One or more of the Following Statements may affect this Document
This document has been reproduced from the best copy furnished by the
organizational source. It is being released in the interest of making available as
much information as possible.
This document may contain data, which exceeds the sheet parameters. It was
furnished in this condition by the organizational source and is the best copy
This document may contain tone-on-tone or color graphs, charts and/or pictures,
which have been reproduced in black and white.
This document is paginated as submitted by the original source.
Portions of this document are not fully legible due to the historical nature of some
of the material. However, it is the best reproduction available from the original
Produced by the NASA Center for Aerospace Information (CASI)
Distribution Category UC-63
ENERGY PAYBACK TIME FOR SILICON PHOTOVOLTAIC SYSTEMS
REPORT N0. SX/111/10
.JOSEPH LI NDMAYER,?
JPL CONTRACT No, 954606
PAYBACK TIME FOR SILICON PHOTOVOLTAIC
57 p HC A04./MF A01?
1335 PICCARD DRIVE
ROCKVILLE, MD. 20850?
It ^.f., ?^^f
PJAnF1 #r! ! ; y; i-
This work was performed for the Jet Propulsion
Laboratory, California institute of Technology,
sponsored by the Natonal Aeronautics and Space
Administration under Contract NAS7-100.
ENERGY PAYBACK TIME FOR SILICON PHOTOVOLTAIC SYSTEMS
FIRST QUARTERLY REPORT
DECEMBER 20, 1976 To MARCH 20, 1977
This first report was prepared by the following
professional staff members of Solarex Corporation:
Joseph Lindmayer, Project Manager
This report contains information prepared
by Solarex Corporation under a JPL Sub-
contract. Its content is not necessarily
endorsed by the Jet Propulsion Laboratory,
California Institute of Technology, the
National Aeronautics and Space Administration
or the Energy Research and Development Admini-
TABLE OF CONTENTS
Definitions for Energy and Payback Time
Energy Assessment of Prevailing Manufacturing
Reduction of Silicon
Crystal (Growth and Wafering
Summary of Energy Assessment
5.Variations of Parameters
Estimation of Indirect Energy
from Product Price
Reduction to Metallurgical Grade
8.4Exhibit D.?Crystal (Growth and Wafering)
Average insolation in the United States
The primary objective of this program of research is the
characterization of the energy requirements and energy pro-
duction potentials of the "Solar Breeder" concept. This
quarterly report documents the results of a careful and exten-
sive study undertaken at the SOLAREX CORPORATION to assess
the energy expenditures of the prevailing manufacturing
technology of terrestrial photovoltaic cells and panels.
The five major production processes in the current technology
are: silicon reduction, silicon refinement, crystal growth,
cell processing and panel building.
One of the most important results of this study is the
fact that the energy payback time for a typical solar panel
produced by the prevailing technology is only 6.4 years.
Furthermore, this value drops to 3.8 years under more favorable
conditions. Since energy payback times as high as 40 years
have been estimated for space cells, this relatively short
payback time reflects the rapid progress made in terrestrial
photovoltaic manufacturing. Moreover, since the major energy
use reductions in terrestrial manufacturing have occurred in
cell processing, this payback time directly illustrates the
areas where major future energy reductions can be made -- silicon
refinement, crystal growth, and panel building.
The comprehensive research approach used in this study
includes the examination in detail of the major production
process and sPq-i?onces of the current technology and the assess-
ment of each of the steps of energy expenditures. Energy
expenditures include direct energy,indirect energy and energy in
the form of equipment and overhead expenses. Payback times were
developed using a conventional solar cell as a "tent" vehicle
which allows for the comparison of its energy generating capability
with the energies expended during the production process. Finally,
the Solar Breeder is described from a systems viewpoint and the
significance of payback time and panel lifetime as important
systems parameters are pointed out.
A comprehensive evaluation of any energy source requires
an assessment of that source's net energy contribution to
In the past,photovoltaics,?
energy source, has not been properly appraised for its
potential in terms of net energy.?
an important alternate
ERDA's National Photo-
voltaic Development Program, which calls for photovoltaic
panels with a twenty-year lifetime, will also require the
use of technologies that allow for energy recovery (or payback
time)? in a fraction of those twenty years.? Early assessments
of photovoltaic energy payback time were based on examina-
tions of space solar cell production; a business that is?
very periodic and extremely inefficient in terms of net
Analysis has shown that the payback time for space
solar cells manufactured under space-demand controlled
situations may be in the neighborhood of forty years. l? With
the advent of terrestrial production, and the introduction
of new technologies and business practices in 1973, the
energy payback time began to decline sharply.? The amount
of change, however, was not quantized to determine what the
actual situation was.?
of a study directed to the examination of the energy pay-
This is the first quarterly report
back time as it evolves in terrestrial manufacturing.
This first report documents that the energy payback
time of terrestrial photovoltaic manufacturing has declined
substantially below that of space production.?
is that in the early part of 1977 the energy payback time
for terrestrial production is less than 6-1/2 years.?
is expected that the energy payback time will continue to drop
as new technologies and improved production techniques are
It is actually expected that the payback time
will ultimately drop below one year.?
general overview of the energy payback time as a function of
to less than one
g? year recovery by the
Figure 1 provides a
The potentially low energy payback time of photovoltaic
systems coupled with the expected long-life of solar panels
allow for the development of nearly energy-independent photo-
voltaic manufacturing plants. These considerations lead
to the Solar Breeder principle which was first proposed by
J. Lindmayer in testimony before the Senate Committee on
Finance in January 1974. 2 It was pointed out that a photo -
voltaic panel manufacturing plant can be made energy independent
by using energy derived from its own roof using its own panels.
