Photoconductivity and Minority Carrier Lifetime

in Tin Sulfide and Gallium Arsenide Semiconductors for Photovoltaics

by

Frances Daggett Lenahan

Submitted to the
Department of Materials Science and Engineering
in Partial Fulfillment of the Requirements for the Degree of

Bachelor of Science

at the

Massachusetts Institute of Technology

June 2016

© 2016 Frances Lenahan

All rights reserved

The author hereby grants to MIT permission to reproduce and to

distribute publicly and electronic copies of this thesis document in whole or in part
in any medium now known or hereafter created

Signature of Author…………………………………………………………………………………
Department of Materials Science and Engineering
April 29, 2016

Certified by…………………………………………………………………………………………
Tonio Buonassisi
Associate Professor of Mechanical Engineering
Thesis Supervisor

Accepted by……...…………………………………………………………………………………
Geoffrey S. D. Beach
Professor of Materials Science and Engineering
Chairman, Undergraduate Thesis Committee
Read by……………………………………………………………………..………………………
Rafael Jaramillo
Toyota Career Development Assistant Professor of Materials Science & Engineering  
Thesis Reader

  2  
 Photoconductivity and Minority Carrier Lifetime
in Semiconductors for Photovoltaics

by

Frances Daggett Lenahan

Submitted to the Department of Materials Science and Engineering

on April 29th, 2015, in partial fulfillment of the
requirements for the degree of Bachelor of Science in
Materials Science and Engineering

Abstract

The growth and maintenance of the modern technological world requires immediate solutions in
the field of clean, renewable energy. One prominent solution is the rapid advancement of solar
cell technologies due to the wide availability of solar energy and the growing versatility of
harnessing it. As efficiencies for these devices creep upwards, it becomes increasingly more
important to find the greatest inhibiting factor. Through a solar cell simulator program (SCAPS),
improvements in the minority-carrier lifetime of cell materials show not only significant
improvements in cell efficiencies, but also an un-masking of improvements by other properties,
which are inhibited when the lifetime is too short. This work aims to calculate the mobility-
lifetime products (µτ) of gallium arsenide (GaAs) and annealed and un-annealed tin sulfide (SnS)
with respective p-doping carrier concentrations of 1018 cm-3, 1016 cm-3, and 1015 cm-3 through
photoconductivity measurements. Films are 1 µm thick and have a four-bar and two-bar contact
configuration to model carrier conductivity as a sheet. For calculations, two methods of
modeling charge carrier generation are considered; a uniform generation throughout the film and
a depth and wavelength-dependent generation. This work found values on the order of 10-1 cm2
V-1, 10-4 cm2 V-1, and 10-5 cm2 V-1, for GaAs, annealed SnS, and un-annealed SnS, respectively,
for both methods of calculation. The simplified approach considering a uniform generation
yielded lower results than the depth and wavelength dependent calculations by about a factor of
two. All values were three to four orders of magnitude higher than those found in the literature.
For this reason, it is believed that the majority-carrier is dominating measurements due to an
inhibited minority-carrier lifetime.

Thesis Supervisor: Tonio Buonassisi

Title: Associate Professor of Mechanical Engineering

  3  
   4  
Acknowledgements

I would like to acknowledge a number of persons and communities that have helped me

not only get through MIT and write this thesis, but develop as a person and a scientist.

Firstly, I would like to acknowledge the unique MIT community for always offering the

means and opportunities to pursue every little passion and the DMSE and SUMS community for

both introducing Materials Science and bolstering a love for the work and each other.

To Dr.Jagadeesh Moodera for the opportunity of my first research experience and

invaluable teachings on curiosity and motivation as the foundations of research.

To Professor Craig Carter for the carefree optimism. I would often find myself learning

only to meet a deadline, but Professor Carter helped me remember that the best thing about

learning is just the new knowledge we obtain.

To the PV Lab Community for always being supportive and collaborative. I immediately

felt like a part of the team, which truly contributed to every positive outlook I had about the

research. In particular, to Professor Tonio Buonassisi for setting one of the greatest examples of

leadership I’ve ever seen. I am so grateful to have been an undergraduate researcher in this lab.

Much gratitude to doctoral candidate Rupak Chakraborty and fellow senior Luisa Barrera

for all their work on the coding for calculating the mobility-lifetime product under non-uniform

generation and an extra special thank you to doctoral candidate Riley Brandt for his endless

guidance, wisdom, and patience throughout my time on the team.

Finally, I am forever grateful for the support and love of my parents. Everything you do

is for the betterment of your children. I can only hope to one day follow your example.

  5  
    

  6  
Contents

1. Introduction .......................................................................................................................12

1.1 Motivation ....................................................................................................................................12

2. Background Physics ......................................................................................................14

2.1 Energy Bands ..............................................................................................................................14

2.2 Doping ...........................................................................................................................................17

2.3 Recombination and Traps ........................................................................................................18

2.4 Basics of a Solar Cell ................................................................................................................20

3. Charge Carrier Lifetime and Photoconductivity ............................................24

3.1 Introduction to Lifetime ............................................................................................................24

3.2 Influence of Lifetime .................................................................................................................24

3.3 Introduction to Photoconductivity .........................................................................................27

3.4 Determining the Mobility-Lifetime Product .......................................................................27

3.4.1 Simplified Approach ..........................................................................................................28

3.4.2 Considering Non-uniform Generation ..........................................................................29

3.5 Methods of Experimental Derivation ....................................................................................31

3.5.1 Snake Configuration ..........................................................................................................32

3.5.2 Van der Pauw Configuration ...........................................................................................33

3.5.2 Bar Configuration ...............................................................................................................34

4. Experimental Procedures ............................................................................................35

  7  
 4.1 Cyrostat Tool ...............................................................................................................................35

4.2 Sample Fabrication ....................................................................................................................36

4.3 Mounting Samples .....................................................................................................................37

4.4 Data Collection and Processing ..............................................................................................38

4.5 Solar Cell Capacitance Simulator (SCAPS) .......................................................................39

5. Results ...................................................................................................................................43

5.1 Test Sample Gallium Arsenide (GaAs) ................................................................................43

5.2 Tin Monosulfide (SnS) .............................................................................................................45

5.2.1 Tests for Sub-band-gap Light ..........................................................................................45

5.2.2 Noise Floor ...........................................................................................................................46

5.2.3 Effects of annealing ...........................................................................................................46

5.2.4 Depth and Wavelength Dependent Generation ..........................................................47

5.2.5 Low Illumination Measurements ...................................................................................49

5.2.6 Seebeck Measurements .....................................................................................................50

6. Conclusions .........................................................................................................................51

6.1 Discussion and Impact ..............................................................................................................51

6.1.1 Doping Effects .....................................................................................................................51

6.1.2 Variations from Literature Data ......................................................................................52

6.2 Further Research .........................................................................................................................53

Bibliography 54

Appendix 57

  8  
   9  
 List of Figures

1-1 Solar Irradiance Spectra 1.5

2-1 Electron Cloud Overlap in Solids
2-2 Direct and Indirect Band-gaps
2-3 Donor and Acceptor Levels of Doped Semiconductor
2-4 p-n Junction Simulation in the Dark
2-5 Current-Voltage Curve for Dark and Illuminated Diode
2-6 Enlarged Current-Voltage Curve for Illuminated Diode
3-1 Graph of Efficiency vs. Log10 (Minority-Carrier Lifetime) for SnS/Zn(O,S)) cell
3-2 Graphs of Efficiency vs. Interface Recombination Velocity for SnS/Zn(O,S)) cell
3-3 Layered Representation of Sample
3-4 Snake Configuration
3-5 Van der Pauw Configurations
3-6 Bar Configurations
4-1 Optics Table Set-up
4-2 Mounted GaAs Sample and Diagram
4-3 Main Screen in SCAPS
4-4 Main Screen of SnS Layer in SCAPS
4-5 Main Screen of a Loaded Cell in SCAPS
4-6 Defect Screen of SnS Layer in SCAPS
5-1 Graph of Mobility-Lifetime vs. Illumination for GaAs
5-2 Graph of Log10 (Conductivity) vs. Temperature for GaAs
5-3 Graph of Log10 (Conductivity) vs. Illumination for Sub-Band-Gap Light
5-4 Graph of Mobility-Lifetime vs. Log10 (Number of Layers) for Annealed SnS
5-5 Graph of Conductivity vs. Illumination for Annealed SnS
A-1 Graph of Log10 (Illumination) vs. Optical Density

List of Tables

5-1 Mobility-Lifetime Product for Uniform and Non-uniform Generation.

A-1 Conductivity for Uniform Generation
A-2 Mobility-Lifetime Product for Uniform Generation

  10  
   11  
1. Introduction

1.1 Motivation

The modern rate of energy and material consumption and waste is unsustainable.

Developed nations currently enjoy a lifestyle only possible through the advancements made

during the industrial age. However, developing nations are fighting to reach this same standard

of living while burdened by the environmental costs of the Western world’s advancement but

unable to make similar sacrifices. Current methods of energy production are moving towards

cleaner and safer sources, but not at rates fast enough to sustain growing global energy demands.

Solar is one of the most promising and fastest growing renewable energy sources in the

field due to its versatile potential. While renewable energy sources such as wind or geothermal

are strongly limited by geographical location, solar energy can be utilized nearly anywhere on

the earth’s surface with around 36,000 TW of available solar resource [2]. Given that the global

energy consumption is around 16 TW [2], the benefit of advancing solar energy harvesting

technologies and pushing up cell efficiencies becomes obvious. With further technological

advancements, solar energy can also be integrated into common items of everyday life.

