Technology and Future of III-V Multi-

Junction Solar Cells

Steven Lansel
School of Electrical and Computer Engineering,
Georgia Institute of Technology, Atlanta, GA
gtg223g@mail.gatech.edu

ECE 6542
Professor Chang
April 21, 2005
Abstract: Multi-junction solar cells created from III-V semiconductor materials exhibit
high efficiencies matched by no other existing photovoltaic technology. Multi-junction
solar cells are composed of 3 layers of material that have different bandgaps. The top
layer has the largest bandgap while the bottom layer has the smallest bandgap. This
design allows less energetic photons to pass through the upper layer(s) and be absorbed
by a lower layer, which increases the overall efficiency of the solar cell. One important
design consideration is that the photocurrent generated in each layer must be the same
since the layers are in series. In addition, the bandgaps of each layer should differ by
approximately equal energies so that the spectrum of incident radiation is most
effectively absorbed. Although multi-junction solar cells are very efficient, they are also
very expensive. Due to their high cost, multi-junction solar cells are primarily used in
systems in outer space and as collector cells where a large amount of sunlight is reflected
onto the cell. The use of multi-junction solar cells made of III-V semiconductor
materials appears to be restricted to limited applications while single crystalline silicon
semiconductors have a wider application due to the lesser cost. This paper will focus on
the present and future design, practical and theoretical efficiency, and applications of III-
V multi-junction solar cells.
 Table of Contents

Introduction 1

Fundamental Solar Cell Concepts 2

Basic Concept of Multijunction Solar Cells 4

Design Considerations 5

Bandgaps 5

Lattice Constant 6

Current Matching 7

Layer Thickness Calculation 8

Current Triple-Junction Solar Cell Performance 10

Future Design Improvements 12

Discussion 13

Conclusions 14

References 16

Appendix A: Layer Thickness Calculation

Introduction

One of the largest challenges mankind will face in the twenty-first century and

beyond is how to supply our increasing need for energy. With the rapid consumption rate

of fossil fuels, we need to consider renewable energy sources such as photovoltaics.

Photovoltaics are a promising technology that directly takes advantage of our planet’s

ultimate source of power, the sun. When exposed to light, solar cells are capable of

producing electricity without any harmful effect to the environment or device, which

means they can generate power for many years while requiring only minimal

maintenance and operational costs. Currently the wide-spread use of photovoltaics over

other energy sources is limited by the relatively high cost and low efficiency of solar

cells.

Multijunction solar cells are a new technology that offers extremely high

efficiencies compared to traditional solar cells made of a single layer of semiconductor

material. Depending on the particular technology, multijunction solar cells are capable of

generating approximately twice as much power under the same conditions as traditional

solar cells made of silicon. Unfortunately, multijunction solar cells are very expensive

and are currently only used in high performance applications such as satellites due to

their cost.

This paper will investigate the design of multijunction solar cells. First,

fundamentals of solar cell operation and performance will be presented. The basic

advantage and operation of multijunction solar cells will be discussed. Then, the paper

will discuss important design issues. Current solar cell design and performance will be

presented. Then, future design improvements will be offered. Finally, the paper will

1
conclude with an argument that the design of multijunction solar cells will improve in the

near future and the technology can become a viable widespread power source.

Fundamental Solar Cell Concepts

Solar cells are semiconductor devices that are designed to generate electric power

when exposed to electromagnetic radiation. The spectrum of light given off by the sun is

shown in Figure 1. The distribution of light in outer space resembles the theoretical

radiation provided by a black body. As the light passes through the atmosphere, some of

the light is absorbed or reflected by gasses such as water vapor and the ozone. For this

reason, the typical distribution of light on the surface of the earth shown in Figure 1 is

different than the distribution of light in space. Engineers must consider the spectrum of

incident light when designing solar cells.

Figure 1. Spectrum of solar radiation in space and on earth [2].