Such a plant becomes not only energy self-sufficient but a
major supplier of new energy, hence the name Solar Breeder.
The second part of this report deals with the basic principles
of a Solar Breeder system in which context the energy payback
time can be re-named as "breeding time." This report will
establish certain mathematical relationships for the Solar
Breeder clearly indicating that a vast amount of net energy
is available from such a plant for the indefinite future.
It should be pointed out that if solar electric plants would
be built based on the Solar Breeder principle, their opera-
tion as a net energy source would be automatically assured.
That is to say, a Solar Breeder facility would not be able
to function if it is unable to produce net energy thereby
introducing an automatic safeguard to assure that it is serving
the needs of society.
This first report assesses the net energy situation for
the early part of 1977. Latex reports will appraise the
modifications in general that can be made to reduce the
energy payback time further. Still later reports will deal
with the energy situation related to new photovoltaic
technologies as they will be expected to come on line at?
later times. Later reports are also expected to refine the
operational mode of the Solar Breeder.
This first report documents not only that there have been
major improvements in the energy payback time, but describes
specificalIy where the improvements have been made and points
out weaknesses that need further attention. This view of
the photovoltaic production from an energy domain also provides?
a new perspective for certain policy decisions in photovoltaic?
Definitions for Energy and Payback Time
The energy expended during the production of photo-
voltaic panels was broklen up into three categories:
Direct Energy . - This quantity is defined
as the amount of energy expended during
the actual production of the cells and
panels; typically involving electrical
Indirect Energy.- This component contains the
energy expended to make raw materials available
for solar panel production. Under this head-
ing we include also major energies expended
in the mining and transportation process of
raw materials as well as their possible
Equipment and Overhead Energy.- The equip-
ment energy is defined as the energy expended
in the manufacture of the production equip-
ment itself. Overhead energy is defined
as the energy expended in lighting, heating
and airconditioning of the manufacturing?
The study will show that the largest energy component
arises from direct energy. However, the indirect energy is
also very significant. The equipment €r overhead energy is
usually the smallest. The determination of indirect and
equipment energies is not always a simple matter because
detailed analyses could ultimately lead to the question of
how much energy was used to create the world.: In order to
cut- off such side roads we have frequently used the price
The prevailing sequence used in the present-day manu-
facture is depicted in Figure II, introducing five basic
operations. The prevailing processes within those operations
are relatively well established. Those are;
Reduction. - In the conventional process quartzite
pebbles are being reduced to metallurgical grade (MG)
silicon by means of carbon-containing agents all in
electric arc furnaces.
Refinement. - Conversion of (MG) silicon to high
purity by means of trichlorosilane gas and sub-
sequent silicon deposition of silicon in poly-
crystalline form. (Semiconductor grade, SeG).
Crystal. - This involves the processing of SeG
silicon into single crystal ingots (usually CZ)
and subsequent slicing of the ingots into wafers.
This consists of the processing
of blank silicon wafers into a finished solar cell.
Panel Building. - A process in which individual cells are
inner-connected and encapsulated to form modules and
The energy payback time will be calculated with the
a. Flat (non-concentrated) panel
b. Panel in fixed position facing true south
at 45° angle.
C. Panel experiences the average U.S. insolation
After the basic payback time has been calculated a section
of this report will elaborate on the potential effects of
FIG. II. PANEL PRODUCTION SEQUENCE AND ENERGIES
increased efficiency, tracking, concentration, pac
density, and geographical location. From this poi
will proceed to estimate the energy payback times
five basic operations and three energy components.
4. Energy Assessment of Prevailing Manufacturing
4.1 Test vehicle
This section will assess the five basic operations for
their energy expenditure in terms of direct, indirect, and
equipment plus overhead energies. As a test vehicle we will
frequently use a 4" diameter solar cell as representative of
the state-of-the-art. The following table lists the basic
characteristics of a 4" cell:
10.16 cm (411)
0.25 mm (0.01011)
4.72 g @ density of 2.33g/cm'
Lifetime of panel?
time per day?
Energy delivered in?
20 years (31,630h)?32kWh?
The energy delivered by such a cell can be readily calcu-
lated for the average U.S. insolation. As for the lifetime
we assume 20 years; however, this is not meant to imply that
the cell has only this limited life. At the present time it
is believed that the life of solar panels is controlled by
the packaging materials in conjunction with the environment.
In setting a 20 year life it becomes possible to express the
energy collected per weight of silicon at the average U.S. location:'
energy delivered per
kg silicon in 20 }Tears = 6,678 kWh
at 100% yield
Since production yields cannot be regarded as 100%,
the following calculations will employ an overall yield of
50% of silicon usage. This means that certain conservation
measures are taken, such as the silicon remaining after CZ
growth is being reused and that the sawing operation is
better than 50% efficient. In addition, it is estimated that
the silicon material yield in. cell production is approximately
90% as a certain portion of reject cells can be reprocessed
and the silicon thereby reclaimed. (This reclamation is not
energy intensive). While we recognize that such yields may
vary depending on individual company practices, we find it
convenient and reasonable to operate with an overall 50%
yield for silicon. (Deviations from this 50% yield can always
be accomodated by a simple scaling factor of the payback times
as will be apparent later). Accordingly, at 50% yield the
energy delivered for one year is:
energy delivered per
kg silicon in 1 year?
at 50% cell yield
4.2 Reduction of Silicon
Although silicon is very abundant, on the earth's crust,
it cannot be found in elemental form. Silicon manufacturing
processes, therefore, must resort to compounds such as the
oxide as the starting material. Because of high purity and
general availability quartz pebbles became the dominant choice
as the starting material for metallurgical grade silicon. The
reduction of the oxide is carriedout in huge electrode arc
furnaces by means of carbon containing agents at high tempera-
tures. Metered amounts of quartz pebbles, coal, coke and wood
chips are loaded into the furnace crucible which may be as
large as 8 meter in diameter and 3.5 meter high.?
supplied to the mix by application of electric power to sub-
merged graphite electrodes,?
silica passes through several modifications until it finally
melts at temperatures in excess of 17000 C and reacts with
the carbon containing additives.?
the mix may reach temperatures up to 3000 0 C, forming elemental
silicon which accumulates at the bottom of the furnace crucible.
