Wavefrontshaping for absorption optimization in a thin film solar cell

Wavefrontshaping for absorption

optimization in a thin film silicon

solar cell.

THESIS

submitted in partial fulfillment of the requirements for the degree of

BACHELOR OF SCIENCE

in PHYSICS

Author : M.J. van Velzen

Student ID : 1046489

Supervisor : MSc. F. Mariani

Prof. Dr. M.P. van Exter 2ndcorrector : Dr. ir. S.J. van der Molen

Wavefrontshaping for absorption

optimization in a thin film silicon

solar cell.

M.J. van Velzen

Huygens-Kamerlingh Onnes Laboratory, Leiden University P.O. Box 9500, 2300 RA Leiden, The Netherlands

December 21, 2015

Abstract

Solar cells work by absorbing light to create electron - hole pairs and use a built-in electric field to separate these pairs to create a

current. We want to create a destructive interference for the reflected light of a thin film solar cell so that we increase the energy inside the solar cell. In our experiment we use a spatial light modulator to control the phase of the light wavefront that reaches the thin film solar cell. We have optimized the lightfront of a random phaseplate on the SLM with a partitioning algorithm.

This caused an 1.2% increase in the ratio between the current of the optimal phaseplate and a control phaseplate in only 4 iterations. After this the ratio became stable which indicates that

we created an optimal phaseplate. It is not yet clear if the increased current is caused by creating destructive interference for

Contents

2.1.3 Thin film solar cells 6

2.1.4 IV-curve 7

2.2 Shaping light to increase absorption 8

2.2.1 Destructive interference 9

3 Methods 11

3.1 setup 11

3.1.1 Controlling the wavefront 12

3.1.2 Imaging 12

3.1.3 Chopper & lock-in amplifier 13

3.1.4 Solar cell specifications 14

3.2 Superpixels & throughput 15

3.3 Algorithm to create the optimal phaseplate 16

4 Results & Discussion 17

4.1 Aligning 17

4.2 Power throughput after SLM 19

4.3 Measuring in the linear regime 20

4.4 Noise 23

4.5 Wavefront optimization 25

Chapter

1

Introduction

For many years the world has been trying to find replacements for fos-sil fuels. The reasons for this are numerous, like trying to combat global warming, being independent of the supply from other countries or prepar-ing for the inevitable when we run out. Many people seek a solution in re-newable energy from wind, the tides or most importantly for our research, the sun.

Solar cells are slowly becoming more common in society. They are a way to transform light into electrical energy and are clearly visible on a lot of buildings. Unfortunately there are not yet enough to fully replace fossil fuels. Their high initial cost and low efficiency make it a costly invest-ment. A lot of research has been going into making them more efficient or cheaper to produce. In our experiment we try to make the solar cell more efficient by changing the light instead of the solar cell.

Nowadays there are ways to control and shape the wavefront of light by spatially modulating its phase. It has already been proven that it is possible to image through random media using this method [1]. We want to control the wavefront of our light to try to decrease the reflected light of the solar cell and increase the absorbed light.

Chapter

2

Theory

2.1

Solar cells

A solar cell transforms the energy from light into electricity. Because of the photovoltaic effect, the energy of incoming photons can raise the electrons inside the solar cell to a higher energy state. These higher energy electrons are then moved through an electrical circuit producing work.

2.1.1

Carrier excitation

The majority of solar cells are made of semiconductors. Because of the Pauli exclusion principle the electrons of the semiconductor can not oc-cupy the same state and organize in bands of allowed levels. The highest accoupied band is called the valence band and is full of electrons. The band above it is empty in first approximation and is called the conduction band (fig 2.1).

The energy needed to excite an electron to the conduction band is de-pendent on the band gap, the energy difference of the conduction and the valence band. An incoming photon can be absorbed by a semiconductor and if the energy of the absorbed photon is equal or bigger than the band gap, the electron can be raised to the conduction band. Raising an elec-tron to the conduction band causes a hole to appear in the valence band. It is easiest to imagine this hole as a positively charged particle. Thus light falling on an semiconductor can create an electron in the conduction band and a hole in the valence band which are free to travel and transport charge and energy.

