated to over 1,200°C. The effluents are vented to the wet scrubber system for neutralization.

At the end of both the growth and etch cycles, an extended N₂ purge is used to flush the reactor chamber of toxic/flammable and corrosive gases.

Reactor cleaning

After each growth cycle, the VPE reactors must be opened, the wafers removed, and both the upper and the lower portion of the reactor physically cleaned. The cleaning process is performed by the operator.

The quartz prescrubber from the reactors is physically moved out of the reactor and placed in an exhausted sink where it is purged with N₂, rinsed with water and then submerged in aqua regia. This is followed by another water rinse prior to drying the part. The intention of the N₂ purge is to simply displace the oxygen due to the presence of unstable, pyrophoric phosphorus. Some residues containing various arsenicals and phosphorus-containing by-products are left on these parts even after the purge and water rinse. The reaction between these residues and the strong oxidizer/acid mixture could potentially generate significant amounts of AsH₃ and some PH₃. There is also exposure potential with other maintenance procedures in the area.

The bottom part of the quartz reaction chamber and the bottom plate (base plate) are scraped clean using a metal tool, and the particulate material (mixture of GaAs, GaAsP, arsenic oxides, phosphorus oxides and entrapped hydride gases) is collected in a metal container positioned below the vertical reactor. A high-efficiency vacuum is used for the final clean-up.

Another operation with potential for chemical exposure is cleaning the reactor’s trap. The trap cleaning is done by scraping the graphite parts from the upper chamber, which have a crust of all the previously mentioned by-products plus arsenic chloride. The scraping procedure generates dust and is performed in a ventilated sink to minimize exposure to the operators. The process exhaust line, which contains all the by-products plus moisture that forms a liquid waste, is opened and drained into a metal container. The HEPA vacuum is used to clean off any dust particles that may have escaped during the transfer of the graphite parts and from the raising and lowering of the bell jar, which knocks off any loose particles.

Metallorganic chemical-vapour deposition

MOCVD is widely used in the preparation of III-V devices. In addition to the hydride gases used as source materials in other CVD systems (e.g., arsine and phosphine), less toxic liquid alternatives (e.g., tertiary butyl arsine and tertiary butyl phosphine) are also used in MOCVD systems, along with other toxics such as cadmium alkyls and mercury (Content 1989; Rhoades, Sands and Mattera 1989; Roychowdhury 1991).

While VPE refers to a compound material deposition process, MOCVD refers to the parent chemistry sources used in the system. Two chemistries are used: halides and metallorganic. The VPE process described above is a halide process. A group III halide (gallium) is formed in the hot zone and the III-V compound is deposited in the cold zone. In the metallorganic process for GaAs, trimethylgallium is metered into the reaction chamber along with arsine, or a less toxic liquid alternative such as tertiary butyl arsine, to form gallium arsenide. An example of a typical MOCVD reaction is:

(CH₃)₃Ga + AsH₃ → GaAs + 3CH₄

There are other chemistries used in MOCVD processing of LEDs. Organometallics used as the group III elements include trimethyl gallium (TMGa), triethyl gallium (TEGa), TM indium, TE indium and TM aluminium. Hydride gases are also used in the process: 100% AsH₃ and 100% PH₃. The dopants used in the process are: dimethyl zinc (DMZ), bis-cyclopentadienyl magnesium and hydrogen selenide (H₂Se). These materials are reacted within the reaction chamber under a low-pressure H₂ atmosphere. The reaction produces epitaxial layers of AlGaAs, AlInGaP, InAsP and GaInP. This technique has been traditionally used in the manufacturing of semiconductor lasers and optical communication devices such as transmitters and receivers for fibre optics. The AlInGaP process is used to produce very bright LEDs.

Similar to the VPE process, MOCVD reactor and part cleaning presents challenges for both the process as well as the occupational hygienist, especially if large amounts of concentrated PH₃ is used in the process. The “cracking” efficiency of these reactors is not as great as that of the VPE reactors. There is a significant amount of phosphorus generated, which is a fire hazard. The cleaning procedure involves the use of dilute hydrogen peroxide/ammonium hydroxide on various parts from these reactors, which is an explosion hazard if, due to operator error, a concentrated solution is used in the presence of a metal catalyst.

Device Fabrication

The GaAs wafer with an epitaxially grown layer of GaAsP on the upper surface proceeds to the device fabrication processing sequence.

Nitride deposition

A high-temperature CVD of silicon nitride (Si₃N₄) is performed, using a standard diffusion furnace. The gaseous sources are silane (SiH₄) and ammonia (NH₃) with a nitrogen carrier gas.

Photolithographic process

The standard photoresist, aligning/exposure, developing and stripping process is utilized as in silicon device processing (see the section on lithography in the article “Silicon semiconductor manufacturing”).

Wet etching

Various mixtures of wet-chemical acid solutions are used in plastic baths in locally exhausted etch stations, some provided with vertically mounted laminar HEPA filtered supply systems. The primary acids in use are sulphuric (H₂SO₄), hydrofluoric (HF), hydrochloric (HCl) and phosphoric (H₃PO₄). As in silicon processing, hydrogen peroxide (H₂O₂) is used with sulphuric acid, and ammonium hydroxide (NH₄OH) provides a caustic etch. A cyanide solution (sodium or potassium) is also used for etching aluminium. However, cyanide etching is slowly being phased out as other etchants are developed for this process. As an alternative to wet etching, a plasma etching and ashing process is used. The reactor configurations and reactant gases are very similar to those utilized in silicon device processing.

Diffusion

A closed ampoule zinc diarsenide solid source diffusion is performed in a vacuum diffusion furnace at 720°C, utilizing a N₂ carrier gas. Arsenic and zinc arsenide are used as dopants. They are weighed in a glove box in the same manner as in bulk substrate.

Metallization

An initial aluminium evaporation is performed utilizing an E-beam evaporator. After backlapping, a last step gold evaporation is performed utilizing a filament evaporator.

Alloying

A final alloying step is performed in a low-temperature diffusion furnace, utilizing a nitrogen inert atmosphere.

Backlapping

Backlapping is done to remove deposited materials (GaAsP, Si₃N₄ and so on) from the backside of the wafer. The wafers are wax mounted to a lapper plate and wet lapped with a colloidal silica slurry. Then the wax is removed by wet stripping the wafers in an organic stripper in a locally exhausted wet chemical etch station. Another alternative to wet lapping is dry lapping, which utilizes aluminium oxide “sand”.

There are a number of resists and resist strippers used, typically containing sulphonic acid (dodecyl benzene sulphonic acid), lactic acid, aromatic hydrocarbon, naphthalene and catechol. Some resist strippers contain butyl ethanoate, acetic acid and butyl ester. There are both negative and positive resists and resist strippers used, depending on the product.

Final test

As in silicon device processing, the completed LED circuits are computer tested and marked (see “Silicon semiconductor manufacturing”). Final inspection is performed and then the wafers are electrically tested to mark defective dies. A wet saw is then used to separate the individual dies, which are then sent for assembly.

Back