According to the Shockley-Queisser Limit, no single p-n junction solar cell, discussed in

Chapter 2.4, will be able to obtain 100% conversion efficiency - the highest theoretical efficiency

is around 33% for a band gap of 1.3 eV, shown by the lowest curve in Figure 1-1. Current single-

crystalline solar cell modules have achieved efficiencies around 25% [14]. Other methods have

sought to surpass this efficiency by utilizing different band gaps and material absorption

properties in layers, called tandem cells. Unfortunately, the production expenses of such cells

often outweigh the benefits of their increased efficiency.

  12  
 The methods pursued in this thesis explore one of the most important and less

documented factors affecting device efficiency – minority-carrier lifetime (τ). This is a measure

of how long the minority charge carrier, an electron or hole, exists within a material. This will be

analyzed in conjunction with the carrier mobility (µ), the speed with which the carrier moves,

through measurements of the material’s photoconductivity (σ), explained further in Chapter 3.

Figure 1-1: Solar Irradiance Spectra 1.5 for 1.1 eV bandgap. The majority of
available light is unusable for solar energy conversion. The beige area corresponds to
the Theoretical Shockley-Queisser Limit. [19]

The minority-carrier lifetime is a very important property for solar cells because it

determines the likelihood that a charge carrier will be collected by a metal contact, or electrode

on the cell, and therefore used for energy conversion. A longer lifetime in a solar cell typically

means a greater likelihood of collection. For this reason, a longer lifetime also correlates to a

larger efficiency. Preliminary work for this thesis to observe the importance of lifetime on cell

efficiency was done on a solar cell simulator called SCAPS. Some of this work can be seen in

Chapter 2.2 and a brief explanation of the SCAPS program is given in Chapter 4.5.

  13  
 Charge carrier lifetime can be difficult to measure, particularly since the time scale is on

the order of a couple of nanoseconds. This thesis will explore one technique to experimentally

determine the photoconductivity of p-doped tin monosulfide (SnS) and gallium arsenide (GaAs)

and computationally acquire the product of charge carrier mobility and lifetime (µτ) of these

materials. The computational approach to determining the mobility-lifetime product will use both

a simplified computation as well as a more complex one, intended to build-upon current

procedures for photoconductivity measurements. It is hoped that this work will contribute to

solutions for efficiency limitations of SnS absorber layers in solar cells.

GaAs has been a popular material for solar research due to its theoretical efficiency limit

around 30% and high carrier mobility [10]. It was chosen as a reference material to assure the

reliability of the machine and techniques.

SnS is not as well researched or processed as GaAs. It has both a direct and indirect band-

gap, which offers the potential for strong optical absorption, characteristic of direct band-gaps,

and longer minority-carrier lifetimes, due to the phonon barrier of indirect band-gaps. The

physics of this will be discussed further in the next section. Finally, SnS is a nontoxic, abundant

material, which does not require very high temperatures for film fabrication.

2. Background Physics

2.1 Energy Bands

First, it is necessary to discuss the conductive mechanisms of semiconductor materials.

According to Bohr’s model of an atom, electrons exist fixed within an orbit of a specific

geometry and energy around a nucleus. These orbits are called electron clouds and movement of

an electron between clouds is only allowable with quantized changes in energy.

  14  
 While Bohr’s atomic model describes electron movement and placement around an atom,

it does not consider how atoms interact when closely packed in a solid. The small atomic spacing

of a crystallographic arrangement causes the electron clouds of atoms to overlap. In metals, this

electron level overlap occurs across so many energy levels, that electrons are able to move freely

throughout the solid. Because of this, metals are strong conductors.

Figure 2-1: Electron cloud overlap in solids. With large enough interatomic spacing,
energy levels begin to split into two distinct bands. [6]

In semiconductor materials, the electron cloud overlap is not so comprehensive, though

there is still considerable interaction. The electron levels of the atoms within the clouds merge to

form two specific bands, called the valence and conduction bands. A schematic of this for

metals, semiconductors, and insulators is shown in Figure 2-1. First, the valence band is the

lower energy grouping, where electrons are immobile. If enough energy is supplied to a bound

electron in the valence band, it can be “excited” into the conduction band, where it is able to

move freely throughout the material. A possible source of energy for excitation is a photon, or

  15  
particle of light. Mobile electrons in the conduction band leave holes in the valence band. Both

charge carriers in their respective bands contribute to a material’s conductivity.

Figure 2-2: Representation of Valance and Conduction bands. An indirect band-gap

(right) will need a phonon contribution in addition to a photon in order for carriers to
traverse the band-gap. This can inhibit absorption. [7]  

The energy spacing between these bands is known as the band gap. In a pure

semiconductor material, no energy states can exist within the band-gap and all electrons in the

conduction band will have energy equal to or greater than this gap energy. The conduction and

valence bands simply identify the boundaries of these available energy states but they are not

themselves discrete energy levels. An insulator has a similar band structure to a semiconductor,

except the band-gap is so large that very few electrons are able to be excited across it at room

temperature.

  16  
 Figure 2-2 shows two possible band structures in a semiconductor; a direct band-gap and

an indirect band-gap. The direct band-gap requires only the energy of this band edge difference

to excite an electron across it. However, an indirect band-gap requires both this energy and a

phonon contribution to overcome the change in momentum required for band transition. A

phonon is a quantum of energy associated with the vibration of a crystal lattice. Excitation or

relaxation across an indirect band-gap is less likely to occur than across a direct band-gap due to

this phonon requirement. As such, direct band-gaps are known for high absorption and indirect

band-gaps are known for longer lifetimes.

As an example, let’s look at the most well-known semiconductor material, silicon (Si),

which has a direct band-gap energy of 1.1 eV, or 1.76×10-19 J. The quantized energy of a single
!∗!
photon of light is , where h is Plank’s constant, c is the speed of light, and λ is the wavelength
!

of the photon. Just within the visible range of the Sun’s spectrum, λ is between 400 and 700 nm,

so the associated energy is between 2.8×10-19 J and 4.9×10-19 J. This is more than enough energy

to generate carriers in Si!

2.2 Doping

In the real case, there are many levels within the band-gap that can influence the

concentration of electrons and holes in the conduction band and the valence band. One way to

intentionally do so is to dope the material. Doping creates donor or acceptor levels, which are

able to contribute an electron or a hole, respectively, to their corresponding conduction bands, as

shown in Figure 2-3.

To dope a material is to introduce atoms of another element with a higher or lower charge

than what exists in the crystal lattice. A common doping pair is Si and boron (B), or Si and

  17  
phosphorous (P). On the periodic table, Si is a group IV element, B is group III, and P is group

V. For this reason, B has one additional positive charge associated with it than Si and P has one

additional negative charge. Therefore, doping Si with B causes the material to be positively

doped, or p-doped, while doping Si with P causes it to be negatively, or n-, doped.

Figure 2-3: Diagram of n-type (a) and p-type (b) doped semiconductors. The donor
energy level within the band-gap of (a) donates electrons to the conduction band.
Similarly, the acceptor level of (b) accepts electrons and leaves conducting holes in the
valence band. [6]  

In the case of SnS, doping tends to come from vacancies, or missing atoms, in the

material. An S vacancy is associated with a negative charge and a Sn vacancy with a positive.

For this reason, when vacancies of one type are reduced, it causes a relative increase in the other

type, effectively doping the material. For one of the SnS samples of this work, annealing in a S

rich environment caused a decrease in S vacancies, which doped the material as p-type [5].

While electrons and holes typically come in pairs, doping allows one to be in excess. The

carrier in excess is called the majority charge carrier, counter to minority charge carrier.

2.3 Recombination and Traps

For a free carrier in the material, there are three possible things that can occur; it can be

collected by an electrode, it can recombine with its charge pair, or it can become trapped. When

  18  
an electron-hole pair recombines, it “no longer exists” because it is no longer contributing to the

conductivity of the material. Recombination occurs when a free carrier electron meets a free

carrier hole and fills it. This electron then drops down to a lower energy level on an atom,

returning to the valence band of the material.

Trapping is similar to recombination in that it prevents the movement of charge carriers,

but unlike recombination, it is not a permanent occurrence. With some thermal energy, the

carrier is re-emitted back into the original band after some time. The probability of re-emission is

often equal to the probability of capture [1].

Trapping and recombination occur at sites within the material dubbed “trapping sites”

and “recombination sites” depending on the type of interaction that will likely take place there. If

a trap is particularly strong, such that the probability of the trapped carrier recombining with a

carrier of opposite sign is greater than the probability of the trapped carrier being thermally freed

from the trap, it will be a recombination site and vice versa for the opposite case. Often times, a

recombination site under one illumination and temperature may be a trapping site at another due

to the change in thermal energy contribution on the sample [1].

Radiative recombination can occur anywhere in the material when an electron meets a

hole and falls back down to the valence band. However, trapping and recombination sites are

present due to defects in the material. A defect is the presence of an imperfection in an otherwise

perfect crystal lattice. This can be the S vacancies present in the SnS material, “dangling” bonds

at the surface of the material, or dangling bonds or lattice mismatch at the interface between two

grains in a crystal, called a grain boundary. Materials with fewer defects will have a lower

probability of recombination for its charge carriers and incidentally, a longer charge carrier

lifetime. If the likelihood of dying is lower, the carrier will live longer.

