• Keine Ergebnisse gefunden

Laser patterning of chalcopyrite and perovskite based solar cells: investigation of the laser-material interaction and laser-induced damages

N/A
N/A
Protected

Academic year: 2021

Aktie "Laser patterning of chalcopyrite and perovskite based solar cells: investigation of the laser-material interaction and laser-induced damages"

Copied!
246
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Laser Patterning of Chalcopyrite and

Perovskite based Solar Cells:

Investigation of the Laser-Material Interaction

and Laser-induced Damages

vorgelegt von

M.Sc.

Christof Schultz

an der Fakultät IV - Elektrotechnik und Informatik

der Technischen Universität Berlin

zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften

-Dr.-Ing.-genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr. Steve Albrecht

Gutachter: Prof. Dr. Bernd Szyszka

Gutachter: Prof. Dr. Bert Stegemann

Gutachter: Prof. Dr. Klaus Lips

Tag der wissenschaftlichen Aussprache: 16. Dezember 2020

Berlin 2021

(2)

Table of Contents

Abstract i 1 Introduction 1 2 Theory 7 2.1 Laser . . . 7 2.1.1 Parameters. . . 7 2.1.2 Laser-material interaction . . . 10

2.1.3 Manuscript: Laser patterning of thin-films . . . 15

2.2 Solar cells - photo-electrical behavior and interconnection . . . 46

2.2.1 Absorption of light . . . 47

2.2.2 Charge carrier separation . . . 49

2.2.3 Recombination . . . 50

2.2.4 Photo-electrical properties . . . 53

2.2.5 Monolithic series interconnection . . . 56

2.3 Solar cells - preparation and material properties . . . 60

2.3.1 CIGSe solar cells . . . 60

(3)

Table of Contents 3 Methods 79 3.1 Laser patterning . . . 79 3.2 Simulation . . . 84 3.3 Measurement techniques. . . 89 3.3.1 Optical microscopy . . . 89 3.3.2 Laser-scanning microscopy . . . 89

3.3.3 Atomic force microscopy . . . 89

3.3.4 Scanning electron microscopy and energy dispersive x-ray spectroscopy. 90 3.3.5 Photoluminescence . . . 93

4 Sample Overview 99 5 Results and discussion 107 5.1 Manuscript: Laser-induced local phase transformation of CIGSe for monolithic serial interconnection: Analysis of the material properties . . . 108

5.2 Manuscript: Revealing and Identifying Laser-Induced Damages in CIGSe Solar Cells by Photoluminescence Spectroscopy . . . 118

5.3 Manuscript: Enhanced Recombination Next to Laser-Patterned Lines in CIGSe Solar Cells Revealed by Spectral and Time- resolved Photoluminescence . . . 128

5.4 Comprehensive calculation of the laser-induced heat flow . . . 136

5.5 Manuscript: Evidence of PbI2-Containing Debris Upon P2 Nano-second Laser Patterning of Perovskite Solar Cells . . . 140

5.6 Manuscript: Ablation mechanisms of nanosecond and picosecond laser scribing for metal halide perovskite module interconnection – An experimental and numerical Analysis . . . 150

5.7 Comprehensive PL investigation of the vicinities of the laser scribed trenches . . 160

(4)

TABLE OF CONTENTS

5.8.1 CIGSe solar cells . . . 163

5.8.2 Perovskite solar cells . . . 166

5.8.3 Application of the laser-material interaction model . . . 168

6 Summary and conclusion 172

A Bibliography 176

A.1 List of tables . . . 176

A.2 List of figures . . . 180

A.3 Bibliography . . . 185

B Symbols, material constants, MATLAB code 211

B.1 Symbols and acronyms . . . 211

B.2 Material constants. . . 214

B.3 Code of the simulation model . . . 215

C Supplementary information 223

C.1 Manuscript contributions . . . 228

C.2 Acknowledgements . . . 235

(5)
(6)

Abstract

In this thesis the laser-material interaction processes for the creation of a monolithic intercon-nection for CIGSe and perovskite solar cells were investigated. By revealing these processes and their influence on the electrical, compositional, and structural properties of the solar cell materials in the vicinity of the laser scribed lines, material-specific ablation models were derived. Thereby, the achieved comprehension of the laser-material interaction processes promotes closing the efficiency gap by adaption of the currently available patterning processes for advancing from single cells to module level.

In order the investigate the laser-material interaction processes fine lines with systematically varied fluence were patterned by a nanosecond laser and picosecond laser with a wavelength of 532 nm. By means of optical microscopy, current-sensing atomic force microscopy, scanning electron microscopy (including energy-dispersive x-ray spectroscopy), photoluminescence spectroscopy and j-V measurements, alterations of the material properties at the laser scribed lines and in its vicinity were revealed.

In the case of CIGSe solar cells, the utilization of ps laser pulses exhibit typically steep edges and a smooth bottom whereas the profiles of the scribe lines, patterned by ns pulses, is different; it changes from slightly bulged edges to rather flat edges with increasing laser fluence. It is concluded that the laser-material interaction processes for patterning of CIGSe solar cells is dominated by thermal processes causing constituent-selective evaporation of the CIGSe, and in the case of ps laser pulses, additional laser-induced stress related damages of the micro-structure. Moreover, irrespective of the utilized pulse duration a region in the range of the laser beam diameter aside the visible scribe line becomes altered by the laser, creating Cu related defect states which promote enhanced recombination resulting in a reduced charge carrier lifetime. Thus, the minimal interconnection width is limited by the width of the laser-affected zones aside the scribed lines.