The picture given so far is for the intrinsic semiconductor, which means that there are no impurities introducing holes or electrons that are able to

Figure 2.1: A Semiconductor electron band structure. The valence band is filled with electrons while the conduction band is empty.

conduct. It is possible to create a semiconductor which has more holes than free electrons or more free electrons than holes by introducing an im-purity in a process that is called doping. By placing an atom in the semi-conductor that has one more valence electron than the rest of the atoms this extra electron is not tightly bound by the crystal lattice. As a con-sequence free electrons are available in what is called n-type doping. By placing an atom in the semiconductor with one less valence electron the opposite type of doping can be obtained and is called p-type doping.

2.1.2

p-n junctions

Semiconductor solar cells use p-n junctions to stop the electron and hole from recombining. A p-n junction is established when a layer of p-type semiconductor and a layer of n-type semiconductor are brought together. In this situation free electrons from the n-type layer will diffuse to the p-doped region while the holes will diffuse to the n-p-doped region. The ex-cess electrons will fill the holes, essentially annihilating each other. This process leaves a depletion layer between the two regions where there are no free holes or electrons.

A p-doped, or n-doped region is neutral, because the free charge carriers cancel the donor or acceptor charges in the semiconductor. In the deple-tion layer though, the free charge carriers are gone. On the n-doped side only the positive donor charges remain. On the p-doped side only the negative ion charges are left. These positive and negative charges create an electric field in the depletion layer. This will cause an electric field in the depletion layer which counteracts the diffusion. The p-n junction thus

2.1 Solar cells 5

reaches an equilibrium and ends in a stable configuration with a built in electric field (figure 2.2).

Figure 2.2: In a p-n junction an n-doped region is in contact with a p-doped re-gion. Diffusion of the free electron and holes leave a depletion layer with a built in electric field. This electric field causes electron-hole pairs formed in the depletion layer to be split before they can recombine.

In a solar cell if a photon with enough energy is absorbed in the p-n junction it will raise an electron to the conduction band creating a hole and a free electron. If the electron-hole pair is formed within the depletion layer the electron will move to the n region and the hole to the p region under the effect of the built in electric field. This process is called the Pho-tovoltaic effect. In an open circuit these charges build-up and counteract the built in electric field. We can use the photovoltaic cell as a current generator by connecting its terminals to an external load.

Only light absorbed within the depletion layer generates a current. A way to increase the depletion layer is spacing the p and n doped regions with intrinsic semiconductor, thus creating a p-i-n junction. This gives the solar cell a bigger region where the electron-hole pairs can be separated to generate a current.

2.1.3

Thin film solar cells

A thin film solar cell is a solar cell with mass-production possibilities. In contrast to silicon solar cells they are produced by physical or chemical deposition techniques which make them a cheaper but less efficient kind of solar cell [2].

A thin film solar cell consists of multiple layers. There are different structures possible but a possible construction is shown in figure 2.3. The first layer consists of glass which acts as a substrate for the solar cell. After that comes a layer of a Transparent Conductive Oxide (TCO) which is the front contact for our electrical circuit and must be transparent to let the light through. Then comes the semiconductor where the light is absorbed and the electron - hole pairs are seperated. The last layer is the back contact and is made of metal.

Figure 2.3: A schematic sketch of a thin film solar cell example from [3]. While the light is absorbed in the semiconductor the Transparent Conductive Oxide and the metal back reflector are neccesary for the contacts in an electrical circuit. The glass supports the solar cell mechanically.

Not every photon is absorbed by the semiconductor. The fraction of absorbed light is dependent on the thickness and properties of the material used. A thicker semiconductor absorbs more light because the photons have a higher chance excite an electron. We define the distance after which the fraction 1

e or 36% of the light is left as the absorption depth of the material. It is dependent on the optical properties of the absorber, which

2.1 Solar cells 7

are typically wavelength dependent.

Another way to increase the absorbed light is to increase the path length of the light through the material. The metal back contact does this by re-flecting the light back into the semiconductor, doubling the path length.

