A nanomaterial and nanophysics 9th semester project – Aalborg University, fall 2008

Characterization of Aluminium Doped Zinc Oxide

Project written by

Nina Brodam Lorentzen and

Morten Vesterager Madsen 2008

Institute of Physics and Nanotechnology Aalborg University - Nanotechnology

Theme: Applications of Nano Materials Title: Characterization of Aluminium Doped Zinc Oxide Thin Film Project Period: P9, 2/9-08 - 19/12-08 Project Group: NM3, Gr. 4.215 Groupmembers: Nina Brodam Lorentzen

Morten Vesterager Madsen Supervisor: Kjeld Pedersen

Number of Copies: 5 Number of Pages: 34 Number of Appendices: 1 Total Number of Pages: 39

Synopsis: The report is written as the first part of a 9th - 10th semester thesis. The report focuses on the possibilities for production and characterization of transparent and conducting zinc oxide films. Chemical vapour deposition and sputter depositions methods are introduced and discussed. On the basis of this discussion, magnetron sputtering is chosen as deposition method. Characterization techniques including transmittance, four point probe, hot probe, photoluminescence, raman, and ellipsometry are presented and discussed. Lastly a roadmap is presented outlining the future development of the project.

Contents

1 Introduction

5

2 Production of Zinc Oxide Films

7

2.1

CVD Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7

2.2

Sputtering Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8

3 Characterization of ZnO Films

10

3.1

Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

3.2

Resistivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

3.2.1

Four Point Probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11

3.2.2

Hot Probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

Film Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.3.1

Profiler Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.3.2

Ellipsometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

3.4

Photoluminescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18

3.5

Raman Characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

3.3

4 End Product 4.1

Material Properties and Surface Texture . . . . . . . . . . . . . . . . . . . . . .

29 29

5 Roadmap

33

A Production of Standard Sample for Probe Distance Settings

35

4

1

Introduction The photovoltaic effect was discovered in 1839 by Edmond Becquerel, and is defined as the emergence of an electric voltage between two electrodes attached to system upon shining light onto it. The photovoltaic effect remained a scientific phenomenon and laboratory curiosity, with few device applications, until the mid-1950s when scientists attempted to power satellites with photovoltaic solar cells. Soon after these were indispensable to satellites and telecommunications equipment in remote areas. The oil crises of the 1970s and the general awareness of the limitations of fossil fuels lead to ambitious programs in search for alternative energy sources. In recent years particularly the atmospheric concentration of greenhouse gasses, such as CO2 , have fuelled the interest for alternative energy. The energy, from the sun, that reaches the earth’s surface in one day exceeds humankind’s total energy requirements for forty years [Nelson, 2003]. As solar cells convert sunlight directly into electricity through the photovoltaic effect, it is recognised to be as one of the most promising solutions. The present photovoltaic market is dominated by so called first-generation photovoltaic product based on the use of silicon wafers. The advantage of wafer based crystalline solar cells is relatively high efficiencies between 12 and 16% for commercial solar cells and laboratory records of 24.4% [Shah et al., 1999]. These cells further excel with excellent stability, reliability, and no deterioration over several decades. The main disadvantage is the manufacturing cost. The use of highly purified silicon wafers cut from a silicon boule is a bottleneck for the end product cost. Thus, high cost for first-generation technology is unavoidable. For this reason, a strongly increasing effort in the field of photovoltaics has been moving forward the development of cost-effective photovoltaic cells in the recent years. To propose an alternative to crystalline silicon solar cells one must keep in mind that at least one shortcoming might be included. Theoretically high efficiencies can be obtained using exotic fabrication techniques and materials. Alternatively cheaper solar cells may be manufactured at a dispense of efficiency. A move toward second generation solar cells means use of thin films consisting of amorphous silicon. The main advantage of this kind of solar cells is simpler production steps, less material consumption, possibility for roll to roll manufacturing. The main drawback is an expected lower efficiency and potential problems with deterioration during operation. The main challenge for realizing low cost solar cells is to reduce the cost of material and processing as well as to retain a high efficiency potential. One solution is to optimise the

5

1. INTRODUCTION

transparent and conductive oxide which is used as front electrode. Present TCOs are made of indium tin oxide. However the limited supply of indium dictates the price of the end product. This is a particular problem for second generation cells which main attraction is low cost. This particular work focuses on zinc oxide as the transparent conductive layer for thin film solar cells of p-i-n type based on layers of doped amorphous silicon. The TCO layer must be made with techniques compatible with the previously deposited layers of amorphous silicon and the polymer film. This creates a series of demands on the deposition technique. This report is meant to review production techniques and characterisation techniques for depositing, doping and characterizing of ZnO films.

Project Limitation The goal of this report is to summarize and present production techniques and characterization techniques for fabricating transparent and conducting doped ZnO films, for use in thin film solar cells. The project is divided into three part objectives, namely the production, characterization, and finally a chapter dealing with the end product, namely ZnO films. The report will be used to review and determine the optimum production techniques and characterization techniques for ZnO.

6

2

Production of Zinc Oxide Films Production of ZnO films can be achieved through a diverse range of production techniques including ˆ Physical methods (sputtering , evaporation , and pulsed laser deposition ) . ˆ Chemical methods (thermal CVD , plasma CVD , MOCVD , sol-gel , and chemical bath deposition , atomic layer deposition).

This chapter deals with the two main categories and focuses on sputtering and CVD processes.

2.1

CVD Processes

Chemical vapour deposition is a chemical process used to produce high-purity, highperformance solid materials. In a typical CVD process, the wafer is exposed to one or more volatile precursors, which react and/or decompose on the substrate surface to produce the desired deposit. Frequently, volatile byproducts are also produced, which are removed by gas flow through the reaction chamber. [Cambell, 2001] The simplest CVD process is thermal CVD, which forms the basis for most epitaxial growth in IC manufacturing. Modifications of the simple thermal CVD system provides alternative energy sources such as plasmas to drive the chemical reactions. This allows the the deposition to occur at low temperatures. Another more advanced process is MOCVD or metal organic CVD. A proposed design of such a system is shown in figure 2.1. The method is based on the use of a metal organic precursor gas such as diethylzinc or dimethylzinc. The main advantage of using such a gas is that it allows deposition directly at the surface. It is possible to mix the gas with other precursors, allowing doping of the sample. Further oxygen can be introduced into the mix, preferably in the form of H2 O or CO2 which does not react at low temperatures. The method allows high deposition rates, for instance 0.85nm/s as reported by [Groenen and van de Sanden, 2000]. Further deposition on large targets is possible. Substrate temperatures of 20 − 300◦ C has been effectively demonstrated and a pressure range from a few miliTorr to atmospheric pressure is possible. At low pressure the process is limited 7

2. PRODUCTION OF ZINC OXIDE FILMS

Argon gas

Shower head Substrate heater DEZ bubbler Pump Doping source Ex.: TMA Oxygen source Ex.: CO2

Figure 2.1: Proposed MOCVD system. The system allows the mixture of reaction gasses before they reach the reaction chamber. Diethylzinc is introduced to the chamber from a bubbler and mixed with inert argon gas. Then oxygen is introduced in the form of ex CO2 and further mixed with a doping source ex trimethylaluminium. Controlling the flow of the individual gasses determine the composition of the deposited layer. by the kinetics of the reactions on the sample, transport of precursors to the surface is fast. At atmospheric pressure the reactants cannot reach the sample as fast. A last thing to consider is the use of a plasma. The plasma adds additional energy and thereby increases the reaction rate at the surface. This further allows deposition at lower temperatures, reducing the chance of a premature reaction between the production gasses. In summary MOCVD is a technique allowing for production scale deposition of ZnO. The main disadvantage of the process is the precursor gas; diethyl zinc. Diethyl zinc is a highly volatile colourless liquid that reacts violently with water and easily ignites upon contact with air. This makes the process more complicated because of increased safety precautions.