The oxygen from the oxide combines with the carbon to form
carbon monoxide gas.?
The molten silicon can be withdrawn
from the bottom of the furnace through a taphole either
As the temperature increases
In the reduction process
continuously or in regular intervals.?
oxygen or an oxygen chlorine mixture to reduce the content of
It may be blown with
metals such as aluminum, calcium or magnesium.?
grade silicon as it is called after the reduction process
attains thus a purity as high as 99.5.
Most of this metallurgical grade silicon is used in the
steel and aluminum industry and in the chemical industry.
Approximately 10 of metallurgical grade silicon is refined
in a separate operation and channeled into the semiconductor
and solar cell industry.
The reduction of silicon as described has been practiced
for many years by the heavy industry and is therefore a highly
developed process which is largely independent of the needs
and demands of the semiconductor industry. Because silicon
is produced in a quantity of approximately 12 tons per day
in a typical plant the necessity to have inexpensive energy
available-for. that process had been recognized by the industry.'
Traditionally these plants have been strategically located in
the vicinity of power stations so that the costs for transmitting
energy over longer distances could be largel y eliminated.
The amount of energy required to make a kg of metallurgical
grade silicon is relatively small. Direct energy is expended
in the form of electrical power to the graphite electrodes
in the amount of about 15 kWh per kg of MG silicon. A more
detailed energy examination is to be found in Exhibit B.
The indirect energy comprises the mining and transportation
efforts expended in the procurement of the raw materials and
in the caloric content of the carbon containing agents. The
indirect energy value is about 31 kWh per kg of MG silicon.
Data for invested equipment and overhead energies were
estimated from plant costs. As Exhibit B indicates the equipment
plus overhead energy is quite low, in the order of 1 kWh per
kg of silicon. This results in a negligible payback time, a
very small fraction of a year.
Based on calculations to be found in Exhibit B we can
summarize the energy cost of the reduction process in TableI
The kWh/kg figure'listed represents the energy actually used,
while the energy payback time is related to present-day cell
TableI. energy in Reduction
The payback times of less than 0.1 and 0.2 years for direct
and indirect energy respectively are quite low. Since equip-
ment plus overhead energy is negligible, a total payback
time of less than 0.3 years for a present-day cell is most
agreeable. Thus, from the standpoint of the photovoltaic
industry the current state of the silicon reduction process
is considered satisfactory with resp&ca to its energy balance
and production capability and is not regarded as an obstacle.
The need for ultrapure starting matei-L.als foi--..the device
development in solid state electronics was recognized -^as soon
as the influence of impurities on the electronic conduction
process was understood and controlled doping techniques were
developed. To fulfill this need, a number of alternative
processes for the preparation of high purity silicon have
been investigated by various laboratories throughout the
world. However, it appears that only the chemical vapor
deposition technology whereby a gaseous compound of silicon
is utilized found its way into a larger scale production
operation. Trichlorosilane is the gas that is used world-
wide today. It is formed by the reaction of hydrogen
chloride and MG silicon at a temperature of approximately
The low boiling point of trichlorosilane at 31.80 C
allows a very effective purification of the gas by means of
fractionated distillation. Practically every impurity displays
a relatively low volatility so that even in large scale pro-
duction processes the final content of electrically active
impurities is typically less than one part per billion atoms.
The preparation of semiconductor grade silicon is carried
out now for over two decades by the reduction of ultrapure
trichlorosilane with hydrogen on a resistance heated silicon
substrate at temperatures exceeding 1000 0 C. A silicon rod
of typically 3/4 11 to 111 thickness is heated directly by current
to temperatures of about 1400 0 C in a gaseous atmosphere
containing a mixture of trichlorosilane and hydrogen. Tri-
chlorosilane reduces on the hot rod to pure silicon and hydro-
chloric acid is formed with the hydrogen.
.The development of the silicon refinement technology was
influenced by the demands of the semiconductor device industry.
This industry developed manufacturing procedures whereby
many chips are produced from a single wafer. Because the
amount of silicon used in the chip is small, primary emphasis
is placed on high purity starting material and homogeneous
quality. Questions with regard to cost played a lesser role
and energy was not even considered. In this economical environ-
ment the installations for the production of ultrapure silicon
reached sizes comparable to small oil refineries. Distillation
columns for trichlorosilane are now several stories high and
the reaction chambers for the silicon deposition accomodate
rod lengths of up to 5 feet. In addition, the demand for
cheap electrical power at high consumption rates led to
strategical plant locations in the vicinity of power stations
where reduced electricity rates could be negotiated.