In addition to this, the surface of the semiconductor is intentionally made rough. As seen in figure 2.3 a rough surface scatters the light, giving it multiple skewed trajectories. These give the light a longer path length and decrease the angle between the light and the back surface. Because of the decreased angle, the chance for internal reflection is increased, which makes it possible to trap light in the semiconductor [4].

2.1.4

IV-curve

You can plot the current generated by the solar cell against the voltage according to the formula

I = Io(eqV/kbT _{−1) −I}

l (2.1)

Here Il is the photocurrent created by illuminating the solar cell and Io is the dark saturation current which is the current in the absence of light and created by diffusion in the solar cell. q is the elementary charge of an electron, V is the applied bias, kb is the Boltzmann constant and T is temperature in degrees Kelvin.

Figure 2.4 shows the IV-curve of a dark and an illuminated solar cell. In the dark a solar cell acts like a diode [2]. A reverse bias which creates a negative voltage hardly changes the current. A forward bias which creates a negative voltage increases the current exponentially.

The diode behaviour is caused by the depletion layer. A forward bias decreases the potential difference over the depletion layer. This makes the electric field weaker, making it easier for electrons to cross, and thins the depletion layer When the forward bias cancels the potential difference, the depletion layer is destroyed and the electrons are free to move, creating the exponential behaviour in the current.

A reverse bias increases the potential difference over the depletion layer. This strengthens the electric field and thickens the depletion layer. There-fore crossing the layer requires a higher energy and there will only be a small current. Only with an extensive reverse bias the solar cell will break down and allow a current increase quickly with the applied voltage [5]. When illuminating the solar cell, the electron-hole pairs created in the

Figure 2.4:The IV-curves show what happens to the current of a dark and illumi-nated solar cell when a voltage is applied. Ilis the current created by illuminating

the solar cell.

depletion layer are separated by the internal field in the depletion layer. When you plot the IV-curves for a dark solar cell and an illuminated solar cell you see that illuminating the cell essentially shifts the IV-curve down (figure 2.4). Applying a forward bias on an illuminated solar cell will thus first decrease the current until it reaches zero.

In our experiment we want to maximize Il without changing the power from the laser. To maximize Il we measure the current generated by the solar cell. With a stable temperature and voltage, the variations of Il lin-early affect the total current and we can thus find a maximum. To mea-sure variations in the total current we use a reverse bias. A reverse bias strengthens the field in the depletion layer which decreases the chance of an electron-hole pair recombining. A reverse bias thus increases the effi-ciency of charge carriers collection.

2.2

Shaping light to increase absorption

To generate a current a photon has to be absorbed by the solar cell, but in thin film solar cells light can also be reflected or transmitted. To in-crease the efficiency it is necessary to minimize the reflected and trans-mitted light. Because of the metal back contact the transmission in a thin film solar cell is almost non-existent. Our challenge is then decreasing the

2.2 Shaping light to increase absorption 9

reflected light. A possible way to do this is by using interference.

2.2.1

Destructive interference

Light reaching the rough surface of the semiconductor is partially reflected and partially scattered inside the semiconductor. In the semiconductor the propagated light will be absorbed while it propagates. Some light will propagate through the semiconductor to eventually be scattered back into free space. Here the field will interfere with the field of the directly re-flected light. These fields are able to constructively or destructively inter-fere, strengthening or weakening the resulting field (figure 2.5).

Figure 2.5: Some light is reflected from the semiconductors front and back inter-face. The two reflected fields interfere in the free space. This results in a strength-ened or weakstrength-ened reflection and, more importantly, more or less light absorbed by the semiconductor.

If we are able to control the wavefront of the light reaching the solar cell we can purposefully make the fields destructively interfere, weakening the resulting field. The removed energy of the weakened field can not disappear because of the preservation of energy, it can only go into the solar cell. By annihilating the fields in free space we thus increase the absorbed energy by the solar cell. We can use a Spatial Light Modulator (SLM) to fully control the wavefront of the light reaching the solar cell.

Chapter

3

Methods

3.1

setup

The setup we used for our experiment is shown in figure 3.1. We use a 4 mW helium-neon laser with λ = 632,8 nm and send light through a chop-per and into a single mode fiber. After the single mode fiber a collimator creates a collimated beam and light is linearly polarized.