  19  
 Another thing to consider when observing defects and trapping/recombination sites, is the

capture cross section (Scapture) of the site. The capture cross section relates to the variation of

potential around the site, which causes trapping to occur. If the binding energy of coulomb

attraction is greater than the thermal energy of the carrier, the carrier will become trapped [1]. In

this way, if the site, or center, is charged, it will preferentially capture the carrier with the

opposite charge. The relationship between Scapture and τ, can be shown by:
!
τ =   !"! (1)
!"#$%&'

where v is the mobility of the charge carrier, and N is the number of recombination centers per

unit volume which can capture a charge carrier. As Scapture increases, or as the preference of

capturing one charge over another increases, the τ of that preferred charge carrier will decrease.

In a semiconductor, measuring the minority-carrier lifetime is possible due to the relative

changes of photo-excited charge carriers compared to intrinsic charge carriers. If the density of

charge carriers in a material is much larger than the density of recombination centers, then the

limiting factor to conductivity will be the minority-carrier. Conversely, if the density of

recombination centers is larger, nearly all minority charges will be recombined and the majority-

carrier will dominate the photoconductivity signal, as is often the case for insulators.

2.4 Basics of a Solar Cell

A solar cell can be formed when a p-type and an n-type semiconductor are placed

together, creating a p-n junction. The difference in charge carrier type causes a flow of carriers

across the interface between the two materials until enough recombination has occurred, that a

neutral barrier forms, called the depletion zone, or space-charge region. An electric field exists

across this zone, which prevents most additional recombination events.

  20  
 Figure 2-4 shows a p-n junction in the dark under no applied voltage. The net flow of

carriers across the junction, or the current, is equal to zero. The total current is the result of

carriers moving in both directions across the space charge region. Diffusion current occurs when

a charge diffuses across the potential barrier, to the side of majority opposite charge – an electron

diffusing from the n to the p side. Drift current occurs when a minority-carrier is generated near

the space charge region, travels into it, and is propelled by the electric field towards the side with

positive like charge – an electron generated on the p side reaches the edge of the space charge

region and is sent to the n side of the device.

Figure 2-4: Screenshot of a p-n junction simulation in the dark. Carriers move across
the depletion, or space charge, region through diffusion and drift. Net current is zero
under zero voltage. [18]  

Applying a voltage across this junction can either raise or lower the depletion zone

barrier for charge flow. A forward bias voltage will decrease the electric field across the barrier

  21  
and thus raise the diffusion current, while a reverse bias will do the opposite. The diffusion

current is driven by recombination events.

The diode equation is given by:

!∗!
𝐼 =   𝐼! 𝑒 !∗! − 1 (2)

where I is the net current, I0 is leakage current in the dark, strongly dependent on

recombination, q is the charge of an electron, V is applied voltage to the junction, k is the

Boltzmann constant, and T is the temperature of the device. The diode equation models the

current of the device in response to an applied voltage. In the light, the equation becomes:

!∗!
𝐼 = 𝐼! 𝑒 !∗! − 1 −   𝐼! (3)

where IL is the photo-generated current. The resulting current verses voltage (IV) curve

for both equations can be seen in Figure 2-5.

2
Current [mA ]

-2

-4
-0.2 0.0 0.2 0.4 0.6
Voltage [V]

Figure 2-5: Current-Voltage (IV) curve for a p-n junction under dark (blue) and
illuminated (orange) conditions. T = 300 K, I0 = 10-8 mA, and IL = 2 mA.  

  22  
 The most important aspects of an IV curve under illumination are the x- and y- axis

intercepts and the point on the curve, which fills the maximum possible rectangle beneath the

curve. The y-axis intercept is called the shore circuit current (ISC) and corresponds to the

maximum possible current the solar cell achieves under zero voltage bias. The x-axis intercept is

the open circuit voltage (VOC) and corresponds to the maximum possible voltage achieved with

zero current. Both are indicated in Figure 2-6. VMP and IMP identify the voltage and current at

which maximum power is achieved. Their product creates the maximum possible rectangle area

beneath the IV curve.

Vmp Voc
0
Current [mA ]

-1

Imp
-2
Isc

-3
-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Voltage [V]

Figure 2-6: Close-up of the illuminated (IV) curve from Figure 2-5.  

Taking the area of this rectangle identifies the maximum power achievable by the cell.

The Fill Factor (FF) is a ratio of this maximum power and the product of VOC and JSC.

Maximizing the FF is desirable since this corresponds to a large generated power for the device.

All basic solar cell concepts, as well as many of the more complex ones, can be found at

  23  
pveducation.org [18]. For a more comprehensive understanding, several excellent textbook

references exist as well [20] [21].

Chapter 3. Charge Carrier Lifetime and Photoconductivity  

3.1 Introduction to Lifetime

The minority-carrier lifetime is simply how long an excited electron or hole exists or

survives in the material before being recombined, collected, or trapped. There are benefits to

having a long or short lifetime, depending on the desired device characteristics. For solar cells, a

long lifetime is better because it will increase the likelihood of the carrier being collected by a

metal contact. A longer lifetime can be achieved with a very pure material with minimal defects

or recombination sites.

3.2 Influence of Lifetime

The following graphs exhibit a few examples of the effect of carrier lifetime on device

efficiency. The change in one parameter, can have a significant effect on the change in efficiency

if a different parameter has already been optimized. However, there may be little to no noticeable

change if this has not yet occurred. The graphs below were modeled by a Solar Cell Capacitance

Simulator (SCAPS). For further information regarding this program, see section 4.5.

  24  
 SnS
α α α
α
15
α
Efficiency [% ]

10 α

β β β β β

5 β

0
-10 -9 -8 -7 -6 -5
2 -1
Log10 (τ) [cm V ]

Figure 3-1: Graph of efficiency vs. defect density for SnS, acquired through SCAPS.
Orange curve indicates cell with parameters reflecting current state-of-the-art SnS
cells. Blue curve taken after cell is optimized for an efficiency of 16.64%.

Figure 3-1 shows the effect of changing the defect density within the p-type layer of the

stack, which directly changes the carrier lifetime by the same order of magnitude. For the SnS

stack, a defect density of 1017 cm-3 corresponds to a lifetime of 0.1 ns, while a defect density of

1012 cm-3 corresponds to a lifetime of 10 µs [13].

Typical good lifetimes for solar cells are around the 100 ns range, so a defect density for

this simulation of 1014 cm-3. The blue curve is after several other parameters in the cell have been

optimized to achieve an efficiency of 16.64%. Maximum theoretical efficiencies for SnS cells

are around 30%, but even in simulation, they can be difficult to achieve.

  25  
 Figure 3-2: Graphs of efficiency vs. interface recombination velocity for SnS, acquired
through SCAPS. Orange curve indicates cell with optimized lifetime of 100 ns. Blue
curve indicates cell with original lifetime of 0.1 ns.

While both curves show a general trend of increased efficiency for increased lifetime, it

is clear that the efficiency increases strongly with small improvements when the material starts

with more defects. Eventually, the curve appears to level out, showing either that it is acceptable

to have a fairly impure material with optimum performance, or the lifetime has reached another

barrier and cannot improve without some additional alterations of other parameters.

The change in efficiency in SnS for the optimized and non-optimized cases, respectively

are ∆𝜂!"#$%$&'( = 6.95% and ∆𝜂!"!!!"#$%$&'( = 1.38%.

While defects exist within the bulk of a material, any surface or interface can be thought

of as one, continuous defect, due to dangling bonds. This means it has prime opportunities for

recombination. The bulk and the surface can be seen as having two separate lifetimes, but for the

purpose of this thesis, only the effective lifetimes will be discussed.

  26  
 It is possible to decrease the recombination velocity of a surface or interface by reducing

the number of dangling bonds with supplemental atoms or better lattice matching. The results of

this are shown in Figure 3-2. For this parameter, τsurface, the bulk lifetime yields some limitations

to its effect on efficiency. For SnS, ∆𝜂!!!.!!" = 1.94% and ∆𝜂!!!""!" = 4.25%.

3.3 Introduction to Photoconductivity

The photoconductivity of a material is the effect of increased bulk conductivity due to the

introduction of photons into the material. Each photon carries a specific energy dependent upon

its wavelength. When the light particle reaches an electron it has the opportunity to excite the

electron to the conduction band by transferring this energy. This excited electron and its

corresponding hole are then free to move about the material in their respective bands,

contributing to the overall conductivity. Because this is a photon-induced effect, it is called

photoconductivity.

3.4 Determining the Mobility-Lifetime Product

For a material under no-light conditions, the conductivity can be modeled by:

𝝈𝒅𝒂𝒓𝒌 = 𝒒 ∗ µμ𝒆 𝒏𝟎 +   µμ𝒉 𝒑𝟎 (4)

where, q is the charge of an electron, µe is the mobility of an electron, µh the mobility of a hole,

and n0 and p0 the intrinsic concentration of electrons and holes, respectively. For a p-doped

material, n0 << p0, so the equation can be simplified to:

𝝈𝒅𝒂𝒓𝒌 = 𝒒µμ𝒉 𝒑𝟎 (5)

Under illumination, carrier concentrations change with the generation of electron-hole pairs.

Since each excited electron generates a hole, ∆n = ∆p:

𝝈𝒍𝒊𝒈𝒉𝒕 = 𝒒 ∗ µμ𝒆 (𝒏𝟎 + ∆𝒏) +   µμ𝒉 (𝒑𝟎 + ∆𝒑) (6)

  27  
where ∆n and ∆p are the change in carrier concentration for electrons and holes, respectively.

From this point, two separate methods were used to calculate µτ.