It is now recognized that the traditional refinement
process as described above is not entirely suitable for the
requirements of the photovoltaic industry. The amount of
material used in a simple solar cell is high compared to the
chip and, therefore, the material costs cannot be ignored
and even constitute an obstacle for the development of the
inexpensive cell. In realization of this fact the solar industry
tries to circumvent the cost and energy expended in the silicon
refinement process by orienting its research efforts towards
the development of an inexpensive solar cell made from less
At the present time, however`, the photovoltaic industry
still uses the same silicon as the integrated circuit manu-
facturer. As might be expected, high energy values and payback
times result from this practice. Details may be found in
Exhibit C? The direct energy is quoted to be 440 kWh per kg
of SeG silicon which alone results in a payback time of about
2.6 years. Equipment plus overhead energies may be estimated
from typical capital investments for refinement plants. The
cost burden per kg of SeG silicon is approximately $11.50
which points to an equipment plus overhead energy of 77 kWh
and a payback time of 0.46 years. The indirect energy is
mainly contained in chemicals used during refinement. Be-
cause the same chemical reaction is passed through in the
forward and reverse direction, namely the formation of pure
trichlorosilane from MG silicon and the subsequent reduction
of the silane back to silicon, little of the material is
expended in the overall process which cannot be recovered.
It is estimated that the indirect energy is of the order of
5% of the direct energy or approximately 22 kWh per kg SeG
silicon resulting in a payback time of 0.13 years. We have
listed the various energies of the refinement process in Table 2
Table 2. Energy in Refinement
Again, we have to emphasize that this energy component
is quite large and will require continuous attention.
4.4 Crystal (Growth and Wafering)
The discovery of the transistor effect approximately thirty
years ago marked the beginning of the semiconductor device tech-
nology based on the single-crystalline state. This state gained
predominance in solid state electronics not only because the
crystalline state could be treated with mathematical rigor but
also because of the early observations that electronic events
were more controlled when the crystallinity was high. In addi-
tion, despite the high symmetry which semiconductors commonly
exhibit, a prominent degree of anisotropy of certain physical
phenomena remained which is exploited in the device technology.
It is therefore not surprising that the device industry requested
single-crystalline wafers already at the time of its infancy
and increased its demand for larger wafers of highest quality
with respect to crystallinity and low dislocation density as
transistors and microcircuits were developed. Because many
chips could be manufactured from a single wafer cost was of
secondary nature and energy considerations nonexistent. Under
these circumstances crystal growth industries encountered a
highly beneficial environment for research on growth methods
and subsequent economical expansion to today's multimillion
Although germanium was the material of early semiconductor
research it was soon replaced by silicon due to its more
advantageous properties. Most growth methods are aimed at pro-
ducing silicon in the single-crystalline form. Of the many
methods developed the Czochralski pulling process gained
worldwide industrial importance although in some instances
crystals obtained by the typically more expensive float zone
technique are preferred,
The Czochralski pulling process starts with a small silicon
seed crystal of predetermined orientation which is lowered into
a molten-silicon-containing crucible until it touches the melt -
surface. As the seed crystal is subsequently pulled from the
liquid surface under -Notational motion silicon from the melt
crystallizes above the solid-liquid interface maintaining the
crystallographic orientation of the seed. After approximately
14 hours of operation a single crystal of up to 4" diameter
and weighing 15-20 kg is obtained. The crystal is subsequently
sliced into thin wafers which are then used by the semiconductor
industry as the starting material for its devices.
The photovoltaic industry is still using the same silicon
wafers in large amounts for the manufacture of individual cells.
However, present wafer costs impede the development of the in-
expensive cell. While solar cells can be made on less orderly
and pure silicon, the present solar industry can only be supplied
from the established CZ technology. In fact, no other form of
silicon is available in quantity in early 1977.
The growth and slicing procedures are now so well established
that their energy requirements can be easily evaluated. Direct
energy is consumed in the pulling process and in the sawing
operation at a combined rate of 42 kWh per kg of crystal result-
ing in a payback time of 0.25 years. Indirect energy is expended
in the form of chemicals, replacement parts, crucibles and
blades for sawing at a rate of 102 kWh leading to a payback
time of 0.61 years. Equipment plus overhead energy is
primarily contained in the cost of Czochralski crystal pulling
equipment at approximately 15 kWh resulting in a payback time
of less than 0.1 years. We summarize these data in Table 3
and refer to Exhibit D for details of the energy analysis.
Table 3. Energy in Crystal
We should emphasize again, that CZ crystal and sawing are
the prevailing technologies in early 1977.?
that significant changes will not occur in this area.
This does not imply
The cell production process starting with the blank pre-
doped silicon wafer and ending with a finished cell consists
commonly of several manufacturing steps, as listed below:
Surface preparation of the wafer.?
usually an etching process to remove the work
damage caused by the saw and to clean the
The formation of the junction typically by
means of diffusion processes.?
Removal of the back junction which can be
done by etching, or alloying an opposite
Formation of the back contacts which is
usually done by evaporation techniques.
Formation of the front contacts.?
typically done by evaporation through a
shadow mask or by application of photo-
Sintering to enforce contact adhesion
Edge clean to eliminate junction
shorting. It is conventionally done
by an etching process.
AR coating of the front surface to reduce
Cell testing and quality control
We have exami 201each manufacturing step with respect to
its direct and ind ect energy expenditure and listed what we
believe are typical ` dustrial values in Exhibit E?
accounted for a direct energy value of 0.42 kWh per standard
cell resulting in 0.26 years of payback time and an indirect
energy value of 0.70 kWh/"test" cell pointing towards a pay-
back time of 0.44 years. Equipment plus overhead energies
were derived from an estimate of the replacement cost of
actual production equipment and from energy expended for
heating, cooling and lighting of the production area. The
combined energy value is approximately 0.08 kWh leading to a
payback time of 0,05 years.The data are summarized in Table
Table 4.Energy in Cell Processing
Payback time in —]
years as of 1977
The important conclusion that results from this -analysis is
the fact that the cell making process is not energy expensive.