2.2

Sputtering Processes

Sputtering is a process where atoms are ejected from a solid target material due to bombardment of the target by energetic ions. There are generally two distinct types of sputtering: DC and AC sputtering. As the coating material is passed into the vapour phase by momentum exchange caused by energetic particle impact any material is a candidate for coating. This would not have been the case if the process depended on thermal or chemical processes. The AC sputtering technique typically operates at 13.56M Hz. The concentration of ions and radicals in this simple capacitative discharge is however a small fraction of the total gas. A way of boosting the efficiency of the technique is forming enriched plasmas. One way of archiving this is to utilize a magnetic field. The application of a magnetic field in a plasma causes the electrons to spiral around the direction of the magnetic field lines. This orbital motion of the electrons increase the probability of collisions with neutral spices and thus the creation of ions. With a magnetic field the Lorentz force will deflect the motion of the electrons perpendicular to both the direction of the velocity and the magnetic field ~ F = q~v × B. 8

(2.1)

2.2. SPUTTERING PROCESSES

With |~v | constant the field will induce a circular motion. Ions move through the field with only minor deflections because of their relatively high mass. Electrons however will follow a helical path making the path of the electron many times larger than without the field. In a plasma with a certain mean free path this clearly increases the change for impact ionisation. Magnetron sputtering is used to deposit a wide range of coatings. The basic sputtering process has been known for many years. At first, the technique used was, the balanced magnetron. Balanced magnetron differentiates from the unbalanced magnetron by the relation between the strength of the magnets. The design of the magnetron contain magnets arranged in such a way that one pole is positioned at the central axis. The second pole is formed by a ring of magnets around the outer edge of the target. Figure 2.2(a) shows a balanced magnetron.

N

Substrate

Substrate

Plasma

Plasma

S (a)

N

N

S

N

(b)

Figure 2.2: (a) Conventional or balanced magnetron. In this arrangement ion current densities of less than 1mA/cm2 is expected. (b) An unbalanced magnetron where ion current densities are expected around 2 − 10mA/cm2 . The unbalanced magnetron often have a configuration where the outer ring of magnets are strengthened with respect to the inner magnet. In this case the field lines are closed between the central point and the outer poles, however, with some directed towards the substrate, see figure 2.2(b). Consequently some secondary atoms are able to follow these field lines. This leads to a less confined plasma at the target area that is allowed to flow out toward the substrate. An unbalanced magnetron provide a high flux of coating atoms compared to a basic sputtering source and also acts as an very effective ion source. Compared to the balanced magnetron with ion current densities below 1mA/cm2 the unbalanced magnetron gives densities around 2 − 10mA/cm2 . The ion current at the substrate is directly proportional to the target current. The deposition rate is also directly proportional to the target current. Ion to atom arrival ratio at the substrate remains constant with increasing deposition rate. Using sputtering one option for producing films of ZnO is to use a zinc target. As the target is conductive a DC plasma can be utilized. This scheme requires the introduction of oxygen during production and control of stoichiometry can be difficult. Another option is to use a ZnO target, however, this rules out the use of DC plasma, since the target is non conductive. Consequently an AC plasma must be used. The ZnO targets can be supplied as either intrinsic or containing a dopant. For the continued project magnetron sputtering has been chosen as production method. The sputter chamber and magnetron is delivered by Polyteknik A/S and expected operational ultimo December 2008.

9

3

Characterization of ZnO Films The main feature of merit for TCO films are the ratio between transparency and resistivity of the films. The higher this ratio is, the better the films generally are. Therefore they are the most important features to characterize. The first two sections deals with methods to determine these features. Furthermore for the production it is important to be able to determine the thickness of the films easily, and preferably even during production or at least without damaging the sample. Finally methods for determining the quality of the films are introduced in the form of photoluminescence and raman characterization.

3.1

Transmission

Transmission is the percentage of energy passing through a system relative to the amount that passes through the reference. Transmission is also used to show the portion of light reflected from a sample. The transmission is expressed as a percentage relative to a standard substance. The transmission measurements of the produced ZnO films on quarts wafers, is made on an Ocean Optics ISSUV-VIS spectrometer. The software, included in the Ocean Optics ISSUVVIS spectrometer, calculates the transmission percentage by the following equation

%T (λ) =

S (λ) − D (λ) · 100%, R (λ) − D (λ)

(3.1)

where S is the sample intensity at wavelength λ, D is the dark intensity at wavelength λ, and R is the reference intensity at wavelength λ [OceanOpticsInc., 2005].

3.2

Resistivity

Electrical resistivity is a measure of how strongly a material opposes the flow of electric current. Low resistivity indicates a material that readily allows the movement of electrical charge. The resistivity of a semiconductor is defined by

ρ=

10

1 [Schroder, 1990]. q (nµn + pµp )

(3.2)

3.2. RESISTIVITY

n and p are free electron and hole concentrations, and µn and µp are the electron and hole mobilities. Thereby the resistivity is both depended on the carrier concentrations and the mobilities. For highly doped materials the majority carrier concentration is by definition much higher than the minority carrier concentration. In those cases, generally only the majority carrier concentration needs to be known.

3.2.1

Four Point Probe

The conductivity of a material can be measured by passing a current through and measuring the potential drop across the surface. The simplest application of this, is to use a two point probe, where each probe serves as a current source and a voltage probe. This scheme has, however, a major drawback. The contact resistance will be measured along with the resistance of the sample. The contact resistance arises when the metal probe is in contact with a semiconductor surface. Such a system may give rise to a Schottky barrier with a large non ohmic contact resistance. This contact resistance may be large enough to dominate the measurements. Figure 3.1(a) shows a two point probe system. The total resistance is given by

U I

(3.3)

U − 2Rwire − 2Rcontact . I

(3.4)

Rtotal = 2Rwire + 2Rcontact + RDU T = RDU T =

V

V

I

I

I

I

Rwire

Rwire

Rwire

Rwire

Rwire

Rwire

Rcontact

Rcontact

Rcontact

Rcontact

Rcontact

Rcontact

I

I

I

RDUT

RDUT

(a)

(b)

Figure 3.1: (a) A two point probe system, and (b) a four point probe system. The wire resistances, and contact resistances are indicated on both figures. RDU T is the resistance of the device under testing. Where RDU T is the resistance of the device under testing. Rcontact is the contact resistance at each probe/semiconductor contact. Thus it is seen that the contact and wire resistance is a part of the measurement. A way to circumvent this problem is to estimate the contact resistance and subtracting it from the result. However, this may prove difficult. Another more elegant solution is to use a four point probe system. By passing the current through the outer two probes and measuring the potential drop between the inner probes, the resistance of the sample becomes independent of the contact resistance. A schematic of this is 11

3. CHARACTERIZATION OF ZNO FILMS

shown in figure 3.1(b). The probes are usually placed in line with equal probe spacing, other probe configurations are possible, but outside the scope of this text. If the input impedance of the voltmeter is assumed to be much larger than RDU T no current of significance will flow through the voltmeter. Therefore the measured potential drop is a result of the known current flowing through RDU T only. The potential drop over the contact resistance and the wire resistance are negligible, as the value of the voltmeter impedance is 109 Ω in our case. Measurement of the resistance of a bulk material yields

ρ = 2πsF

V [Schroder, 1990, eq.1.10]. I

(3.5)

Here the probe spacing plays an important role. F is a correction factor correcting for edge effects, thickness effects, and probe placement effects. For very thin samples the resistance is

ρ = 4.532t

V . I

(3.6)

The sheet resistance is found by measuring the ratio of the voltage drop to the forced current multiplied by a geometric correction factor. In the case of infinite films on an insulating substrate F = π/ ln (2) = 4.532 [Schroder, 1990, p. 5]. Thin layers are often characterized by their sheet resistance expressed in units of ohms per square. The sheet resistance is given by

ρs = 4.532

V [Schroder, 1990, eq.1.16]. I

(3.7)

