EPA/600/R-02/089
1998
REDUCTION OF ARSENIC WASTES
IN THE
SEMICONDUCTOR INDUSTRY
By
Joseph T. Swartzbaugh, Ph.D.
And
Jeffrey A. Sturgill
University of Dayton Research Institute
Environmental Science and Engineering Group
Dayton OH 45469-0132
Cooperative Agreement No. CR 821808-01
Project Officer
Paul Randall
Sustainable Technology Division
National Risk Management Research Laboratory
Cincinnati, Ohio 45268
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
NOTICE
This publication was developed under Cooperative Agreement No. CR 821808-01
by the U.S. Environmental Protection Agency. EPA made comments and suggestions on
the document intended to improve the scientific analysis and technical accuracy of the
document. However, the views expressed in this document are those of the University of
Dayton and EPA does not endorse any products or commercial services mentioned in this
publication. This document is intended as advisory guidance only to the wood preserving
industry in developing approaches to waste reduction. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
11
-------
FOREWORD
The U. S. Environmental Protection Agency is charged by Congress with
protecting the Nation's land air and water resources. Under a mandate of national
environmental laws, the Agency strives to formulate and implement actions leading to a
compatible balance between human activities and the ability of our natural systems to
support and nurture life. To meet this mandate, EPA's research program is providing
data and technical support for solving environmental problems today and building a
science knowledge data base necessary to manage our ecological resources wisely,
understand how pollutants affect our health, and prevent or reduce environmental risks in
the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks from
threats to human health and the environment. The focus of the Laboratory's research
program is on methods and their cost-effectiveness for the prevention and control of
pollution to air, land, water and subsurface resources; protection of water quality in
public water systems; remediation of contaminated sites, sediments and ground water;
prevention and control of indoor air pollution; and restoration of ecosystems. NRMRL
collaborates with both public and private sector partners to foster technologies that reduce
the cost of compliance and to anticipate emerging problems. NRMRL's research provides
solutions to environmental problems by: developing and promoting technologies that
protect and iprove the environment; advancing scientific and engineering information to
support regulatory and policy decisions; and providing the technical support and
information transfer to ensure implementation of environmental regulations and strategies
at the national, state, and community levels.
This publication has been produced as part of the Laboratory's strategic long-term
research plan. It is published and made available by EPA's Office of Research and
Development to assist the user community and to link researchers with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
in
-------
ABSTRACT
The research described in this report was aimed at initiating and developing
processes and process modifications that could be incorporated into semiconductor
manufacturing operations to accomplish pollution prevention, especially to accomplish
significant reduction in the quantity of arsenic waste generated in that industry. The
effort resulted in the development of processes for the recovery of both gallium and
arsenic from gallium arsenide semiconductor crystal manufacturing. Recovery of
materials from both solid and aqueous waste streams was achieved and the solids
recovery process was demonstrated at an operating semiconductor manufacturing plant.
The processes developed herein are applicable to other types of III-V semiconductor
manufacturing, including indium phosphide, gallium phosphide and indium arsenide
manufacturing.
The two processes developed include processes for recovery of materials from
both solid and aqueous waste streams. The solid waste recovery process a thermal
process for separation of gallium and arsenic from each other and from process
contaminants with subsequent thermal refining of the captured gallium and arsenic. The
aqueous waste recovery process incorporates sequential precipitation of the arsenic and
gallium to allow for their recovery and reuse. This report was submitted in partial
fulfillment of the requirements of Cooperative Agreement No. CR 821808-01
IV
-------
TABLE OF CONTENTS
Section Description Page
Notice ii
Foreword iii
Abstract iv
Acknowledgements x
1.0 Conclusions and Recommendations 1
2.0 Introduction 3
2.1 Why Arsenic and Other Toxic Elements are
Important in Semiconductor Device
Manufacturing
2.2 "Strategic" Elements Used in Semi-conductor 5
Devices
2.3 General Summary of Manufacturing Processes 7
for Compound Semiconductors
2.4 Summary of Pollution Prevent!on Processes 13
Developed Under This Grant
3.0 Recycling and Recovery of Materials from Solid GaAs Wastes 18
3.1 Current Disposal/Recycling Methodology 18
3.2 Recovery Process Development 19
3.3 Prototype System 21
4.0 Recovery and Recycling of Gallium and Arsenic from Crystal 30
Polishing Wastewaters
4.1 Current Treatment Methodology 30
4.2 Approach for Metals Recovery 32
4.3 Methods for Arsenic Recovery 35
4.4 Methods for Gallium Recovery 39
4.5 Process for Recovery of Materials from Aqueous 42
Waste
5.0 Economic Assessment 43
5.1 Factors Affecting the Economics of GaAs and
Other III-V Material Recovery Systems
5.2 Important Economic Factors for Solid and 45
Aqueous Gallium Arsenide Recovery
v
-------
TABLE OF CONTENTS (continued)
Section Description Page
5.3 Important Economic Factors for Solid and 45
Aqueous Indium Phosphide Waste Recovery
5.4 Important Economic Factors for Recovery of 46
Other III-V Compounds
5.5 Economic Factors for Recovery of Wastes from 46
III-V Epitaxial Processes
5.6 Summary 46
References 48
VI
-------
LIST OF FIGURES
Figure No. Description Page
2.1 Process Flow Diagram for Compound Semiconductor Device 8
Fabrication
3.1 Schematic of the Recovery Process for III-V Solid Wastes 21
3.2 Cross-section of Unit Operation 1 - Thermal Separation Furnace 22
3.3 Schematic of Existing Equipment for Low-Temperature 23
Purification of Gallium
3.4 Schematic of Existing Equipment for Sublimation /Purification 24
of Arsenic
3.5 EDS Analysis of Unreacted GaAs 25
3.6 EDS Analysis of Volatized Material 26
3.7 EDS Analysis of Residue 26
3.8 EDS Analysis of Volatile Material after 2-hour Run 26
3.9 Analysis of Residue from 1-hour Run with Continuous 27
Evacuation
3.10 Analysis of Residue from 2-hour Run with Continuous 27
Evacuation
3.11 SIMS Analysis of Pure Gallium Fraction Resulting from 28
Thermal, Low-Pressure Recovery of GaAs
3.12 SIMS Analysis of Gallium "Slag" Resulting from Thermal, 29
Low-Pressure Recovery of GaAs
4.1 Current Treatment Approach for GaAs Polishing Wastes 31
4.2 Comparison of Metal Arsenate Systems 36
4.3 Experimental Residual Concentrations of As as a Function of pH 37
and Metal-Arsenate System at 25°C
vn
-------
LIST OF FIGURES (continued)
Figure No. Description Page
4.4 Experimental Residual Concentrations of As in Calcium 38
Arsenate Precipitates as a Function of pH and Temperature
4.5 Measured Residual Ga Concentrations as a Function of 40
Metal-Arsenate System and pH at 25°C
4.6 Measured Residual Ga Concentrations Following Calcium 41
Arsenate Precipitation as a Function of Temperature and pH
4.7 Developed Process for the Sequential Recovery of Gallium and 42
Arsenic from GaAs Polishing Wastes
Vlll
-------
LIST OF TABLES
Table No. Description Page
2-1 Compound Semiconductors 5
4-1 Arsenic and Gallium Concentrations in a Typical Industrial 31
Filtrate and Filter Cake using Ferric Hydroxide Coprecipitation
Methodology
4-2 Treatment/Recovery Procedures for As Considered and Tested in 33
Phase I
4-3 Treatment/Recovery Procedures for Ga Considered and Tested in 34
Phase I
4-4 Comparison of Metal-Arsenate Systems for As and Ga 39
Removal/Recovery from GaAs Polishing Wastewaters at Different
Metal-As Ratios
5-1 Estimated III-V Raw Material Costs 43
5-2 Estimated Annual Tonnage of Bulk III-V Crystals Produced in 44
1995 in the U.S.
5-3 Estimated Annual Costs of Raw Material Wasted in the Form of 45
Solids from Gallium Arsenide Crystals
5-4 Estimated Annual Costs of Raw Material Wasted in the Form of 45
Aqueous Streams from Gallium Arsenide
5-5 Estimated Annual Costs of Raw Material Losses in Epitaxial 46
Growth Processes
5-6 Total Estimated Annual Dollar Value of Wasted Raw Materials 47
from III-V Semiconductor Growth Operations
IX
-------
ACKNOWLEDGEMENTS
This research was performed under U.S. EPA Cooperative Agreement No.
CR821808-01. The researchers would first like to thank the staff of the U.S.
Environmental Protection Agency for the opportunity to perform this much-needed
research in this vital industry. Special thanks go to Mr. Paul Randall, the EPA's technical
monitor for his many useful suggestions and for his understanding and support during
difficult phases of the effort. Thanks go also to Dr. Morris Young, president of American
X-tal Technology (AXT) and to the staff at AXT for their help and input in this research
effort. Finally, the help and guidance of Ms. Laura Rae of the U.S. Air Force Wright
Laboratories, Materials Lab-Electromagnetic Materials and Mr. Bob Gedrich of the U.S.
Navy Surface Warfare Center are appreciated.
x
-------
1.0 CONCLUSIONS AND RECOMMENDATIONS
Gallium arsenide (GaAs)-based semiconductor devices are used for a multitude of
military and commercial applications in the United States and throughout the world,
including lasers, light-emitting diodes, and communications. Manufacturing processes
devoted to the fabrication of these devices generate large volumes of wastes which
contain the toxic metal arsenic, as well as the economically valuable metal gallium.
Arsenic is currently regulated under a number of federal and state laws, including
legislation that makes generating companies liable for environmental cleanup at waste
disposal sites, even if these wastes have been manifested and disposed of in an approved
manner. In addition, even though many of the wastes currently being disposed by the
industry are unlisted (e.g. solid GaAs), the toxic arsenic contained therein is regulated,
should it be released to the environment (e.g., through the action of acids, present in
many landfills). Current gallium prices make recovery of wastes containing this metal
economically viable if the recovery process is sufficiently low cost. Therefore, recovery
of these metals (As and Ga) from GaAs processing wastes is economically advantageous.
One process has been developed for the on-site recovery of both arsenic and
gallium from gallium arsenide (GaAs) solid wastes. The process described herein first
involves the thermal separation of GaAs solid wastes into their constituent elements (with
a minimum of energy input or additional handling). Each of the separated elements is
then purified to the required levels for further crystal growth using low-cost procedures.
Because of this three-step approach, the developed procedure can accommodate a wide
range of input material characteristics. Prior work with GaAs thermal separation and
constituent element purification provided a template for the development of this process,
and subsequent thermodynamic consideration of each of these unit operations provided a
theoretical basis for process optimization.
A second process was developed for the recovery of both arsenic and gallium
from gallium arsenide polishing wastes. The economics associated with the current
disposal techniques utilizing ferric hydroxide precipitation dictate that sequential
recovery of toxic arsenic and valuable gallium, with subsequent purification and in-house
reuse of both, is to the benefit of the gallium arsenide crystal grower. The developed
process involves first the removal of the majority of the arsenic and suspended polish as a
mixed precipitate of calcium arsenate and polish. This first process step is performed at
ambient temperatures and at a pH > 11 using NaOH. At these pH regimes, gallium is
retained in solution as a sodium gallate species. Precipitation of virtually pure gallium
hydroxide is then accomplished in the next process step through pH adjustment to
between 6 and 8 with waste acids. The commonly used ferric hydroxide coprecipitation
step is retained as a final treatment step, but because of the removal of the majority of the
arsenic, gallium, and polish in the two prior steps, far less waste is land disposed. A
patent application has been filed with the United States Patent Office.
In summary, the authors recommend that the processes developed under this
cooperative agreement be considered for implementation as in-plant pollution prevention
techniques. It is believed to be to the ultimate economic advantage of existing GaAs
-------
fabrication companies to minimize or altogether eliminate the amount of toxic arsenic
which is disposed of from their manufacturing operations. This not only eliminates
"short-term" costs such as manifesting and disposal, but also the much more costly "long-
term" liability costs associated with environmental cleanup. Payback for gallium
recovery is "immediate", in terms of reduction of operating costs. The payback
associated with arsenic recovery is an avoidance of future costs that might be incurred for
environmental cleanup. The processes developed will allow recovery and reuse of these
materials in a cost-effective and environmentally responsible manner.
-------
2.0 INTRODUCTION
The majority of prior research devoted to pollution prevention in the vital
semiconductor industry has focused on the replacement of ozone-depleting
chlorofluorocarbon (CFC) precision cleaners and solvents with environmentally-benign
chemicals which can function in the same capacity. With the guidance of such entities as
SEMATECH (SEmiconductor MAnufacturing TECHnology, a research consortium of
the 10 largest U.S. semiconductor manufacturers) and successful development programs
by the industrial vendors, a host of proven replacements have been developed and are
beginning to be implemented throughout the semiconductor manufacturing industry.
Ongoing efforts in this area will undoubtedly continue as further improvements in
cleaning processes are tested and marketed.
However, there are ongoing environmental threats in certain types of
semiconductor manufacturing that have been largely ignored until the present. For
example the toxic element arsenic is widely used as a principle component in important
semiconductor substrates such as GaAs and as a dopant for modifying the electronic
characteristics of other substrates. At present, much of the waste produced in arsenic-
based semiconductor manufacturing goes to land disposal, while arsenic-laden
wastewaters produced in some operations are released (at low concentrations) to local
POTWs. Arsenic is a relatively low cost material and by itself, offers little economic
incentive for the implementation of pollution prevention.
There is, however, an incentive for the development of waste minimization and
materials substitution practices because of the ongoing use of certain, relatively scarce
materials (e.g., gallium and indium) in arsenic-containing semiconductor device
manufacturing procedures. The worldwide scarcity of such materials, as well as the lack
of any domestic ores or suppliers, implies that the U.S. semiconductor industry must
continue to pay high prices for imported raw materials. A number of very possible
political or economic scenarios in foreign countries (or even in the U.S.) could drive the
price of these materials to exorbitant levels, or could cause them to become essentially
unavailable. Therefore, these materials truly can be classified as "strategic" metals,
because of their use in a high technology, defense-related capacity, as well as their
overall availability.
This report includes an introduction to the current status of the U.S.
semiconductor industry from a materials-related pollution prevention standpoint. The
U.S. Environmental Protection Agency (U.S. EPA) has an interest in the minimization of
wastes (such as arsenic) from semiconductor manufacturing operations, while at the same
time doing so in a way that will not hamper this important industry. Under the subject
project, the University of Dayton Research Institute (UDRI) has conducted specific
research into pollution prevention and waste minimization methods for this industry from
a materials recovery perspective.