2
 Solar cells consist of one or more p-n junctions. Light enters the semiconductor

material through the n region and generates an electron-hole pair (EHP) in the material

due to the photoelectric effect. The n region is designed to be thin while the depletion

region is thick. If the EHP is generated in the depletion region, the built-in electric field

drifts the electron and hole apart. The result is a current through the device called the

photocurrent. If the EHP is generated in the n or p regions, the electron and hole drift in

random directions and may or may not become part of the photocurrent [1].

The following terms deal with the performance of a solar cell:

• short-circuit current, Jsc: The current of a solar cell when the top and bottom

(negative and positive leads) are connected with a short circuit. This is the

horizontal intercept on the I-V curve shown in Figure 2.

• open-circuit voltage, VOC: The voltage between the top and bottom of a solar

cell. This is the vertical intercept on the I-V curve shown in Figure 2.

• power point: The point on the I-V curve of a solar cell at ( J PP , VPP ) that

generates the maximum amount of power for the device. This is the point that

encloses the most amount of area in the first quadrant when vertical and

horizontal lines are drawn from the point. This represents power since the area is

equivalent to the current times voltage of the cell. The power point is shown in

Figure 2.

• fill factor, FF: A percentage given by Equation 1 that describes how close the I-

V curve of a solar cell resembles a perfect rectangle, which represents the ideal

solar cell.

VPP * J PP
Fill Factor = (Eq. 1)
VOC * J SC

3
 Figure 2. Current versus voltage curve (I-V curve) for a typical solar cell [3].

• quantum efficiency: The number of EHPs that are created and collected divided

by the number of incident photons [1]. This is a percentage since each photon can

produce at most one EHP.

• overall efficiency: The percent of incident electromagnetic radiation that is

converted to electrical power. Often the overall efficiency for a given solar cell

depends on many factors including the temperature and amount of incident

radiation.

Basic Concept of Multijunction Solar Cells

With a traditional single layer solar cell, much of the energy of incident light is

not converted into electricity. If an incident photon has less energy than the bandgap of

the semiconductor material, the photon cannot be absorbed since there is not enough

energy to excite an electron from the conduction band to the valence band. Therefore,

none of the light with less energy than the bandgap is used in the solar cell. If an incident

photon has more energy than the bandgap, the excess energy will be converted into heat

4
since the electron can only absorb the exact amount of energy required to move to the

valence band.

Multijunction solar cells can make better use of the solar spectrum by having

multiple semiconductor layers with different bandgaps. Each layer is made of a different

material, which usually is a III-V semiconductor, and absorbs a different portion of the

spectrum. The top layer has the largest bandgap so that only the most energetic photons

are absorbed in this layer. Less energetic photons must pass through the top layer since

they are not energetic enough to generate EHPs in the material. Each layer going from

the top to the bottom has a smaller bandgap than the previous. Therefore, each layer

absorbs the photons that have energies greater than the bandgap of that layer and less

than the bandgap of the higher layer. The most common form of multi-junction solar cell

consists of three layers, which is called a triple-junction solar cell.

Design Considerations

Bandgaps

Since the bandgaps of the materials used in a multijunction solar cell determine

which layer a photon is absorbed in, the bandgaps determine how much energy can be

obtained from each photon. Ideally the difference between adjacent layers of the solar

cell is approximately constant so that each layer can absorb an equal amount of the

spectrum of incident light shown in Figure 1. Since the amount of excess energy from

light converted to heat is equal to the difference between the photon energy and the

bandgap of the absorbing material, the difference between bandgaps should be made as

small as possible. Also the solar cell should take advantage of as much of the spectrum

5
as possible so the top layer should have a high bandgap and the bottom layer should have

a small bandgap that can absorb as much of the spectrum as possible. Clearly there is a

design tradeoff for a given number of layers of a multijunction solar cell between having

the bandgaps differ by a small amount and have the bandgaps cover a large range of the

spectrum. Triple-junction solar cells currently in production are made of GaInP, GaAs,

and Ge, which have bandgaps of 1.8 eV, 1.4 eV, and 0.7 eV, respectively [2].