Figure 3.1: Schematic of the experimental setup. A laser beam goes through the chopper into a single mode fiber. After the fiber the beam goes through a collima-tor and a polariser to reach the SLM. We use an imaging system consisting of a tube lens and an objective to image the wavefront after the SLM on the solar cell. Camera 1 and 2 are used for aligning purposes.

3.1.1

Controlling the wavefront

A Spatial Light Modulator (SLM) is a device that is able to spatially control the phase or intensity of light. The SLM we use consists of multiple pix-els of a translucent liquid crystal that has a controlled birefringent effect. This allows every pixel to adjust the phase of light falling on it. The device can be controlled via a computer to create different phaseplates and shape the wavefront reflected from the SLM. There are multiple possibilities for the use of an SLM like adaptive optics, speckle interferometry or seeing through scattering media [1]. Using the SLM to control the wavefront of the light falling on the solar cell, we want to make the reflected fields de-structively interfere in the free space.

The SLM works by changing its pixels refractive index and is optimized for linearly polarised light. We thus place horizontal polariser before the light reaches the SLM. After the SLM the light reaches a beamsplitter which sends 30% of the light intensity to camera 1. To focus the other 70% of the light intensity on the 4x4 mm solar cell we use an imaging system con-sisting of a lens and an infinity-corrected objective. Reflected light from the solar cell is sent back to the beamsplitter. The beamsplitter then sends light to camera 2 which is placed the same distance from the lens as our SLM.

To fully control the wavefront falling on the solar cell we need to know if our solar cell is in the focus of the objective. We can check this by using the conventional use of the objective as a microscope. If we shine some light on the solar cell while it is in the focus of the objective, it gives an image on camera 2. We use camera 1 to see if the phaseplate off the SLM is in the middle of our light beam.

3.1.2

Imaging

The light reflected from the SLM is not a plane wavefront. Each pixel on the SLM acts as a point source with a different phase. By placing a tube lens with its focal point on the SLM we shape the light from each pixel into a collimated beam. The light is then focused by the objective on its focal plane (figure 3.2). The objective we use has a 40 times magnification and a numerical aperture (NA) of 0.6.

3.1 setup 13

Figure 3.2: Our imaging system consisting of a tube lens and objective focusses the light from the SLM on the solar cell.

3.1.3

Chopper & lock-in amplifier

We want to accurately measure only the photocurrent generated by the wavefront shaped light. All other light or sources of fluctuations are noise. A way to separate the signal from the noise is by using a chopper and a lock-in amplifier. The chopper is a small rotating fan which shapes the light excitation into square wave pulses.

The lock-in amplifier can measure the photocurrent component oscil-lating at the chopping frequency [6] by multiplying the input signal from the solar cell by a reference signal at the same frequency. The resulting sig-nal is proportiosig-nal to the photocurrent generated by the laser. Our chop-per and reference signal have a frequency of 333 Hz.

3.1.4

Solar cell specifications

The solar cell we use in our experiment consists of an aluminium back reflector, an amorphous silicon p-i-n semiconductor, TCO and a glass sur-face. The silicon has a roughness of σ = 40nm and lc = 200 nm where σ is the root mean square deviation of the height and lcis its lateral correlation lenght (figure 3.3).

Figure 3.3:Roughness of the solar cell.

We measure the current from the solar cell according to the electronic circuit in figure 3.4. The volt generator is connected to the solar cell to supply a reverse bias of 0.5 V. The lock-in amplifier measures the voltage over the load resistor and sends it to the computer. We will use Ohm’s law

I = V

RL (3.1)

I is the current generated by the solar cell, V is the measured voltage over the load and RL the resistance of our load. We can now calculate the cur-rent generated by the solar cell when measuring the voltage using the lock-in and knowlock-ing that RL =500Ω.

3.2 Superpixels & throughput 15

Figure 3.4: Our solar cell cell is in series connected with a volt generator and a RL= 500Ω load. The lock-in amplifier is paralel connected to the load. This way

it can measure the voltage over the solar cell.