-------
2.1 WHY ARSENIC AND OTHER TOXIC ELEMENTS ARE IMPORTANT
IN SEMICONDUCTOR DEVICE MANUFACTURING
Silicon (Si) has been, and will continue to be, the dominant material used for the
overwhelming majority of semiconductor device applications. Silicon itself is
environmentally-benign, and is toxic only when in the form of gaseous silane or as
certain organosilanes. In the last ten to twenty years, however, there has been a
tremendous upsurge in the use of compound semiconductors (i.e., semiconductors whose
crystalline structure contains two or more elements) for both commercial and military
applications, as these materials have moved from the laboratory to specific applications.
The usage and demand for compound semiconductors will continue to increase, in much
the same way that the demand for silicon-based devices has continued to increase. Many
of these compound semiconductors utilize chemical elements or precursor materials that
exhibit varying degrees of toxicity (e.g., arsine, phosphine, stibine, etc.) For this reason,
compound semiconductors present an opportunity to perform pollution prevention and
waste minimization on a materials recovery and reuse basis.
2.1.1 Compound Semiconductors in Use Today
Table 2-1 presents a very general summary of the compound semiconductors
which are in use today, or which are rapidly approaching widespread acceptability.
Arsenic, as a group V element, is used in the manufacture of certain so-called III-V
compound semiconductor materials (i.e., semiconductors composed of elements from
group III and group V of the periodic table). The last three "families" shown in Table 2-1
have not as yet made a significant impact on the U.S. semiconductor industry, but should
be expected to do so within the next ten to twenty years. The first three compound
semiconductor families already are in widespread use.
The III-V semiconductors are used in a multitude of device applications,
including light-emitting diodes (LEDs), lasers, detectors, and communication devices.
For example, every cellular phone presently utilizes a gallium arsenide (GaAs) chip
because the transfer of electrons through GaAs results in microwave oscillations.
Because III-V pnictides can be readily alloyed with one another, it is possible to fabricate
light emitters or detectors capable of functioning from far-infrared through green spectral
wavelengths. And, with the continuing development of nitride-based devices (e.g. GaN,
A1N), it will very soon be possible to fabricate solid-state devices that can function well
into the ultraviolet.
The majority of II-VI semiconductors have traditionally been used for
photovoltaic applications, wherein light energy is converted into electrical current.
Mercury cadmium telluride (HgCdTe), when cooled to liquid nitrogen temperatures, is an
excellent detector for use in the far-infrared spectral regime, and as such, is used on
satellites for weather or spy applications. The IV-VI chalcogenides such as lead sulfide
(PbS) are also utilized as infrared detector devices.
-------
Table 2-1
Compound Semiconductors
Compound Family
III-V
II-VI
IV-VI
IV-IV
I-III-VI2
II-IV-V2
Semiconductor Examples
GaAs, InP, InSb, GaP, GaN
CdS, CdSe, HgCdTe, ZnSe
PbSe, PbS
SiC, Si-Ge
CuInSe2
ZnGeP2
U.S. Production/Usage
Large
Large
Medium
Small
Small
Small
2.1.2 Toxic Elements and Precursors Used in Compound Semiconductor Fabrication
Lead, mercury, and cadmium have been targeted by the Pollution Prevention Act
as being among seventeen chemicals for which waste reduction options must be
attempted for those manufacturing entities that utilize them. Therefore, fabricators of
cadmium- and mercury-based II-VI compounds, as well as lead chalcogenide devices, are
required to seek methodologies to reduce emissions of these chemicals.
In addition to these three chemical elements, arsenic and selenium are currently
regulated under such legislation as the Drinking Water Act. Therefore, disposal of
wastes containing lead, cadmium, mercury, arsenic, or selenium carries the potential for
future legal liabilities. Somewhat less toxic species such as antimony, copper, zinc, or
tellurium also may be regulated on a state or local basis. Finally, it must be remembered
that dopant or epitaxial sources may also be regulated. An example of the former
includes beryllium; examples of the latter include phosphine (PH3), ammonia (NH3),
silane (SiH4), or hydrogen sulfide (H2S). Many of these epitaxial sources are among the
most toxic substances known to man. For example, the dopant and epitaxial source
arsine (AsH3) is instantly lethal in concentrations of only 250 parts per million in air.
Lower concentration exposures result in chronic effects and often subsequent death.
2.2 "STRATEGIC" ELEMENTS USED IN SEMICONDUCTOR DEVICES
The U.S. semiconductor industry currently imports the majority of its silicon from
Australia, due to the high purity reserves that are present in that country. If needed, the
industry could turn to lower-grade, domestic reserves for its source of silicon. There are,
however, three chemical elements that must be classified as "strategic" because of the
current lack of domestic reserves or suppliers, their worldwide scarcity, and their
important use in U.S. defense-related applications. These are gallium, indium and
germanium. Of these, gallium and indium are currently utilized almost exclusively in III-
V semiconductor applications, although the future will also see their use in I-III-VI2
semiconductors. Germanium was one of the first semiconductors used for
communications; its use in "cats-whisker" crystal radios even predated our understanding
of semiconductors. However, newer semiconductor applications (e.g. gallium-doped
germanium, silicon-germanium, and II-IV-V2 compounds) are causing a revival of its
usage. It must also be remembered that these three chemical elements have other, albeit
-------
specialized applications - gallium in magnesium gallate phosphors for photocopying,
germanium in bismuth germanate scintillators for astral sensor devices, and indium for
specialized plating applications. A true appreciation of the relative scarcity of these
materials and the precarious U.S. supply situation can only be realized by quickly
reviewing the sources (or lack thereof) for each of these three chemical elements.
Gallium (Ga) Sources
In terms of its abundance in crustal rocks, gallium is not that rare a chemical
element, being 30th in terms of abundance at an average concentration of 19 ppm.
However, what makes gallium so rare in terms of availability is that it is very uniformly
distributed throughout a large number of rocks at these concentrations. Thus, there are no
concentrated ores from which gallium can be extracted, as is often the case with many
chemical elements which are present in lower concentrations in crustal rocks, such as
silver, gold, or the platinum metals. There are no gallium-containing minerals of any
economic significance. The few minerals that do contain appreciable concentrations of
gallium (e.g. germanite and gallite) are so rare that they can be considered nothing more
than mineralogical curiosities.
Contacts made with semiconductor manufacturers indicated that the U.S.
semiconductor industry obtains most of its high-purity gallium from either Japan or
Germany. Gallium is concentrated as a result of the processing of other materials whose
ores contain low concentrations of gallium and thus is derived from wastes of other
industrial processes, such as flue dusts from the zinc industry or sludges from the
aluminum industry. For example, bauxite (the primary aluminum ore) typically contains
0.003 to 0.01% Ga. Concentrations in zinc ores (e.g. sphalerite) are comparable.
Because of such scarcity, industry sources say that the price of semiconductor grade
gallium recently has ranged from $0.50 to $1.50 per gram.
Indium (In) Sources
Like gallium, there are no indium minerals of any economic significance. Those
indium minerals that do exist in nature (e.g. roquesite, indite, and dzhalindite) are
exceedingly rare. Indium is one of the rarest of the commonly-used compound
semiconductor constituents in terms of its crustal abundance (61st in abundance at an
average concentration of 0.24 ppm). However, from a practical standpoint, it is more
readily available because it occurs not only in zinc and tin ores, but also in association
with lead, iron, and copper sulfide. Therefore, indium can be derived from the flue dusts
and sludges of these industries. High purity indium for semiconductor fabrication is most
often obtained from Germany or Japan, at an average cost of $50 to $150 per kilogram,
depending on purity.
Germanium (Ge) Sources
Germanium is also a rare element in the earth's crust, being 53rd in order of
abundance at an average concentration of 1.5 ppm. As with gallium and indium, no
-------
significant ores or minerals exist for commercial mining, and germanium minerals (e.g.
germanite, argyrodite, argutite, renierite, and briartite) are merely collector items. At
present, germanium is principally derived from flue dusts from the zinc or tin industries,
and, according to manufacturers, its cost approaches $300 per kilogram in raw,
unpurified form. One of the highest natural concentrations of germanium is in coal
(0.01%), and coal ashes can contain up to 1% germanium as GeO2. Some work was
conducted by the British in the 1950s and 1960s to develop methodologies for the
extraction of germanium from coal ash, but there are no such sources being used in the
U.S. at present.
2.3 GENERAL SUMMARY OF MANUFACTURING PROCESSES FOR
COMPOUND SEMICONDUCTORS
The ultra-precise manufacturing steps involved in fabricating semiconductor
components from raw materials represents one of the greatest achievement by materials
science and chemistry to date. In order to achieve materials with the final desired
electrical characteristics and properties, a large number of manufacturing steps are
necessary. Although the exact manufacturing steps are largely governed by the final
desired properties of the material, as well as the initial starting materials, a general
overview of the manufacturing steps can be summarized (Figure 1.1). These include:
• Growth of bulk substrate crystals;
• Cutting, polishing, and etching of substrates;
• Epitaxial growth of circuit constituents on substrate (if desired);
• Multisequenced masking and doping of atoms into substrate (if desired);
• Metallization;
• Alloying/annealing; and
• Final lapping and separation of individual semiconductor "chip" devices.
Typically, the first three operations are performed at facilities that grow the initial
crystal of semiconductor material. These crystal growers are referred to as
semiconductor "foundries" and they provide the basic substrates on which specific
devices can be "grown" in a controlled manner. Subsequent operations
are then carried out by one or more specialty device manufacturing houses until
semiconductor "chips" are prepared for the final user market.
Of course, each one of these general processes may involve numerous preparation and
handling steps. Also, all of these steps may not be necessary for the fabrication of a
specific electronic component, or the order in which they are carried out may be varied.
Each of these manufacturing steps is summarized below, with a particular emphasis on
the types of input materials used, as well as waste products that are typically generated.
Finally, it should be remembered that the conclusion of each of these manufacturing steps
represents a quality control checkpoint from the standpoint of meeting product quality
-------
INPUT
MATERIALS
MANUFACTURING
PROCESS
WASTES
GENERATED
Si. Ga. As, Ge. etc.
(Solids)
GROW
SUBSTRATES
Negligible
Water, etchants
( Liquids) ^
A wide range ar solids,
wide number Dfelemints
Cleaners, plastics,
Etchants, Etc. *"
A wide range nf SD lids.
wide number df elements
Negligible
Negligible
GRINDING, CUTTING
AND SLABBING
i
r
ETCH ING AND
POLISHING
1
EPITAXIAL
GROWTH
1
MAS!
1
DOPING
1
r
METALLIZATION
1
i
ALLOYING AND
ANNEALING
1
r
FINAL TESTING
AND SEPARATION
Negligible
Solid Substrate material
( e.g.. SI, GaAs. inP, etc.)
(Volume: Very Large)
Water contaminated with
^_ elchants and chemical
byproducts frnm
(Volume : Large )
Waste input materials that did
not react (Volume: Medium)
Undesirable solid materials from
reactorwaJls. (Volume; Large)
Used Input materials
contaminated with
ssmiconductDr materials
(Volume: Large )
Waste input materials that did
not react ( tfol: Medium )
Waste efchants contaminated with
5Bff!i ecnductn r m alerialS
( Volume : Small)
Negligible
Negligible
Solid material
( Volume: Very large)
Figure 2.1 - Process Flow Diagram for Compound Semiconductor
Device Fabrication
(Note-volume descriptors are relative and specific to a particular semiconductor)
-------
objectives. Any devices that do not meet specifications at each of these points will be
discarded. Such discards are a significant contribution (at least at the present time) to the
overall semiconductor waste stream.
2.3.1 Growth of Bulk Substrate Crystals
The quality of bulk substrates is an important aspect of electronic device design.
Significant improvements have been made in the areas of bulk crystal growth with regard
to uniformity, reproducibility, thermal stability, diameter control, and impurity and
dopant control. Substrate crystals of silicon or germanium have conventionally been
prepared by either the Czochralski or float-zone methodologies. However, compound
semiconductors, because they involve the joining of two or more elements, must use
more sophisticated techniques. Although many different methodologies are currently
being used in the industry today, two examples stand out as being representative of these
methodologies. The horizontal gradient freeze technique is a static technique where the
melt is gradually solidified by movement of a temperature gradient along the melt.
Vertical Bridgeman furnaces utilize a similar temperature gradient movement (in a
vertical direction, however) to achieve controlled crystal growth. Average growth rates
using these methodologies are from 1 to 5 mm per hour. Typical crystal dimensions
produced by these methodologies range from one to six inches in diameter, and from two
to thirty inches in length.
The actual growth of bulk crystals (also called boules or ingots) generates very
little waste, since starting materials are fed into the system in exact quantities. Should
excess starting materials result, then they are almost always reused in the production of
other crystalline boules. While a defective boule would represent a significant mass of
waste materials, their incidence of occurrence is very low, and even if a defective boule
were to be generated, usable portions of the material are frequently salvaged.
2.3.2 Cutting, Polishing, and Etching of Bulk Crystals
The semiconductor boule obtained has a generally cylindrical shape with
somewhat conical ends. After removing the ends (which, since they have smaller
diameters than required are usually wasted), the first operation is often surface grinding.
This process is used to precisely define the diameter of the material and is accomplished
using a rotating cutting tool (i.e., a lathe) that makes multiple passes down the rotating
boule until the desired diameter is obtained. A flat is then ground along the entire length
of the ingot, and the surface orientation is determined by cutting several slices and
measuring their crystalline orientation using an x-ray diffraction method. The cutting
saw is then reset so that the proper orientation (i.e., the desired crystal faces) is achieved
for subsequent cutting of wafers.
Upon proper orientation, the crystal is cut into thin slices called wafers. The
slicing is accomplished using either of two common wafer sawing procedures. In one
procedure, the inside diameter of a ring-shaped saw blade made of stainless steel with
diamond impregnated on the inner rim. Newer, more efficient methods, utilize a series of
-------
rapidly moving abrasive-coated steel wires over the ingot, so that multiple cuts can be
performed simultaneously. Both such cutting processes are liquid cooled, and a volume
of material is lost during this process equal to the width of the saw blade or wire. In fact,
approximately one-third of the total crystal mass can be lost as sawing fines during the
cutting process. This waste stream is in the form of fine powders suspended in an oil or
water matrix (depending on the coolant liquid used for the cutters.) The sawing operation
also leaves a damaged layer of about 20 to 50 microns thick on the wafer that is later
removed by lapping and etching.
2.3.3 Wafer Lapping, Etching and Polishing
The final operation performed at a crystal foundry is polishing. The wafers are
mounted onto large circular stainless steel polishing plates (lap plates), and either wax or
vacuum is used to hold them in place. These plates are then mounted on a polisher, and
the wafers are pressed against a tough polishing pad. A polishing agent such as alumina
and an etchant that contains a chemical oxidizer are used simultaneously, and the surfaces
are continuously flushed with water as they are polished. The etchant (oxidizer) is used
to aid in polishing by slowly dissolving some of the semiconductor material from the
surface being polished. Either one or both sides of the wafer are polished to a mirror-like
finish. After a thorough cleaning and subsequent inspection, the wafers are ready for
device fabrication. Device fabrication usually is performed at specialty job-shops, while
only the early stages of semiconductor manufacturing (i.e., boule growth, wafer cutting
and polishing) are performed at semiconductor foundries.