The criticism that a predominant amount of energy is tied up
in the cell making and which is still prevailing originated at
times when cells were ?Wade solely for space applications and
were indeed very energy intensive.?
However, major technological
advances have been made in the last few years which changed
the situation dramatically.?
For example, the diffusion process
was always believed to require unusual amounts of electrical
As we show in Exhibit E?
the whole diffusion process
requires only about 0.16 kWh per cell burdening the payback?
In addition, we believe that the fulltime only by?
potential of the diffusion process has not yet been completely
utilized in a production environment.?
The same is true for
other processes in the cell manufacture.?
Part of the data
listed in Exhibit E is the result of our directly monitoring
energy inputs to the Solarex production process and therefore
represent factual energy figures.?
It is expected that the
energy balance of the cell making process will further improve,
however, at the present time a large part of the blame for
high energy expenditures of the overall panel production
process rests mostly with silicon refinement and crystal growth.
The apparent fact is that cell and panel processing has gone
through many changes in the last three years, resulting not only
in lesser cost, but also in great reduction of energy use.
The individual cell is well equipped to fulfill its
energy delivering task but major power can only be derived
from the formation of many cells into the solar panel.?The
backbone of the panel consists typically of a sheet of plastic
or metal which is strong enough to provide structural support.
Individual cells are arranged on this board in a geometric
fashion with efficient area utilization and electrically
surface covering the cells and, after curing,P rotecting them
from future environmental impact.
Silicone rubber is then poured over the whole
The direct energy expended in the manufacture of a typical
panel is approximately .09 kWh per standard cell equivalent
to our test vehicle yielding a payback time of less than 0.1
years as shown in Exhibit F.?
The indirect energy expended
in the form of panel hardware and encapsulant amounts to
approximately 1.67 kWh per cell which may be converted into a
payback time of 1.0 years. Equipment plus overhead energy
are typically 0.17 kWh per cell pointing towards a payback
time of 0.11 years. As can be seen from these data which are
listed in Table 5.
Table S. Energy in Panel Building
Payback Times in
Years as of 1977
Direct energy? 0.09?
panel building requires little electrical energy which is reflected
in the low payback times of direct and equipment plus overhead
energy respectively. The relatively high payback time of the
indirect energy is due to the calculated energy content in
materials used to make the panel, although the total cost of
these materials is a fraction of a dollar.
Summary of Energy As_ses'sment
The overall payback time of our test vehicle is the sum
of the individual payback times as derived in the preceding
sections. In order to visualize their significance they are
shown in Figure III in the form of a vertical bar pattern
and in an accumulating fashion along the panel building train.
REDUCTION REFINEMENT CRYSTAL?
FIG. III. PAYBACK TIME VS. TI-fE
5.? Variations of Parameters
The appraisal of the energy payback time as documented
in the preceding sections deals primarily with the details
of the prevailing manufacturing process. In order to arrive
at a quantitative value of the payback time we based our
calculations on a well defined cell as a test vehicle and
assumed certain operational conditions of the final panel.
These assumptions were basically as follows:
cell efficiency is 12.5%
insolation per day is 4.33 sun hours
cell thickeness is 10 mil
the packaging factor of the cells in the panel
is about 70%
the flat panel is in a fixed position facint
true south at 45 0 angle and concentration is not employed
Based on these assumptions we derived a payback time of 6.4
years. However, it is clear that this value can change as the
above assumptions are allowed to vary. The payback time then
becomes not only a function of the details of the manufacturing
process but depends also on conditions surrounding the panel
As can be seen in Exhibit G the da 4 ly average insolation
in the United States varies with the location and can be as
high as 6 sun hours. In addition, the 12.5% efficiency value
of our test vehicle may rise as high as 15% under certain
manufacturing conditions. If only these two new data are
introduced into the former analysis the payback time would
reduce to 3.8 years.
Further improveirents with respect to shorter payback times
will be introduced. when a higher utilization of silicon in
the form of thinner wafers becomes standard practice. In addition,
the circular shape of the wafers limits area utilization in
the-panel. When rectangular cells find their way into the
production process, a significant saving in indirect panel
energy will occur. Of course, concentration or tracking
would also reduce the energy payback time.
In summary, the practices prevailing today project
6.4 years of energy recovery or as low as 3.8 years under
Appraisal of the energy payback time for the early part
of 1977 reveals some interesting facts.?
Most of all,?
clearly demonstrates that those areas of production technology
that were heavily cultivated by the terrestrial photovoltaics
manufacturexers are the areas where tremendous reductions in
energy consumption have occurred.?
This is surely the case
for solar cell processing; the terrestrial production that
began in 1973 was extremely sensitive to the energy question
and abruptly reduced its energy consumption when compared
to the frequently quoted 20-30 year energy recovery in space
At the same time, however, not much energy
reduction occurred in the basic silicon production asit
remained essentially the same as for the semiconductor industry.
The solar cell manufacturers have not practiced the production
of. inexpensive silicon for their own use; accordingly, they
inherited a high energy process from the semiconductor industry.
Therefore, we believe that the key to future reductions in
energy consumption (and cost reduction) is the extension of
their sensitivity for low energy /low cost processing techniques
to all aspects of production technology.