Lattice Constant

In monolithic multijunction solar cells, the different semiconductor layers are

grown directly on top of the other layers using the same substrate. As a result of this

method, the lattice constant, which describes the spacing of the molecules of a crystal

structure, must be the same for all of the layers. Research at the Natoinal Renewable

Energy Laboratory (NREL) showed that a lattice mismatch as small as 0.01%

significantly decreases the current produced by the solar cell [4]. The restriction of each

semiconductor material having the same lattice constant significantly decreases the

number of materials that may be used. Figure 3 shows the lattice constant and bandgap

of common semiconductor materials. Lines between different materials represent

semiconductors that are created by combining different amounts of the two materials.

The vertical line passing through Ge represents the materials that are used to create the

current triple-junction cells made of GaInP, GaAs, and Ge.

6
 Figure 3. Lattice constant and bandgap of common semiconductor materials [5].

Current Matching

Since the current flows through a solar cell from the top to the bottom, the layers

of a multijunction solar cell are in series. Therefore, the current passing through each

layer must be the same and the current produced by the solar cell is limited by the layer

that produces the least amount of current. For maximum efficiency, the cell must be

designed so that each layer produces the exact same current. The current is proportional

to the number of photons absorbed in each layer. The two most important factors in

determining the thickness of each layer is the number of photons in the spectrum that the

layer should absorb and the absorption constant of the material. The light intensity

decreases exponentially with penetration depth into a material where the exponential

constant is called the absorption constant [1:221]. A layer with a low absorption constant

must be made thicker since on average a photon must pass through more of the material

7
before it is absorbed. Properly designing the thickness of each semiconductor material

based on these factors will match the current produced by each layer.

Layer Thickness Calculation

After materials are selected with desired bandgaps and lattice constants, the

thickness of each layer must be determined so that each layer will generate the same

photocurrent. In this section, I will go through a rough calculation to determine the

proper thickness of each layer of a GaInP/GaAs/Ge triple-junction solar cell. The

calculation would be much more precise and useful if experimental data was used in the

calculation. Unfortunately, I do not have access to actual experimental data but can use

graphs of experimental data to arrive at a rough calculation that should be somewhat

accurate.

In determining the thickness of each layer, the most important difference between

semiconductor materials is the material’s absorption constant and the number of incident

photons with a given energy. The absorption constant as a function of photon

Figure 4. Absorption coefficient vs wavelength for various 8

semiconductor materials [1].
wavelength for various semiconductors is provided in Figure 4. The absorption constant

as a function of photon energy for (AlxGa1-x)0.52In0.48P is provided in Figure 5. When

x=0, the graph shows the alpha for GaInP. The photocurrent generated by each layer is

approximately proportional to the number of photons absorbed in the layer. I am

assuming that the quantum efficiency of the different materials is made to be

approximately equal, which will be justified later. Therefore, we want to determine the

thickness of each layer so that the number of photons absorbed in each layer is equal,

which is a function of alpha for each material.

In the calculation, I assume that the incident solar radiation is given by Planck’s

Law for blackbody radiation, which is given in Equation 2 where T=6000 K for the sun.

2h ν 3
I(ν ) =
 hν  (Eq. 2)
c 2 * exp  - 1
 kT 

The blackbody assumption is fairly accurate for outer space but is inaccurate on the

earth’s surface due to absorption in the atmosphere as shown in Figure 1. The

Figure 5. Absorption coefficient vs photon energy for

(AlxGa1-x)0.52In0.48P [9].
9
assumption could be used with some accuracy for design of solar cells for use in space

but would prove poor for terrestrial solar cells.