The correction factor depends on the probe geometry and the ratio of the probe spacing and the thickness. In order to keep the validity of the correction factor it is therefore necessary to use large samples compared to the probe spacing, to make the measurement near the centre of the sample, and to have the spacing between the individual probes be as equal as possible. Further the underlying substrate must optimally be insulating or be of much higher resistivity than the layer to be measured. [Cambell, 2001, ch. 3.6] The main drawbacks of the four point probe method lies in its need for physical contact with the sample. The method does damage to the sample under testing. This damage may not be very large, but it rules the technique out as a test used for device fabrication. The actual setup used for the four point probe measurements is build around a SUSS MicroTech PM5 probe station in a SUSS SE1000 MicroTech ShieldEnclosure. The probe station allows four needles to be placed individually on the sample, with micrometer screws. To ensure acruate placement a standard sample was fabricated with indicated distances at 2mm, 1.5mm, 1mm, 250µm, and 125µm. The manufacturing of this standard sample is explained in appendix A. The four point probe measurements were made with a Keithley Instruments 2182a nanovoltmeter and a Keithley Instruments 6221 current source. A program in C# was written to control and read data from the two instruments for automated IV curve recording. A series of measurements were carried out on samples coated with gold in tabletop sputter coater. Eight different samples were made with thicknesses in the range from 10 to 300nm. The resistivity was measured for with probe distances of 125µm, 250µm, 500µm, and 1000µm, as shown on figure 3.2. The expected result was that the resistivity was independent of the 12

3.2. RESISTIVITY

250

Resistivity [[Ωcm]

200

150 1 mm 500 um 100 250 um 125 um 50

0 0

50

100

150

200

250

300

Film Thickness [nm]

Figure 3.2: The resistivity of eight gold films with thicknesses varying from 10nm to 300nm. The resistivity was measured for four different probe distances, namely 125µm, 250µm, 500µm, and 1mm. probespacing and the thickness of the sample. However, the results show strong fluctuations. There are number of reasons for this. Firstly the films were sputtered in a table top sputter coater which did not produce uniform coatings. This means that fluctuations in film thickness were to be expected and that the stated thickness may not be entirely correct. Secondly the correction factor in equation 3.5 is only valid for films of uniform thickness and infinite extent. In reality when the films extend is large compared to the spacing this assumption is acceptable. In the specific case this was not the case. Every sample were scratched in order to carry out profiler measurements, and this scratch significantly reduced the size of the sample. In conclusion it is important to conduct measurements near the centre of the sample, and be careful of scratches in the surface.

3.2.2

Hot Probe

After the resistivity has been measured it will be valuable to determine the conductivity type of the semiconductor. A relatively simple technique utilizes the four point setup by heating the sample and determining the sign of the thermal emf or Seebeck voltage generated by a temperature gradient. An AC current flows between the probes 1 and 2 and cause a joule heating of the semiconductor. Between probes 3 and 4 the Seebeck voltage is generated by the diffusion of the thermally generated carriers from the hot region of the sample to the cold region, see figure 3.3(b). A measurement of the potential drop across probe 3 and 4 can reveal the carrier type. The majority carrier currents for n- and p-type materials are

dT dx dT Jp = −qpµp Pp , dx

Jn = −qnµn Pn

(3.8) (3.9) 13

3. CHARACTERIZATION OF ZNO FILMS

Soldering Iron

HI

A

HI

LO

1

(a)

AC

LO

HI

2

3

V

LO

4

(b)

Figure 3.3: The thermoelectric method. In (a) a soldering iron acts as heating source and in (b) two probes act as a heating source. An alternative here is to utilize another source of heat, for example a soldering iron. The last two probes measures the Seebeck voltage generated.

Pn < 0 and Pp > 0 are the differential thermoelectric power when no external electric field is applied [Sze, 2007, eq. 134-135]. Since the region around probe 1 and 2 is hot a temperature gradient will be established and dT /dx > 0. Then the electron current in an n-type sample flows from left to right in the geometry of figure 3.3 and in the opposite direction for p-type samples. This type of measurement is effective for samples with resistivity in the range of 10−3 to 103 Ω · cm. An alternative to using two of the probes for heating is to use a heater directly. A simple example is the use of a soldering iron. This technique has be tested successfully with ITO samples.

3.3

Film Thickness

The production of TCO for solar cell applications require a fixed thickness of the films. Therefore it is important to be able to determine the thickness of the films easily, and preferably during production or at least without damaging the sample.

3.3.1

Profiler Measurements

The profiler works in principle like a atomic force microscope. It scans a needle mounted on a cantilever over the surface and measures the force on the needle. The only difference to a AFM is that the profiler only scans one direction. The profiler used in this project was a XP - 2T M Stylus Profiler from Ambios Technology Inc. To make the profile measurements of the films a scrape was made with a scalpel and the needle was scanned over this step. 14

3.3. FILM THICKNESS

3.3.2

Ellipsometry

Ellipsometry is an excellent method for determining film thickness, and it is useful for measuring the complex index of refraction of a given material. The main attraction of the technique is that it is generally non-invasive and non-destructive. Ellipsometry makes use of the fact that the polarization state of light may change when the light beam is reflected from a surface and thereby making it possible to deduce information about the film properties. The main advantage of ellipsometry is that, in opposition to other optical techniques that are inherently diffraction limited, ellipsometry exploits phase information and the polarization state of light, and can achieve ˚ angstr¨om resolution. An ellipsometry measurement records the parameters Ψ and ∆ given by

rp = tan (Ψ) ei∆ [Azzam and Bashara, 1977, eq.3.67], rs

(3.10)

tan (Ψ) is the amplitude ratio upon reflection, and ∆ is the phase shift. Since ellipsometry is measuring the ratio of two values rather than the absolute value of either, it is a very robust, accurate, and reproducible method. However, as ellipsometry is an indirect method, where the measured Ψ and ∆ cannot be converted directly into the optical constants of the sample, a model analysis must be performed. Direct conversion into real data is only possible in simple cases of isotropic, homogeneous and infinitely thick films. In all other cases a layer model must be established, which considers the optical constant and thickness parameters of all individual layers of the sample including the correct layer sequence. Then using an iterative procedure unknown optical constants and or thickness parameters are varied, and Ψ and ∆ values are calculated using the Fresnel equations for rs and rp . The calculated Ψ and ∆ values, which best matches the experimental data, provide the optical constants and thickness parameters of the sample. To obtain the ellipsometry parameters one may use rotating angle ellipsometry and meassurering the intensity at at least three different angles. In order to calculate the physical properties such as the complex index of refraction or the film thickness a model must be set up in order to link the ellipsometric parameters to the actual physical properties. The ellipsometer used for experiments in this project is the spectroscopic ellipsometer SE 850 from Sentech. The SE 850 is computer controlled via the Sentech software Spectraray2. The ellipsometer has a range of wavelength from 350 to 850nm.

Infinite Film For a sample with an infinite film thickness the refractive index, n1 , can directly be calculated from the ellipsometric parameters. Figure 3.4 shows the basic setup. The ellipsometry equation, equation 3.10, is

rp = tan (Ψ) ei∆ , rs

(3.11)

which contains ellipsometric parameters and the Fresnel reflection coefficients given by 15

3. CHARACTERIZATION OF ZNO FILMS

Figure 3.4: Reflection and transmission of an incident light wave at an surface boundary or infinite film.

n0 cos (θ0 ) − n1 cos (θ1 ) [Klein and Furtak, 1986, eq.2.62a], n0 cos (θ0 ) + n1 cos (θ1 ) n0 cos (θ1 ) − n1 cos (θ0 ) rp = [Klein and Furtak, 1986, eq.2.70b]. n0 cos (θ1 ) + n1 cos (θ0 )

rs =

(3.12) (3.13)

The only unknown is the angle θ1 , which can be determined using Snell’s law. n1 can then be evaluated as

s n1 = n0 sin (θ0 )

tan (Ψ) ei∆ − 1 1 + tan (Ψ) ei∆

2

tan2 (θ0 ) + 1.