3.4.1 Simplified Approach

In the simplest condition, full absorption and a uniform generation is assumed over the

entire volume of the sample and µτ is calculated from the conductivity data acquired from the

experiments. The experiments were done at low injection levels and for the simplified case,

generation depends only on illumination. There is neither depth nor diffusion dependence. For

the change in conductivity caused by illumination:

𝝈𝒍𝒊𝒈𝒉𝒕 −   𝝈𝒅𝒂𝒓𝒌 = ∆𝝈 =  𝒒 ∗ µμ𝒆 ∗ ∆𝒏 + µμ𝒉 ∗ ∆𝒑     (7)

It is known that the change in charge carriers can be calculated by:

∆𝒏 = 𝑼 ∗ 𝝉 = 𝑮 ∗ 𝝉 (8)

where U is the recombination rate, 𝜏 is the lifetime, and G is the generation rate. For the

simplified case, the recombination rate is equal to the generation rate, which is calculated as the

sum of all photons generated by energies above the band-gap energy on the AM1.5 spectra.

Under 100% illumination of one sun, or 1000 W cm-2, for the 1 µm thick samples, the generation

rate is 2.85×1027 m-3 s-1 for SnS and 2.08×1027 m-3 s-1 for GaAs.

At lower illuminations, the new generation rate is simply reduced by the same fraction as

the light. For example, a sample under 500 W cm-2 illumination will observe a halved generation

rate from that calculated for 1000 W cm-2.

Combining Equations 7 and 8 yields:

∆𝝈
µμ𝝉 = 𝑮∗𝒒 (9)

  28  
Determining just the 𝜏 of a material is possible if µ is known, but that is beyond the scope of this

thesis. The plots presented in Chapter 5 show µτ products acquired through the simplified

approach.

3.4.2 Considering Non-uniform Generation

At every point within the sample, there are varying quantities of generated carriers due to

the non-uniform penetration depth of photons and the diffusion of generated carriers. In this case,

the depth dependence is considered, but not diffusion. This method calculates a µτ considering a

depth and wavelength dependent generation and then uses an iterative process to individually

calculate the µ and τ of the material considering the resistance acquired experimentally and

bounds of µ and τ found in literature.

In the dark, the conductivity of the material doesn’t change. The dark resistance, which

is related by the geometric constraints of the sample, is also constant:

𝒅∗𝒕 !𝟏 𝒒∗𝒅∗𝒕 !𝟏
𝑹𝒅𝒂𝒓𝒌 =   ∗ 𝝈𝒅𝒂𝒓𝒌 =   ∗ µμ𝒉 ∗ 𝒑𝟎   ( 10 )
𝑳 𝑳

where d is the length of the contacts, L is the length between contacts, and t is the thickness of

the slab. In the light, the conductivity can now consider the depth dependence of carrier

generation, as modeled by:

𝝈𝒍𝒊𝒈𝒉𝒕 = 𝒒 ∗ µμ𝒆 (𝒏𝟎 + ∆𝒏(𝒙)) +   µμ𝒉 (𝒑𝟎 + ∆𝒑(𝒙)) ( 11 )

where ∆n and ∆p become functions of the sample depth, x.

  29  
 Photon
s  

  N  =5  

Figure 3-3: Diagram of layered sample. Longer wavelength photons tend to penetrate
deeper into the sample. Depth is split into N = 5 slabs.

Since the light reaches the top of the sample first, most of the higher energy photons are

absorbed here. Generation decreases as the lower energy photons penetrate into the sample and

interact with electrons. To model this, the sample can be split into N slabs down its thickness,

shown in Figure 3-3. The generation rate becomes:

∆𝒏 𝒙
𝑮 𝒙 = 𝜶 ∗ 𝑵𝟎 ∗ 𝒆!𝜶∗𝒙 =           ( 12 )
𝝉

where α is the absorption coefficient, and N0 is the photon flux, found from the AM1.5 spectra.

Generation rate is still equal to the recombination rate, U(x), though now both are depth

dependent. To model this dependence in the resistance under illumination, a summation of the

resistance of each individual slab is taken:

𝟏   𝑵 𝟏
𝑹𝒍𝒊𝒈𝒉𝒕
=   𝒊!𝟏 𝑹 ( 13 )
𝒊

and:
𝒕
𝟏   𝝈𝒍𝒊𝒈𝒉𝒕 ∗𝒅∗
𝒊
=   ( 14 )
𝑹𝒊 𝑳

Substituting values into the Rlight equation and taking the limit as number of slabs, N, approaches

infinity:

  30  
 𝒕
𝟏   𝒅
=   ∗ 𝒒 ∗  µμ𝒆 (𝒏𝟎 + ∆𝒏(𝒙)) +   µμ𝒉 (𝒑𝟎 + ∆𝒑(𝒙))   𝒅𝒙
𝑹𝒍𝒊𝒈𝒉𝒕 𝑳
𝟎

𝒒∗𝒅 𝝈𝒅𝒂𝒓𝒌 𝒕
=   ∗ ∗ 𝒕 +   µμ𝒆 ∗ 𝟎
∆𝒏 𝒙  𝒅𝒙 ( 15 )
𝑳 𝒒

𝑞 ∗ 𝑑 𝜎!"#$
=   ∗ ∗ 𝑡 +   µμ! ∗ 𝜏 ∗ 𝐺
𝐿 𝑞

which gives:

𝟏 𝟏 𝒒∗𝒅
  −   𝑹   =   ∗ µμ𝒆 ∗ 𝝉 ∗ 𝑮         ( 16 )
𝑹𝒍𝒊𝒈𝒉𝒕 𝒅𝒂𝒓𝒌 𝑳

therefore:

𝟏 𝟏 𝟏
µμ𝒆 𝝉 =   𝒒∗𝒅 ∗   −   𝑹   ( 17 )
∗𝑮 𝑹𝒍𝒊𝒈𝒉𝒕 𝒅𝒂𝒓𝒌
𝑳

The coding process to calculate a µτ product under non-uniform generation is briefly

discussed in Chapter 4.4.

3.5 Methods of Experimental Derivation

Several methods of photoconductivity measurements currently exist which all follow the

same general principles of measurement and calculation. The next sections will briefly discuss

the workability and shortcomings of three methods that were considered for this thesis. Due to

time constraints for both experimental work on sample fabrication and measurement as well as

computational work on simulation and modeling, the simplest albeit least accurate method was

chosen. These shortcomings are minor and do not significantly affect the data. The other two

configurations are discussed to elaborate on the versatility of photoconductivity measurements as

well as offer alternative methods for anyone hoping to take similar type measurements.

  31  
 3.5.1 Snake Configuration

The snake configuration was a method utilized by JT Sullivan, a previous doctoral

student at the MIT Photovoltaic Laboratory [9]. A schematic of his device, taken from his thesis,

is shown in Figure 3-4.

Figure 3-4: Schematic of Snake configuration used by Joseph Timothy Sullivan.

Actual samples had eight fingers. [9]

This device set-up creates a minimal spacing between contacts and minimal exposed

surface area of the semiconductor material to quickly capture carriers as soon as they are

generated while minimizing the effect of thermal excitation. Additional measures for preventing

heating effects, such as chopping of light and a thermal sink beneath the sample, eliminates the

false positives for photoconductivity associated with thermally excited carriers. The area to

length ratio in this configuration is maximized to allow for very small ∆σ measurements.

Unfortunately, the dimensions of such a device cause it to be both difficult to fabricate

and difficult to handle and measure.

  32  
 3.5.2 Van der Pauw Configuration

The van der Pauw method is used to calculate the resistivity of the material, where the

conductivity is the reciprocal of the resistivity. An image of possible van der Pauw

configurations is shown in Figure 3-5. It is important that the sample is as symmetric as possible

and the contacts are as small as possible while remaining on the edge of the sample.

Figure 3-5: Possible van der Pauw sheet configurations where the darker squares are
the contacts. [8]

In the van der Pauw configuration, a current is run along one length of the sample while a

voltage is measured across the other length. Together, these can be combined to find the

resistance measurement, R. Assigning the contacts as 1, 2, 3, 4, starting with the upper right

contact and moving clockwise on the disk configuration of Figure 3-5, if a current is run along

points 1 and 2, and a voltage is measured along 3 and 4, then a resistance can be measured as:

𝑽𝟑𝟒
𝑹𝟏𝟐,𝟑𝟒 = 𝑰𝟏𝟐
( 18 )

To account for asymmetric qualities of the sample, reversed polarity measurements are utilized

to acquire vertical and horizontal resistances across the sample:

𝑹𝟏𝟐,𝟑𝟒 !𝑹𝟑𝟒,𝟏𝟐 !𝑹𝟐𝟏,𝟒𝟑 !𝑹𝟒𝟑,𝟐𝟏

𝑹𝒗𝒆𝒓𝒕𝒊𝒄𝒂𝒍 = ( 19 )
𝟒
𝑹𝟐𝟑,𝟒𝟏 !𝑹𝟒𝟏,𝟐𝟑 !𝑹𝟑𝟐,𝟏𝟒 !𝑹𝟏𝟒,𝟑𝟐
𝑹𝒉𝒐𝒓𝒊𝒛𝒐𝒏𝒕𝒂𝒍 = ( 20 )
𝟒

  33  
and then substituted into the van der Pauw formula:
!𝝅∗𝑹𝒗𝒆𝒓𝒕𝒊𝒄𝒂𝒍 !𝝅∗𝑹𝒉𝒐𝒓𝒊𝒛𝒐𝒏𝒕𝒂𝒍
𝒆 𝑹𝑺
+𝒆 𝑹𝑺
=𝟏 ( 21 )
where RS is the sheet resistance of the material.