The thickness calculation performed in Maple is provided in Appendix A. After

deriving expressions for the photocurrent generated in each layer, the currents were

matched. A reasonable cutoff for the Ge layer was then calculated to be approximately

40 μm. In practice, terrestrial solar cells are designed so that the Ge layer is fixed at 150

μm and the thickness of the other two layers is designed to match the current [8].

Current Triple-Junction Solar Cell Performance

The most powerful multi-junction solar cells in commercial production today are

triple-junction cells made of GaInP, GaAs, and Ge. Only two companies, Emcore and

Spectrolab, are currently producing these highly efficient cells and are licensing

technology from the National Renewable Energy Laboratory (NREL). The first cell

design created by Emcore was the 3J or TJ (triple-junction) cell. Later improvements

resulted in the more efficient ATJ (advanced triple-junction) cell [6]. The I-V curve and

performance parameters for typical 3J and ATJ solar cells are shown in Figure 6. Typical

10
Figure 6. Typical I-V curves for 3J and ATJ solar cells [6].
 Figure 7. Quantum efficiency for each layer of TJ and ATJ cells [7].

efficiencies for these types of cells are 26% and 27.5%, which is far greater than the

typical efficiency of 16% for terrestrial silicon cells [6].

Viewing the quantum efficiency of each layer of the ATJ cells, which is shown in

Figure 7, gives insight into the performance of each layer. Overall the quantum

efficiency is high for the solar cell since it is near 90% for most wavelengths. The

transitions between which layer absorbs a light of a given wavelength is due to the fact

that the wavelength dictates the photon’s energy and which layer can absorb it. There is

a sharp transition between the GaAs and Ge layers; however, the transition between the

InGaP and GaAs layers is gradual and can be improved. The energy of photons that have

wavelengths below approximately 650 nm and pass through the InGaP into the GaAs

layer is not being efficiently captured. Ideally the InGaP layer would absorb all photons

that are above the wavelength corresponding to the bandgap of GaAs. It can be seen that

11
the germanium layer absorbs a larger portion of the spectrum then the other two layers,

which is a result of the large difference between the bandgaps of GaAs and Ge.

Future Design Improvements

The current use of GaInP, GaAs, and Ge for the layers of a multi-junction solar cell can

be improved. The germanium layer absorbs too much of the spectrum, since the

difference between the bandgap of the top two layers is 0.4 eV while the difference

between the bottom two layers is 0.7 eV. Using a new semiconductor layer that has a

bandgap of 1.25 eV will make the cell more efficient since the bandgaps will differ by a

constant 0.55 eV. Another possible design improvement keeps the three GaInP, GaAs,

and Ge layers but adds another layer of a material with a bandgap of 1.0 eV. This new

four-junction solar cell will have a difference of 0.3 eV or 0.4 eV between each adjacent

layer. The distribution of the spectrum into which layer absorbs each wavelength for

these proposed solar cell designs is provided in Figure 8. The theoretical and practical

Figure 8. Efficiencies and spectrum division based on layer for

current and future multi-junction solar cells [2].

12
efficiencies for these new technologies under standard terrestrial illumination are shown

to exceed 40% and 32%, respectively, with higher efficiencies possible for increased

illumination [8]. In addition, to the bandgap requirements of a new semiconductor

material, it also must match the lattice constant of germanium in order to be feasible.

Discussion

Multijunction cells made of III-V semiconductor materials offer much higher

efficiencies then single layer cells and especially traditional silicon solar cells. Right now

these cells are primarily used in space applications since they are so expensive. Currently

solar cells under typical illumination levels found on the surface of the earth produce

efficiencies in the high 20’s. With improvements to multijunction solar cell design and

better semiconductor materials, these efficiencies can be increased to the high 30’s in the

near future. Although the production and materials cost will decrease with further

advances and large-scale production, multijunction solar cells will still be far more

expensive than regular silicon solar cells. This is due to the fact that the multijunction

cells are much more complex and use materials that are rarer than silicon. As a result,

multijunction cells must be used differently than other solar cells. Multijunction cells

can be used as viable power sources by building an apparatus that directs a large amount

of incident light onto a small area of solar cells, which is referred to as a collector cell. A

picture of a collector cell apparatus is provided in Figure 9. In this application, the solar

cells must be extremely efficient and the cost is not a consideration since only a small

number of cells are used to generate a large amount of electricity.

