Processing of advanced two-stage CIGS solar cells

MANIKANDAN SAMPATHKUMARMajor Professor :

Dr. Don Morel

Committee members : Dr. Chris Ferekides Dr. Andrew Hoff

A thesis presented in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering

1

Outline• Overview of two-stage CIGS processing• Advancement of the two-stage process• Fabrication• Characterization

• Important Parameters • Controlled Se flux• Changes in growth recipe• Gradients in sample

• Results• I-V characteristics of the device

• Conclusion and Recommendations 2

Two-stage process• The two-stage process favors commercial manufacture, the growth mechanisms

are developed to have optimum elemental usage and to enhance throughput .

Commercial manufacture of CIGS solar cells. ( Courtsey : Miasole )

3

• In commercial manufacture, a conveyor belt with the substrate is moved along at a steady pace and a series of effusion cells are placed to coat the substrate with the elements.

• The idea was to get the same thickness of CGS and CIGS layers, but divide the deposition into many cycles to increase the throughput .

Advancement of the two-stage process

Glass substrates travelling in conveyor belts to be coated with CIGS. ( Courtsey : Miasole )

4

• In a two stage deposition we have a CGS layer and a CIGS layer. The combined thickness of the two layers is the total thickness of the absorber (~2µm).

• The advanced two stage process is based on the idea that the thickness of the CGS and CIGS layers (say 1µm each) instead of getting deposited in one single go, is deposited by series of effusion cells in the order of 200nm of thickness 5 times which sums up to the total thickness of 1µm .

• This will allow the speed of the conveyor belt to increase by a factor of 5.

• Number of cycles = Total thickness of the CGS (or CIGS) layer / Thickness of one layer.

• More the number of cycles, higher the throughput.

• Our aim was to study device performance for different cycles.

• The reproducibility of the solar cells is important to the success of the process.

• We have successfully reproduced 1 cycle of the two stage process.

Advancement of the two-stage process

5

Multisource ThermalEvaporator

• Our multisource thermal evaporator emulates a scenario similar to the commercial manufacture of CIGS solar cells.

• Deposition of elements is controlled by the sub shutters covering the effusion cells.

• One cycle refers to deposition single layer of CGS and CIGS.

Thermal Evaporator for CIGS growth

6

Fabrication Process

Glass Substrate

Back contact Mo

P-type CIGS

N-type CdS

Buffer layer ZnO

Top Contact ZnO: Al

7

Sample storage and cleaning• The storing and cleaning of glass substrates is a sensitive part of the solar cell development.

• Storing the glass samples in polymers, even prior to its cleaning process, can cause the samples to get contaminated. This has a negative effect on the molybdenum back contact, which has a cascading effect on the absorber layer.

(a) Gradients observed in the Mo back contact.

(b) Contamination observed at 30000X on the Mo back contact.

(c) Imperfections in the Mo back contact as a result of improper glass cutting and storage.

8

Nuances in chemical bath deposition• There are two types of growth of CdS films; they are Ion-by-ion growth and cluster-by-cluster

growth. Ion-by-ion growth give rises to A-quality films and cluster-by-cluster growth give rises to B-quality films .

• The speed of the magnetic spinner vital to the outcome of deposition. An increase in the stirring speed causes the arrival rate of Cd+2 and S-2 ions to increase.

a) Uneven grains due to low spinner speed

b) Even grain growth due to increase spinner speed

c) Non uniform deposition due to cluster-by-cluster growth

d) Uniform deposition due to Ion-by-ion growth

9

EDS depth penetration• EDS was extensively used for determining the atomic composition of elements in

the CIGS solar cells.

• The beam voltage determines the penetration depth of the electrons, which gives us an estimate of how far the x-rays are emitted from.

• For the atomic elements present in a sample, EDS beam voltage can be between ten times the lowest peak of interest and two times the highest peak of interest for accurate results.

• Simulations for different beam voltages as a function of depth are done and they are compared to an EDS quantitative analysis with an equivalent beam voltage.

• Parameters such as the elements in the sample, thickness of the sample and the beam voltage are inputted for the simulation of depth profiling.

• This gives us the atomic percentage of elements at various depths.

• 0.9 < Cu/(In+Ga) < 1 and 0.2 < Ga/(In+Ga) < 0.3 are required for good device performance.