The basic processes today can be broken up into three
Production-of silicon, sheet, cell production
In an overview they rank as follows:
practiced today is inherited from the semi -
The inexpensive and low
The whole technology as
energy demanding reduction into MG silicon
is completely distorted by the high energy
cost of the refinement. The trichloro-
silane process is basically unacceptable not
only for its high price, but also for its high
energy content. Even if every gram of silicon
can be utilized, it requires an energy
payback time of 1.6 years. Clearly, new?
approaches in refinement must be introduced
and practiced in terrestrial production,
Cell Production. With the introduction of
terrestrial production, major changes occurred
in this area. The payback time has been re-
duced dramatically when compared to the previous
space cell production. We can see an order of
magnitude reduction (at least at this company)
in this area; so much so that the indirect
material energy now exceeds the direct energy
used in cell production. Cell production is
clearly in good shape as far as energy is con-
cerned and further improvements will occur.
Panel Building. Terrestrial panel building is
a new activity and is distinctly separate from
space panel building. It is now becoming clear
that certain terrestrial environments are more
hostile than the space environment and, therefore,
this area will experience continued changes.
The present study indicates that the packaging
materials contribute a significant amount of
indirect energy, even though the direct
energy is small. The energy requirement of
the packaging materials is a new parameter
to consider in the development of high reliability
panels. It should be noted that the present
circular cell requires a disporportionate amount
of packaging material and energy; therefore,
close packing is a basic requirement also_
It is expected that as time goes by the technology will
change, resulting in a continuously
We propose that new technologies must be appraised for their
energy content, particularly in the silicon refinement and
decreasing payback time.
The Solar Breeder is an energy self-sufficient plant pro-
ducing new energy in the form of solar electric panels that
are available for external use. This section discusses some
general relationships for such a Solar Breee.3r. Clearly, the
critical parameters that play important roles in the Solar Breeder
operation are the lifetime of the panel, TL , and the energy pay-
back time which in the present context can be called the T B or
the breeding time.
For convenience, the basic process is represented by five
operations as shown in Figure IV. For simplicity, the energy
is subdivided into two main parts; the direct and indirect
components. While in the previous text a term "overhead energy"
is also used, this component is relatively small and can be
included into the indirect energy component. The breeder plant
is operated as a vertically integrated unit where all operations
are based on direct electrical energy. It will be clear that
the lifetime of the panels, T L , must exceed the breeding time,
T B , in order to produce net energy to society.
When the operation of the breeder begins it must borrow
energy from conventional sources. The power flow can then
be formulated as
where Putility represents the direct use of energy while At
has the dimension of power representing all indirect energies.
Based on a constant production rate the Solar Breeder will
reduce its external power needs linearly in time as its own
power supply increases. Accordingly, we may write (for a closed
feedback loop in Figure IV)
AE indirect - Rt?
where R is the rate of production in units of power per time.
Pu z1 ty?
Power flow from energy in materials, AE ndirect/At
Fig.IV. Schematic Diagram for Solar Breeder.
and when power in
= T + Ti?
3ependence is achieved T will be zero and therefore
= T + Ti?
Self-sufficiency will be reached when Pexternal reaches zero.
At that time t =TB where TB is the breeding time. To a first
approximation, it may be said that the utility and indirect
power components are proportional to the rate of production R.
If all quantities are redefined as power/unit production rate and
denoted by T (having time units) we find from (2)
In other words, the breeding time is determined by
= Putility ^t =?
electrical energy used
solar watts produced
solar watts pio uced
The relationship.s . in (5)?
breeding time as
allow us to write the
direct energy + indirect energy
solar watts produced
At the average U.S. location, the peak sun power is
available for 4.33 hours/day.? In 7 days,?
30.31 hours are
available; assuming an 80% interface efficiency, we can
calculate with 24.2 sun hours/week.?
A 40 hour work-week
yields 0.6 W (peak) solar power, which factor allows us to
express the breeding time in W (peak) units:
TB? 1.65 direct energ?
+ indirect energy
It is apparent that in order to have a reasonably short
breeding time, such as a few years, not more than a few kWh
energy can be invested in producing one W(peak) soar power.
The breeding time is the length of time required to
reach self-sufficiency when the photovoltaic production plant
is put into the breeder or "closed feedback loop" mode of
operation. The meaning of breeding time remains the same
even if the manufacturing plant is allowed to operate in the
"open loop" mode of operation; i.e., when it does not retain
any of its produced panels, but delivers them to users out-
side of the plant. Accordingly, the breeding time is a
basic operational parameter of any photovoltaic production
The importance of the breeding time assumes even more
serious dimensions when the Question of net energy is raised.
It is obvious that a new energy source should be developed
only if it has the potential of becoming a net energy source.
Since solar energy is continuously derived from previously
produced panels, their lifetime is of great importance.
Denoting the lifetime of a panel by T L , let us examine
This plot shows power as a function of time..
On the time scale, two important points are denoted, TB
for the breeding time and, TL , for the panel lifetime.
For a constant production rate the power needed by the
facility is constant, while the solar power increases
linearly as panels are produced at a constant rate, R.
The following calculations can be derived from Figure v if
we write the solar power,p s, supplied by the panels as:
NET GAIN IN
•.^'? Power needed for operation
T I ME --^-
,Fig.. V. Power and Energy for Calculating Energy Gain.
The net energy supplied by the facility is represented
by the area shown in Figure V, which excludes the fossil fuel
investment and all energies required to operate the facility.
The net energy in the first cycle of operation is:
R? 2 _ T 2
Enet -f 'fit dt - Efossil?
The invested fossil energy is readily available from
Figure V also
Rt dt =?
The net energy gain in the first breeder cycle defined
as TL is
Enet? _ TL?
sho B ( c
In order to be in the net energy mode it is required that
TL > N/2 TB?