Initial test measurements were taken on samples of this configuration but acquiring a µτ

beyond the simplified case of generation would have required modeling in 3D. None of this

initial data is presented in this thesis.

3.5.3 Bar Configuration

All samples in this thesis are in the bar configuration. Dimensions are shown in Figure 3-

6. This is the simplest method examined for measuring photoconductivity and can be modeled as

a 1D or 2D problem. Electrons and holes conduct across the material as a sheet between the two

contacts and the depth of the sample can be approximated as individual layers to account for the

depth dependence of generation. Additionally, contacts in this configuration are easy to fabricate.
10.4 mm

I+  ,  V+    
750 µm

GaAs    
13 mm

7.22 mm

I   V V I
                 SnS    
-­‐   -­‐   +   +  

I-­‐  ,  V-­‐  

Figure 3-6: Bar configurations and dimensions used for experiments. GaAs is in the
two-bar configuration (left) and SnS in the four-bar configuration (right).

  34  
 The four-probe sample set-up is preferred because it eliminates the resistive contribution

of the interface between the sample and the contact, as well as the interface between the contact

and the conducting wires for measurement. In the two-bar configuration, four conducting wires

are utilized, so that the latter interface contribution is still eliminated, but the resistive barrier

between the contact and the sample still contributes. Even so, this contribution is very small and

likely does not affect the data in this work.

Chapter 4. Experimental Procedures

4.1 Cryostat Tool

All measurements were done on an optics table using an ARS cryostat tool with NesLab

chiller and a Newport LCS-100 lamp. The light intensity of the lamp was found to be 1250 W

cm-2 as calibrated from a Si calibration sample. The measured photocurrent can be found in

Figure A-1 the appendix. Two Thorlabs FW102C filter wheels are able to reduce the optical

density down to 0.02% of the original light source. The set-up can be seen in Figure 4-1.

The cryostat tool is capable of reducing the temperature within the chamber to around 9

K and the temperature of the sample mounted in the chamber to around 100 K. A full cool-down

requires around one hour. Precise temperatures are acquired with a feedback loop between the

LabView program and a Lakeshore temperature controller. The chamber is heated using a

resister coil.

There are two temperature sensors used for readings: temperature sensor A, which is

placed on the central finger within the chamber and temperature sensor B, which is placed

directly on the sample with a small amount of vacuum grease and an electrically and thermally

isolated copper prong to hold it in place.

  35  
 Figure 4-1: Optics table set-up. Vacuum pump is unplugged and canister is on.

4.2 Sample Fabrication

The sample stack for the GaAs sample is GaAs wafer / insulating layer (MOVCD) / p-

GaInP / GaAs film / p-GaInP / (Ti/Au) contact. The p-GaInP barriers passivize the surface of

GaAs to reduce surface recombination effects but otherwise do not contribute to the performance

of the material. The film is 1 µm thick, with contact lengths of 10.4 mm and distance between

contacts measured at 13 mm. A schematic can be seen in Figure 3-6. GaAs is a direct band-gap

semiconductor of 1.42 eV. In the literature, µτ values are found to be on the order of 10-4 cm2 V-1

for un-doped GaAs [3] and 1.3×10-6 cm2 V-1 for p-doped GaAs at a carrier concentration of

5×1018 cm-3 [11]. The GaAs sample used in this work is highly doped up to 1018 cm-3.

  36  
 For the SnS samples, both are 1 µm thick with a contact length of 7.22 mm and a distance

between contacts of 750 µm, shown in Figure 3-6. The stack is; Si substrate / insulating layer

(SiO2) / SnS / (Ti/Au) contact, where the SnS layer was deposited by thermal evaporation. SnS

has both a direct and indirect band-gap of 1.3 eV and 1.1 eV respectively. Literature values of µτ

for SnS did not exceed 10-7 cm2 V-1 [4] [22]. The annealing process for the doped sample was

4% hydrogen sulfide (H2S) in nitrogen gas (N2) for 10 minutes at around 80 Torr. Annealing a

sample will increase the crystalline grain diameter; reducing the number of edge defects and

effectively improving the conductivity of a material. Additionally, the S rich environment of the

annealing process causes some S vacancies in the material to be filled, effectively increasing the

Sn vacancies, which are acceptors in the p-type material [5]. The un-annealed sample is p-type

with a carrier concentration around 1015 cm-3 while the annealed is 1016 cm-3. Future research

may consider fabricating samples with passivation layers to reduce surface recombination. It is

not explored in this thesis.

4.3 Mounting Samples

Samples are mounted to the sample holder with some vacuum grease on the back of the

sample and held in place with a thermally and electrically isolated copper prong, as shown in

Figure 4-2. Wires are carefully placed on the contacts and bonded with silver paint. The sample

and mount are then moved to the cryostat, where it is screwed into the central finger.

Temperature sensor B is placed directly on the sample with a small amount of vacuum grease

and held in place with another thermally and electrically isolated copper prong. The wires are

connected and the corresponding positions on the sample are noted. The outer canister is

carefully secured and an Edwards TIC pumping station is turned on to create a vacuum

environment.

  37  
  
 

Figure 4-2: GaAs sample mounted and screwed onto cryostat cold-finger (left) and
schematic of GaAs sample (right). The sample is held in place with a plastic screw
and washer as well as an isolated Cu prong. The prong also holds temperature sensor
B in place during measurements.

Resistances are measured thrice during this process to assure contacts are still working:

directly after the application of the silver paint, directly after mounting into the cryostat, and

soon after fitting the canister and putting the sample under vacuum. Once the system has reached

pressure below 0.1 mTorr, the cooling system can be turned on.

4.4 Data Collection and Processing

The sample is cooled to very low temperatures and then heated incrementally by 20 K.

The tolerance is 0.5 K and must be maintained for 60 s before each illumination measurement.

Light hysteresis effects do not seem to be a problem, since the change in conductivity is sudden

and returns to within 5% of the original value in the dark in less than 1 s. Even so, illumination

measurements are done from dark to full illumination. A Keithley SourceMeter sources a voltage

  38  
across two contacts and the photoconductive response is measured. IV curves are generated with

each measurement and all subsequent data is derived from these.

All data processing was done on the coding platform MatLab. The simplified approach to

calculating µτ is taken to acquire an approximate estimate for each temperature and illumination

measurement and generate the plots for the results section of this thesis. A more complex

MatLab code created by a colleague considers the depth and wavelength dependence of

generation, discussed in Chapter 3.2.2.

The code takes an Rdark, a µτ acquired through the simplified approach, and bounds for µ

and τ. Through an iterative process, it is able to determine a µ and a τ such that the calculated

Rlight in the program equations is as close as possible to the Rlight found experimentally. For

several data sets, the calculated µτ have been about a factor of two les than those found through

the simplified modeling approach. A table of calculations can be found in Chapter 5.2.4.

Since the values for µτ acquired through the two methods correlate strongly and because

the more complex approach requires each measurement be calculated separately, in the interest

of time all graphs were created with µτ values from the simplified approach.

4.5 Solar Cell Capacitance Simulator (SCAPS):

The program used for beginning research in this work was the Solar Cell Capacitance

Simulator (SCAPS), which is available, free of cost, on the web. It is “a one dimensional solar

cell simulation program developed at the Department of Electronics and Information Systems

(ELIS) of the University of Gent, Belgium” [24]. Through this program, a photovoltaic

researcher is able to upload material properties and generate layers of a cell to simulate a p-n

junction. Material parameters and device properties can be set and manipulated within each layer

  39  
as well as at all interfaces. Through SCAPS, an SnS solar cell with an n-type layer of zinc oxy-

sulfide (Zn(O,S)) was analyzed.

Figure 4-3: Main screen of operation in SCAPS program. Simulation parameters can
be inputted, such as; operating voltage, generated graphs, or saved folder location.
 
While certain qualities, such as a thin n-type layer, high doping concentrations, and a low

defect density, positively contributed to an increasing efficiency, it is important to note, that any

quality could be impossible to achieve in the laboratory. For example, very thin layers may be

impossible to fabricate, or a doping concentration that is too high will destroy the charge barrier

across a p-n junction.

  40  
 Figure 4-4: Main screen of SnS layer in SCAPS program. Previously loaded
characteristics of the material layer can be individually edited.
 
For many material applications, it is most desirable to work with a pure material, where

there are little to no defects. For photovoltaic research, this translates into little to no traps or

recombination sites for carriers.

Figure 4-5: Main screen for the loaded cell. A 1D depiction of the cell is given on the
right. Each individual layer and interface can be altered, including the front and back
contacts.
 

  41  
 Within the SCAPS program, defects can be created and altered at each editable position

to simulate this characteristic of the material. Through these defects, the lifetimes of electrons

and holes can be manipulated, together or independently.

Figure 4-6: Screen of defect one of SnS layer. Characteristics specific to this defect
can be edited. Changing “Nt total (1/cm3)” directly alters both carrier lifetimes
equally. This effect can be seen on the main screen for the layer, represented in Figure
4-4, above. To change the carrier lifetimes separately, capture cross section can be
edited.
 
Unfortunately, determining the most ideal qualities and quantities on SCAPS does not

guarantee the perfect solar cell. While some parameters may be difficult to fabricate, other

factors may hinder experimental findings, such as the interactions between two materials during

deposition or strong tunneling effects at contacts. It becomes necessary to input more layers or

  42  
generate more defects within the simulation to create a more accurate image of real-life results.

For example, Figure 4-5 shows a layer of ZnO between the Zn(O,S) and indium tin oxide (ITO)

layers, because it forms naturally to create a stable interface.