10

Real time vs Simulation• EDS analysis on the left hand side shows the atomic percentage of elements for an input

voltage of 15 KeV. It has a Cu/(In+Ga) = 0.99 and Ga/(In+Ga) = 0.42.

• The total CIGS thickness is 2000 nm, simulation shows that the obtained atomic information for a 15 KeV input voltage is from the first 720 nm of the 2000 nm thick CIGS sample.

EDS quantitative analysis for an input voltage of 15KeV

Casino simulation for an input voltage of 15KeV 11

Real time vs Simulation• The bandgap changes as a function of the Ga content.

• EDS analysis on the left hand side shows the atomic percentage of elements for an input voltage of 25 KeV. It has a Cu/(In+Ga) = 0.96 and Ga/(In+Ga) = 0.35.

• Simulation shows the results are from the first 1700nm of the sample. The increase in beam voltage has increased the penetration depth, the higher beam voltage gives an analysis of the bulk.

EDS quantitative analysis for an input voltage of 25KeV Casino simulation for an input voltage of

25KeV 12

Two-Step growth recipe • The two step growth recipe involves the deposition of Copper Gallium Selenide

(CGS) followed by Coppper Indium Gallium di Selenide (CIGS), a copper rich CGS will provide the subsequent layer of CIGS with a larger grain size.

• The figure below shows the metal ratio present with respect to the atomic % of Se. These data points were obtained as a result of series of runs which had varying selenium fluxes.

• At low selenium flux, there may be loss of copper. This is a menace, especially when we are trying to get a copper rich CGS base.

52 53 54 55 56 57 58 59 60 61 620

1

2

3

4

5

6

Cu/In

Cu/Ga

Atomic % Se

Metal ratio

Cu/In and Cu/Ga as a function of Se content13

52 53 54 55 56 57 58 59 60 61 620

1

2

3

4

5

6

Cu/In

Cu/Ga

Atomic % Se

Metal ratio

Two-Step growth recipe

• The Se fluxes were unmonitored during the course of the run and the varying data supports the fact that the process was not reproducible due to uncontrolled Se flux rate.

• Under conditions that give rise to excessive Se incorporation there is apparent loss of Ga as well.

• A low Se flux will deter a copper rich base and reduce device performance and a high Se flux will give rise to loss of Ga. It is very important to keep the Selenium to metal flux ratio at 3-5.

Cu/In and Cu/Ga as a function of Se content

14

Controlled Selenium Flux• The metals namely copper, indium and gallium, after reaching their evaporation temperature

have an evaporation rate which is linearly proportional to temperature.

• The metals practically require no monitoring; they just have to be kept at a constant temperature to provide the necessary flux.

• Selenium after reaching its evaporation temperature, has an evaporation rate which increases exponentially with rise in temperature.

• The Figure below shows the atomic percentage of selenium for eight different depositions, four which were done when the selenium flux was not monitored and four which had their flux rate monitored.

3.6 3.8 4 4.2 4.4 4.6 4.8 5 5.245

48

51

54

57

60

63

Monitored and con-trolled Se flux deposi-tion

Unmonitored Se flux deposition

(Cu+In+Ga)/10

Inc

orpo

rate

d Se

(%)

Monitored vs Unmonitored Se flux depositions

15

Gradients along the sample• After the absorber deposition, there were a few gradients observed across the sample.

• These gradients are due to the thickness variation of the elements along the sample which have the tendency to interfere with the elemental ratios across the sample.

• A recipe prepared for an optimum CIGS solar cell might not be the yield the same device performance throughout the sample.

• The orientation of copper and selenium effusion cells are such that they are at the extremes and tilted more than the other effusion cells. The thickness variation of copper and selenium across the sample was studied.

12

34

0

500

1000

1500

2000

2500

3000

Cu

Se

CuSe

Sample points

Th

ick

nes

s(Ȧ

)

Cu and Se gradients across the sample 16

Device performance as a result of thickness variation

Device performance as a result of thickness variation

• An EDS measurement done after the deposition of cadmium sulphide is prone to a slight error and due to this the EDS data tends to deviate from the normal readings a little.

• This can be seen in figure 32, where a Cu/III ratio of 1.2 has good device properties, but in reality a Cu/III ratio of 1.2 should not yield working results.