The high percentage of wastes associated with foundry operations makes their
wastes the largest mass of waste for the compound semiconductor industry. Toxic wastes
from these operations can be loosely categorized into two forms:
1. liquid wastes that contain dissolved metal ions from the etching and polishing
operations, and
2. solid wastes, i.e., large pieces and cutting fines from the cutting and shaping
operations.
The wastes from crystal polishing consist of the flushant wastewaters containing
dissolved substrate materials with suspended polishing agent. For example, the polishing
wastewaters from GaAs manufacturing contain dissolved arsenic and dissolved gallium at
concentrations of between 200 and 400 ppm. These wastewaters are difficult to treat
because of stringent arsenic discharge limits and because of the difficulties introduced
into any sludge dewatering operations by the extremely fine, suspended polishing agent
particles.
Solid wastes from cutting can range in size from whole segments of ingots (e.g.
6" diameter wafer pieces) to fine powders of a few microns in size suspended in the
saw's coolant liquid. Out-of-spec wafers (along with the discarded boule ends and ingot
segments) compose another form of solid wastes generated at semiconductor foundry
10
-------
operations. These latter solid wastes are physically large and are of high purity. Thus,
they have been the only part of a foundry's waste stream that has been captured for
recovery and reuse. Often, these wastes are sent to off-site contractors who process the
waste for recovery of gallium alone (with any arsenic being wasted). No effort has
previously been made to recover materials from the sawing fines or from the polishing
wastewaters.
2.3.4 Epitaxial Growth
Epitaxial growth is the means whereby ultrathin layers of exact chemical
composition are laid down on substrate wafers that have been prepared by the
methodologies described above. In particular, this is a useful means to prepare
semiconductors that are to be used for laser or LED applications. There are four general
means by which epitaxy can be achieved. These are:
• Liquid-phase epitaxy (LPE);
• Vapor-phase epitaxy (VPE);
• Organometallic vapor-phase epitaxy (OMVPE); and
• Molecular beam epitaxy (MBE).
LPE was the first commercially used epitaxial growth process and it involves the
growth of an epitaxial layer on a single crystal substrate from a solution saturated or
supersaturated with the material to be grown. VPE utilizes vaporized metal chlorides or
hydrides that are transported under controlled conditions (e.g., temperature, pressure,
flow rate) to the metal substrates. Unlike the LPE processes, smooth surfaces are
attainable, and several processing runs can be performed in an eight hour day.
Unfortunately, the toxicities associated with the hydride species used are always high.
OMVPE is an improvement over VPE because the reactions are irreversible and this
allows very abrupt transitions in composition of epitaxial structures. Such structure is a
necessity for the fabrication of digital or analog alloy systems. Another advantage is that
lower temperatures can be used for the growth processes and this minimizes the effects of
interdiffusion. Additionally, the organometallic substances used in OMVPE are less
toxic than hydrides resulting in lower toxicity starting materials and waste products alike.
MBE is the process of depositing epitaxial films from molecular or atomic beams on a
heated substrate under ultrahigh vacuum (UHV) conditions. The beams are thermally
generated from elemental feedstocks in Knudsen-type effusion cells. The thermal beams
travel in rectilinear paths to the substrate where they condense and grow under kinetically
controlled growth conditions.
With the exception of LPE, the wastes generated from epitaxial growth processes
are gaseous or solid in nature. The gaseous wastes are the "exhaust" vapors that are
drawn off from the epitaxial reactors. These are either passed through "hot boxes" where
they are mixed with oxygen and burned, or are oxidized using often proprietary catalyst
materials. Typically, the oxidized gases are scrubbed out of the effluent gas stream and
added to the plant's wastewaters for treatment. These waste materials are disposed of
with little or no effort directed towards recovery. It is reported that epitaxial growth
11
-------
processes utilize only 20 to 25 percent of the input materials for final product fabrication.
Such process "efficiency" is dependent upon both the design of the reactor and the
product being fabricated. For example, existing nitride fabrication systems utilizing
ammonia feedstocks frequently operate at only 1% efficiency in terms of input ammonia.
Solid wastes constitute those materials that are deposited on the epitaxial system's
reactor walls. It is frequently necessary to remove these wastes prior to the next
fabrication run because their presence could interfere with achieving the requisite partial
pressures necessary for each constituent in the subsequent run. In fact, fully 50 percent
of the labor hours associated with such epitaxial growth processes is devoted to cleaning
waste solids off of walls and other reactor parts. This material is not currently recycled
back into the overall manufacturing scheme partly because it is usually contaminated
with dopant and maskant materials.
2.3.5 Masking and Doping
For establishing specific circuits on a chip, it is common to add dopant atoms into
the substrate or to add epitaxial layers in specific configurations in order to produce
regions of controlled electronic behavior. This is done by first masking those regions for
which no dopant atoms are desired, and then using doping techniques for the regions that
are still exposed.
Masking can be divided into two distinct processes, both of which are necessary
for the successful transfer of an image to the surface of a semiconductor wafer. These
processes include the generation of the "mask", whose image is transferred to the wafer;
and the process of transferring the image from the mask to the surface of a wafer through
the use of a sensitized layer called a photoresist. Masking results in aqueous streams that
may have measurable concentrations of metals that had been deposited on the photoresist
when the photoresist is subsequently removed from the substrate wafer with chemical
agents.
Doping involves the emplacement of dopant atoms into selected regions of a
semiconductor crystal and this is accomplished using either diffusion doping or ion
implantation. Diffusion doping involves two distinct processes: predeposition, wherein a
carefully controlled amount of dopant is placed onto the surface of the semiconductor;
and drive-in, which uses a thermal process to cause diffusion of the dopant into the
overall crystal bulk. Today, most doping is effected using a technique called ion
implantation. This process takes ions of a desired dopant, accelerates them using an
electric field, and scans this "ion beam" across a wafer to obtain a uniform predeposition
with subsequent thermal drive-in. Older doping operations generated wastes similar to
those described under epitaxial deposition. Ion implantation techniques generate far less
wastes.
12
-------
2.3.6 Metallization/Alloying/Annealing
After the devices have been fabricated, they must be connected together to
perform electronic circuit functions. The process of implanting electrical connections is
called metallization. Alloying and annealing involve a low temperature heating to ensure
low-resistance contact between the deposited metal and the fabricated electronic devices.
These processes generate little toxic waste.
2.3.7 Final Lapping and Separation
The backside of a wafer may have to be altered in order to prepare it for
subsequent processing steps. Backside lapping of a wafer is used to remove diffused
layers that interfere with the electrical properties, to thin the wafer, or to prepare the
backside for subsequent metal deposition. When lapping the backside of wafers,
approximately ten thousandths of an inch of material is removed from each wafer. This
results in fine powders in a water carrier. Currently, these powders are filtered or settled
out of solution, and are subsequently land disposed.
The wafers are now ready for final quality control testing. To determine their
acceptability, the wafers are placed on a probe and each microchip device on the wafer is
tested. Those devices on the wafer that function properly are left alone; those that fail are
typically marked with a spot of ink. Separation into individual devices is then achieved
with a wafer scribe. This ultrathin saw or laser separates the substrate material into
square or rectangular components. Any device that does not function properly is
discarded. Finally, there is a large wastage of edge material from each wafer. Because
each wafer is round (from the crystal growing process), and because the devices
themselves are square or rectangular, there will be some solid waste generated when the
devices are separated, even if all devices contained on it pass performance inspections.
2.4 SUMMARY OF POLLUTION PREVENTION PROCESSES DEVELOPED
UNDER THIS RESEARCH EFFORT
It can be seen that in its current state, the U.S. semiconductor industry offers a
multitude of opportunities for waste reduction. Indeed, although the industry itself is
perceived as "clean", due to the ultrahigh purity required of input materials and
processing equipment, the amount and nature of the waste materials are "dirty" in
comparison to many other manufacturing industries. The semiconductor manufacturing
industry has devoted most of its resources to building better devices and electronic circuit
components, and not to minimization of wastes. Because of the high value of part of
their waste streams, combined with the environmental threats posed by other parts of
their wastes, U.S. semiconductor fabricators have a very real (but currently unrecognized)
economic incentive to implement pollution prevention and waste minimization strategies.
Current Superfund and RCRA legislation makes generating companies liable for
environmental cleanup at waste disposal sites, even if wastes have been manifested and
disposed in an approved manner. In addition, even though many of the wastes currently
13
-------
being disposed are unlisted (e.g., GaAs is not a specifically regulated waste except in
California) and usually will pass a TCLP test, the toxic metals contained therein still can
constitute an environmental threat. In such circumstances, disposal of wastes that contain
toxic metals can carry with it potential legal liabilities for environmental cleanup, even if
the wastes themselves are unlisted. Finally, shipment of wastes to off-site "recyclers",
wherein "strategic" metals are recovered, but toxic species are not, still leaves some legal
responsibility with the generating, semiconductor manufacturer should any uncontrolled
release (e.g., of unrecycled arsenic) subsequently occur.
The industry's process lines are currently geared towards the manufacture of ultra-
precise, miniaturized components, but not towards the recovery of waste fractions. The
industry generally views the many processing steps associated with device fabrication (as
well as the procedures necessary for environmental compliance) as a cost of doing
business, and this cost is passed on to the customer. When compared to other U.S.
manufacturing schemes, the overall process efficiencies are terribly low. For example,
just in the first three unit operations shown in Figure 1.1 (the crystal foundry operations),
total material wastage is approximately 50 percent of the original input material.
At the device fabrication operations (i.e., unit operation four and beyond in Figure
1.1), only about 17% of their input materials (wafers) will ultimately be usable as final
devices. In other words, for the entire process, from original crystal growth to final
device testing and separation, only approximately 8.5% of the input materials will be
used in final electronic components. The remaining material (nearly 92 percent) is
currently discarded as wastes.
In spite of the very large apparent paybacks associated with implementation of
pollution prevention and waste minimization in the industry, there are certain
characteristics unique to U.S. semiconductor manufacturing operations that need to be
addressed. These characteristics set the industry apart from other manufacturing
operations in the U.S.
2.4.1 Material Substitution Difficulties
In many manufacturing operations, substitution of environmentally-benign
materials for toxic or harmful materials has been and is being conducted. Indeed, the
successful replacement of CFC precision cleaners by other agents or processes in the
semiconductor industry has resulted in a significant decrease in regulated emissions.
However, replacement of "toxic" semiconductors (e.g. those compounds containing lead,
cadmium, mercury, arsenic, or selenium) or "strategic" semiconductors (e.g. those
containing gallium, germanium, or indium) with less toxic or lower-cost semiconducting
compounds would be a far more time-consuming process than for other manufacturing
operations. For example, it is theoretically possible that the III-V compound
semiconductor material aluminum antimonide could be grown and doped to perform
many functions of more expensive and more toxic equivalent III-Vs, but there are
difficulties associated with such material replacement concepts.
14
-------
The development period for semiconductor compounds and devices is far longer
than for most products developed by other manufacturing industries. For a given
semiconductor compound, the development period is measured in terms of years (e.g.,
twenty years) as opposed to months for most other manufactured products. And, because
the semiconductor industry is so interconnected in terms of small companies (see below),
validation of the effectiveness of replacement compounds and devices would be needed
across a number of manufacturing firms.
This is not to say that materials substitution is not possible in terms of
replacement compounds. Rather, it is the opinion of the authors that one of two scenarios
would be necessary in order to overcome the development time necessary to bring a new
product on-line, and to gain industry acceptance. In one such scenario, if a replacement
product already had a long history of development work having been performed on it,
this would significantly shorten the development time necessary for device development,
and may also help to foster industry acceptance. An example would be the material
aluminum antimonide mentioned above. Conceivably such material could replace
higher-cost, higher toxicity III-V materials such as GaAs and InP for some device
applications, and aluminum antimonide does have a long history of development work
associated with it. Under a second scenario, if the materials to be replaced were still
"new" in terms of their industry implementation, then waste minimization efforts could
more readily be incorporated into production schemes. An example would be the Ill-
nitride wide band gap devices that are just beginning to see large-scale industrial
application. Because this industrial sector is still so "new", it is still possible to
implement good waste minimization and pollution prevention practices without seriously
affecting the industry's concepts about "proper" manufacturing methods.
2.4.2 Difficulties with Small Company Size
A further impediment to implementing pollution prevention in the U.S.
semiconductor industry is the fact that unlike many other manufacturing schemes, the
entire processing operation is not done under a few, relatively large roofs. Rather, the
industry consists of many small "fab-shop" companies performing one or two of the
manufacturing steps shown in Figure 1-1, and then selling their products to companies
which are involved in additional fabrication operations further down the overall
manufacturing scheme. Pollution prevention implementation must therefore enlist the
cooperation of several companies in order to be effective for the whole industry.
Because of the small size and operating budgets of such small firms, economic
resources simply don't exist to perform research on a company-by-company basis in
order to improve process efficiencies, design new environmentally benign or lower-cost
products, etc. The cost of environmental compliance is simply passed onto the next
customer rather than implementing pollution prevention steps that would eliminate or
reduce compliance costs.
15
-------
2.4.3 Purity Concerns
Recycling in the semiconductor industry entails different concerns than for other
manufacturing industries. This is because the purity standards for most input materials to
the various process unit operations are extremely rigid. For example, materials with a
purity less than 99.999% will generally not be tolerated because they will result in
manufactured products that will not meet performance specifications. For most other
manufacturing industries, recycled wastes with a 99% purity is exceptionally good, and
materials of 90% purity can usually be reused. The question is now being raised in the
semiconductor industry, however, as to what material purity is really necessary if the
material is being recycled back into its generating process. For example, "wastes" of
GaAs consist of nearly 50 percent gallium atoms and the same number of arsenic atoms,
with minor amounts of other impurities that may be picked up during the processing
steps. If the other, minor impurities are successfully removed, does the purity of the
gallium have to be 99.999% with absolutely no arsenic present, provided that the material
is to be reused for the growth of GaAs? Would 99.9% be sufficient? Such questions can
only be answered by growth and testing of semiconductor compounds and devices using
recycled materials in which some residue of the related material is present.
2.4.4 Material Recovery Processes Developed
At the initial stages of the research effort, it was our intent to first address the
waste resulting from vapor phase epitaxial growth processes (both VPE and OMVPE).
These wastes are in the form of highly toxic chemicals (e.g., arsine, phosphine,
organoarsine, etc.) which is usually treated on-site by oxidation in a "hot box" with the
oxidized materials scrubbed from the effluent using a water scrubber. This results in
generation of wastewaters that must undergo treatment for arsenic removal prior to
release to a POTW. However, as our research efforts progressed, it soon became obvious
that there were two major reasons for focusing on waste streams other than the wastes
from epitaxial growth operations. First, other waste streams, specifically those from
foundry operations, constituted a larger and more important "target" for pollution
prevention efforts. Second, the development of recovery techniques for the foundry
wastewaters also would result in methods for recovery and reuse of materials dissolved in
the epitaxy hot-box's scrubber waters. In that way, adoption of a technique with broad
applicability in the semiconductor foundries could also allow recovery of epitaxial wastes
through an add-on process to the existing waste control methods.