Equation 12 showsa delightful conclusion: The Solar Breeder
in its first cycle of operation will reproduce the invested
energy many times. For example, if the panel life is 20 years`
and the breeding time is one year, the solar breeder
will produce in its first cycle 400 times the energy invested!
In its second cycle of operation, it must reproduce its own
panels, but in the above example, it will take only a small
fraction of its energy output to do this. In other words,
the net energy gain is already very large in the first cycle
of operation and `even beyond that the energy produced grows
quadratically, ad infinitum. Clearly, the solar breeder is
a tremendous source of new energy.
The solar breeder approach is a method in a new type
of system analysis in evaluating photovoltaic production
plants. The breeder employs a feedback principle to test
the inherent characteristics of tare facility. In the open
loop situation, it simply remains a production facility
to the outside world, but its inherent characteristics
Let us also examine briefly the intermediate situation,
namely when the loop is only partially closed. If the solar
breeder is allowed to be the only panel user, it will go
into the self-supporting state in TB time and then abruptly
appears on the market with its excess capacity. This is
shown in Figure VI. -If, on the other hand, the facility is
allowed to sell a portion of its products within TB , self-
sufficiency is delayed beyond TB , but its products can appear
on the market more gradually. This is also shown in Figure
VI. While none of the basic system parameters change, the
solar br e eder operational mode offers a new degree of freedom
in planning market entry. Its mode of operation actually
allows proportioning external and internal use of panels at
will. The only restriction is that after self-sufficiency
all but a small fraction of its produced panels must go to
with abrupt market with gradual market
Fig.B1. Abrupt and Gradual Market Entry of Solar Breeder
There are clearly further sophistications and generalities
in the solar breeder. We will pursue this new type of system
analysis and expand on its scope and dimensions. It may well
be that large scale expansion of manufacturing should indeed
be based on the solar breeder approach which will assure
delivery of net energy to society.
8.1 EXHIBIT A - Estimation of Indirect Energy from Product Price
The determination of indirect and equipment energies is
not always a simple matter because detailed analyses lead in
too many directions in the search for expended energy. In
order to cut off such side roads we have frequently used
the purchase price of a product for guidance of its energy
content. We base the validity of this procedure on the
results of a research document 3) which reports that on the
average 2% of the purchase price of items such as equipment
or materials reflect the cost of energy expended in the
manufacturing of the item.
Equipped with this assumption we could determine the cost
of the expended energy in equipment and materials but not the
energy value itself. The missing conversion factor-of energy
vs. price was taken from a recently published study 4) where
it is pointed out that the composite price per million BTU
.s $0.879. The word composite means that the quoted price
is composed of the prices of various energy sources weighted
by the relative importance of the individual source. In
more practical terms the average cost for one kWh is thus
Based on these two assumptions it is now possible to
derive an energy value from the equipment or material purchase
price at a rate of 6.67 kWh per price dollar. We have adopted
this procedure frequently except in cases where we were aware
that this simple formula does not apply. For example, the
price for photoresist is based largely on initial research
costs, quality control and "on the fact that practically
only one manufacturer has succeeded in making it" as we
were informed. In ^gases like this we derived energy from
approximately one-third of t
1..,e purchase price.
8.2 EXHIBIT B - Reduction to Metallur gical Grade Silicon
The manufacture of metallurgical grade (MG) silicon
is carried out on a large scale by the reduction.of
quartzite with carbon-containing agents. The process
occurs in huge electrode arc furnaces at high tempera-
tures according to the overall equation
Sio 2 (s ) + 2C (s ) -4Si (1) + 2CO (g)
Metered amounts of quartz pebbles, coal, coke and wood chips
are loaded into the furnace crucible which may be as large
as 8 meter in diameter and 3.5 meter high. Heat is supplied
to the mix by application of electric power to submerged
graphite electrodes. As the temperature increases silica
passes through several modifications until it finally melts
at temperatures in excess of 1700 0 C. and reacts with the carbon-
containing additives. In the reduction process the mix may
reach temperatures as high as 30000 C, forming elemental
silicon which accumulates at the bottom of the furnace
crucible. The molten silicon can be withdrawn from the
bottom of the furnace through a taphole either continuously
or in regular intervals. Metallurgical grade silicon, as it is
called after the reduction process, attains thus a purity
as high as 99.5%.
The yearly production of MG silicon in the United States
has now exceeded 140,000 short tons5) Most of it is used
in the steel, aluminum and chemical industry. Approximately
1% of MG silicon is refined in a subsequent operation and
channeled into the semiconductor and solar cell industry.
Direct energy is supplied to the smelting process in
the form of electric power to the graphite arc electrodes
The electric energy consumption6) per . gross ton is 13,952
kWh or 15.4 kWh per kg MG-silicon resulting in a payback time
of 0.09 years..
Indirect energy is consumed in the form of miring efforts
and rail transportation of the raw materials and in the form
of the caloric content of some of the raw materials themselves.
The amount of raw materials which constitute a typical mixture
to yield I kg of MG silicon may be listed as follows.
caloric energy content
pet. coke briquettes
raw petr. coke
The caloric energy content of the carbon-containing
raw materials has been calculated using the following conversion
wood chips? 4000 kcal/kg
Thus, the combined caloric energy content expended in the
carbon-containing raw materials is
19281 kcal which is equiva-
lent to 22.4 kWh.
Additional energy is consumed in the mining,production
and transportation process of the raw materials. According
to a study by the Battelle Columbus Laboratories 8) these
energies have been determined as follows:
106 BTU per ton of item
silica pebbles, mining?
rail transportation (300 miles)?
rail transportation (300 miles)?
rail transportation (300 miles)?
wood chips, sawing and chipping?
truck transportation (50 miles)?