(3.14)

Calculation of Film Thickness Another case of importance in ellipsometry is an optical system consisting of an ambient-filmsubstrate system as shown in figure 3.5. It is possible to determine the thickness of the film in such a system, if the refractive indices are known. In order to do this it is essential to link the optical parameters to the ellipsometric parameters in equation 3.10 to the actual physical parameters. This time the reflection coefficients are given by

Rp =

r01,p + r12,p e−i2β [Azzam and Bashara, 1977, eq.4.37], 1 + r01,p r12,p e−i2β

(3.15)

Rs =

r01,s + r12,s e−i2β [Azzam and Bashara, 1977, eq.4.38], 1 + r01,s r12,s e−i2β

(3.16)

where r01 and r12 are the reflection parameters for the ambient film and the fim substrate respectively. β is the phase angle containing the thickness of the film and is given by 16

3.3. FILM THICKNESS

Figure 3.5: A film substrate optical system. The system consists of three parts; the ambient environment, the film, and the sample. Modified from [Azzam and Bashara, 1977]

d β = 2π n1 cos (θ1 ) [Azzam and Bashara, 1977, eq.4.31]. λ

(3.17)

The ellipsometric formula, equation 3.10, becomes Rp = tan (ψ) ei∆ . Rs

(3.18)

To determine the relation between the physical parameters and the ellipsometric paramters the reflection ratio must be evaluated Rp r01,p + r12,p e−i2β 1 + r01,s r12,s e−i2β = Rs 1 + r01,p r12,p e−i2β r01,s + r12,s e−i2β
2 r01,s r12,s r12,p e−i2β + (r12,p + r01,p r01,s r12,s ) e−i2β + r01,p = 2 r01,p r12,p r12,s (e−i2β ) + (r12,s + r01,s r01,p r12,p ) e−i2β + r01,s =

AX 2 + BX + C . DX 2 + EX + F

(3.19) (3.20) (3.21)

This equation can be rewritten into 

     Rp Rp Rp 2 D−A X + E−B X + F − C = 0. Rs Rs Rs

(3.22)

The equation is a complex quadratic equation which can be solved in X. Now the film thickness can be expressed as X, given by X = e−i2β , and β, given by β = 2π λd n1 cos (θ1 ).

d=

i ln (X) λ . 4πn1 cos (θ1 )

(3.23)

Only one solution can be valid for the film thickness, which should be real and positive. 17

3. CHARACTERIZATION OF ZNO FILMS

3.4

Photoluminescence

Photoluminescence, PL, is the spontaneous emission of light from a material under optical excitation. PL is used to characterize a number of material parameters and providing electrical characterization. The emission spectrum can be used to identify surface, interface, and impurity levels. The intensity of the PL signal provides information on the quality of the surfaces and interfaces. Since the sample is excited optically, electrical contacts and junctions are unnecessary and materials of high resistivity pose no practical difficulty. For this reason photoluminescence is a widely used experimental method for the study of semiconductors. Photoluminescence is a technique in which a substance absorbs photons and then re-emits photons. It relies on the creation of electron-hole pairs by incident radiation and subsequent radiative recombination photo emission. Any non radiative recombinations emit no light and are therefore not detected. PL provides the transition energies, which can be used to determine the electronic energy levels. Since the absorption depends strongly on the energy, the penetration depth of the incident light will depend on the excitation wavelength. Lasers are the instruments of choice for photoluminescence excitation as they are monochromatic, intense and readily focused. Further the excitation energy selects the initial excited state in the experiment. The initial photoexcited state is short lived and relaxation to within kT of the lowest available state is a fast process usually orders of magnitude faster than recombination. Therefore the excitation energy may, not play an all that important role since ordinary PL emission only reveals the lowest energy states. Often if the bandgab energy is exceeded the energy does not play a significant role. On the other hand, the excitation intensity will influence the result of the photoluminescence experiment. The intensity determines the density of photoexcited electrons and holes, which governs the behaviour of these carriers. The electron-hole recombination mechanism has a distinct dependence on carrier density. For example the number of interface and impurity states are finite, and recombination at these sites will saturate at high excitation. In addition the photoexcited carriers can alter the distribution of interface states. Impurities gives rise to peaks in the PL spectrum at transitions energies in the material. Identification of these impurity energies is precise as the energy resolution of PL is high. The emission properties of ZnO films are strongly dependent on growth conditions of the films. Emission in the visible spectrum have been reported in most microcrystaline ZnO films grown at low temperature. The emission is normally contributed different intrinsic defects, such as oxygen vacancies, zinc vacancies, zinc interstitials, and oxygen interstitials. The exact origin of the visible emission is not yet fully understood and is debatable. The following will contain the expected transition energies in ZnO found in literature. First and foremost a band is expected around the bandgap energy, resulting from electron transfer from valance band to conduction band. Further a number of defects resulting primarily from non stoichiometry of the films is expected. Table 3.1 lists the energies of the different transition energies reported in literature. The table only takes into account transitions observed in intrinsic ZnO. Using the table peaks in a photoluminescence spectrum can be assigned to the different transitions. It is, however, important to keep in mind that further transition energies may be possible due to unaccounted mechanisms. 18

3.4. PHOTOLUMINESCENCE

Description Fundamental bandgap Neutral zinc vacancy

Value 3.31eV ; 3.37eV 3.06eV ; 3.07eV

Zni

Neutral zinc interstitial

0.47eV

VO

Neutral oxygen vacancy

1.62eV

Oi

Oxygen interstitial

2.47 and 2.97eV

OZn

Oxygen antisite defect

2.36 − 2.38eV ; 1.87 and 2.37eV

VO Zni Complex of an oxygen vacancy and zinc interstitial

0.97 and 2, 17eV

VB VZn

Source Determined by [Madsen and Brodam, 2008]; Theoretical (pseudo po[Ellmer et al., 2008] tential) ; Experimental [Lima et al., 2001]; Experimental (excitation [Xu et al., 2003] spectrum); Theoretical (FP-LMTO, corrected to actual bandgap) [Xu et al., 2003] Theoretical (FP-LMTO, corrected to actual bandgap) [Xu et al., 2003] Theoretical (FP-LMTO, corrected to actual bandgap) [Xu et al., 2003] Theoretical (FP-LMTO, corrected to actual bandgap) [Lin et al., 2001]; Experimental (estimate [Xu et al., 2003] concentrations of intrinsic defects post annealing and evaluation PL spectra); Theoretical (FP-LMTO, corrected to actual bandgap) [Xu et al., 2003] Theoretical (FP-LMTO, corrected to actual bandgap)

Table 3.1: The values are primarily based on Full-Potential Linear Muffin-Tin Orbital calculations by [Xu et al., 2003]. These energies are all multiplied by a correction factor to take into acount that the calculation underestimate the fundemental bandgab. The values are, however, supported by experimental results obtained by [Lin et al., 2001] and [Lima et al., 2001]. Energies are given as distance from the conduction band.

19

3. CHARACTERIZATION OF ZNO FILMS

Photoluminescence Setup The photoluminescence setup centres around a fibre photometer, see figure 3.6. By measuring the intensity of the scattered light from the sample a photoluminescence spectrum can be obtained if the directly scattered light can be deducted. This is achieved with a filter. A good practice is to avoid the directly reflected laser beam. This setup is build around a 405nm laser. The particular laser is chosen as it has been demonstrated that 405nm is sufficient to induce green luminescence, even though the energy is less than the bandgap.

Sample Laser Lens

Filter

Detector Figure 3.6: The arrangement is build on an optic table. After the 405nm laser the sample is placed. After the sample a lens is placed. The purpose of this lens is to enlarge the signal from the sample. A fibre photometer, used for detection, is placed in the lens’ focal point. Between the lens and the fibre photometer a filter is placed. The filter ensures that the signal measured by the fibre photometer is the directly scattered light from the sample. Figure 3.7 shows a photoluminescence spectrum taken from a ZnO sample. The sample was borrowed from Kjeld Pedersen. The spectrum shows clear luminescence around 470 − 650nm. As the sample consists of both film and nanowires it was difficult to ascribe the transition energies to specific defects. The experiment demonstrated the ability to obtain photoluminescence spectra from films of ZnO with a 405nm laser. A PL spectrum made under similar conditions were reported by [Jeong et al., 2003], see figure 3.8(a). Their ZnO film were made by rf magnetron sputtering and spectra were obtained for varying oxygen to argon ratios. Their findings suggested that PL spectra are more dependant on film stoichiometry than film crystallinity. Films grown in more oxygen rich environment show improved stoichiometry with less oxygen vacancies and zinc interstitials. This indicates that the green emission originates from the oxygen vacancy or Zn interstitial related defect. The spectra obtained by [Jeong et al., 2003] differers from the spectra seen in figure 3.7 by the fact that this ZnO sample contained nanostructures, primarily nanowires. As demonstrated by [Djurisic and Leung, 2006] different nanostructures yield different PL spectra, see figure 3.8(b).