Focusing on the large quantities of arsenic-bearing wastes generated in the III-V
foundries, we have developed two processes that can be implemented at foundry sites for
the recovery and in-plant reuse of both the very valuable gallium and the highly toxic
arsenic that are presently wasted. The first process developed was a thermally-based 3-
unit operation process for recovery of both gallium and arsenic from the solid wastes.
The wastes processable with this technique include both the large-sized wastes such as
boule ends and waste wafers, but also the difficult-to-treat saw fines. Together, these two
waste streams account for two-thirds of the arsenic wasted at semiconductor foundries
16
-------
(approximately 30 to 35% of the arsenic wasted in all of the GaAs semiconductor
industry.)
The other process that was developed under this grant is a method for capturing
and recovering both gallium and arsenic from the polishing wastewaters. Through use of
specific reagents, combined with careful pH and temperature control, the process
sequentially removes arsenic from the waste stream and then gallium. Upon conversion
back to metallic arsenic and gallium, the recovered wastes can be further treated for
purification using the process developed for the solid waste streams. In fact, the same
process used for recovery of oxidized gallium and arsenic from foundry wastewaters can
be applied to vapor phase epitaxial scrubber waters. Both of these recovery processes for
foundry wastes are described in detail in following sections of this report.
17
-------
3.0 RECYCLING AND RECOVERY OF MATERIALS
FROM SOLID GaAs WASTES
Due to the economic value of gallium, a number of recovery methodologies have
been developed and tested over the years, but none have been applied for in-plant
pollution prevention. The first step in these processes is typically to separate the gallium
and arsenic. A well-documented method to accomplish this separation is to contact the
solid waste stream with an aqueous solution or a heated bath, allowing a chemical
reaction to facilitate separation. These separation media consist either of an oxidizing
species such as hydrogen peroxide l'2 or nitric acid 3"6, or of molten sodium hydroxide.7
Once in solution or the heated bath, the gallium is then sequentially removed using a
number of additional methodologies. While these practices have been demonstrated as
workable, they all involve the introduction of a very large volume of an additional media
in order to effect separation. This larger-volume media then must itself be treated (e.g.,
to preclude the release of toxic arsenic or to recover gallium). Thus, rather than merely
separating the constituent elements from one another, these approaches result in the
requirement for further processing (for metal recovery) from a much larger waste stream.
3.1 CURRENT DISPOSAL/RECYCLING METHODOLOGY
On a material weight basis, approximately 50% of solid GaAs wastes generated
by semiconductor foundries are currently disposed of, and 50% are treated off-site for
recovery of the gallium fraction. Because GaAs is not (at this time) a listed hazardous
waste, disposal usually implies that the waste materials are placed into 55-gallon barrels,
and disposed of in landfills. When GaAs wastes are sent to the only existing U.S.
recycler, only gallium is presently being recycled and arsenic is presently not recycled.
One major problem with the existing disposal approach is that GaAs may be
converted to the insidious gaseous toxin arsine (AsH3) under acidic conditions. (Arsine
gas is immediately lethal in concentrations as low as 250 ppm; lower concentrations
result in chronic effects. The LDso for arsine is 0.5 ppm.) It is well documented that
landfills are typically anaerobic (reducing), with the simultaneous formation of organic
acids such as acetic and formic acids. Therefore, GaAs exposed to typical landfill
conditions could theoretically be easily converted to arsine gas. Even under less
catastrophic circumstances, such as the oxidation of the released arsenic to the trivalent or
pentavalent state, aqueous-phase arsenic will still represent a measurable toxic threat.
The prevailing attitude within the GaAs bulk crystal industry is that the existing
disposal and recycling approaches are satisfactory. Under existing U.S. environmental
laws, the original GaAs crystal grower is liable for any future environmental cleanup
costs related to releases of the arsenic into the environment, with or without recycling of
the gallium from the waste stream. Perhaps the only reason why such costs have not
begun to be incurred by today's crystal growers is because the industry is only
approximately twenty years old, and so arsenic contamination that is directly attributable
to GaAs production has not yet been observed. Judging from the large monetary sums
currently being awarded to localities (for liabilities) and environmental contractors (for
18
-------
cleanup) from disposal of arsenic-containing wood preservatives, it is safe to predict that
the future holds some very unpleasant economic surprises for today's GaAs crystal
growers if disposal approaches are not altered to allow for arsenic recovery and reuse.
Therefore, recycling of both gallium and arsenic from GaAs manufacturing wastes offers
both short-term (gallium recovery) and long-term (minimizing arsenic-related liabilities)
economic benefits
3.2 RECOVERY PROCESS DEVELOPMENT
Thermal processing of GaAs solid wastes to recover gallium has also been
demonstrated in the past. While thermal separation under air has been achieved for
1110
GaAs, that procedure results in the formation of arsenic and gallium oxides. ' These
oxide "slags" require an additional processing step (reduction) to obtain reusable metals.
Therefore, from an in-plant pollution prevention approach, separating under an inert
atmosphere or under vacuum is more desirable in order to minimize the number of
processing steps (and thus the overall cost of the recovery operation). This too has been
attempted, and many of the processes described are very exact with respect to necessary
conditions to achieve thermal separation.
Initial studies of the effects of high temperature conditions (above 950°C) showed
that thermal cracking of the GaAs takes place until the partial pressure of arsenic vapor in
the head space prevents further sublimation of arsenic. Thus, a conceptual process was
proposed in which the GaAs solids would be subjected to high temperatures at reduced
pressure with a continual draw-off of released arsenic vapors. Continued operation of
such a process would ultimately result in removal of most of the arsenic leaving a residue
that would be high in gallium, and which would contain any unmelted (or high-boiling)
contaminants. However, it was expected that such thermal separation alone would not
produce gallium or arsenic products of sufficient purity for reuse in semiconductor crystal
growth. Further processing steps would be required whereby the arsenic-rich vapors and
the gallium-rich residue could be further purified to acceptable levels for reuse.
3.2.1 Purification Thermodynamics of Arsenic
Processes for recovery and purification arsenic are not as well developed as those
proposed for gallium. In fact, no recovery/purification of arsenic is attempted in any of
the gallium arsenide recovery processes. This is the case in all reported thermal
separations - liquid gallium is recovered, but condensed arsenic is disposed of,
presumably due to its low raw material cost.8"12
Arsenic purification has been achieved through the zone refining of such arsenic
compounds as arsenic trioxide,13'14 arsenic trichloride,15 or arsine gas,16 followed by
reduction to the elemental state with a number of reducing agents, including hydrogen17'18
or metallic species such as aluminum and calcium.14 Arsenic has also been purified
through a zone refining process whereby the arsenic is heated to 814°C at a pressure of
36 atmospheres, so as to achieve a solid-liquid transition.19 Using any of these concepts
within a GaAs recovery process would mean that the compound purification processes
19
-------
would require three processing steps (conversion of elemental arsenic to a compound,
followed by purification and then re-reduction to the elemental state). Such an approach
also would present inherent toxicity concerns (since the only truly successful reduction is
that using hydrogen gas, of which arsine is a thermodynamically-feasible byproduct).
The high-pressure approach to arsenic purification also bears significant hazards for in-
house recovery operations.
Purification of arsenic using sublimation mechanisms has been achieved in the
past,20'21'22'23 and because it is a one-step process with minimal energy or toxicity (only
metallic arsenic is involved and not the more toxic oxides or hydrides) concerns, this
approach was chosen for incorporation into the overall recovery scheme. The prior art
suggests that the use of an inert carrier gas (e.g., nitrogen) aids in the separation of
arsenic from impurities.
3.2.2 Purification Thermodynamics of Gallium
Gallium purification requires a significantly different purification methodology
than those traditionally applied to other semiconductor materials (e.g. silicon or
germanium) because gallium is present in the liquid phase at ambient temperatures and
pressures. For this reason, gallium purification has been attempted in a multitude of
fashions. The methodology that was used for a number of years involved the conversion
of (impure) gallium to gallium trichloride (which is a solid), followed by zone refining of
the gallium trichloride, and reduction to pure metallic gallium.24"30 This methodology
was not considered further for our recycling process because, like arsenic purification
utilizing arsenic trichloride, it is a three-step process which requires material addition.
Additional purification methodologies that were rejected because they involve a three-
step purification process included electrolysis,31"33 hydride reduction,34 and nitride
separation.35 A one-step physical separation process of gallium from its impurities was
desired.
Because gallium has a large liquidus range (b.p. = 2403°C), a considerable
amount of energy would be required to separate gallium from other metals and from
refractories (e.g. diamond) utilizing a liquid-vapor separation methodology. Therefore, a
separation based on liquid-solid phase transitions is required, of which three procedures
are documented in the literature: single crystal lifting process, fractional crystallization,
and zone melting. The single crystal lifting process achieves purification by contacting a
seed crystal with a gallium melt, with gradual lifting,36,37 but due to its low
productivity, this physical separation methodology was rejected. Fractional
crystallization exhibits a similarly low productivity and was also rejected.
Zone melting of gallium appears to offer a reasonable and cost effective method
for purification of gallium from a thermal separation process, provided that the initial
separation results in a gallium-rich stream that is better than 90% pure gallium. At such a
purity level, the gallium-rich product should have a melting point close to the melting
point of pure gallium (approximately 35°C). Under such conditions, the gallium-rich
product would be a solid at room temperature and a low-energy heat source could easily
20
-------
raise the product temperature to the melting point allowing for an inexpensive zone
refining operation.
3.3 PROTOTYPE SYSTEM
Based on the initial studies, a conceptual recovery process was developed that
could conceivably be operable in a small space and at reasonable costs. The proposed
process also seemed likely to achieve product purities sufficient to recycle the recovered
materials back into the semiconductor crystal growing operation. The process was
envisioned to include an initial thermal separation of GaAs into an arsenic-rich vapor
stream and a gallium-rich residue. These product streams would then undergo further
processing for purification to necessary levels.
A major result of the research performed under the subject grant has been the
development of a three unit operation procedure in which the solid wastes are thermally
separated into their constituent elements (with a minimum of energy input or additional
handling). Then each of the separated elements (gallium and arsenic) is purified to the
required levels for further crystal growth. Prior work with GaAs thermal separation and
constituent element purification provided a template for the development of this
"optimum process", and subsequent thermodynamic consideration of each of these unit
operations provided a theoretical basis for implementation into the developed process.
Figure 3.1 shows a schematic of the developed process for solid III-V materials as
it currently exists. As a result of this research, it is now known that the lowest cost means
input GaAs
solid wastes
impure
arsenic
Arsenic
condenser
refined
arsenic
Thermal
Separation
Furnace
impure
gallium
Zone
Refiner
refined
gallium
Figure 3.1 - Schematic of the Recovery Process for III-V Solid Wastes
21
-------
to separate III-V solids is through the low-pressure, high-temperature process shown as
the first unit operation in Figure 3.1. It is also now known that some further processing
of recovered gallium or arsenic is necessary in order to maximize the quantity of
recovered material for reuse.
3.3.1 Unit Operation 1 - Thermal Separator
The first and most critical operation in the developed recovery process is the
thermal separation furnace. Figure 3.2 shows a cross-section of the thermal separation
furnace that was constructed for laboratory and field trials of the proposed concept. The
unit can be operated at temperatures above 950°C and at reduced pressure with an inert
atmosphere. The off-gases are continually pumped through a series of condensers for
capture of the arsenic-rich vapors. The residue in the reactor contains the gallium-rich
fraction mixed with a separable slag of other contaminants.
It was known that thermal separation alone could not achieve product purities
great enough to allow for immediate reuse of the recovered products. In addition to
contaminants introduced into the waste stream, some contaminants could be intentionally
introduced into the crystals as dopants for specific control of the crystal's electronic
characteristics. Dopants commonly found in GaAs include Si, Zn, and C or Cr. These
dopants typically occur in concentrations of about 10 18 atoms/cc. Some of these dopants
(e.g., C at about 1015 atoms/cc) are utilized to make GaAs semi-insulating, and some are
utilized to make GaAs semiconducting n-type (Si) or p-type (Zn). The physical
characteristics of each dopant are important because they will dictate where the dopant
Thermocouples
Fill Tube
and
Viewport
Up-to-Air
Valve *-»-
Pressure Guage
Stainless
Steel
Shell
Resistance
Heating
Elements
Thermal Wrap
Insulation
Bottom Plate
Figure 3.2 Cross-section of Unit Operation 1 - Thermal Separation Furnace
22
-------
will likely occur in the product streams resultant from the thermal separation. For
example, silicon and iron are very soluble in liquid gallium, and so these elements can
be expected to remain with the gallium, forming an impure "slag". Simultaneously,
"volatile" dopants such as sulfur and selenium are expected to partition into the arsenic
fraction. For this reason, further processing steps were incorporated into the process for
purification of the gallium-rich and arsenic-rich product streams.
3.3.2 Unit Operation 2 - Low-Temperature Zone Refining of Gallium
Because the impurity levels in the product gallium from doped GaAs sources are
expected to be fairly low (because dopant concentrations are relatively low), then the
melting temperature of the "impure" gallium should approach that of pure gallium (i.e., ~
35°C.) UDRI has developed a system for the purification of nearly pure gallium with
small concentrations of impurities (Figure 3.3). Specifically, it involves chilling the
gallium with ice water or refrigerant to 0°C or less, and then using a controlled heat
r
Low-speed
Reversible Motor
Spiral
Groove
Figure 3.3 - Schematic of Existing Equipment for Low-Temperature Purification of
Gallium
23
-------
source (heat lamp) to heat specific zones of the gallium sample. (Note: Although gallium
is a solid at ambient temperatures of 20-25°C, this is not a sufficiently cold temperature
for efficient zone refining due to supercooling effects. In other words, zone refining from
ambient room temperature will not be successful because the temperature difference
between solid and liquid will not be sufficient to effect impurity segregation.) The
molten zone is allowed to pass through the gallium by slowly rotating the pan containing
the gallium. Based upon the segregation coefficients of each of the contained
"impurities", the dopant elements will segregate to both ends of the spiral groove in the
gallium refining pan shown in Figure 3.3 and can then be removed.
3.3.3 Unit Operation 3 - Sublimation Refining of Arsenic
As noted previously, the arsenic fraction can be contaminated with the more
volatile dopants, especially carbon. UDRI has achieved some purification of arsenic
through the use of a repeated sublimation/condensation process as is shown in Figure 3.4.
Specifically, the arsenic in the first condenser (e.g. evolved from the low-pressure, high-
temperature process in Figure 3.2) is heated to slightly above the sublimation temperature
of arsenic (610°C) in an inert gas stream such as nitrogen, and recondensed in a second
condenser. This thermally separates the arsenic from impurities due to differences in
partial pressure and volatility. Further processing can be achieved through additional
sublimation/condensation (e.g., heating from condenser 2 into condenser 3, etc.).