Using the conversion factors of 907.2 kg/ton and 3410
BTU/kWh the indirect energy content in the raw materials for
the smelting process aside from their caloric value is
The combined indirect energy content in the raw materials
for the smelting process is thus the sum of 22.4 kWh and 9.0
kWh, i.e., 31.4 kWh which results in a payback time of 0.19
Data for invested equipment and overhead energy cannot be
readily found in the literature. However, an announcement of
the National Metallurgical Corporation 5) to expand the
production capability of one of their plants from 4,500
tons to 13,000 tons annually at a cost of $5.5 Million
allows us to estimate these energies. If we assume a
return of their investment in 10 years during which time
approximately 85,000 additional-tons of MG-silicon are
produced, the cost per kg silicon is $0.071. If we
further make the assumption that 50 of the invested cost
constitutes an energy cost (which is high) and that this
cost is converged into energy units at a rate of $0.003/kWh
the invested energy per kg MG-silicon is 1.18 kWh. The
payback time for this energy amount is of the order of
7.0x10 - years.
8.3 EXHIBIT C? Refinement
Up to the present time the preparation of semiconductor
grade (SeG) silicon appears to be impossible without resort-i
ing to ultrapure gaseous silicon compounds from which the high
purity silicon can be reclaimed. Amongst the many silanes
which could be used for that purpose trichlorosilane is
preferred worldwide because it can be employed at lower
temperatures and faster rates. It is formed in high yields
by the interaction of MG silicon powder and hydrochloric
acid at a temperature of 300 0 C. The exothermic process
occurs in a fluidized bet reactor according to the chemical
Si(s) + 3HC1 (g) —; SiHC1 3( g) + H2 (g)(1)
To obtain the desired purity trichlorosilane must be separated
from metal chlorides and other silanes such as SiC1 4' Tri-
chlorosilane has a low boiling point of 31.8 0 C which allows
a very effective purification by means of fractionated distilla-
tion due to the fact that all other byproducts display low
Ultrapure silicon is obtained from the purified tri-
chlorosilane via chemical vapor deposition, whereby trichlorosilane
reduces in the presence of hydrogen to silicon. Simply speaking,
the chemical reaction is the reverse of the fluidized bed
reaction of Eq. (1). The reduction occurs at temperatures
exceeding 10000 C on a resistance heated starting rod
(poly-rod) made from silicon having a purity comparable to
the deposit. Due to demands for large wafer sizes polyrods
now reach diameters of 4 inches and more during reaction
times on the order of a hundred hours
The production of trichlorosilane requires relatively little
energy due to the exothermic nature of the fluidized bed
reaction. However,.direct energy is required in the dis-
tillation process for the purification of the gas. The
value quoted 9) is 40 kWh per kg of SeG silicon. The domi-
nant part of direct energy used in the refinement process
is expended in the silicon deposition process which occurs on
the current heated starting rod. 400 kWh per SeG silicon 9)
is consumed in this process so that the total direct energy
expended in refinement reaches 440 kWh per kg SeG silicon
resulting in a payback time of 2.63 years.
The indirect energy is small compared to the direct energy
expended. Most of the indirect energy is contained in hydro-
chloric acid and hydrogen gas. However, because the same chemical
reaction is passed through in the forward and reverse direction
little of the raw materials are actually expended in the whole
process. In order to account for material losses we make the
assumption that the indirect energy is of the order of 50
of the direct energy or 22 kWh resulting in a payback time
of 0.13 year.
E ui ment and Overhead energies were derived from
industrial expansion estimates for the production of SeG
expects to enlarge its production
capability at a cost of $46 million. The typical output
silicon. Dow Corning?
of polysilicon after such an expansion is 200 metric tons
per yeas. Assuming a 10 year lifetime of such an investment
the cost contribution to the price of lkg SeG silicon would be
$11.50 representing an energy expenditure of 76.7 kWh
which is equivalent to a payback time of 0.46 y2ars,.
8.4 EXHIBIT D Crystal (Growth and Wafering)
The prevailing Method for the production of single-
crystalline silicon is based on the Czochralski pulling process
whereby the crystal is drawn from the melt contained in a
quartz crucible. At the start of the process a small seed
crystal of predetermined crystallographic orientation is
lowered onto the melt surface. As the seed is subsequently
pulled from the surface under a rotational motion additional
silicon from the melt crystallizes above the liquid solid
interface whereby the crystallographic orientation of the
seed is maintained. Pulling times of 100 hours or more
result in crystals exceeding 4 inches in diameter and over 30
inches long. The crystals are then sliced into thin wafers
and sold to the semiconductor and photovoltaic industry.
Direct energy requirements for the pulling of a crystal of
15 kg in weight are reported 7,'T0 )to be 610 kWh or 40.7 kWh
per kg SeG silicon. Approximately 7.4 kg of ingots can
be processed in a typical slicing operation yielding 600
wafers in 16 hours. The energy required to power the3/4
HP motor commonly installed in a slicing machine is 8.8 kWh
or 1.2 kWh per kg silicon ingot. The total direct energy
for pulling and wafering is thus 41.9 kWh per kg SeG silicon
resulting in a payback time of 0.25 years.
Indirect energy is contained in materials such as argon,
quartz crucibles, replacement parts, wafering blades and slurry.
The costs of some of the materials have been reported to be:
$1.21 /kg SeG-silicon
quartz crucibles 6.25
12.01 /kg SeG-silicon