20

3.4. PHOTOLUMINESCENCE

300.000

250.000

Intensity [counts]

200.000

150.000

100.000

50.000

0 350

450

550

650

750

850

950

Wavelength [nm]

Figure 3.7: Photoluminescence spectrum of a ZnO sample, borrowed from Kjeld Pedersen. The spectrum shows clear luminescence around 470 − 650nm. As the sample is believed to primary consist of wire it is difficult to ascribe the transition energies to specific defects.

(a)

(b)

Figure 3.8: (a) Photoluminescence spectrum for ZnO films prepared by rf magnetron sputtering at various oxygen to argon ratios [Jeong et al., 2003]. (b) PL spectra for different nanostructures: 1 - thetrapods, 2 - needles, 3 - nanorods, 4 - shells, 5 - highly faceted rods, and 6 ribbons [Djurisic and Leung, 2006].

21

3. CHARACTERIZATION OF ZNO FILMS

3.5

Raman Characterisation

Raman spectroscopy is a fast, non-destructive and convenient diagnostic tool for characterizing the structural properties of films as well as any stress or strain that may be present in films. This method is based on the phenomenon of inelastic scattering of light. The incident light interacts with phonons or other excitations in the system, resulting in the energy of the photons being shifted up or down. The shift in energy gives information about the phonon modes in the system. The Raman spectrum represents the intensity of the shifted or Raman-scattered light as a function of the Raman shift expressed in absolute wavenumber shift.Infrared spectroscopy yields similar, but complementary, information.

Raman Spectra in ZnO The wurtzite lattice of ZnO can be considered as two interpenetrating hexagonal close-packed lattices, as shown on figure 3.9. It can be seen that layers of zinc atoms alternate with layers of oxygen atoms. The wurtzite structure has a tetrahedral arrangement of four nearest neighbours. In the bonding geometry of ZnO, each zinc atom bonds to four oxygen atoms, and each oxygen atom bonds to four zinc atoms. With four atoms per unit cell there are 12 phonon branches; 9 optical and 3 acoustic. The wurtzite structure optical phonons at the centre of the Brillouin zone, consist of an A1 branch, a doubly degenerate E1 branch, two doubly degenerate E2 branches, and two B branches, belonging to the following irreducible representation

Γopt = 1A1 + 2B1 + 1E1 + 2E2 . [Ellmer et al., 2008]

(3.24)

Figure 3.9: Crystal structure of a supercell for ZnO with aluminium doping [Song, 2005]. Both A1 - and E1 -modes are polar, and split into transverse optical and longitudinal optical phonons with different frequencies, because the macroscopic electric fields associated with the longitudinal phonons. The short-range interatomic forces cause anisotropy, and A1 - and E1 -modes possess, therefore, different frequencies. For the lattice vibrations with A1 - and 22

3.5. RAMAN CHARACTERISATION

E1 -symmetry, the atoms move parallel and perpendicular to the c-axis, respectively, see figure 3.10. [Ellmer et al., 2008]

[0001]

B1

E1

E2

A1

[2110]

Figure 3.10: Displacement patterns of the optical phonons of a lattice with wurtzite crystal structure. Modified from [Ellmer et al., 2008]. Both A1 - and E1 -modes are Raman and IR active. The two nonpolar E2 -modes are Raman active only. The B1 -modes are IR and Raman inactive. For crystals with wurtzite structure, pure longitudinal or pure transverse phonons of well defined symmetry can be observed only if the phonon propagating is along or perpendicular to the c-axis. The frequencies of the different modes, as reported by [Damen et al., 1966], can be seen in table 3.2. The specific scattering configuration is also determining to which modes are Raman active. Table 3.3 shows possible scattering configurations with matching allowed Raman modes. Modes E2 (high) E2 (low) A1 (T O) A1 (LO) E1 (T O) E1 (T O)

Frequency 437cm−1 101cm−1 380cm−1 574cm−1 407cm−1 583cm−1

Table 3.2: The table shows the Raman active fundamental optical phonon modes in ZnO, as reported by [Damen et al., 1966]. Scattering configuration x(zz)x’ x(zy)x’, x(yz)x’ x(yy)x’ z(xx)z’ z(xy)z’

Allowed Raman modes A1 (T O) E1 (T O) E2 , A1 (T O) E2 , A1 (LO) E2

Table 3.3: Raman selection rules of optical phonon modes. The values are valid for wurtzite crystal structures where the optical c-axis is parall el to the z direction. The notation follows the Porto notation, where the first and last letter denotes the direction of the incident and scattered light respectively [Damen et al., 1966]. The letters inside the parenthesis indicate the polarization of incident and scattered light. A typical Raman spectra of a doped ZnO thin film shows additional modes occurring. It is suggested that the additional modes are related to defect-induced modes. Actually the additional modes can be related to modes of ZnO which are Raman inactive within a perfect crystal. Upon doping-induced defect formation, the translational crystal symmetry can be broken and Raman inactive modes becomes Raman active. [Ellmer et al., 2008] 23

3. CHARACTERIZATION OF ZNO FILMS

Besides the Raman spectrum of ZnO crystals it might be important to take into account the material beneath. For example silicon, which has diamond structure, a peak near 521cm−1 is observed [Song, 2005]. Additionally weak peaks at 305cm−1 and 900 − 1100cm−1 are present due to two-phonon transitions [Song, 2005]. For amorphous materials the Raman spectrum show a few broad bands, corresponding roughly to peak in the broadened phonon density of states for the crystalline state. This is attributed to the fact that the selection rules do not apply after the loss of long range order. [Song, 2005] investigated aluminium doped films of ZnO deposited by rf magnetron sputtering and examined the films with Raman in the backscattering mode. In this mode the two E2 modes and the A1 longitudinal optical modes are observed while all other modes are forbidden. In the Porto notation this particular scattering configuration is denoted as z(xx)z’, see table 3.3. The modes expected are the high-frequency E2 -mode at 437cm−1 , and the longitudinal optical A1 -mode at 574cm−1 , see table 3.2. Figure 3.11 shows room-temperature Raman spectra obtained by [Song, 2005] from a set of sputter coated ZnO:Al films prepared at 1.0P a and 250◦ C, with alternating sputtering power from 50 to 200W . An additional curve from a sample deposited at 0.2P a, 250◦ C and 100W is also presented. Peaks appear at around 439cm−1 for all samples. These come from the ZnO E2 vibration modes. The small shift towards higher frequency can be a result of residual stress in films. Finally the fact that the E2 -modes are present in all the films suggest that the films have an preferred c-axis orientation.

Figure 3.11: A set of Raman spectra of ZnO:Al films on glass measured at room temperature. The films were sputtered at 250◦ C, and other deposition parameters are labelled on the curves. The vertical dotted lines are drawn at Raman shifts of about 380, 439, 501, and 579cm−1 . [Song, 2005] In addition to the above discussed E2 -modes vibrations, peaks around 573 − 579cm−1 were observed for all samples, roughly corresponding to the longitudinal A1 -mode. The shift in these peak positions are attributed to structural disorder and crystal defect in the film. Hence 24

3.5. RAMAN CHARACTERISATION

the intensity of the A1 optical mode is influenced by the sputtering parameters. From figure 3.11, the relative intensity of the A1 -mode to the E2 -mode decreases with increasing rf power. Taking into account that the conductivity of the produced films increases with rf power, the A1 -mode is sensitive to film conductivity, which is associated with the change of the free carrier concentration in the films. Thereby the A1 mode gives a measure of the conductivity of the film. The same behaviour has been observed for doped GaN by [Wieser et al., 1998]. The comparison is compelling as GaN has the same wurtzite structure as ZnO. In addition to the peaks associated with the E2 -mode (439cm−1 ) and the A1 -mode (573 − 579cm−1 ) peaks at about 380 and 501cm−1 were observed. The 380cm−1 correspond to the translational A1 -mode, which should be restricted by selection rules, but that may occur due to the polycrystaline structure of the sample. [Song, 2005] concludes that the last mode at 501cm−1 is related to the films conductivity, due to the fact that the highest peak corresponds to the film with the lowest resistivity. This suggest a possible relation to the carrier concentration in Al doped films.