Resistance
Healing Eements
Figure 3.4 - Schematic of Existing Equipment for Sublimation /Purification of
Arsenic
24
-------
3.3.4 Analytical Results
Initial studies were performed to evaluate the basic thermal separation concept
that ultimately was incorporated as unit operation number 1 in the concept process.
Analyses of samples were conducted using Energy Dispersive Spectroscopy (EDS). This
surface analysis technique is capable of detecting the presence of atomic constituents in
solid samples down to the parts per million (ppm) level. Thus, EDS will show the
presence of unacceptable contaminants at the ppm level and was suitable for initial
assessments of the performance of the recovery processes. However, EDS is not a truly
quantitative technique and it was known that a more sensitive technique would ultimately
be required in order to demonstrate that the recovered material has sufficient purity for
reuse.
Figure 3.5 shows an EDS analysis of a slab of unreacted GaAs solid waste typical
of the feed material used in subsequent studies. Initial studies of the thermal separation
concept were performed under an inert atmosphere (TSte) but at atmospheric pressure. It
was soon realized that as the sample released volatile arsenic, the partial pressure of
arsenic rose until further arsenic sublimation stopped. Continued operation at high
temperature allowed other, less-volatile materials to escape with no real increase in the
arsenic removal
i tea~ ^ !
t
i 1
i 1 : :
i
• i
i
j
. i
i
\
As
' A •
Ga/ \ .
/v \
J \ J \ X
69 10 11 1£
KeV
Figure 3.5 EDS Analysis of Unreacted GaAs
rate. Figure 3.6 shows the EDS analysis of a recondensed solid sample of the volatile
fraction resulting from a 30 minute run at approximately 1000°C. The presence of an
indium peak results from a low-level (< 0.3%) of indium dopant present in the original
material. Figure 3.7 shows that, for such operating conditions, high levels of arsenic
remain in the residues. In fact, continued operation at high temperature and at
25
-------
IN
i As
it
It
As
GaJL i
f n
.A j i n
i
3 45 6 7 83 10 1112 13 14 15
^ i
_> v
J
10
KeV
11
12
Figure 3.6 EDS Analysis of Volatiized
Material
Figure 3.7 EDS Analysis of
Residue
atmospheric pressure and above can result in the eventual volatilization of gallium so that
the recondensed solid will contain significant proportions of gallium as a contaminant in
the arsenic-rich fraction (Figure 3.8).
,Ga
\As
f\
A
10
KeV
1 1
12
Figure 3.8 EDS Analysis of Volatile Material after
2-hour Run
At this point, it became obvious that, for the process to achieve effective
separation of the gallium and arsenic fractions, the evolved arsenic must be continuously
removed and recondensed outside of the thermal separation reactor. A thermal reactor
was developed that could be continuously evacuated through condenser units for capture
of the evolved arsenic. The EDS analysis results on the residual solids are shown in
26
-------
Figures 3.9 and 3.10 for two runs wherein the evolved gases were pumped out through
cooled condensers. In each of these samples, some low level of unvolatilized arsenic is
shown (the small "knee" to the right of the smaller gallium peaks in each figure)
indicating that the gallium-rich residue would probably still require further purification.
1 .......
j
1
I . ...
i
1
! i
;Ga
i •
I
-. Ga
\ A?* .
i
i
i
j
i
J
1
1
t
Ga
i
i
I ...
'/Us ' • '
I / Vf
KeV
1 £
45 6 7 8 9 10 1112 13 14 15
KeV
Figure 3.9 Analysis of Residue from 1- Figure 3.10 Analysis of Residue from 2-
hour Run with Continuous Evacuation hour Run with Continuous Evacuation
Because of the presence of the low levels of arsenic (approximately 3 to 4%) in
the residue after thermal separation, it was decided that an additional process step would
be required to achieve adequate gallium purity for its reuse. Thus, a low-temperature
zone refiner was chosen as the probable best method for purification of the residual
gallium-rich material. Similarly, because the volatile fraction could contain materials
other than arsenic, it was decided that a sequence of condensation steps followed by
volatilization steps would purify the arsenic-rich fraction evolved in the original thermal
separation. That logic resulted in the development of the concept process shown in
Figure 3.1 at the beginning of this section.
2.3.5 Field Trials at a Semiconductor Foundry
The process for recovering reusable materials from solid GaAs wastes was tested
at the AXT Fremont facility in August of 1996. While the primary goal of the field
testing was to establish that the proposed recovery process could be effected without
disturbing normal GaAs foundry operations, a secondary objective was to establish the
purity of the recycled materials.
Waste GaAs kerf and wafer pieces were subjected to the low-pressure, low-
temperature thermal process utilized as the first unit operation in the recovery process for
solids. Specifically, batches of waste material were placed into a graphite/SiC crucible
within the recovery reactor, the pressure reduced to < 1 torr, and the temperature raised to
> 1050°C. The waste material was processed for 2-3 hours under these conditions, and
arsenic separated out as a condensable vapor, leaving a gallium-rich residue in the
crucible. The arsenic and gallium fractions were collected and it was seen that the
residue was composed of two different gallium-rich fractions. The purity of the two
different gallium fractions from this process is shown in Figures 3.11 and 3.12 which
27
-------
present SIMS (Secondary Ion Mass Spectrograph) analyses of the product streams.
Figure 3.11 is a SIMS analysis of a low-melting fraction that results from this first unit
operation. As can be seen, the resultant material is virtually pure gallium. Figure 3.12
shows the purity of a higher-melting "slag" that typically forms around this pure gallium.
As can be seen, this material contains substantial amounts of iron, silicon, and other
detrimental elements, which account for the higher melting temperature of this fraction.
Equivalent quantities of arsenic were recovered from the condenser as a finely
divided powder. No reliable method of measuring the purity of arsenic to the four-9s
level was available for the study. The only reliable method to measure the purity of
recovered arsenic, is by using it in combination with gallium of certified purity to grow
crystals of gallium arsenide. The purity of the recovered arsenic can then be determined
by measuring the characteristics of these crystals.
Because of the extremely hjigh purity of the gallium fraction and the difficulties
of measuring arsenic purity, the arsenic purification step and the gallium zone refining
step were not performed during the field test of the process. Discussions with the
foundry operator indicated their desire to participate in future development activities
aimed at a commercial process for in-plant pollution prevention.
10-
10H
10"
10'
Ga
O
O
GaO
/
iim" i "I
0 20 40 60 80 100
FigureS. 11 - SIMS Analysis of Pure Gallium Fraction Resulting from Thermal,
Low-Pressure Recovery of GaAs
(note - presence of oxygen and GaO peaks results from use of
oxygen as primary ion source)
28
-------
o
10
100
Figure 3.12 - SIMS Analysis of Gallium "Slag" Resulting from
Thermal, Low-Pressure Recovery of GaAs
29
-------
4.0 RECOVERY AND RECYCLING OF GALLIUM AND ARSENIC FROM
CRYSTAL POLISHING WASTEWATERS
Facilities that grow bulk crystals of GaAs perform a number of subsequent
processing steps wherein the boules (or "ingots") are slabbed into wafers, and etched and
lapped to remove surface damage. The final operation performed by these facilities is
polishing, to achieve a mirror finish on one or both faces of the wafer. The wafers are
mounted onto large circular stainless steel polishing plates (lap plates) and either wax or
vacuum is used to hold them in place. These plates are then mounted on a polisher and
the wafers are pressed against a tough polishing pad. The polishing is done "wet" in
which a very fine polishing agent such as alumina and a mild etching agent that contains
an oxidizing species are used to remove surface materials through a combination of
mechanical and chemical action.
Use of an oxidizing species as the polishing etchant results in solubilized metal
ions according to the following general reactions:
GaAs + "oxidizer" > Ga+3 + As+5 + "residual oxidizer" (Eq. 1)
A number of chemical oxidizers have been used as polishing etchants in the laboratory
and in industry to polish the respective metals38'39'40. In general, it is desirable to utilize
chemical species which will aid the polishing operation by oxidizing arsenic to the water-
soluble +5 valence state, because the use of acids (without oxidizer) will lead to the in-
plant generation of toxic arsine (AsH3) gas. The most commonly used oxidizer species
are hydrogen peroxide, chlorinated compounds (especially hypochlorite), and nitric acid.
Typical concentrations of oxidizer are 30% in water, depending on oxidizer species.
The wet polishing process results in an aqueous wastestream that contains from
200 to 400 ppm of each dissolved metal, as well as residual oxidizer concentrations of
from 3 to 10%. At such concentrations, those wastestreams require subsequent treatment
for arsenic removal prior to discharge of the water. The polishing wastestream itself has
a "milky" appearance, due to the large concentrations of very fine polish (e.g. < 0.5
micron) suspended within it. Although some of this suspended polish will settle after
time, most remains suspended; hence the white appearance. The pH of these solutions is
dependent upon the initial oxidizer solution used; but the resultant wastestream is
generally more basic than the initial solution, due to a number of factors (e.g. presence of
oxide polish, generation of soluble gallium, etc.). Any process for precipitation and
removal of the arsenic concomitantly will also remove the suspended polishing agent.
4.1 CURRENT TREATMENT METHODOLOGY
The current treatment approach for GaAs polishing wastes is shown in Figure
4.1. A soluble ferric iron species (e.g., ferric chloride or ferric nitrate) is added to the
polishing wastewaters, and the pH is adjusted so as to precipitate insoluble ferric
30
-------
NaOH
200-400ppm Ga3+
200-400ppm As5+
@10% residual oxidizer
Ferric Hydroxide
Co-precipitation
to POTW
Precipitated Fe(OH)3
with absorbed Ga and As
Filtration
-^ to Landfill
Figure 4.1 - Current Treatment Approach for GaAs Polishing Wastes
hydroxide. The toxic arsenic is "coprecipitated" with the ferric hydroxide. Coagulating
and flocculating agents are added to aid in the physical removal of the resultant
precipitate.
Table 4-1 shows the resultant arsenic (and gallium) concentrations for a typical
industrial filtrate and filter cake using this approach. It is immediately obvious from
inspection of the appearance of the filter cake that a tremendous excess of ferric iron is
necessary on a weight-to-weight basis with respect to arsenic. This results in a large
volume of waste solids which must be disposed, and which could readily leach toxic
arsenic. The colloidal nature of the polishing
Table 4-1
Arsenic and Gallium Concentrations in a Typical Industrial Filtrate and Filter Cake
using Ferric Hydroxide Coprecipitation Methodology
Arsenic (ppm)
Gallium (ppm)
Filtrate Discharge*
1.8-2.5
8.5- 10.5
Filter Cake**
10.3 -24.6
7.5- 18.4
* Filtrate concentrations determined following methods described in EPA 600/4-79/020 . Samples taken from industrial discharge in
August 1996.
** Metal concentrations determined using a technique in which weighed, pulverized sample of filter cake was placed into 25.0 ml of
70% nitric acid solution and allowed to sit in this solution for 24 hours at ambient temperature, but with no agitation. Metal
concentrations in the extracting acid were determined using Perkin-Elmer Model 3030B AA Spectrophotometer, and measured
concentrations correlated back to filter cake concentration.
agent adds difficulty to the physical separation process (as will be discussed later). For
this reason, it is difficult to obtain consistent arsenic concentrations in the discharged
filtrate on a day-to-day basis.
31
-------
An additional concern for the disposed filter cake, from a materials recovery
standpoint, is that the two materials for which recycling is desirable (arsenic and gallium)
are now intimately mixed with a tremendous excess of a third material (iron). Therefore,
recovery and recycling of arsenic and gallium from the current filter cake would be
extremely difficult. This, combined with the problems sometimes encountered with
meeting arsenic discharge limits for the filtrate, led the authors to develop a process that
not only will treat for arsenic, but will do so in a way to allow for recovery of both
arsenic and gallium.
4.2 APPROACH FOR METALS RECOVERY
UDRI has found that for the development of many pollution prevention and
treatment systems, especially in areas (such as this) where there is no proven solution to
the problem, it is best to begin with almost an "Edisonian" approach wherein the widest
possible selection of realistic concepts are evaluated. The advantage to this approach is
that it precludes inherent prejudices for or against options which may have problems
associated with implementation. The disadvantage to this approach is that the researcher
is confronted with the need to "start from the ground up" in the development of a
treatment solution. This disadvantage can be overcome through the use of a simple,
straightforward phased test approach (as described in this section). This testing approach
will quickly weed out those potential candidates (or variables) which appear to have
shortcomings associated with their continued use in the program.
A literature search revealed only one previous paper detailing "recovery" of
gallium and arsenic from polishing waste waters.42 The described process, involving two
evaporation steps (and thus having high energy demands), did not result in a separation of
gallium and arsenic from one another. Therefore, it truly was necessary to start from the
"ground up" for the development of a recovery approach.
Phase I testing involved the preparation of "surrogate" solutions containing 250
ppm gallium (as Gads) or arsenic (as NasAsO/t), half of the solutions contained oxidizer
(H2O2) and the other half did not. (No polishing agent was added to these Phase I
solutions - as the intent was merely to identify technically feasible chemical
treatment/recovery options.) Candidate treatment options (listed in Table 4-2 and 4-3)
were then evaluated using the prepared "surrogate" solutions. The residual metal
concentrations (after settling) were measured using a Perkin Elmer AA
spectrophotometer. Table 4-2 details those options which were tested in Phase I for
arsenic separation and recovery, while Table 4-3 details those options which were tested
for gallium separation and recovery. Observations and comments about each of the
candidate treatment/ recovery options tested are also included in Tables 4-2 and 4-3.
Phase I testing indicated that the most likely candidate for successful arsenic
removal/recovery is a procedure that results in an arsenate species precipitate.
Operational difficulties were encountered with sulfide precipitation and stannous
reduction procedures in the presence of 5% oxidizer. The shortcomings described in
Table 4-2 for other approaches indicated that arsenate precipitation is the only reasonable
32
-------
treatment/recovery option that could be used if arsenic recovery after removal is a goal.
With this in mind, an attempt was made to coprecipitate both gallium and arsenic as
"gallium arsenate" from simulated polishing wastes using a number of techniques (e.g.,
pH and temperature control, "salting out" of solution, etc.). If successful, "gallium
arsenate" species could then be reduced back to gallium arsenide. All attempts to prepare
43
this species were unsuccessful - gallium arsenate exhibits too high a solubility.
Table 4-2
Treatment/Recovery Procedures for As Considered and Tested in Phase I
Procedure
1) Pyrogallol chelation
2) Gallic acid chelation
3) Sulfide precipitation
a)NaHS
b) Thioacetamide
4) Arsenate precipitation
5) Chemical reduction
a) Stannous ion
b) Sulfite ion
c) Hypophosphite ion
d) Aluminum metal
e) Magnesium metal
6) Ion exchange
7) Ferric hydroxide
(baseline)
Cited References by Number
(See Reference Section)
44,45,46
44,455,46
44,47,48,49,50
51,52,53
54
55
Comments/Observations
Successful for +3 valence state of
As; unsuccessful for As+5 and in
presence of oxidizer.