Chapter 5. Results
 
5.1 Test Sample GaAs

To assess the reliability of our photoconductivity measurements, a GaAs thin-film sample

was tested due to the well-known photoconductivity and mobility-lifetime values in literature.

Un-doped GaAs has a µτ on the order of 10-4 cm2 V-1 [3]. Doped GaAs with p-type carrier

concentration on the order of 1018 cm-3 has a µτ value of 10-6 cm2 V-1 [11]. The experiments,

however, found the GaAs sample to have a µτ product three to four orders of magnitude higher,

as shown in Figure 5-1.

0.6 O

0.5
O
0.4
µτ [cm 2V -1 ]

O O
0.3 O
O O
O
O O O
O O
O
0.2 O
O

O O
0.1
O O

0.0
0 20 40 60 80 100
          Illumination [% of one Sun]            

Figure 5-1: µτ vs. Illumination % of one sun for GaAs sample.

  43  
 The GaAs µτ values show a general increase for an increase in temperature, but a

decrease for increasing illumination. The latter trend appears as ∆σ does not drop as quickly as

the generation rate at lower illumination values.

Additionally, for GaAs a conductivity floor exists around 3×103 Ω-1 m-1, where the

number of photo-generated carriers does not exceed the number of intrinsic carriers. Figure 5-2

shows an experimentally obtained curve comparable to the generic semiconductor carrier

generation curve, where a “freeze-out” occurs at lower temperatures, an “extrinsic region” occurs

around the flat, middle part, and an “intrinsic region” begins at high temperatures. While the

graph may be showing a freeze out at lower temperatures, the increase in conductivity near 300

K is not caused by intrinsic carriers, as this typically happened around 600 K. Instead, this may

be due to another dopant level, which becomes more accessible at higher temperatures. Data for

σ and µτ for GaAs can be found in the appendix.

 
Figure 5-2: Log scale of the Conductivity vs. Temperature for GaAs sample. The
missing data points for the 61% and 112% curves were caused by a data collection
error in the Keithley SourceMeter.
 
 
 
 

  44  
5.2 Tin Monosulfide (SnS)

The main purpose of this thesis is to determine the µτ of SnS and assess the technique

and reliability of the characterization tool used to acquire it.

5.2.1 Tests for sub-band-gap Light

Initial measurements of SnS samples showed higher conductivity than anticipated. In

case this was a result of sub-band-gap light exciting electrons to or from defects states within the

band-gap, multiple filters were tested. Figure 5-3 shows the results of this data.

1.0 +*
+*
Δ
+* Δ
0.5 +
*
Δ

+Δ
*
0.0
Δ
+
Log10(Δσ )

* O
-0.5 Δ O

-1.0
O

-1.5
O
-2.0 O
O

0 20 40 60 80 100
  Illumination [% of one Sun]

Figure 5-3: Log scale of the change in conductivity vs. linear scale of illumination for
three different filters and the regular filters for the set-up. Notice how below band-gap
light, represented by the 1250 nm long pass filter, does not offer a significant
conductivity contribution. All measurements were done on the SnS annealed sample.

The direct band-gap of SnS is around 1.1 eV, which corresponds to 1127 nm wavelength

light. The long pass filter at 610 nm wavelength let in much of the above band-gap light and

shows nearly the same conductivity as the unfiltered light. Similarly, the band pass filter between

315 nm and 710 nm showed only a slightly lower conductivity, as expected. The long pass filter

  45  
at 1250 nm did show an increase in conductivity. However, over all illuminations, this increase

was not more than 0.5 Ω-1m-1, which suggests that unfiltered photoconductivity is most likely

due to above-band-gap excitation.

The tests for sub-band-gap light showed that lower energy excitation is not a significant

problem for the conductivity measurements performed on this tool.

5.2.2 Noise Floor

In order to understand the sensitivity of the set-up, a noise floor was established. Multiple

dark measurements were taken at 290 K on the annealed SnS sample. Conductivity was

measured at 7.11 ± 0.106 Ω-1m-1. Several of the measurements at lower illuminations show a ∆σ

value within this error range and such, cannot be trusted as accurate data. They have been

omitted from the work presented in this thesis.

5.2.3 Effects of Annealing

For both SnS samples, µτ values were mostly on the order of 10-4 cm2 V-1, which is

around three orders of magnitude larger than what was found in the literature. Compared to each

other, the µτ values for the annealed and un-annealed SnS samples differ by about a factor of

two, where the annealed sample shows larger µτ values.

The annealing process for the SnS sample both fills S vacancies and causes grain growth

in the film. Reduced defect density will lead to an increase in the carrier lifetimes. S vacancies

in particular are known to trap majority-carrier holes and catalyze recombination in SnS [5].

All trends for µτ calculations are similar to those found for GaAs. The µτ increases for an

increasing temperature and a decreasing illumination. Data for µτ and σ for both samples can be

found in the appendix.

  46  
 5.2.4 Depth and Wavelength Dependent Generation

As mentioned before, two processes were used to determine the µτ for the samples: a

simplified approach assuming uniform generation throughout the thickness of the material and a

more complex modeling of the depth and wavelength dependence of charge carrier generation.

Table 5-1 shows the results of this dependence compared to the simplified approach for a few

temperatures at 1120 W cm-2 illumination.

Table 5-1: Data set of µτ, fitted µ, and fitted τ values for all three samples at temperatures from
230 K to 310 K. Values of µτ calculated from the uniform generation method and depth and wavelength
dependent generation method are about a factor of two different from each other. The fitted µ and fitted τ
values are not exact. Repeating calculations can cause the fitted values to change by an order of
magnitude, but their product will always be nearly exactly that of the calculated µτ value in the case for
non-uniform generation.

GaAs
Temperature [K] 230 250 270 290 310
µτ at Uniform Generation 2.84E-01 2.93E-01 2.95E-01 2.95E-01 2.84E-01
µτ at Non-uniform Generation 6.84E-01 7.03E-01 7.08E-01 7.09E-01 6.82E-01
µ Fit 9.30E+01 7.90E+01 7.20E+01 7.70E+01 7.70E+01
τ Fit 7.30E-03 9.00E-03 9.90E-03 9.30E-03 8.80E-03
SnS#un%annealed
Temperature [K] 230 250 270 290 310
µτ at Uniform Generation - 3.22E-05 4.61E-05 5.42E-05 6.19E-05
µτ at Non-uniform Generation - 7.74E-05 1.11E-04 1.30E-04 1.49E-04
µ Fit - 1.40E-01 4.07E+00 2.81E+00 9.00E-01
τ Fit - 5.48E-04 2.72E-05 4.64E-05 1.66E-04
SnS#annealed
Temperature [K] 230 250 270 290 310
µτ at Uniform Generation 2.08E-04 2.21E-04 2.29E-04 2.27E-04 2.12E-04
µτ at Non-uniform Generation 5.00E-04 5.30E-04 5.51E-04 5.46E-04 5.08E-04
µ Fit 1.60E+01 2.20E+01 1.16E+00 1.30E+01 1.30E+00
τ Fit 3.13E-05 2.36E-05 4.75E-04 4.21E-05 3.84E-04

The values calculated by including the depth and wavelength dependent generation are

about a factor of two higher than those calculated assuming uniform generation. This makes

sense, because the integrated non-generation is smaller than the uniform generation case and

since generation is in the denominator for determining µτ, Equation 9 in Chapter 3.4.1, µτ will

be higher.

  47  
 Since the non-uniform generation is calculated with an integral, the µτ values generated

are strongly dependent upon the discretization used. For example, if the number of discrete

layers, N is set to 100, this gives a layer thickness of 10 nm. Due to the exponential dependence

of generation, Equation 12 in Chapter 3.4.2, the surface of the material will generate several

times more, even several orders of magnitude more, charge carriers than further down into the

sample. For this reason, initial calculations at N = 100 overestimated generation and therefore

underestimated µτ to be less than that calculated in the simplified case. The influence of N on µτ

calculations can be seen in Figure 5-4.

O O O O O O O O
0.7 O
O O O
O
O
0.6 O
O

0.5 O
µτ [cm 2V -1 ]

O
0.4
O
Uniform Generation
0.3 -•- -•--•--•-
-•-O -•- -•--•--•- -•--•--•- -•--•--•--•- -•--•- -•- -•- -•- -•-
-•--•- -•-
O
O
0.2 O
O

0.1 O

0.0
2.0 2.5 3.0 3.5 4.0 4.5
Log10 (N)

Figure 5-4: Log scale of the number of layers modeled into which the sample is split
for modeling, N vs. µτ. All values were taken at 290 K and an illumination of 1120 W
cm-2 for the GaAs sample.

As N increases, the calculation of depth and wavelength dependent generation becomes

more precise, similar to taking a Riemann Sum with a smaller and smaller ∆x. At values around

N = 5000, the values for µτ begin to plateau around 7.1×10-1 cm2 V-1 for GaAs, 5.5×10-4 cm2 V-1

  48  
for the annealed SnS, and 1.3×10-4 cm2 V-1 for the un-annealed SnS sample. At N = 5000, layer

thickness is around 2 Å. Additionally, this shows the incredible influence of surface defects. At

the atomic scale, just the first 10 nm accounts for more than 95% of overall generation. If the

defect density at the surface of the material is large, most of these generated carriers will be lost.

5.2.5 Low illumination measurements

For the annealed SnS sample, the conductivity at lower illumination values was higher

than expected, while at very high illumination values, the conductivity was found to be

somewhat smaller than expected. This trend is similar for the un-annealed sample, although less

pronounced. A plot of σ vs. illumination for the annealed SnS sample can be seen in Figure 5-5.