Raman in this project The Raman microscope used was a Renishaw inVia Raman microscope using a 785nm 500mW laser. A 1(LO)

Raman intensity [arb.u.]

E2

200

400

600

800

1000

1200

-1

Wavenumber [cm ]

Figure 3.12: A backscatter Raman spectrum obtained for the z(xx)z’ configuration. Modes are expected for E2 (high) at 437cm−1 and A1 (LO) at 574cm−1 , see table 3.3. Further multiphonon modes at 208cm−1 , 334cm−1 , a broad band from 540 − 670cm−1 , 986cm−1 , and a broad band starting from 1050cm−1 , peaking at 1084cm−1 and 1149cm−1 are indicated. A single crystal ZnO sample was tested in the z(xx)z’ scattering configuration. The obtained spectrum can be seen in figure 3.12. The expected modes are the E2 (high) mode and the A1 (LO) mode. E2 (high) can clearly be identified, however the A1 (LO) can not be seen. Besides these first order modes, multi phonon modes are expected at 208cm−1 , 334cm−1 , a broad band from 540−670cm−1 , 986cm−1 , and a broad band starting from 1050cm−1 , peaking at 1084cm−1 and 1149cm−1 as reported by [Damen et al., 1966]. The 334cm−1 mode is very prominent and all other modes except 986cm−1 could be identified. The Raman spectrum closely resembles the Raman spectrum obtained by [Bundesmann, 2005], shown in figure 3.13. 25

3. CHARACTERIZATION OF ZNO FILMS

In both the spectra obtained in this project and by [Bundesmann, 2005] the E2 (high) can clearly be identified. Further the multiphonon modes at 208cm−1 , 334cm−1 , a broad band from 540 − 670cm−1 are visible in both spectra. The E2 (low) mode at 101cm−1 seen in figure 3.13 is not visible in the spectra obtained in this project as it lies outside the sensitive range of the instrument. Alternative Scattering Configurations

Figure 3.13: Raman spectrum obtained by [Bundesmann, 2005]. The spectrum is recorded at room temperature, in the z(xx)z’ configuration. In order to examine Raman spectra at different scattering configurations three different configurations were investigated. These are x(zz)x’, x(yy)x’, and z(xx)z’. As can be seen from table 3.3 different modes are expected to be active. One problem with the experiment was that it was not immediately possible to polarize the scattered light. Thereby the peaks from the alternative polarization are expected. Specifically a E1 (T O) peak is to be expected in x(zz)x’ and x(yy)x’. The results can be seen in figure 3.14(a). As can be seen the E1 (T O) is visible, because of the lack of polarization. Besides this all expected peaks can be found in the spectra. The only missing peak is A1 (LO) which is generally reported to be much smaller compared to ¨ ur et al., 2005]. The results are generally consistent with those reported by other peaks [Ozg¨ [Bundesmann, 2005], see figure 3.14(b).

26

3.5. RAMAN CHARACTERISATION

A1(TO) MP E (TO)

MP

1

MP

Raman intensity [arb.u.]

MP

200

300

400

500

600

700

800

Wavenumber (cm-1)

(a)

(b)

Figure 3.14: (a) Raman spectra obtained for x(zz)x’, x(yy)x’, and z(xx)z’ respectively, x(zz)x’ being topmost. (b) Spectra obtained by [Bundesmann, 2005]. In both spectra relevant modes are indicated by dotted lines.

27

3. CHARACTERIZATION OF ZNO FILMS

ZnO Bulk and Film Spectra [Bundesmann, 2005] examined Raman spectra for both film and bulk ZnO samples. The resulting Raman spectra for five different scattering configurations can be seen in figure 3.15. The described spectral features of in the Raman spectra of ZnO bulk also appear in Raman spectra for the c-plane ZnO thin film, expect for the A1 (LO) which in most cases was difficult to detect anyway. In film spectra several tops appear corresponding to phonon modes of sapphire, see figure 3.15(b). The part of the spectra resulting from the base will be large for thin films. Especially for ZnO films which are transparent as the scattering volume will include considerable parts of the base material. The ZnO film in figure 3.15(b) is roughly 2µm thick.

(a)

(b)

Figure 3.15: Raman spectra obtained for (a) bulk ZnO and (b) c-plane ZnO film on c-plane sapphire [Bundesmann, 2005].

28

4

End Product In the previous chapter, the experimental procedure and results of ZnO:Al films made by magnetron sputtering were described in detail. By optimising the parameters of the sputtering process, a set of deposition conditions can found to produce high-quality films with good crystallinity, low electrical resistivity, and uniform lateral parameters. [Chen et al., 2004] found that ZnO:Al films with electrical and optical properties compatible to the expensive ITO material can be obtained by the magnetron sputtering technique. The objective of this chapter is to investigate sputtered ZnO films properties, for application in photovoltaic cells. One of the most important properties is the light trapping of the TCO layer, this light trapping leads to a better quantum efficiency of the solar cell.

4.1

Material Properties and Surface Texture

Textured ZnO films prepared by sputtering and post deposition wet chemical etching has been comprehensively studied in the parts years. The combination of sputter deposition and etching technique opens a variety of possibilities to optimise light trapping in silicon thin films. For silicon based solar cells, transparency, conductivity, and light scattering ability of highly conductive glass/ZnO substrates have have to be optimised. All material properties strongly depend on the structure of the ZnO film. Most important for the development of surface textured ZnO films by chemical etching is the fact that the surface morphology of the etched ZnO depends on the initial film structure. The film structure itself is mainly influenced by the deposition condition. Hence, it follows that a detailed understanding of the basic relationship between deposition parameters, film growth and etching behaviour is a necessary prerequisite. [Wagner et al., 2001] found the relationship between sputter deposition parameters, film growth, film structure and the surface structure after etching in hydrochloric acid. [Kluth et al., 2003] used scanning electron microscopy, SEM, and atomic force microscopy, AFM, to characterise the surface textures obtained after etching. 29

4. END PRODUCT

Modified Thornton Model [Kluth et al., 2003] observed a strong decrease in the carrier mobility and an increase in etching rate with increasing deposition pressure. The surface morphology obtained after etching changes from crater-like to hill-like surfaces with increasing pressure, see figure 4.1. In general, the initial film structure controls the surface texture obtained after etching. This correlation between sputter parameters, film growth and structural properties can be discussed in terms of a modified Thornton model, which is shown in figure 4.1.