Unsuccessful under all application
conditions.
Precipitation without oxidizer
resulted in formation of very fine
particle size which was difficult to
settle or filter; attempted
precipitation in presence of
oxidizer was unsuccessful.
Successful (using calcium and
magnesium) with and without
oxidizer.
Stannous ion successful both with
and without oxidizer; although
required concentration was much
higher for oxidizer. Sulfite and
hypo-phosphite unsuccessful for
all conditions. Aluminum and
magnesium resulted in highly
toxic arsine formation.
Successful with no oxidizer;
unsuccessful in presence of
oxidizer.
Successful with and without
oxidizer.
A number of potential candidates were found which appeared suitable for gallium
recovery; the most favorable being hydroxide or phosphate precipitation. Although
fluoride precipitation also appeared favorable for gallium recovery, there was some
concern on the part of the researchers about adding large concentrations of a fluoride
33
-------
Table 4-3
Treatment/Recovery Procedures for Ga
Considered and Tested in Phase I*
Procedure
1) Chelation
a) Hydroxyquinoline
b) Tannic acid
c) Trioctylamine
d) Pyrogallol/gallic acid
2) Hydroxide precip.
a) NaOH
b) NH4OH
c) Urea
3) Phosphate precipitation
a) sodium phosphate
b) triethyl phosphate
4) Oxalate precipitation
a) sodium oxalate
b) diethyl oxalate
5) Malonate precipitation
6) Sulfide precipitation
7) Fluoride precipitation
8) Chemical reduction
a) stannous ion
b) hypophosphite ion
c) metallic zinc
9) Ion exchange
10) Ferric hydroxide
(Baseline)
Cited References by
Number
44,56,57,58,59
44,60
60,61,62,63,64,65
66
67,68,69
70
71,72
73
74
75
54,76,77,78
55
Comments/Observations
Hydroxyquinoline and
dibromohydroxyquinoline
precipitation was effective in absence
of oxidizer; consistent results were
difficult to obtain in presence of
oxidizer.
Other chelating agents were
ineffective.
Gelatinous precipitates produced
under all conditions with NaOH and
NH4OH. Urea failed to precipitate
Ga in presence of oxidizer, possibly
due to oxidation of urea.
Precipitation using sodium phosphate
successful under all conditions.
Triethyl phosphate failed to
precipitate Ga.
Precipitation unsuccessful under all
conditions.
Precipitation unsuccessful under all
conditions.
Fine particle size of precipitates made
physical separation difficult; no
precipitation occurred in presence of
oxidizer.
Precipitation successful under all
conditions for Ga.
Removal using stannous ion
successful under all conditions; higher
concentrations necessary in presence
of oxidizer. Zinc and hypophosphite
removal unsuccessful.
Removal unsuccessful in presence of
oxidizer.
Successful with and without oxidizer.
* Note: One reagent cited as a gravimetric reagent for gallium, cupferron, was not tested due to its high
toxicity.
compound (e.g., NaF) to a wastewater system, as this water would then require treatment
for fluoride removal prior to discharge.
Following review of the phase I results, a sequential precipitation process was
proposed for GaAs polishing wastes, wherein the arsenic species was first removed as an
34
-------
insoluble arsenate with the gallium then being removed either as a hydroxide or
phosphate. These proposed processes were evaluated more thoroughly in Phase II
testing.
4.3 PHASE II TESTING OF METHODS FOR ARSENIC RECOVERY
Arsenic removal efficiency using metal arsenate precipitation is dependent upon
the following process variables:
• Specific metal employed;
• System pH;
• System temperature; and
• Reactant concentrations.
In order to minimize the costs associated with arsenic removal, inexpensive metal-
arsenate systems were chosen for the treatment/recovery process. An additional
restriction required that the metal additive not be toxic, so as to preclude subsequent
treatment requirements for the wastestream. Phase I testing indicated the feasibility of
using calcium- or magnesium as the metal additives in arsenate precipitation systems.
Additional metal additive systems considered included iron-arsenate and manganese-
arsenate precipitation. These were not tested because the calcium and magnesium
systems appeared to be both technically successful and cost effective
The literature indicates that the ferric hydroxide coprecipitation process currently
used in industry for waste treatment is in actuality a combination ferric arsenate-ferric
hydroxide precipitation process; that is, the arsenic is primarily precipitated in the form of
ferric arsenate. Optimum pH conditions for the calcium, magnesium, and iron systems
were derived from the literature (Figure 4.2). Optimum pH conditions for ferric-arsenate
removal is in the acidic regime (pH 2-3). Optimum pH conditions for calcium-arsenate
(pH 11-12) and magnesium-arsenate (pH 9-11) are basic. Figure 4.2 also indicates that
the magnesium- and calcium-arsenate systems are at least as effective as the ferric-
arsenate system for arsenic removal. Additional process-related concerns can be found in
the existing literature.51"53'79"81
Samples of GaAs polishing wastewaters were obtained from American
X-tal Technology (AXT) in Fremont, California in August, 1996. Measured arsenic
concentrations in these wastewaters at the time of their sampling were 210 to 215 ppm.
(Slightly lower initial gallium concentrations of 145 to 150 ppm were measured in these
same wastewaters, as will be discussed.) The pH of the solutions varied from 8.3 to 8.8,
depending upon the concentration of alumina polish present in the samples.
An initial evaluation of the effectiveness of the four metal systems was conducted
by adjusting the pH of 25 ml samples of this wastewater to the desired pH with NaOH or
HNOs, adding the metal salt (as chlorides) for a 10:1 metal to As ratio on a mole basis,
centrifuging for 10 minutes, and then extracting the top portion for As analysis using a
35
-------
• 3
I
! »
I ,
2 3
pH
10 12
pH
Figure 4.2 - Comparison of Metal Arsenate Systems
47
Perkin-Elmer Model 303OB AA spectrophotometer following procedures outlined in
EPA Methods for Water and Wastewater, part 200.43 The pH meter used was calibrated
with 4.00, 7.00, and 10.00 standards, and the AA spectrophotometer was calibrated using
10, 100, and 250 ppm standards. System temperature during these tests was maintained
at25°C +1°C.
The plots of residual As concentrations versus pH for each of these metal systems is
shown in Figure 4.3. Several important issues were determined from this study:
1) Given the proper pH conditions, the removal effectiveness of Fe, Ca, and Mg
arsenates are comparable. The fluctuations in residual metal concentrations at
identical pH values for identical metals indicates that further optimization is a
function of physical separation, not of chemistry. For example, 7 calcium
arsenate samples were filtered through a 0.1 micron filter, while 7 others were
centrifuged for 10 minutes. The resultant residual concentration profiles confirm
the hypothesis that the physical separation method is as important as the
chemical process used.
2) Manganese offers no advantage over the use of iron, calcium or magnesium.
3) At higher pH values, gallium (e.g. gallium hydroxide) resolubilizes. Therefore,
for those systems which utilize high pH values for As removal (especially
calcium-arsenate), gallium will remain in solution, allowing for subsequent
removal of gallium containing very low concentrations of arsenic.
4) The sludge volumes produced are very high, but this is due in large part to the
large amount of polish present. In fact, once precipitation occurs (i.e., after the
addition of the soluble metal chlorides), the solution is so "thick" that air bubbles
can become "trapped" in the resultant slurry mix. Although this effect was not as
noticeable at lower pH values (e.g., < 9), a definite "thickening" of the solution
still occurs. This probably contributes to the separation difficulties encountered
in industry with the ferric hydroxide precipitation.
36
-------
o
to
O
•05
Oi
CC
200
180
160
140
120
100
80
60
40
20
0
10
11
12
PH
Figure 4.3 - Experimental Residual Concentrations of As as a Function of pH and
Metal-Arsenate System at 25°C.
The first issue is especially important for this wastestream. As mentioned previously,
even prior to treatment, the solution has a milky appearance due to the high
concentrations of very fine alumina polish suspended within. A number of techniques
(e.g., centrifugation, filtration through 0.1 micron filters, use of settling aids) were
attempted to remove this polish prior to treatment for arsenic removal, but all were
unsuccessful. This suggests that the polish itself imparts colloidal properties to this
solution, and this strongly affects any subsequent physical separation process. In fact,
variability in residual arsenic concentrations, as was seen from sample to sample in our
experiments, is to be expected from day to day in an industrial setting due to the nature
of the colloidal suspension of polish. None of the treatment approaches tested resulted in
arsenic concentrations acceptable for discharge, primarily because flocculating agents
and coagulants typically used in actual industrial practice were not included in the tests.
The literature indicates that for the arsenate systems of interest, higher
temperatures reduce the solubility of the resultant precipitates. Therefore, a subsequent
test program similar to that described above was performed wherein calcium
precipitations were performed at two higher temperature values (40 and 60°C).
Experimental results for these tests are plotted versus the 25°C data in Figure 4.4. This
information suggests that the increase in arsenic removal achieved at elevated
temperatures for this wastestream may not be worth the energy cost of raising the
temperature. In addition, no significant decrease in the sludge volume due to elevated
temperatures was observed.
37
-------
130
120
110
100
90
80
70
60
50
40
30
20
10
10
11
12
Figure 4.4 - Experimental Residual Concentrations of As in Calcium Arsenate
Precipitates as a Function of pH and Temperature.
Finally, the effect of additive metal concentration upon residual arsenic (and
gallium) concentrations was studied using the procedure described above, with a fixed
pH value for each metal (11.8 for Ca, 10.5 for Mg, 9.6 for Fe). The experimental results
are shown in Table 4-4.
The lowest metal-to-arsenate ratio utilized for calcium or magnesium is 1.5:1, whereas a
1:1 ratio is used for iron. These ratios were selected based on the precipitation reactions:
+2
3Ca+z + 2AsO4"
AsO
+3
-3
Ca3(AsO4)2
FeAsC>4
(Eq. 2)
(Eq. 3)
The data in Table 4-4 suggest that a metal-to-arsenic ratio of 2:1 is sufficient for both
calcium and magnesium to remove arsenic, whereas much higher amounts of iron are
necessary to achieve comparable removal efficiencies. In addition, it was found that for
magnesium, and especially calcium, increased metal-to-arsenic ratios noticeably increase
the volume of sludge produced (hence, the majority of the sludge for greater-than-
stoichiometric concentrations is likely the metal hydroxide of the additive metal).
38
-------
Table 4-4
Comparison of Metal-Arsenate Systems for As and Ga Removal/Recovery from
GaAs Polishing Wastewaters at Different Metal-As Ratios.
Metal
Sample
Ca-32
Ca-33
Ca-34
Mg-8
Mg-9
Mg- 10
Fe- 14
Fe- 15
Fe- 16
Metal: As
Ratio (mole)
1.5:1
2:1
5:1
1.5:1
2:1
5:1
1:1
2:1
5:1
Final
pH (±0.2)
10.3
10.3
10.5
8.4
8.5
8.8
7.1
7.4
7.6
Sludge
Vol. (%)*
24.3
25.8
36.9
22.8
24.5
31.7
-
-
-
Residual As
cone.
(ppm ±1)
25
25
21
16
18
17
78
57
39
Residual Ga
cone.
(ppm ±1)
154
176
158
36
34
35
16
10
31
* Sludge removed was transferred to graduated cylinders to measure volume remaining. This number
compared to the original volume of wastewater tested is the reported sludge volume percent. Sludge
volumes for the ferric hydroxide process were not measured.
4.4 PHASE II TESTING OF METHODS FOR GALLIUM RECOVERY
The efficiency of gallium removal from the polishing wastewater is dependent
upon the following process variables:
• The particular anionic species chosen in which to precipitate gallium;
• System pH;
• System temperature; and
• Reactant concentrations.
Phase I studies indicated that two anionic species were suitable for gallium precipitation;
hydroxide and phosphate. Because the intended final fate of these precipitates is
conversion (through a reduction process) to elemental gallium, it was decided that use of
gallium hydroxide as a precipitate would be less problematic. That is, reduction of
gallium phosphate would likely yield gallium phosphide which would then in turn have
to be converted to gallium metal, or even elemental phosphorus. Neither of these
outcomes is desirable. Therefore, it was decided that phosphates would not be added to
the wastewaters, since they would potentially interfere with conversion of the
precipitates to a recoverable material.
The concentrations of gallium in the as-received wastewater samples from AXT
were slightly lower (e.g., 50 ppm less) than the arsenic concentrations in the same
samples. This is attributable to the fact that at the observed pH of these samples (8.3 to
8.8), gallium hydroxide is relatively insoluble in water. The samples received from AXT
were taken from the top of the collection tank for the polishing waste; therefore, any
39
-------
precipitated gallium hydroxide had a chance to partially settle in the collection tank prior
to sample collection.
The literature contains sufficient information about the solubility of gallium
hydroxide to allow an understanding of what is happening to the gallium during the
arsenic recovery processes, and thus to optimize the subsequent gallium recovery
process.67'82'83 When sodium (or potassium) hydroxide is applied to a solution
containing gallium, various hydrated oxides and hydroxides of gallium begin to
precipitate as the pH rises. At a certain pH, however, a new species, sodium gallate
[NaGa(OH)4] is favorable, which is soluble in water. The pH at which this species is
formed depends greatly upon the type and concentration of other ions, but it is generally
in the pH range of 9.7 to 11,82 In other basic environments, such as in NH4OH, the
gallate species does not occur. The literature suggests that the minimum solubility of
gallium hydroxide species is at a pH of 5 to 7, depending again upon ionic conditions.
Finally, the literature also suggests that the solubility of gallium hydroxide decreases with
decreasing temperature. Based on this information, a metal gallate/ gallium hydroxide
based recovery concept was chosen for testing.
Figure 4.5 illustrates the measured amount of residual gallium remaining in
solution as a function of metal-arsenate system and pH, while figure 4.6 shows the effect
of temperature during metal-arsenate precipitation upon residual gallium concentrations
in this waste stream. The effect of calcium, magnesium, and iron concentrations upon
residual gallium concentrations were shown above in Table 3-4.
180
160
140
1=
o
~g 120
o 100
o
60
40
20
0
10
11
12
Figure 4.5 - Measured Residual Ga Concentrations as a Function of
Metal-Arsenate System and pH at 25°C
40
-------
1BO
160
140
a
o
I 120
,i,
o 100
o
"co
1 80
en
,-i,
CL
60
40
20
0
25"C
60DC
10
11
12
Figure 4.6 - Measured Residual Ga Concentrations Following Calcium Arsenate
Precipitation as a Function of Temperature and pH
This information suggests that the optimum conditions to solubilize gallium
during arsenic precipitation for this wastestream is to utilize a calcium-arsenate
precipitation at room temperature, at pH conditions of 11 or higher. It was surprising that
in this concentration regime, gallium concentrations were not higher for the magnesium-
arsenate removal. (Perhaps because magnesium gallate precipitates.)