17

Growth cycle

Se-5

Å/s

CGS

Ramp Up

CIGS

Cool Down Se-ON

Se OFF

Cu- 2Å/s

Ga-1.5Å/s

Se- 15Å/s

Cu- 2Å/s

In-

2.5Å/s

Ga-1.5Å/s

Se-25Å/s

Se-15 Å/s

CIG - OFF Se - ON

Time (minutes)

 

34

 

27

 

56

 

66

 

86

 

Substrate Temperature

(°C)

 

 

550

 

300

 225

 

• Cu is cut off 5 minutes before the total run time of 56 minutes.

0

 

CIGS growth cycle

18

Growth recipe changes

• The previous two stage growth recipe developed in our lab was poor in stoichiometry reproducibility.

• The assumption was that all the elements had a constant flux rate for a fixed temperature.

• The ongoing formation of secondary thermocouple effects under the effusion cell worsened the condition more as the flux rate of Se was constantly jumping around.

• When the Se flux was better monitored, meaning a certain amount of power was given to maintain a constant flux rate, the growth recipe needed certain changes.

• The changes to the growth recipe are named as process 1 and process 2.

• Process 2 is an advancement of process 1.

19

Process 1• A constant Se flux of 25 Å/s was maintained for process 1.

• The initial CGS deposition was cut off by 5 minutes, making the first stage of deposition close to 24 minutes and the second stage of CIGS deposition was made to run for 27 minutes Cu was cut off 4 minutes before the total run time of 56 minutes.

• Even though reproducible results were obtained, Se flux was used in excess.

0 1 2 3 4 5 60

0.5

1

1.5

2

2.5

3

Cu/InCu/Ga

Sample points

Met

al r

atio

Growth recipe results for process 1

• Process 1 had more Ga than required. It also had Se incorporation in excess.

• Few more changes to the growth recipe were done to reduce selenium wastage and get the optimum ratios, this growth recipe is termed process 2. 20

Process 2• Process 2 has CGS has running for 34 minutes and CIGS for 20 minutes. Cu is shut off 2 minutes

before the total run time of 56 minutes.

• Instead of constantly maintaining the selenium flux at 25 Å/s, we maintained selenium to metal flux ratio of 5.

• The initial CGS run had a selenium flux of 15 Å/s and the subsequent CIGS run had a flux of 25 Å/s.

0 1 2 3 4 5 60

1

2

3

4

5

6

Cu/InCu/Ga

Sample points

Met

al r

atio

Growth recipe results for process 2

• A decrease in Ga and In can be observed and these are in favor of the growth recipe.

• Gradients are an outcome of variation in elemental thickness.

21

Metal/III ratios for process 2

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.50

0.2

0.4

0.6

0.8

1

1.2

Cu/IIIGa/III

Sample points

MetalGroup III

• A Cu/III ratio of (0.9 - 1) and a Ga/III ratio of (0.2 - 0.3) are expected for a good device performance.

• A majority of the Cu/III ratios are below 0.9 due to variation in elemental thickness across the sample.

Metal/III ratios for process 2

22

Cu cut off times• The Cu cut off time towards the end is adjusted such that all the metals are shut during

the end of the run making it an ideal two stage process.

• The growth recipe has a CGS deposition for the first 34 minutes, followed by a CIGS deposition for the next 22 minutes.

• The run is engineered to get a Cu/III ratio of 0.9-1 and a Ga/III ratio of 0.2-0.3.

51 52 53 54 55 56 57 580

0.2

0.4

0.6

0.8

1

1.2

Cu/IIIGa/III

Time (mins)

MetalGroup III

Copper cut off time

• The figure shows how the Cu cut off time is adjusted to get an optimum Cu/III ratio and also have a two stage deposition.

Process 1 Process 2

23

Process 1 vs Process 2• The background of the growth recipe change was to reduce selenium wastage

and increase the atomic percentage of copper, overall increasing the copper to group III ratio.

• Process 2 was successful in producing reproducible stoichiometry with 0.9>Cu/III >1 and 0.2>Ga/III>0.3.

Cu In Ga Se0

5

10

15

20

25

30

35

40

45

50

55

60

65

Process 1Process 2

Elements

Atomic% of elements

Atomic % of elements present in process 1 and process 224

Process 1 vs Process 2

Cu In Ga Se

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

% change in atomic percent between process 1 and 2

% change in atomic percentage of elements between process 1 and

process 2• We can see from the above figure that there is a 24% increase in Cu, an unchanged In, a 47% decline in the Ga content and a 3% fall in the Se content.