200 O

O
O

150
O O O
σ [Ω -1m -1 ]

O
O O
O
O O
100 O
O O

O O
O
O
OO
50
O
O
O

0 20 40 60 80 100
Illumination [% of 1 Sun]

Figure 5-5: Conductivity vs. Illumination for SnS annealed sample. Minimal Δσ at
low illumination possibly due to sensitivity of machine and unfilled traps.

At low light intensity, the conductivity of the sample becomes larger very rapidly but

then transitions into an almost linear region. This could be due to the sensitivity of the

instruments picking up a low signal. It may also be considered, that traps will have a greater

  49  
influence on conductivity at lower light intensities because there are not as many carriers

available to both fill the traps and continue conducting through the material. At higher light

intensities, more traps are filled, allowing other carriers to move about.

Questioning the large µτ produced at low illumination requires re-examining Equation 9

from section 3.2.1. While the Δσ of the sample is decreasing with decreasing illumination values,

it is not doing so proportional to the reduction in generation rate, which is in the denominator of

this equation. The generation rate is reduced to a percent of the total generation rate equal to the

illumination percent value. This can cause the denominator to become very small, while the Δσ

doesn’t change very much, leading to a very large µτ. For this reason, it is better to consider µτ

values acquired from larger light intensities, which show a linear relationship with σ.

5.2.6 Seebeck Measurements

In an effort to determine the type of carrier dominating conductivity measurements, an

experiment to determine the Seebeck coefficient of the material was done. Unfortunately, this

did not yield tangible results.

Samples were arranged such that one side would be heated and a temperature gradient

would form. A voltage was then measured across this gradient. The sign of the measured change

in voltage would determine the dominant carrier type. Unfortunately, the measured ∆V values

were too small to overcome the noise of the experiment.

A typical Seebeck measurement in on the range of 100 µV K-1, therefore the temperature

gradient across the sample would need to be several K in order to acquire a good signal. The

length across the sample is only 8.6 mm, so acquiring a several K difference would be very

difficult, especially considering that the Si substrate of the film is a fairly thermally conductive

  50  
material. It is likely that the thermal conduction was through the substrate rather than the film

itself, making a Seebeck measurement difficult.

Further experimentation on this tool would benefit greatly from a successful Seebeck

measurement or if possible, a Hall measurement, to determine the dominant charge contribution

to the measured conductivity values.

Chapter 6. Conclusions

6.1 Discussion and Impact

These experiments have yielded a much higher µτ product than anticipated, on the order

of three to four orders of magnitude higher for all three samples. There could be a number of

factors contributing to this, but ultimately, it is believed that this measurement technique is

showing a majority-carrier µτ.

The calculations done considering depth and wavelength dependent charge carrier

generation have yielded results only about a factor of two different than those found using a

simplified approach. Additionally, the code is not too computationally intensive; several

thousand layers can be modeled in seconds. The depth and wavelength dependent generation

offers more precise measurements for µτ, but uniform generation still yields reasonable results.

6.1.1 Doping Effects

Originally, it was expected that the measured change in carrier concentrations for holes

and electrons would be equal and that the doping of the sample would not have a significant

effect on the measured values using this method. This is not the case. Doping has had a notable

effect, particularly in the GaAs sample.

  51  
 Typically, as a material becomes more doped, the carrier mobility decreases [12]. In the

case of literature values of the µτ for GaAs, heavily doping the material can decreases the µτ by

two orders of magnitude. This is likely also due to an increased chance of recombination and

resulting shorter lifetime for minority-carriers. A strong argument that the methods discussed in

this thesis measure the majority-carrier is the measured µτ values for GaAs.

In a GaAs sample doped up to 1018 cm-3, µn = 1.1×103 cm2 V-1 s-1 [12], µp = 1.8×102 cm2

V-1 s-1 [12] and τn = 1.2×10-9 s [11]. Values for τn were assumed to equal τp. With these

measurements, a doped GaAs sample would need a value of τp = 1.57×10-3 s to acquire the

measured µpτp = 2.8×10-1 cm2 V-1 at 1120 W cm-2 illumination and 310 K. It is likely that while

the increased chance of recombination due to the increased number of holes significantly reduces

the minority-carrier lifetime, the majority-carrier lifetime could actually increase because so

many of the carriers will not be recombined and will instead be collected by the contacts.

The SnS samples are doped two to three orders of magnitude less than the GaAs samples.

However, the effect on µτ is the opposite case when they are compared with each other. The

annealed SnS sample is doped an order of magnitude more than the un-annealed sample and this

leads to a factor of two increase in µτ. This improvement may just be a result of reduced bulk

defects by annealing.

6.1.2 Variations from Literature Data

From the literature, µτ products for SnS not greater than 10-7 cm2 V-1 were acquired from

two different techniques: a similar photoconductivity measurement to the one discussed in this

work of sourcing a voltage and measuring the photoconductive response of extended light pulses

[4] and an optical pump, terahertz-probe photoconductivity (TPC) measurement [22]. For the

latter technique, TPC, it is known that the minority charge carrier was dominating µτ

  52  
measurements and that the measured SnS samples were grown and processed using the same

methods as this work. The former technique was performed on cubic-crystalline SnS with a

band-gap of 1.7 eV, while the films in this work and for the TPC measurements have an

orthorhombic crystalline structure. Additionally, the SnS films with the cubic-crystalline

structure were grown in a chemical bath and likely had much smaller grains, contributing to

higher recombination rates and a lower µτ.

For both SnS samples, the effects of doping are not likely to be the cause of such a

significant difference between the τn and τp. Determining what has caused this disparity is

beyond the scope of this thesis, though it may be linked to a recombination-active defect present

in the material. Currently, very little is known about defects in SnS specific to recombination.

6.2 Further Research

Further research suggestions would explore varying annealing parameters, doping levels,

and growth techniques of SnS films. An un-doped GaAs sample could be tested on the tool and

compared with recorded literature values to further assure the quality of data collected.

Additionally, a successful Seebeck coefficient or Hall measurement would determine the

dominant contributing carrier to photoconductivity and confirm that the majority-carrier yields

the results presented in this thesis.

   

  53  
Bibliography
[1] R. H. Bube, Photoconductivity of Solids. New York: Wiley, 1960. Print.

[2] “Our Clean Energy Future.” Oil Drum – ANZ. N.o., 19Aug. 2013. Web. 20 May 2016.
http://todanz.blogspot.com/2013/08/our-clean-energy-future.html

[3] G.C. Valley, H. Rajbenbacj, and H.J. von Bardeleben. Mobility-Lifetime Product Of
Photoexcited Electrons in GaAs. Applied Physics Letters 56.4 (1990): 364-366. Inspec.

[4] A. R. Garcia-Angelmo, R. Romano-Trujillo, J. Campos-Álvarez, O. Gomez-Daza,

M.T.S. Nair, P.K. and Nair (2015), Thin film solar cell of SnS absorber with cubic
crystalline structure. Phys. Status Solidi A, 212: 2332–2340. doi:10.1002/pssa.20153240
http://onlinelibrary.wiley.com/doi/10.1002/pssa.201532405/abstract

[5] P. Sinsermsuksakul, L. Sun, S. W. Lee, H. H. Park, S. B. Kim, C. Yang, and R. G.

Gordon, Adv. Energy Mater. 4, 1400496 (1–7) (2014).
http://faculty.chemistry.harvard.edu/files/gordon/files/prasertadvenergymater2014.pdf

[6] A. W. Bott. Electrochemistry of Semiconductors. Current Separations, 17:3, 1998.

[7] B. Madore. A Knowledgebase for Extragalactic Astronomy and Cosmology. California

Institute of Technology. Retrieved May 6, 2016 from
http://nedwww.ipac.caltech.edu/level5/Sept03/Li/Figures/figure8.jpg

[8] “Performing Van Der Pauw Sheet Resistance Measurements Using the Keithley, S530
Parametric Tester.” Application Note Series. Keithley, 2012. Web. 2 Feb. 2016.
http://www.tek.com/sites/tek.com/files/media/document/resources/S530_VanDerPauwSh
eetRstnce.pdf

[9] J. T. Sullivan. Understanding the Viability of Impurity-Band Photovoltaics: A Case Study

of S-doped Si. Thesis. Massachusetts Institute of Technology, 2013.

[10] B. M. Kayes, Hui Nie, R. Twist, S. G. Spruytte, F. Reinhardt, I. C. Kizilyalli, and G. S.

Higashi. 27.6% Conversion Efficiency, a New Record for Single-junction Solar Cells
under 1 Sun Illumination. 2011 37th IEEE Photovoltaic Specialists Conference (2011): n.
pag. Web.

[11] D. M Kim, S. Lee, M. I. Nathan, A. Gopinath, F. Williamson, K. Beyzavi, and A. Ghiasi.

Minority electron mobility and lifetime in the p+GaAs base of AlGaAs/GaAs
heterojunction bipolar transistors. Applied Physics Letters, 62, 861-863 (1993),

[12] C. Fonstad. 6.012 – Electronic Devices And Circuits. Lecture 2 – Uniform Excitation;
Non-uniform Conditions (2009): 8. 15 Sept. Web. 18 Apr. 2016.
http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-012-
microelectronic-devices-and-circuits-fall-2009/lecture-notes/MIT6_012F09_lec02.pdf

  54  
[13] M. M. Niall, R. E. Brandt, V. Steinmann, R. Jaramillo, C. Yang, J. R. Poindexter, R.
Chakraborty, H. H. Park, X. Zhao, R. G. Gordon, and T. Buonassisi. Framework to
Predict Optimal Buffer Layer Pairing for Thin Film Solar Cell Absorbers: A Case Study
for Tin Sulfide/zinc Oxysulfide. Journal of Applied Physics 118.11 (2015): 115102. Web.