3.2

Superpixels & throughput

Light coming from different pixels must be distinguishable when it reaches the solar cell. We can calculate the size of the diffraction limited spot on the solar cell using the Abbe diffraction limit:

λ

2∗N A =d (3.2)

Here λ is the wavelength of our laser, NA is the numerical aperture of our system and d is the maximum diameter of the spot we can image that is distinguishable. The NA of our system is determined by our objective which has an NA of 0.6 and we use a laser with λ=632, 8nm. The diffrac-tion limited spot is thus 632, 8

2∗0.6 = 527nm. The pixels on the SLM have a diameter of 8 µm and our imaging system has a 40 times de-magnification which leaves the spot of a single pixel at 200 nm. This means we can not distinguish the spot of a single pixel. We need to create ”superpixels” from groups of pixels with the same phase to control the wavefront. By creating superpixels consisting of 4 pixels we know for certain that the spots on the cell are distinguishable. Furthermore smaller superpixels causes more light loss in our throughput because of diffraction as shown in chapter 4.2.

3.3

Algorithm to create the optimal phaseplate

To create this phaseplate we need an algorithm that has multiple iter-ations with a feedback loop. In these iteriter-ations pixel(s) cycle from 0 to 2π in n steps and at the end the optimum phaseplate is chosen. There are multiple algorithms that do this like sequential algorithms that cycle only one pixel or partitioning algorithms that cycle multiple pixels [7].

For choosing an algorithm we have to take into account that the change in photocurrent should be measurable for every step in the cycle. An al-gorithm with a fast initial increase thus seems best to prove that we can increase the current by shaping the light. A partitioning algorithm has the highest initial increase when trying to focus light through a scattering medium. In this algorithm in each iteration half of the pixels are randomly selected to cycle in n steps and after every step the photocurrent is mea-sured. At each cycle an optimized phaseplate is calculated and another iteration starts with the new phaseplate. For our experiment we chose iterations with n=10 steps.

Chapter

4

Results & Discussion

4.1

Aligning

To be able to increase the current from our solar cell we need to have con-trol of the wavefront that reaches the cell. This means that our setup must be fully aligned and the laser spot on the SLM must be fully encompassed by the phase plate. The easiest way to ensure that the laserspot is encom-passed is to have the center of the phase plate in the center of the laser spot. We use the aligning phase plate shown in figure 4.1 to test this. When the center of the phase plate is in the center of the laser spot we see a certain symmetric light spot on camera 1.

Once we have full control of the wavefront we have to be sure we con-trol it on the spot where it hits the semiconductor of the solar cell. Camera 2 is placed on the same distance from the solar cell as the SLM. Thus when the surface of the semiconductor is imaged in camera 2, the hologram on the SLM is imaged on the semiconductor (figure 4.2).

Figure 4.1: Procedure used to align the SLM (a) The aligning phaseplate on the SLM. (b) A picture from camera 1 of the laserspot without an phaseplate on the SLM. (c) A picture from camera 1 with the aligning phaseplate on the SLM. The symmetric lightspot means the center of the phaseplate is on the center of the laserspot.

Figure 4.2:We know the solar cell is correctly aligned when we see the surface of the semiconductor. This is a 40 times magnification of the surface of the solar cell. The spots are the parts of the rough surface that are in the focus.

4.2 Power throughput after SLM 19

4.2

Power throughput after SLM

The SLM is not a perfect device, it has a fill factor of 93%. Because we can not control 7% of the SLM even a flat phaseplate creates an diffraction pat-tern. Only the center spot of this diffraction pattern falls in our throughput which leaves us with 60% of the lightpower.

Random phaseplates on the SLM can change the power throughput by diffracting light outside of our optics. We use a photodiode to measure the throughput dependance on the size of our superpixels. We first performed a calibration curve of the photodiode (figure 4.3). By fitting the results we can now use the output of the photodiode to calculate the power of the light.

Figure 4.3:A calibration of the photodiode shows that its voltage output changes linearly with the power on the photodiode.