13
 Figure 9. Example of a collector cell apparatus [2].

Conclusions

Multijunction solar cells work by layering semiconductor materials that have

different bandgaps. Light enters through the layer that has the largest bandgap.

Depending on the energy of the photon, it penetrates the solar cell until it reaches a layer

that has a smaller bandgap than the photon’s energy. Using this concept, multijunction

solar cells are more efficient than single layer solar cells. This is because less of the

energy of a photon exceeds the bandgap of the absorbing semiconductor, which is energy

lost to heat.

There are a number of factors involved in the design of multijunction solar cells.

The materials used in the cell must all have the same lattice constant. In addition, the

bandgaps of the semiconductors should divide the spectrum of incident light into

approximately equal regions and also span most of the spectrum so that all incident light

14
is absorbed efficiently. The thickness of each layer in the solar cell must be adjusted so

that each layer produces the same amount of current since the layers are in series.

The most efficient solar cells in production today are triple-junction cells made of

GaInP, GaAs, and Ge that achieve typical efficiencies of 27.5%. The current design can

be improved by finding a semiconductor material with a bandgap of approximately 1.0

eV or 1.25 eV that has the same lattice constant as germanium. Although the price of

multijunction solar cells will decrease in the future, they will remain expensive but can

still be used in collector cells for effective power production.

15
 References
[1] S. O. Kasap, Optoelectronics and Photonics: Principles and Practices, New York:
Prentice Hall, 2001.

[2] B. Burnett. (2002). The Basic Physics and Design of III-V Multijunction Solar Cells.
Retrieved April 1, 2005 from http://www.nrel.gov/ncpv/pdfs/11_20_dga_basics_9-
13.pdf

[3] D. R. Carroll, The Winning Solar Car, Warrendale: SAE International, 2003.

[4] R. R. King, R. A. Sherif, C. M. Fetzer, and P. C. Colter, “Advances in High-

Efficiency Multijunction Terrestrial Concentrator Cells and Receivers,” in Proc. Of
NCPV and Solar Program Review Meeting 2003, 2003, pp. 211.

[5] J. M. Román, “State-of-the-art of III-V Solar Cell Fabrication Technologies, Device

Designs and Applications,” Advanced Photovoltaic Cell Design, 2004.

[6] M.A. Stan, D.J. Aiken, P.R. Sharps, N.S. Fatemi, F.A. Spadafora, J. Hills, H. Yoo,
and B. Clevenger, “27.5% Efficiency InGaP/InGaAs/Ge Advanced Triple Junction
(ATJ) Space Solar Cells for High Volume Manufacturing” in Proc. of the 29th IEEE
Photovoltaic Specialists Conference, 2002, pp. 816-819.

[7] M.A. Stan, P.R. Sharps, N.S. Fatemi, F.S. Spadafora, D.J. Aiken, and H.Q. Hou,
“Design and Production of Extremely Radiation-Hard 26% InGaP/GaAs/Ge Triple-
Junction Solar Cells” in Proc. of the 28th IEEE Photovoltaic Specialists Conference,
2000, pp. 1374-1377.

[8] S.R. Kurtz, D. Myers, and J. M. Olson, “Projected Performance of Three- and Four-
Junction Devices Using GaAs and GaInP,” in Proc. of the 26th IEEE Photovoltaic
Specialists Conference, 1997, pp. 875-878.

[9] Y. N. Mohapatra. Epitaxy. Retrieved April 19, 2005 from

http://home.iitk.ac.in/~ynm/People/YNM/Epitaxy.ppt.

16