Figure 4.1: The modified Thornton model. Sample A was deposited with a high sputter pressure of 40mbar and without intentional heating, while sample B was deposited at a medium pressure of 2.7mbar and 150◦ C substrate temperature. Sample C was sputtered with the lowest pressure of 0.4mbar and at 270◦ C. [Kluth et al., 2003] Originally, the Thornton model was developed to describe the growth of sputtered metals depended of substrate temperature and sputter pressure. However there are some fundamental differences between TCO and metals. Firstly, ZnO exhibits a distinctly higher melting point, 1975◦ C, than typical metals. To account for this [Kluth et al., 2003] suggested some modifications of the original Thornton model to make it applicable to rf sputtered ZnO:Al films on glass substrates. Because of the higher melting point of TCO the ratio of substrate temperature and melting temperature considered in the Thornton model becomes very small, the typical range of substrate temperature is 80 − 400◦ C. [Kluth et al., 2003] therefore suggested that the substrate temperature is taken into account instead of the normalised value. [Kluth et al., 2003] showed the variations in bulk and surface structure by the example of three representative ZnO:Al samples, which were deposited in three different deposition regimes. Sample A was deposited with a sputter pressure of 40mbar and without intentional heating, 30

4.1. MATERIAL PROPERTIES AND SURFACE TEXTURE

while sample B was deposited at 2.7mbar and at 150◦ C. Sample C was sputtered at 0.4mbar and at 270◦ C. The samples was characterised by a cross-sections SEM scan. Figure 4.1 shows SEM images of Sample A, B, and C before and after etching. On Sample A the ZnO grains extend from the substrate to the top of the film. Their c-axis is oriented parallel to the substrate normal and a fibre texture is present. On sample B and C the crystallites are also highly oriented, but their c-axis exhibits a small inclination with respect to the substrate normal. Single crystallites can be distinguished in sample A by the SEM cross-sections scan. This, however, is more difficult for sample B and C. This indicates an increasingly compact and denser structure for the transition from Type A to Type C ZnO. After etching structural differences between the three samples are more obvious, see figure 4.1. Entirely different surface morphologies are observed. Type A ZnO was homogeneously etched with high rate, chemical etching had no impact on the surface morphology but only reduced the film thickness. Type B and C ZnO were anisotropically etched leading to a crater-like surface structure. The regular, rough morphology of the Type B film indicates a homogeneous attack of the complete initial surface by the etchant. Due to its extremely compact film structure Type C material is only partly etched leading to a few randomly spread craters on the film surface. Sample A can be identified as a Zone 1, as indicated on figure 4.1, film with typically low compactness. The more compact film structure of sample B and the highly dense film structure of sample C can be found in the middle and at the edge of Zone 2, respectively. Highly compact polycrystalline ZnO:Al films exhibit to an increasing extent the properties of a single crystalline ZnO. This is reflected in increased hall mobility due to an improved film structure with less structural defects and by a higher mechanical and chemical stability. The highly dense film structure reduces the physisorption of oxygen at grain boundaries. A compact film structure also prohibits the penetration of fluid etchant. Consequently, the highly oriented ZnO:Al films are only attacked by the etchant from one crystallite site. This leads to an anisotropic etching process, which produces the observed crater structure. While the films prepared in Zone 1 show poor electrical properties and only small features. Films of Zone 2 with high and medium compactness provide both, high conductivity and larger surface features after etching. Especially Type B films exhibit a regular crater structure. Such surface structures are an example for a morphology which leads to an effective light trapping in silicon thin film solar cells. Note that due to the high melting point of ZnO, Zone 3 of the original Thornton model is not present in the applied temperature range. Recrystallisation, which is typical for Zone 3, appears at much higher substrate temperatures. However the general statement of the original Thornton model is maintained: increasing the substrate temperature and reducing the sputter pressure leads to a more compact and dense film structure.

Application in Silicon Thin Film Solar Cells The most sensitive detector to evaluate the suitability of textured TCO substrates for application in solar cells is the solar cell itself. [Wagner et al., 2001] examined glass/ZnO substrates with distinctly different surface morphology in p-i-n solar cells and measured their quantum efficiency. Quantum efficiency is the percentage of photons hitting the photoreactive surface that will produce an electron-hole pair. The improvement [Wagner et al., 2001] found in quantum efficiency obtained by the introduction of a ZnO:Al film with crater-like surface structure is shown in figure 4.2. Smaller overall quantum efficiency is obtained on the type C ZnO film, which has a regular surface texture compared to type D film, which has a crater-like texture, as seen on figure 4.2. Actually the 31

4. END PRODUCT

Figure 4.2: Quantum efficiency of two solar cells co-deposited on smooth and texture-etched ZnO/glass substrate. [Wagner et al., 2001] difference in efficiency between type C and D films is 2.4 percentage point, type D films having the highest efficiency of 9%. [Wagner et al., 2001]

32

5

Roadmap For the continuation of the project a number of goals must be meet. The primary objective of the second half of the project is to produce and characterise several different types of ZnO films. This include obtaining precise control over production parameters for intrinsic film production as well as films doped with aluminium. In the case of doped films, control over degree of doping must be established. The chosen method of production is magnetron sputtering. The work needed include ˆ Gain control over thickness as a function of time, plasma intensity, pressure, and temperature. ˆ Optimising growth to ideal films. ˆ Introducing a dopant.

These objectives will be meet by making a number of production runs varying the different parameters. Table 5.1 details the planned production runs. It is planned that each run will be made for 10 samples, meaning that 10 different thicknesses, pressures, O2 ratios will sampled. In addition each production must be made at three different substrates, including glass, highly doped Si, and SiO2 covered Si. This is to insure that every characterization technique can be carried out. This means that a total of 360 samples must be manufactured. By running the different films produced through the characterization techniques it should be possible to identify the most optimum films with respect to primarily transmittance and conductivity. Secondly the structure of the film surface is an important factor in the final solar cell efficiency, and will be included in the investigation through SEM characterization, by measurement of dielectric function via ellipsometry, PL, Raman, and optical microscopy. When a conclusion has been reached on optimum film conditions, it will be time to introduce doping. The doping will be included by adding aluminium to the ZnO target.

33

5. ROADMAP

Run 1

Parameter Pressure, annealing (300 to 600◦ C)

Constant Thickness, effect, sample temperature, O2 ratio

2

O2 ratio, annealing (300 to 600◦ C)

Pressure, thickness, effect, sample temperature

3

Effect, annealing (300 to 600◦ C)

Pressure, thickness, sample temperature, O2 ratio

4

Thickness (100 to 1000nm), annealing (300 to 600◦ C)

Pressure, effect, sample temperature, O2 ratio

Investigate Electrical Properties (FFP, Hot Probe), Optical (Ellipsometri, Raman, PL), and Quality (SEM, Optical Microscopy) Electrical Properties (FFP, Hot Probe), Optical (Ellipsometri, Raman, PL), and Quality (SEM, Optical Microscopy) Electrical Properties (FFP, Hot Probe), Optical (Ellipsometri, Raman, PL), and Quality (SEM, Optical Microscopy) Thickness (Profiler, Ellipsometri), Electrical Properties (FFP, Hot Probe), Optical (Ellipsometri, Raman, PL), and Quality (SEM, Optical Microscopy)

Table 5.1: Overview over planned experiments.

34

Appendix A

Production of Standard Sample for Probe Distance Settings In order to investigate the change in resistivity measurements as a function of the probe distance, it is necessary to precisely set the probes at a certain mutual distance. To do this a standard sample was made, on which points of certain mutual distance is indicated. Using this standard sample probes can be placed. The standard sample were made with a lithography system, described in section A.

Sample Design

Figure A.1: Design of standard sample. Distances are marked at 2mm, 1.5mm, 1mm, 500µm, 250µm, and 125µm.

Production Method To produce the standard samples, it is necessary to master a number of different procedures, including cleansing of the glass plates, sputtering of metal, and treatment of the photoresist. The total procedure is illustrated on figure A.2. In the following the preparation of the glass plates, the treatment of the photoresist and etching of the metal are further described. 35

APPENDIX A. PRODUCTION OF STANDARD SAMPLE FOR PROBE DISTANCE SETTINGS

Cleansing

Sputtering

Softbaking

Exposur

Etching

Lacquering

Spincoating

Rehydration

Development

Hardbake

Figure A.2: The steps followed during the production of the standard sample for probe distance stetting. The glass plates are cleansing in an ultrasonic bath. After which they are sputter coated with gold. Before the lithography process can be carried out the sample has to be coated with a photoresist. In this project this is done on a Arias multifunction cupboard. After the resist coating, the sample should be kept at room temperature for at least 15 minutes to allow most of the solvent to evaporate before it is soft baked on a hotplate. If the solvent is not evaporate the resist surface will dry fast, and trapped solvent remaining in the film and the solvent may form bubbles in the film. Soft baking prevents bubbling in the film by nitrogen created during exposure and improves resist adhesion on the plate. After soft baking most of the water in the sample is evaporated, but since the photo reaction during exposure requires a certain amount of water, in order to allow the creation of carboxylic acid, the rehydration is important as it improves the development rate. [Clariant, 2008] The exposure is performed with a lithography arrangement build by the group [Lorentzen et al., 2006]. After exposure the photoresist releases nitrogen which in the ideal case diffuses through the photoresist to the film surface without forming bubbles. In the case of insufficient soft bake, the nitrogen release during exposure leads to popping of the photoresist film and create tension crackles. [Clariant, 2008] The development process is performed in 30s. The hard bake performed after development tends to increase the stability of developed photoresist structures for subsequent processes. After hardbaking the sample is placed in aqua regia to etch the exposed gold area. When the gold is etched the remaining resist is removed with acetone. Finally to protect the surface of the gold is treated with lacquer.