As a final "proof-of-concept" for the arsenic and gallium recovery process from
these wastewaters, five 25 ml replicate samples of polishing wastewaters were first
subjected to a simulated treatment/recovery system from this process. The first unit
operation in the tested concept was a calcium arsenate precipitation process at 25°C, with
a calcium-to-arsenic ratio of 5:1, and in pH ranges from 11-12. The resultant precipitate
was allowed to settle, the sludge volume and pH measured, and a 5 ml sample of the
filtrate drawn off for Ga and As analysis using AA spectrophotometry. The filtrate was
then treated with sulfuric acid (to simulate etching acids, another wastestream available at
these facilities) in the second unit operation to adjust the pH to conditions at which
gallium hydroxide will precipitate. The volume of the gelatinous precipitate of gallium
hydroxide sludge was then measured as well as the solution pH. The resultant precipitate
was filtered, and a 10 ml sample of the filtrate was drawn off for Ga and As analysis.
Finally, the third unit operation featured further treatment with ferric iron, to simulate the
existing coprecipitation process that can be used for a final polishing step in the treatment
prior to discharge. Sludge volumes, final pH, and residual As and Ga concentrations
were measured for these samples. All five trials resulted in residual arsenic
concentrations less than 5 ppm, with recovery of the majority of the original dissolved
arsenic (as calcium arsenate) and gallium (as gallium hydroxide).
41
-------
4.5 PROCESS FOR RECOVERY OF MATERIALS FROM AQUEOUS WASTES
As a result of the above studies, a process has been developed and tested on actual
GaAs polishing wastes which not only allows for the treatment of a toxic species
(arsenic), but does so in a way that allows for its recovery and reuse. The process also
recovers the "strategic" metal gallium from these waste streams. The developed process
is shown in Figure 4.7. In the first step, the pH of the wastewaters is adjusted to between
11.5 and 12 with sodium hydroxide. A soluble calcium salt is added in metal-to-arsenic
ratios of no more than 5:1, but preferably 2:1 (to minimize sludge volume). The resultant
precipitate is passed into a centrifuge which removes precipitated calcium arsenate and
polish. Centrifuged sludge possessed an average solids concentration of 8.4% without
further filtration. The centrate liquid is added to precipitation supernatant liquid and then
passed into a second reaction tank wherein the pH is adjusted to between 6 and 8 through
the use of waste etching acids. The resultant fine, gelatinous precipitate of gallium
hydroxide is allowed to settle, and is filtered through a 0.1 to 0.5 micron filter. The final
supernatant and filtrate liquids are then sent to the (existing) ferric hydroxide
coprecipitation system for final treatment of residual arsenic. Removal of the colloidal
polish particles in the first (calcium arsenate) precipitation results in greater ease for
physical separation of precipitates in the two subsequent treatment processes.
Centrifugation is the preferred separation technique for the first unit operation in the
process due to the nature of the suspended solid material; filtration is acceptable for the
two subsequent processes.
ZOO - 400 ppm of
Group III element
8
200 - 400 ppm of
Group V element
plus residual oxidizer
To Recovery
or Disposal
ADJUST pH TO 9.5-12.5
USING ALKALI METAL
HYDROXIDE
ADD SOLUBLE
ALKALINE METAL SALT
CENTRIFUGE
sludge
_y
SOLID ALKALINE METAL
Group V OXYANION
ADJUST pH TO 6 - 8
USING WASTE ACIDS
liquid
_y
FERRIC HYDROXIDE
COPRECIPITATION
liquid
W
SETTLING / FILTRATION
sludge
GELATINOUS GROUP II!
METAL HYDROXIDE
To Reduction
Process
sludge
_V
FERRIC HYDROXIDE
FILTER CAKE
To Disposal
Figure 4.7 - Developed Process for the Sequential Recovery of Gallium and Arsenic
from GaAs Polishing Wastes.
42
-------
5.0 ECONOMIC ASSESSMENT
5.1 FACTORS AFFECTING THE ECONOMICS OF GaAs AND OTHER III-V
MATERIAL RECOVERY SYSTEMS
Discussions with semiconductor manufacturers indicated that similar levels of
wastage were to be seen for all types of semiconductors. Because of similarities in the
chemistry and physics of III-V semiconductors (including InP, GaP and InAs) it soon
became obvious that with only slight changes, all of the processes developed for GaAs
could be applied to other III-V manufacturing. Thus, as part of the evaluation of the
pollution prevention concepts developed under this research effort, the researchers
performed an estimate of the economic benefits associated with the recovery and
recycling technologies for the III-V semiconductor manufacturing industry. The
economic calculations that are made here are confined to the U.S., and only to the III-V
semiconductor manufacturing industry.
Three cost contributors make up the majority of the cost benefits associated with
recovery of toxic and/or strategic materials. These are:
1) elimination of short-term environmental costs associated with waste disposal,
2) elimination or minimization of long-term liabilities that could be incurred in
the event of a release of hazardous wastes, and
3) elimination of some raw materials costs due to the reduction in usage because
they are replaced with recycled materials.
It was quickly found that item #3 offers significant economic benefits due to the high
costs of gallium and indium. Raw material costs for III-V elements at present can be
estimated as shown in Table 5-1.
Table 5-1 Estimated III-V Raw Material Costs
Material Estimated Present Cost
Gallium $ I/gram = $900,000/ton
Indium $0.25/gram = $225,000/ton
Aluminum Negligible
Arsenic $0.05/gram = $45,000/ton
Phosphorus Negligible
Antimony $0.06/gram = $54,000/ton
43
-------
The above costs do not take into account the additional processing that is
necessary for purification of raw materials to level of purity that is required for
semiconductor growth. Nor do they take into account the additional processing to
convert these metals into organometallic species for epitaxial growth processes. In other
words, benefits accruing from recovery of materials that do not require these additional
processing steps are not considered here, making this analysis very conservative.
In order to estimate an annual benefit from materials recovery, it was first
necessary to ascertain the volumes of III-V materials currently being produced in this
country. No definitive volume numbers by year could be found in the literature for the
year 1995. It is known that in 1985, 20 tons of bulk gallium arsenide were produced in
the U.S. However, not only was 1985 more than 10 years ago, it was rather early in the
"lifetime" of this semiconductor material. Since 1985, a multitude of additional uses for
gallium arsenide have been developed. No U.S. Bureau of Mines information could be
found for annual usage of gallium (or arsenic, for that matter), but a 1992 value is
available for germanium usage, which is used almost exclusively in some special
semiconductor devices. The 1992 usage of germanium (7 tons), along with the 1985
value of gallium arsenide produced, provides a basis for estimating the bulk tonnage of
GaAs crystals produced in this country in 1995. These estimated values are shown in
Table 5-2 below. The 1995 usage of GaAs was estimated to be 20 to 30 times greater
than the usage of germanium. However, to be conservative, a multiplier of only 15 is
used. Discussions with industry sources provided the basis for estimating other III-V
usage.
Table 5-2 Estimated Annual Tonnage of Bulk III-V Crystals
Produced in 1995 in the U.S.
Gallium Arsenide - 100 tons
Indium Phosphide - 10 tons
All other III-V compounds - 0.5 to 1 tons (total)
Although these III-V compounds are formulated with 1:1 atomic ratios of Group III and
Group V elements, one must recall that the weights of these elements differ. Therefore,
although 10 tons of gallium arsenide contain roughly 5 tons each of gallium and arsenic,
10 tons of indium phosphide contain approximately 8 tons of indium and 2 tons of
phosphorus.
In 1995, 2.5 tons of arsine and 6.5 tons of phosphine were produced and used in
the U.S. in 1995 for epitaxial growth and doping operations. It is estimated that
approximately 0.5 tons each of organoarsines and organophosphines as well as some
organoantimony compounds were used in these processes, thereby contributing a total of
10 tons of gaseous Group V source materials used in 1995. Because the ratio of Group V
to Group III elements is typically 2:1, it is further estimated that 5 total tons of Group III
elements (gallium, indium, aluminum) were used in epitaxial growth processes in 1995.
Because gallium and indium are heavier than aluminum, it is estimated that 2 tons each of
44
-------
gallium and indium were used in organometallic form in 1995. In addition, molecular
beam epitaxy processes typically utilize the elemental species of these Group III and V
elements. Therefore, it is estimated that 3.5 tons of arsenic, 7.5 tons of phosphorus, 0.5
tons of antimony, 2.5 tons of gallium, 2.5 tons of indium, and 1.5 tons of aluminum were
used in all aspects of epitaxial growth and doping operations in 1995.
5.2 IMPORTANT ECONOMIC FACTORS FOR SOLID AND AQUEOUS GALLIUM
ARSENIDE RECOVERY
It is estimated that 100 tons/year of bulk gallium arsenide crystals are produced in
this country, and it is known that from crystal growth to final chip production
approximately 75% of these materials are wasted (sawing wastes, wafer breakage, out-of
spec pieces and chips. Then an estimated 75 tons of gallium arsenide are wasted
annually, or approximately 37.5 tons each of gallium and arsenic. Therefore, the total
cost of raw materials that become wastes are as shown in Table 5-3.
Table 5-3 Estimated Annual Costs of Raw Material Wasted in the Form of Solids
from Gallium Arsenide Crystals
Gallium Costs: 37.5 tons X $900,000/ton = $33,750,000
Arsenic Costs: 37.5 tons X $45,000/ton = 1,687,500
TOTAL $35,437,500
In communications with industry sources, it was indicated that the metals lost
during aqueous etching and polishing operations equal approximately 2% of bulk crystal
raw material inputs. Therefore, 2 tons per year of gallium arsenide can be assumed to be
lost during aqueous etching and polishing operations. Estimates based on these numbers
are shown in Table 5-4.
Table 5-4 Estimated Annual Costs of Raw Material Wasted in the Form of
Aqueous Streams from Gallium Arsenide
Gallium Costs 1 ton X $900,000/ton = $900,000
Arsenic Costs 1 ton X $45,000/ton = $45,000
TOTAL $945,000
5.3 IMPORTANT ECONOMIC FACTORS FOR SOLID AND AQUEOUS INDIUM
PHOSPHIDE WASTE RECOVERY
Similar calculations were performed for raw material costs for indium phosphide
semiconductor manufacturing, since that is another important III-V semiconductor whose
production volume is expected to grow in the near future. The estimated annual cost of
raw material wasted in the form of solids from indium phosphide crystal production is
based on an estimate of 6 tons wasted at a cost of $225,000 per ton for a total of
45
-------
$1,350,000. The estimated annual cost of wasted raw materials as aqueous wastes from
indium phosphide production is based on an annual estimate of 0.16 tons wasted. Using
the same cost per ton of raw materials, this gives $36,000 per year worth of raw materials
wasted in the aqueous waste streams of indium phosphide producers.
5.4 IMPORTANT ECONOMIC FACTORS FOR RECOVERY OF OTHER III-V
COMPOUNDS
Similar calculations were also performed for the other III-V compounds that are
currently being produced in the U.S. The calculated values for the raw material losses
and potential savings are shown in the final summary table.
5.5 ECONOMIC FACTORS FOR RECOVERY OF WASTES FROM III-V
EPITAXIAL PROCESSES
The reactor systems currently used for III-V epitaxial growth processes currently
operate at about 25% efficiency - that is, 75% of the input material ends up as wastes in
the form of solid wastes on reactor walls (which is then land disposed), or in the form of
scrubber waters or oxidized species from treatment operations to the reactor exhausts.
Cost calculations for these waste streams are shown in Table 5-5.
Table 5-5 Estimated Annual Costs of Raw Material Losses in Epitaxial Growth
Processes
Gallium losses 2.5 tons X 0.75 X $900,000/ton = $1,687,500
Indium losses: 2.5 tons X 0.75 X $225,000/ton= $ 421,900
Aluminum losses: negligible cost
Phosphorus losses: negligible cost
Arsenic losses: 3.5 tons X 0.75 X $45,000/ton= $ 118,100
Antimony losses: 0.5 tons X 0.75 X $54,000/ton = $ 20,000
TOTAL $2,247,500
5.6 SUMMARY
The numbers presented in this section are disturbing in that they represent very
appreciable dollar costs in an industry that is still in its first couple decades of existence.
There are three means to accomplish reductions of these costs. The first involves the
work that UDRI has performed under the subject grant involving capture and recovery of
the toxic and valuable metals from the processing steps. The second involves the
development and construction of more efficient processing steps that do not result in 75
to 85% wastage. The third alternative may be the development and substitution of lower
cost alternatives for some III-V semiconductor materials.
46
-------
The potential dollar savings presented here are probably low. A great deal of
simplicity was used in order to estimate these values. For example, the costs associated
with the extensive purification and synthesis processes to produce semiconductor-grade
material sources were not considered. One must also consider that this industry is
expected to double in size before the end of the century due to the tremendous increase in
demand from the private sector for III-V devices. Some of the materials that are
currently being produced in the one ton per year range are expected to undergo dramatic
increases in demand. Other III-V materials which now only appear in research
laboratories will begin to be commercially produced before the end of the century. For
example, gallium nitride has been proven as a blue LED material, and commercial
production of this material should soon begin. Finally, one needs to consider the large
quantities of wastes from prior years that have either already been land disposed or are
being stored in warehouses because the parent companies do not wish to dispose of them.
All of these factors will raise the total dollar value of the materials that can be recovered
using the technologies developed under this grant. A summary of the estimated value of
these is presented in Table 5-6.
Table 5-6 Total Estimated Annual Dollar Value of Wasted Raw Materials
from III-V Semiconductor Growth Operations
Source Dollar Value - 1995
Gallium Arsenide - all solid wastes $35,437,500
Gallium Arsenide - aqueous wastes $945,000
Gallium Antimonide - all solid wastes $178,900
Gallium Phosphide - all solid wastes $168,700
Indium Phosphide - all solid wastes $1,350,000
Indium Phosphide - aqueous wastes $36,000
Indium Antimonide - all solid wastes $52,300
Indium Arsenide - all solid wastes $50,600
Aqueous wastes exclusive of GaAs and InP $10,000
Epitaxial Growth and Doping - all III-V elements $2,247,500
Evaluating the worth of the recoverable materials requires further estimations. If it is
assumed that in the early stages of implementation, only 10% of the available material
would be processed, then, based on gallium arsenide alone, $3.5 million worth of
materials could be recovered annually. This quantity would be appropriate for a typical
foundry. Capital and operating costs should be minimal for the processes sized for a
typical GaAs foundry, with capital equipment estimated at less than $500,000 per plant
and annual operating costs of less than $150,000 per plant. In such a case, extremely
favorable cost recovery rates could be achieved for implementation of the developed
processes.