By replacing the solar cell with the photodiode, we measure the power throughput of 10 different random phaseplates for superpixel sizes rang-ing from 1 to 20. Figure 4.4 shows the average power reachrang-ing the photodi-ode for every superpixel size. Pixelsize of zero corresponds to a flat phase-plate. This has the maximum throughput and we use this to normalize our results. It is clear that small superpixel sizes cause less power througput and that bigger superpixels slowly converge to the reference point. For our experiment we use a superpixel size of 4. This leaves 66% of the through-put from a flat phaseplate. This leaves us with 66%∗60%

100 = 40% of the original light power before the SLM.

Figure 4.4:The superpixels diffraction reduces the throughput of our optical sys-tem: Smaller superpixel sizes causes more light loss while bigger superpixels slowly converge to the maximum of a flat phaseplate.

4.3

Measuring in the linear regime

We want the photocurrent to be proportional to the absorbed light, so we need to check if the solar cell has a linear response to lightpower. To find this linear regime we will look at multiple IV-curves.

Looking at the IV-curves of our solar cells we noticed they deteriorated over time, sometime assuming the shape expected for resistive behaviour. This means that we somehow broke our solar cells (figure 4.6). To explain the reason our cells broke we looked at multiple causes including our bias source. After measuring the output of the bias source (figure 4.5) we found that peak voltages appeared when we turned the bias source on or off. We obtained stable IV-curves when we made sure that the solar cell was never connected to the bias source when this was switched on/off.

With stable IV-curves we made measurements for different light power reaching the solar cell (figure 4.7). Though the measurements with a for-ward bias have a clearly different slope than measurements for a reverse bias and the curves show the behaviour of a solar cell. The used solar cell does not have the best IV-characteristics. By plotting the laser power reaching the solar cell against its output at a bias of -0.6 volt (figure 4.8), we do show that the photocurrent depends linearly on the incident power. Thus we can still work with the cell and adjustments in the output of the solar cell are now easy to correlate to differences in the absorbed power.

4.3 Measuring in the linear regime 21

Figure 4.5: A voltage measurement of our bias source. (1) The bias source is turned on. (2) The output of the bias source is turned on. (3) The bias source is set to a 1 V output. (4) The bias source is turned off.

Figure 4.6: IV-curves of the cell we use and of a broken cell. While the broken cell is a straight line, the cell we further used in our experiment clearly showed an IV-curve not perfect but still similar to that expected for a diode.

Figure 4.7: IV-curves of the cell we used with different laserpowers on the cell. With more power on the cell the lines shift downwards according to the formula [2.1]. That there is a current for a reverse bias without light indicates that the solar cell used has a greater than normal dark saturation current (.

Figure 4.8:The light power reaching the solar cell is plotted against its output at a bias of -0.6 V. The graph shows that by measuring with a negative bias we are still working in the linear regime

4.4 Noise 23

4.4

Noise

To be able to use our algorithm, the increase of the output of the solar cell must be bigger than the noise in our system. To find the noise in our sys-tem we first measured the light power of our laser for 100 seconds with a powermeter (figure 4.9). With the laser [70,2 mA] we measured an average of 55.6 µW. The signal has a standard deviation of 0.0338 µW which gives us a signal to noise ratio of 0.06%.

Figure 4.9: The fluctuations of the laser output measured with a powermeter on the spot of the solar cell.

With the laser focused on the solar cell we measured its output for 30 seconds using the lock-in (figure 4.10). The average current from the solar cell is 4.841 µA with a standard deviation of 0.005 µA which gives us a signal to noise ratio in our whole system of 0.11%. This is the threshold for noticeable changes in our system.

Amorphous silicon has light induced degradation. To quantify this we looked at the long term behaviour of our solar cell (figure 4.11). When we look at the part where the laser is turned on we clearly see that the output decreases. After 15 minutes the output has decreased by 2,0%.

Figure 4.10: The output of the solar cell over 30 seconds. The measurement is made with a lock-in amplifier while a constant lightpower of 76µW falls on the cell.

Figure 4.11: The output of the solar cell over 20 minutes. Every 30 seconds a measurement is made. After 90 seconds the laser is turned on with a constant output and 142 µW reaching the cell. At the end of the measurement the current has decreased by 2.4% from when the laser was turned on.