[14] “Solar Efficiency Limits.” Solar Efficiency Limits. Four Peaks Technologies. n.d. Web.
12 Mar. 2016. http://solarcellcentral.com/limits_page.html

[15] Arafat, Yeasir, F. M. Mohammedy, and M. M. Shahidul Hassan. Optical and Other
Measurement Technologies of Carrier Lifetime in Semiconductors. International Journal
of Optoelectronic Engineering IJOE 2.2 (2012): 5-11. Web. 25 Feb. 2016.

[16] United States. U.S. Department of Energy. Office of Integrated and International Energy
Analysis. U.S. Energy Information Administration (EIA). N.p., Sept. 2014. Web. 20 May
2016. http://www.eia.gov/forecasts/ieo/more overview.cfm

[17] N. V. Joshi, Photoconductivity: Art, Science, and Technology. New York: Marcel
Dekker, 1990. Print.

[18] C. Honsberg, and S. Bowden, PVEducation. N.p., n.d. Web. 5 Feb. 2016.
http://pveducation.org/.

[19] "Singlet Exciton Fission." Rao Group - Optoelectronics. University of Cambridge, n.d.
Web. 26 Mar. 2016. http://www.rao.oe.phy.cam.ac.uk/Research/singlet-fission

[20] M. A. Green, Solar Cells: Operating Principles, Technology, and System Applications.
Englewood Cliffs, NJ: Prentice-Hall, 1982. Print.

[21] P. Würfel, Physics of Solar Cells: From Principles to New Concepts. Weinheim: Wiley-
VCH, 2005. Print.

[22] R. Jaramillo, S. Meng-Ju, K. Benjamin. Ofori-Okai, V. Steinmann, V. Yang, K. Hartman,

K. A. Nelson, A. M. Lindenberg, R. G. Gordon, and T. Buonassisi. Transient Terahertz
Photoconductivity Measurements of Minority-carrier Lifetime in Tin Sulfide Thin Films:
Advanced Metrology for an Early Stage Photovoltaic Material. J. Appl. Phys. Journal of
Applied Physics 119.3 (2016): 035101. Web.

[23] J. Linnros, and V. Grivickas, 2002. Carrier Lifetime: Free Carrier Absorption,
Photoconductivity, and Photoluminescence. Characterization of Materials.

[24] A. Niemegeers, M. Burgelman, K. Decock, S. Degrave, and J. Verschraegen. Simulation

Programme SCAPS-1D for Thin Film Solar Cells Developed at ELIS, University of
Gent. University of Gent, Dept. Electronics and Information Systems (ELIS), n.d. Web.
http://scaps.elis.ugent.be/

  55  
   56  
Appendix

Photocurrent
OO
2 O
Log 10 (Illumination) [% of one Sun ]

O
O
OO
O
O
O
O
OO
1 O
O
O
O
O
OO
O
0 O
O
O
OO
O
O
O
-1 O
OO
O
OO
O

0 1 2 3 4
Optical Density of Filters

Figure A-1: Calibrated photocurrent in Suns with Si calibration sample on a log scale. The light
intensity of the Newport LCS-100 lamp used for the experiments was found to be 1250 W cm-2.
Filter wheel #1 is capable of attenuating light by an optical density of 0.1, 0.2, 0.3, 0.4, and 0.6.
Filter wheel #2 can do 0.5, 1.0, 2.0, 3.0, and 4.0. The two filter wheels could combine any set of
filters to attenuate from an optical density from 0 to 4.6.

  57  
Table A-1. Values for σ calculated using the uniform-generation approach. Missing values are a
result of a momentary data collection error in the Keithley SourceMeter during measurements.
No or incomplete data was taken.

GaAs Illumination [% of one Sun]

Temperature [K] 0 0.289 1.81 9.67 61.7 112
150 2.56E+03 2.74E+03 2.86E+03 3.10E+03 4.06E+03 4.98E+03
170 2.92E+03 3.07E+03 3.20E+03 3.54E+03 5.27E+03 7.47E+03
190 3.10E+03 3.22E+03 3.34E+03 3.74E+03 - -
210 3.16E+03 3.25E+03 3.37E+03 3.83E+03 8.05E+03 1.29E+04
230 3.16E+03 3.25E+03 3.38E+03 3.93E+03 8.69E+03 1.38E+04
250 3.17E+03 3.26E+03 3.41E+03 4.02E+03 9.05E+03 1.41E+04
270 3.35E+03 3.47E+03 3.65E+03 4.36E+03 9.53E+03 1.43E+04
290 4.41E+03 4.67E+03 4.92E+03 5.80E+03 1.10E+04 1.54E+04
310 8.18E+03 8.73E+03 9.08E+03 1.01E+04 1.48E+04 1.88E+04
SnS un-annealed Illumination [% of one Sun]
Temperature [K] 0 0.289 1.81 9.67 29.6 112
210 - - 4.26E+00 5.26E+00 7.30E+00 1.14E+01
230 1.03E+00 3.81E+00 4.18E+00 6.44E+00 - -
250 3.04E+00 - - 9.06E+00 1.19E+01 1.75E+01
270 4.17E+00 7.55E+00 1.03E+01 1.42E+01 1.80E+01 2.49E+01
290 8.09E+00 1.19E+01 1.52E+01 1.99E+01 2.43E+01 3.24E+01
310 1.53E+01 1.92E+01 2.30E+01 2.85E+01 3.38E+01 4.31E+01
SnS annealed Illumination [% of one Sun]
Temperature [K] 0 0.289 1.81 9.67 50.6 112
130 1.52E+01 2.38E+01 4.00E+01 6.47E+01 1.03E+02 1.28E+02
150 1.40E+01 2.12E+01 3.47E+01 5.47E+01 8.60E+01 1.08E+02
170 1.49E+01 2.20E+01 - 5.25E+01 8.11E+01 1.01E+02
190 1.76E+01 2.48E+01 - 5.52E+01 8.32E+01 1.03E+02
210 2.24E+01 3.04E+01 4.35E+01 6.23E+01 9.09E+01 1.11E+02
230 3.03E+01 3.91E+01 5.33E+01 7.32E+01 1.03E+02 1.24E+02
250 4.25E+01 5.20E+01 6.72E+01 8.84E+01 1.20E+02 1.42E+02
270 6.26E+01 7.23E+01 8.80E+01 1.10E+02 1.43E+02 1.65E+02
290 9.49E+01 1.04E+02 1.19E+02 1.41E+02 1.74E+02 1.97E+02
310 1.44E+02 1.51E+02 1.64E+02 1.84E+02 2.16E+02 2.39E+02

  58  
Table A-2. Values for µτ calculated using the uniform-generation approach. Missing values are a
result of a momentary data collection error in the Keithley SourceMeter during measurements.
No or incomplete data was taken

GaAs Illumination [% of one Sun]

Temperature [K] 0.289 1.81 9.67 61.7 112
150 1.82E+00 5.04E-01 1.66E-01 7.32E-02 6.49E-02
170 1.54E+00 4.74E-01 1.94E-01 1.15E-01 1.22E-01
190 1.20E+00 4.03E-01 1.98E-01 - -
210 9.46E-01 3.51E-01 2.08E-01 2.38E-01 2.62E-01
230 9.01E-01 3.61E-01 2.36E-01 2.69E-01 2.84E-01
250 9.31E-01 4.00E-01 2.67E-01 2.87E-01 2.93E-01
270 1.23E+00 4.98E-01 3.13E-01 3.01E-01 2.95E-01
290 2.62E+00 8.38E-01 4.30E-01 3.19E-01 2.95E-01
310 5.74E+00 1.50E+00 5.93E-01 3.23E-01 2.84E-01
SnS un-annealed Illumination [% of one Sun]
Temperature [K] 0.289 1.81 9.67 29.6 112
210 - - - - -
230 2.40E-03 4.34E-04 1.40E-04 - -
250 - - 1.55E-04 7.46E-05 3.22E-05
270 2.92E-03 8.47E-04 2.60E-04 1.17E-04 4.61E-05
290 3.30E-03 9.80E-04 3.04E-04 1.37E-04 5.42E-05
310 3.31E-03 1.05E-03 3.40E-04 1.56E-04 6.19E-05
SnS annealed Illumination [% of one Sun]
Temperature [K] 0.289 1.81 9.67 50.6 112
130 7.43E-03 3.42E-03 1.28E-03 4.32E-04 2.51E-04
150 6.29E-03 2.86E-03 1.05E-03 3.55E-04 2.09E-04
170 6.14E-03 - 9.72E-04 3.27E-04 1.92E-04
190 6.29E-03 - 9.73E-04 3.24E-04 1.89E-04
210 6.89E-03 2.90E-03 1.03E-03 3.38E-04 1.97E-04
230 7.61E-03 3.17E-03 1.11E-03 3.60E-04 2.08E-04
250 8.23E-03 3.40E-03 1.19E-03 3.82E-04 2.21E-04
270 8.35E-03 3.50E-03 1.23E-03 3.97E-04 2.29E-04
290 7.57E-03 3.28E-03 1.18E-03 3.90E-04 2.27E-04
310 6.11E-03 2.75E-03 1.04E-03 3.57E-04 2.12E-04

  59