47
-------
REFERENCES
1) Jadvar, R., et al., "Recovery of Gallium and Arsenic from GaAs Wafer
Manufacturing Slurries," Env. Progress 10(4): 278-81, 1991.
2) Coleman, J.P., and B.F. Monzyk, "Oxidative Dissolution of Gallium Arsenide and
Separation of Gallium from Arsenic," U.S. Patent 4,759,917; to Monsanto Co., July,
1988.
3) Yamashita, H., and K. Kato, "Recovery of Individual Metals from Gallium-Arsenic-
Indium Semiconductor Wastes by Flotation," Anal. Sci. 6: 783-4, 1990.
4) Charlton, T.L., "Recovery of Gallium from Gallium Compounds," U.S. Patent
4,094,753; to Cominco Ltd., June, 1978.
5) Showa Keikinzoku K.K., "Recovery of Gallium," Jpn. Kokai Tokkyo Koho JP
59,213,622; December, 1984.
6) Inooka, M., "Metal Impurity Removal from Scrap Gallium," Jpn. Kokai Tokkyo
Koho JP 63,270,428; to Chiyoda Chem. Eng. and Construct. Co., Ltd., November,
1988.
7) Nippon Mining Co., Ltd., "Recovery of Indium and Gallium from Scrap
Semiconductors Containing III-V Compounds," Jpn. Kokai Tokkyo Koho JP
60,027,613; February, 1985.
8) Abrjutin, V.N., et al., "Process for Producing High-Purity Gallium," U.S. Patent
4,362,560; December, 1982.
9) Inooka, M., "Vacuum Pyrolysis of Scrap Gallium," Jpn. Kokai Tokkyo Koho JP
63,270,430; to Chiyoda Chem. Eng. and Construct. Co., Ltd., November, 1988.
10)Hirayanagi, K., and A. Sakamoto, "Recovery of Gallium," Jpn. Kokai Tokkyo Koho
JP 61,076,627; to Showa Keikinzoku K.K., April, 1986.
11) Inooka, M., and M. Soma, "Method and Apparatus for Continuous Recovery of
Gallium," Jpn. Kokai Tokkyo Koho JP 01 47,825; to Chiyoda Chem. Eng. and
Construct. Co., Ltd., February, 1989.
12) Clauss, D., et al., "Recovery of Highly Pure Gallium from Thermally Decomposable
Gallium Compounds," East German Patent 120,861; July, 1976.
13) Wolff, G.A., "Purification of Arsenic and Antimony," Patent US 2,944,885; 1960.
14)Geach, G.A. etal, "The Purification of Pure Arsenic," J. Chem. Soc. p. 1207, 1950.
15)Harada, R.H. and A.J. Strauss, "Preparation of Indium Arsenide," Appl. Phys. 30:
121, 1959.
16)Effer, D., "Purification of Arsenic," J. Electrochem. Soc., 108: 357, 1961.
17) Antell, G.R. and T.W. Roberts,, "Method for Purification of Arsenic by Hydrogen
Reduction," Patent US 2,975,048; 1961.
18)Mehal, E.W., "A Novel Approach to Purification of Arsenic," Patent US 3,039,755;
1962.
19) Weisberg, L.R. and P.R. Celmer, "Arsenic Purification by Crystal Growth from the
Melt," J. Electrochem. Soc. 110: 56-9, 1963.
20)Weisberg,L.R. andF.D. Rossi, ""Vapor Zone Refining," Rev. Sci. Instrum. 31: 206-7,
1959.
21) Siemens - Schuckertwerke A.G., "High-Purity Elements and Compounds by
Sublimation," Patent Belgium 661,023; 1965.
48
-------
22)Baba, K. et al, "Muffle Reactor for Sublimation of Arsenic," Jpn. Kokai Tokkyo
KohoJP 62,156,237; 1985.
23) Becker, K.P. et al, "High Purification of Volatile Substances," Patent Ger. (East) DD
234,410; 1986.
24) Richards, J.L., "Purification of a Metal by Zone-Refining One of its Salts," Nature
111: 182-3, 1956.
25) U.K. Admiralty, "Preparation of Pure Gallium," Chemistry & Engineering News,
p.2887, 1956.
26) Goldsmith, N. et al, "The Preparation of High-Purity Gallium," J. Less-Common Met.
4: 564-71, 1962.
27) Kern, W., "Zone Refining of Gallium Trichloride," J. Electrochem. Soc. 110: 60-5,
1963.
28)Hutter, J.C. and A. Peyron, "Purification of Gallium by Halogenation and
Electrolysis," Patent US 3,075,901; 1958.
29) Mayer, A., "Pure Gallium by Chloride Reduction," Patent US 3,150,965; 1961.
30)Kulifay, S.M., "Purifying Gallium by Washing with Acids," Patent US 3,166,405;
1965.
31) Stearns, R.I., "Purification of Gallium," Nature 203: 749-50, 1964.
32)Lysenko, V.I. and P.P.Tsyb, "Purification of Gallium from Microimpurities,"
Zh. Prikl. Khim. 38: 488-94, 1965.
33)Mikhailov, V.A. et al, "Removal of Impurities from Liquid Gallium by
Electrodiffusion in Static Cells," Dokl. Akad. Nauk SSSR 184: 158-60, 1969.
34) Okamato, Y. et al, "Preparation of High-Purity Gallium by Hydride Reduction,"
Proc. Conf. Ultrapurif. Semicond. Mater., Boston, MA, pp. 94-105, 1961.
35)Merkel,H., "Spectroscopically Pure Gallium," Patent US 2,927,853; 1960.
36)Zimmerman, W., "Gallium Purification by Single Crystal Growth," Science 119:
411-2, 1954.
37)Degilh R. and C. Blaizat, Patent France 1,382,904; 1964.
38) Sinha, M.P. and S. Mahapatra, "SEM study of etching of GaAs substrates in the
H3PO4:H2O2:H2O system", Microelectron. J. 18(2): 48-51, 1987.
39) Wismayer, A.C. and B.L. Weiss, "The Etching of Gao.yAlo.sAs Using KI-I2," Mat.
Lett. 6(8,9): 284-6, 1988.
40)Hseih, H.F. et al, "A Comparison of the Electrochemical Behavior of Illuminated n-
InAs and other III-V Semiconductors in Concentrated Nitric Acid," J. Electrochem.
Soc. 139(7): 1897-1902, 1992.
41) US Environmental Protection Agency, Methods for Chemical Analysis of Water and
Wastewater. EPA-600/4-79/020, March 1983
42) Jadvar, R. et al, "Recovery of Gallium and Arsenic from GaAs Wafer Manufacturing
Slurries," Environ. Prog. 10(4): 278-81, 1991.
43) Sebba, F. and W. Pugh, "Gallium IV. Phosphates and Arsenates of Gallium," Trans.
Roy. Soc. S. Africa 25: 391-401, 1938.
44) Burger, Organic Reagents in Metal Analysis, Pergamon Press, 1973.
45) Suzuki, N. et al, "Liquid-Liquid Extraction of Arsenic(III), Arsenic(V),
Methylarsenate and dimethyl arsinate in Various Systems," Anal. Chim. Acta 185:
239-48, 1986.
49
-------
46) Votava, J. and M. Bartusek, "Reactions of Arsenic Acid with Pyrocatechol and
pyrogallol," Collect. Czech. Chem. Commun. 42(2): 620-6, 1977.
47) Ruiz Cano-Cortez, J., "Solubility Products of Arsenic, Antimony and Bismuth
Sulfides," Dyna 46(9): 405-6, 1971.
48)Fedor, J. et al, "Evaluation of the Possibility of Arsenic Storage in the Form of
Sulfides," Rudy 37(2): 37-40, 1989.
49) Milyutina, N.A. et al, "Solubility of Sulfides and Oxides of Some Minor Metals in
Aqueous Solutions of Sodium Sulfide," Tr. Inst. Met. Obogashch., Akad. NaukKaz.
SSR 21: 14-19, 1967.
50) Sedlacik, J. et al, "Detoxification of Arsenic-Containing Wastewaters from
Desulfurization of Coke-Oven Gas," Czech Patent, CS 261,397; 1989.
51)Mahapatra, P.P. et al, "Solubility of Calcium Hydrogen Arsenate in Aqueous
Medium," Indian. J. Chem. 25A: 647-9, 1986.
52)Robins, R.G., "The Solubility of Barium Arsenate: Sherritt's Barium Arsenate
Process," Metall. Trans. B 16: 404-6, 1985.
53)Mahapatra, P.P. et al, "Physicochemical Studies on Solid Solutions of Calcium
Phosphorus Arsenic Hydroxyapatites," Bull. Chem. Soc. Jpn. 62: 3272-7, 1989.
54) Bard, et al., Standard Potentials in Aqueous Solution, Marcel Dekker, 1985.
55) CRC Handbook of Ion Exchange, Boca Raton, FL, 1986
56)Rane, A.T. and D.R. Nepali, "A Comparative Solvent Extraction Study of Indium(III)
withS-Mercatoquinolinol and Its Analogues," Indian J. Chem. 25 A: 846-51, 1986.
57) Mohammad, B. et al, "On-Line Preconcentration of Aluminum, Gallium and Indium
with Quinolin-8-ol for Determination by Atomic Absorption Spectrometry," J. Anal.
AtomicSpect. 8: 325-31, 1993.
58) Saito, K. and M. Tamura, "Slow Isotopic Exchange of Chelate Compounds of Indium
in Water and in Anhydrous Organic Solvent - letter," J. Inorg. Nucl. Chem.
13:334-5, 1960.
59) Lewis, R.G., "Process for the Extraction of Gallium from Aqueous Solutions,"
U.S. Patent, US 4,965,054; 1990.
60) Abisheva, Z.S. et al, "Mechanism of Gallium Extraction by tri-n-octylamine,"
Process Metall. 7A: 949-54; 1992.
61)Fischer, C. et al, "On the Extraction of In(III) with tri-n-octylamine from HC1
Solutions,"/. Inorg. Nucl. Chem. 39:513-7, 1977.
62) Fischer, C. et al, "On the Extraction of In(III) with tri-n-octylamine from HBr and HI
Solutions,",/. Inorg. Nucl. Chem. 42: 891-5, 1980.
63) Yakabe, K. et al, "Liquid-Liquid Extraction of Gallium(III) Complex Ion from
Aqueous Oxalic Acid Solutions with High Molecular Weight Amine," J. Inorg. Nucl.
Chem. 39: 871-5, 1977.
64)Bagreev, V.V. and Yu.I Popandopulo, "Co-extraction and Suppression of Extraction
of Thallium in the Presence of Indium, Gallium and Iron(III) in the Systems
Trioctylamine-hydrchloric Acid and Aliquat 336-Hydrochloric Acid," Proc. Anal.
Div. Chem. Soc. 15(1): 1-4, 1978.
65) Sato,T. et al, "Liquid-Liquid extraction of Gallium(III) from Hydrochloric Acid
Solutions by Organophosphorus Compounds and High-Molecular Weight Amines,"
Solvent Extr. lonExch. 2(2): 201-12, 1984.
50
-------
66) Ambulkar, R.S. et al, "Complexes of Gallium(III) with Pyrocatechol, Pyrogallol and
Protocatechuic, a-Resorcyclic, Gallic and 2,3-Dihydroxynaphthalene-6-sulphonic
Acids in Aqueous Medium," Indian J. Chem. 20A: 1044-6, 1981
67) Sato, T. and T. Nakamura, "Studies of the Crystallization of Gallium Hydroxide
Precipitated from Hydrochloric Acid Solutions by various Alkalis," J. Chem. Tech.
Biotechnol. 32: 469-75, 1982.
68) Yura, K. et al, "preparation and Properties of Uniform Colloidal Indium Compounds
of Different Morphologies," Colloids and Surfaces 50: 281-93, 1990.
69) Thompson, L.C.A. and R. Pacer, "The Solubility of Indium Hydroxide in Acidic and
basic Media at 25°C," J. Inorg. Nucl. Chem. 25(8): 1041-4, 1963.
70) Goiffon, A. et al, "Solubilites des Phases MinXvO4 (M=Al,Ga; X=P, As) dans les
Solutions d'Acide H3XO4 sous Conditions Hydrothermales: Application A la
Crystallogenese de 1'Arsenate d'Aluminium AlAsO/t," J. Cryst. Growth 71: 763-70,
1985.
71)Dixit, K.R. et al, "Potentiometric Studies on Some Mixed Ligand Complexes of
Gallium(III) and Indium(III) with Ethylenediaminediacetic Acid as Primary Ligand,"
J. Indian Chem. Soc. LXI: 199-201, 1984.
72)Kul'ba, F.Ya. et al, "Complexing of Gallium(III) with Oxalate Ions in an Aqueous
Solution in Relation to Acidity," Zh. Neorg. Khim. 19(3): 674-80, 1974.
73)Kulshrestha, Renu, et al, "Polarographic Reduction of Indium(III) in its Mixed
Ligand Complexes with Picolines and Carboxylate Ions," Indian J. Chem. 26A: 845-
8, 1987.
74) Tunaboylu, K. and G. Schwarzenbach, "Solubility of Indium Sulfide," Chimia
24(12): 424-7, 1970.
75) Tananaev, IV. and N.V. Bausova, "The Chemistry of Gallium Fluorides and Their
Utilization for the Separation of Gallium from Other Metals," Khim.Redkikh
Elementov, Akad. Nauk SSSR, Inst. Obschei Neorg. Khim. 2: 21-36, 1955
76) Vanhees, J. et al, "Standard Reduction Potential of the Indium-Indium(III)
Electrode," J. Phys. Chem. 85: 1713-8, 1981.
77) Chao, Li-Chung and R. D. Rieke, "Preparation and Reactions of Highly reactive
Indium Metal," J. Organomet. Chem. 67: C64-6, 1974.
78) Shimizu, Takafumi, "Manufacture of Fine Indium Powder," Jpn. Kokai Tokkyo Koho
JP 63,270,404; 1988.
79) Harris, G.B. and S. Monette, "The Stability of Arsenic-Bearing Residues," Arsenic
Metall. Fundam. Appl, Proc. Symp. pp. 469-88, 1988.
80) Stefanakis, M. and A. Kontopoulos, "Production on Environmentally Acceptable
Arsenites-Arsenates from Solid Arsenic Trioxide," Arsenic Metall. Fundam. Appl.,
Proc. Symp. pp. 287-304, 1988.
81) Takahashi, T. and K. Sasaki, "Systems of ZnO-As2O5-H2O and Mg-As2O5-H2O," J.
Chem. Soc. Japan 56: 843-6, 1953.
82) Ivanov-Emin, B.N. and Ya.I. Rabovik, "The Acid Properties of Gallium Hydroxide,"
J. Gen. Chem. U.S.S.R. 14: 781-5, 1944.
83)Nazarenko, V.A. et al. "Spectrophotometric Determination of the Hydrolysis
Constants of Gallium Ions," Zh. Neorg. Khim. 13(6): 1574-80, 1968
51
------- |