4.5 Wavefront optimization 25

4.5

Wavefront optimization

Figure 4.12 shows the measured current for every step in multiple itera-tions. The iterations show a waveform with a clear minimum and maxi-mum. Iteration 1 has a maximum at 70.9 µA and a minimum at 69.8 µA. The variance in the iteration is 1,6% of the average output which is a fac-tor of 10 bigger than the signal to noise ratio in our system. This shows that the modulation is an effect of the projected phaseplate. The difference between the average output of iteration 1 and 5 is twice as big as the dif-ference between the average of iteration 10 and 15. This indicates that the output is converging to a maximum.

Figure 4.12: In every iteration half of the pixels are cycled from 0 to 2π in 10 steps. The plot shows the measured voltage output of the solar cell for every step in iteration 1, 5, 10 and 15. The four iterations have a sinusoidal fit and all show a clear maximum.

Figure 4.13 shows the output of the solar cell after every iteration for a flat and optimized phaseplate and their ratio. The output for the flat phaseplate is used as a control signal to monitor fluctuations of the laser and the solar cell.

There is a visible, steap, first increase in the output for the optimized phaseplate and for the ratio between the optimized and flat phaseplate. After four iterations the ratio between the optimized and flat phaseplate

has increased by 1.2%. This proves that it is possible to increase the out-put of a solar cell by optimizing the phaseplate of the SLM. Unfortunately we do not know if the difference in output is caused by an optimized ab-sorbance of the solar cell, or an optimized throughput which causes in-creased light power falling on the solar cell. We also do not know if differ-ent parts on the solar cell give other results. These are natural question for future research.

4.5 Wavefront optimization 27

Figure 4.13: The current from the solar cell is measured for a flat phaseplate and the optimized phaseplate. The output of the solar cell shows a clear increase in the first 4 steps. The ratio between the measurement for the optimized and the flat phaseplate is also shown. We clearly see an increase at the beginning after

Chapter

5

Conclusion

In our set-up we control the wavefront of the light that reaches our thin film solar cell. We have shown that using the partitioning algorithm it is possible to optimize a random phaseplate to increase the output of a solar cell. We have shown that we are in the linear regime of our solar cell where an increased output is directly correlated to the increased light power on the solar cell and that the increase in the ratio by 1.2% is bigger than the noise in our system. The sinusoidal waveform that is visible in every iteration of our algorithm is an indication that there is an optimal phaseplate.

Light power is lost in the throughput of our system because of diffrac-tion. This also means that we have not yet proven that we can increase the absorption of the thin film solar cell. It is possible that we are optimizing the throughput of our system. Further research where the throughput of the system and the reflection of the solar cell is measured is necessary.

References

[1] A. P. Mosk, A. Lagendijk, G. Lerosey, and M. Fink, Controlling waves in space and time for imaging and focusing in complex media, Nature Photon-ics 6, 283 (2012).

[2] J. Nelson, The Physics of Solar Cells, Series on Properties of Semicon-ductor Materials, Imperial College Press, 2003.

[3] U. Palanchoke, V. Jovanov, H. Kurz, P. Obermeyer, H. Stiebig, and D. Knipp, Plasmonic effects in amorphous silicon thin film solar cells with metal back contacts, Optics Express 20, 6340 (2012).

[4] O. Isabella, J. Krˇc, and M. Zeman, Modulated surface textures for enhanced light trapping in thin-film silicon solar cells, Applied Physics Letters 97, 101106 (2010).

[5] O. Breitenstein, J. Bauer, K. Bothe, W. Kwapil, D. Lausch, U. Rau, J. Schmidt, M. Schneemann, M. C. Schubert, J.-M. Wagner, and W. Warta, Understanding junction breakdown in multicrystalline solar cells, Journal of Applied Physics 109, 071101 (2011).

[6] P. A. Temple, An introduction to phasesensitive amplifiers: An inexpensive student instrument, American Journal of Physics 43, 801 (1975).

[7] I. M. Vellekoop and A. P. Mosk, Phase control algorithms for focusing light through turbid media, Optics Communications 281, 3071 (2008).