Litography Arrangment The lithography arrangement, utilizes direct writing. Direct writing is based on a focused laser beam, which is able to move across the surface of a sample, to write patterns of the desired shape. The focused laser beam exposes the photoresist, and it is then possible to remove the exposed or non exposed regions depending on the type of photoresist. The arrangement has been developed by this group in an earlier project. The main function of the arrangement is the ability to write structures in the order of micrometres on a plane surface, additionally the arrangement is required to be capable of writing several structures on the same sample in layers. The final arrangement is shown in figure A.3. 36

All the functions are controlled by a computer.

Figure A.3: The arrangement is build on an optic table. After the UV laser there is an array of lenses, placed with a mutual distance of 150mm. 125mm from L1, a 20µm pinhole, P1, is placed in the lenses focal points. The purpose of these lenses is to enlarge the laser beam so it fills out the 100× zoom objective. After this set of lenses a beam splitter is placed. The distance between L2 and the beam splitter is not important because the laser beam after L2 is collimated. To use autofocus and scan the sample, the helium neon laser is used, to use these functions the mirror is positioned so the helium neon laser beam hits the objective. The reflection from the sample then goes through the beam splitter and is collimated in lens L3. In the focal point of L3 a 200µm pinhole, P2, is placed, which ensures that maximum intensity reaches the detector when the sample is in focus. The polariser is placed in the arrangement to control the intensity of the laser beam, this ensures that the detector is not over exposed. When writing, a 405nm, 25mW UV laser with an internal shutter is used. Due to exposure when using this wavelength, a helium neon laser with a wavelength of 633nm is used for automatic focusing by moving the sample closer or further away from the objective. The sample is placed in the arrangement and an autofocus procedure is started. This is to ensure that the sample is in focus. To make sure that the sample is in focus at all time under the writing process, an auto focus slope correction function can be used. A file with the wanted pattern is loaded and the sample is exposed. The exposure time used was 30ms.

37

Bibliography

[Azzam and Bashara, 1977] Azzam, R. M. A. and Bashara, N. M. (1977). Ellipsometry and Polarized Light. North-Holland Personal Library. 0-444-87016-4. [Bundesmann, 2005] Bundesmann, C. (2005). Phonons and plasmons in ZnO-based alloy and doped ZnO thin films studied by infrared spectroscopic ellipsometry and Raman scattering spectroscopy. Shaker Verlag. 978-3-8322-4730-0. [Cambell, 2001] Cambell, S. A. (2001). The Science and Ingineering of Microelectronic Fabrication, second edition. Oxford University Press, Inc. 0-19-510508-7. [Chen et al., 2004] Chen, H., Qiu, C., Peng, H., Xie, Z., Wong, M., and Kwok, H. S. (2004). Co-sputtered aluminium doped zinc oxide thin film as transparent anode for organic lightemitting diodes. Center of Display Research, The Hong Kong University of Science and Technology:489–492. [Clariant, 2008] Clariant (2008). Microchemicals. http://www.microchemicals.com. [Damen et al., 1966] Damen, T. C., Porto, S. P. S., and Tell, B. (1966). Raman effect in zinc oxide. Physical Review, 142:570–574. [Djurisic and Leung, 2006] Djurisic, A. B. and Leung, Y. H. (2006). Optical properties of zno nanostructures. Small, 2:944–961. [Ellmer et al., 2008] Ellmer, K., Klein, A., and Rech, B. (2008). Transparent Conductive Zinc Oxide. Springer. 978-3-540-73611-0. [Groenen and van de Sanden, 2000] Groenen, R. and van de Sanden, M. C. M. (2000). Expanding thermal plasma cvd of textured zno with focus on solar cell applications. Photovoltaic Specialists Conference, 2000. Conference Record of the Twenty-Eighth IEEE, 28:822– 824. [Jeong et al., 2003] Jeong, S.-H., Kim, B.-S., and Lee, B.-T. (2003). Photoluminescence dependence of zno films grown on si(100) by radio-frequency magnetron sputtering on the growth ambient. Applied Physics Letters, 82:2625–2627. [Klein and Furtak, 1986] Klein, M. V. and Furtak, T. E. (1986). Optics. John Wiley and sons, Inc. 0-471-87297-0. 38

BIBLIOGRAPHY

[Kluth et al., 2003] Kluth, O., Sch¨ope, G., H¨ upkes, J., Agashe, C., M¨ uller, J., and Rech, B. (2003). Modified thornton model for magnetron sputtered zinc oxide: film structure and etching behaviour. Thin Solid Films, 442:80–85. [Lima et al., 2001] Lima, S. A. M., Sigoli, F. A., Jr, M. J., and Davolos, M. R. (2001). Luminescent properties and lattice defects correlation on zinc oxide. International Journal of Inorganic Materials, 3:749–754. [Lin et al., 2001] Lin, B., Fu, Z., and Jia, Y. (2001). Green luminescent center in undoped zinc oxide films deposited in silicon substrates. Applied Physics Letters, 79:943–945. [Lorentzen et al., 2006] Lorentzen, N. B., Madsen, M. V., Nørgaard, J. M., and Jensen, J. M. (2006). Laserwrite manual. [Madsen and Brodam, 2008] Madsen, M. V. and Brodam, N. (2008). Transparent and conductive films of zinc oxide. [Nelson, 2003] Nelson, J. (2003). The Physics of Solar Cells. Imperial College Press. 1-86094349-7. [OceanOpticsInc., 2005] OceanOpticsInc. (2005). Ooichem software http://www.oceanoptics.com/technical/chem2000Manualwithooichem.pdf.

user’s

guide.

[Schroder, 1990] Schroder, D. K. (1990). Semiconductor Material and Device Characterization. John Wiley and Sons, Inc. 0-471-51104-8. [Shah et al., 1999] Shah, A., Torres, P., Tscharner, R., Wyrsch, N., and Keppner, H. (1999). Photovoltaic technology: The case for thin-film solar cells. Science magazine, 285:692–698. [Song, 2005] Song, D. (2005). Zinc oxide tcos (transparent conductive oxides) and polycrystalline silicon thin-films for photovoltaic applications. [Sze, 2007] Sze, S. M. (2007). Physics of semiconductor devices. John Wiley and Sons, Inc. 0-471-84290-7. [Wagner et al., 2001] Wagner, H., Kluth, O., and Rech, B. (2001). Sputtered zinc oxide films for silicon thin film solar cells: Material properties and surface texture. European Photovoltaic Solar Energy Conference. [Wieser et al., 1998] Wieser, N., Klose, M., Dassow, R., Scholz, F., and Off, J. (1998). Raman studies of longitudinal optical phonon-plasmon coupling in gan layers. Journal of Crystal Growth, 189/190:61–665. [Xu et al., 2003] Xu, P. S., Sun, Y. M., Shi, C. S., Xu, F. Q., and Pan, H. B. (2003). The electronic structure and spectral properties of zno and its defects. Nuclear Instruments and Methods in Physics Research B, 199:286–290. ¨ ur et al., 2005] Ozg¨ ¨ ur, ., Alivov, Y. I., Liu, C., Teke, A., Reshchikov, M. A., Doan, S., [Ozg¨ Avrutin, V., Cho, S.-J., and Morko¸c, H. (2005). A comprehensive review of zno materials and devices. Journal of Applied Physics, 98:2–95.

39