• For a CGS deposition, selenium must form various compounds with copper and gallium. The loss of gallium points to the direction of the formation of volatile gallium (GaxSey).

• Even though there is a fall in gallium content, the flux rate of gallium is same in process 1 and process 2. This means the amount of gallium used up is the same in both the process.

• The Ga incorporation loss is in favor of the two step growth recipe as it brings down the Ga/III ratio from 0.34 to 0.24. 25

Atomic composition of elements for good efficiency CIGS solar cell

• The composition for the best device results got from a two stage CIGS solar cell is shown in the figure below.

• The results are from an EDS measurement taken after CdS deposition.

S - 4.58%Cd - 3.23%In - 13.15%Cu - 20.1%Ga - 7.28%Se - 51.66%

Atomic Composition of elements for a good efficiency CIGS solar cell

26

I-V characteristics for SC-53• The solar cells were made with the procedure where the Se flux was unmonitored.

• The result were not exactly reproducible as the stoichiometry of the devices kept changing from run to run.

• SC-53 has a Cu/III ratio of 0.78 and a Ga/III ratio of 0.34.

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6-0.00005

0

0.00005

0.0001

0.00015

0.0002

0.00025

0.0003

0.00035

Dark I-V

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

-0.003

-0.002

-0.001

0

0.001

0.002

0.003

0.004

Light I-V

Dark I-V for SC-53 Light I-V for SC-53

• The device has an efficiency of 5.35 %.

27

I-V characteristics for SC-56• A highest Voc of 520 mV was obtained for Cu/III of 0.66 and a Ga/III of 0.43

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7-0.00005

0

0.00005

0.0001

0.00015

0.0002

0.00025

0.0003

0.00035

Dark I-V

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

-0.003

-0.0025

-0.002

-0.0015

-0.001

-0.0005

0

0.0005

0.001

Light I-V

Dark I-V for SC-56 Light I-V for SC-56

• The efficiency of the device is 3.83 %.

• The increase in Voc is attributed to the increase in Ga/III ratio.

• An increase in Ga/III ratio will increase the bandgap of the device thereby increasing the Voc .

• Even though there is an increase in Voc , the current and fill factors take a hit and reduce the device efficiency.

• This highlights the need for the right stoichiometric balance to get optimum device performance.

28

I-V characteristics for SC-47

• The efficiency of the device is 6.33%.

• The results of the dark and light I-V measurements for SC-53, SC-56 and SC-47 were a combined effort of my lab mate Ryan Anders and me.

• The best efficiency curves were obtained for Cu/III of 0.98 and a Ga/III of 0.35.

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6-0.00010

0.00010.00020.00030.00040.00050.00060.00070.0008

Dark I-V

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

-0.0035-0.003

-0.0025-0.002

-0.0015-0.001

-0.00050

0.00050.001

0.0015

Light I-V

Dark I-V for SC-47 Light I-V for SC-47

29

I-V characteristics for Process 2

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6-0.0000020

0.0000020.0000040.0000060.000008

0.000010.0000120.0000140.0000160.000018

Dark I-V

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

-0.000025-0.00002

-0.000015-0.00001

-0.0000050

0.0000050.00001

0.0000150.00002

Light I-V

• The I-V characteristics show a problem with the top contacts.

• The reduced device efficiency can be attributed to the currents which are in the micro-ampere range.

• The devices with process 2 can be tweaked further with good top contact.

Dark I-V for MSC-11 Light I-V for MSC-11

30

Conclusions and Recommendations• The objective was to control the processing of two stage CIGS solar cells.

• The control of the process minimizes error and enhances throughput during commercial manufacture.

• Different processes were tried to get the right recipe useful for scaling up the process.

• Process 2 was successful in having the right stoichiometric balance with optimum elemental usage .

• The reproducibility and performance depend on parameters such as cleaning and storing samples, maintaining a constant selenium flux, paying attention to the nuances of CBD etc.

• Better devices with process 2 can be achieved with better conductivities of the top contact.

• Reproducibility of a single cycle of the advanced two-stage process was achieved.

31

Thank You !!!

32