In the case of perovskite solar cells, the experiments showed that patterning with laser fluences below the ablation threshold of the perovskite results in a notably amounts of PbI2 debris at the laser scribed line. It was shown that by means of ps laser pulses improved electrical

(7)

ii

performance based on lower contact resistances and higher fill factors compared the ns laser scribed lines can be achieved. Moreover, while the morphologies of the scribed lines patterned by ns laser pulses exhibit a strong thermal impact of the laser pulses on the scribe line edges and the underlying front contact layer, the application of ps laser pulses for patterning leads to well-defined sharp and steep scribe line edges with constant width without any sign of thermal treatment. Thus, it is concluded that the utilization of ns laser pulses results in thermally dominated ablation processes whereas the usage of ps laser pulses facilitates mechanically stress-assisted material ablation. In contrast to the CIGSe, in case of perovskite solar cells the analysis of the vicinity of the scribe lines does not indicate laser-affected zones aside, neither for ns laser pulses nor for ps laser pulses.

Based on the derived ablation model for perovskite solar cells, the applicability of this model is demonstrated and a well agreement of the experimental results and the model is achieved. In accordance with the revealed laser-material interaction processes and the derived ablation model optimized P2 and P3 patterning resulted in mini-modules efficiencies of about 19.4%.

Kurzfassung

In dieser Arbeit wurden die Laser-Material-Wechselwirkungsprozesse zur Herstellung einer monolithischen Verschaltung für CIGSe- und Perowskit-Solarzellen untersucht. Anhand der un-tersuchten Laser-Material Wechselwirkungsprozesse und ihres Einflusses auf die elektrischen, kompositionellen und strukturellen Eigenschaften der verwendeten Solarzellenmaterialien im Umfeld der Laser-strukturierten Linien wurden materialspezifische Ablationsmodelle abgeleitet. Durch das dadurch erreichte Verständnis der Laser-Material-Wechselwirkungsprozesse können die derzeitigen Laser-basierten Strukturierungsprozesse gezielt verbessert werden.

Zur Untersuchung dieser Wechselwirkungsprozesse wurden feine Linien mit systematisch variierter Fluenz mit einem Nanosekundenlaser und einem Pikosekundenlaser mit einer Wellenlänge von 532 nm ablatiert. Mittels optischer Mikroskopie, Rasterkraftmikroskopie, Rasterelektronenmikroskopie (einschließlich energiedispersiver Röntgenspektroskopie), Pho-tolumineszenzspektroskopie und elektrischer Messungen wurden Änderungen der Materi-aleigenschaften im Umfeld der mittels Laser strukturierten Linien aufgedeckt.

Im Fall von CIGSe-Solarzellen weisen die Lasergräben bei der Verwendung von ps-Laserpulsen typischerweise steile Kanten und einen glatten Boden auf, während die Profile der mit dem ns Laser strukturierten Linien leicht gewölbte bis flache Kanten aufweisen. Anhand der erzielten Ergebnisse wird gefolgert, dass die Laser-Material-Wechselwirkungsprozesse zur Strukturierung von CIGSe-Solarzellen von thermischen Prozessen dominiert werden, die eine Material-selektive Verdampfung der einzelnen Komponenten des CIGSe verursachen. Darüber hinaus wurden zusätzlich Veränderungen in der Mikrostruktur neben den

(8)

sicht-Abstract iii

baren Grabenkanten, in einem Bereich der dem Laserstrahlduchmesser entspricht, beobachtet die bei der Verwendung von ps-Laserpulsen ausgeprägter sind als bei der Verwendung von ns-Laserpulsen. Diese nicht-sichtbaren Schädigungen konnten Kupfer-bezogenen Defekten zu-geordnet werden die widerum erhöhte Rekombination von Ladungsträgern und infolgedessen verringerte Ladungsträgerlebensdauern hervorrufen. Somit ist die minimale Verschaltungsbre-ite durch die BreVerschaltungsbre-ite der durch den Laser beeinflussten Randbereiche begrenzt.

Im Fall von Perowskit-Solarzellen zeigten die Experimente, dass die Strukturierung mit Laser-fluenzen unterhalb der Ablationsschwelle des Perowskit Materials zu einer beträchtlichen

Menge von PbI2-Resten im Lasergraben führt. Es wurde gezeigt, dass diese Materialreste

mittels ps-Laserpulsen reduziert werden können was eine verbesserte elektrische Verschaltung aufgrund niedrigerer Kontaktwiderstände und damit höhere Füllfaktoren ermöglicht als im Verglich mit ns-Laserpulsen. Die Morphologien der mittels ns-Laserpulse strukturierten Lin-ien und deren Ränder sowie die darunter liegende Kontaktschicht deuten auf einen starken thermischen Einfluss hin. Im Gegensatz dazu weisen die mittels ps Laser struturierten Gräben definierte scharfe und steilen Kanten mit konstanter Breite auf. Anzeichen eines thermischen Einflusses konnten nicht beobachtet werden. Daher wird geschlussfolgert, dass die Verwen-dung von ns-Laserpulsen zu thermisch dominierten Ablationsprozessen führt, während die Verwendung von ps-Laserpulsen mechanische Spannungen hervorruft, welche die Materialab-lation unterstütz. Im Gegensatz zur CIGSe zeigt die Analyse der Umgebung der Lasergräben bei Perowskit-Solarzellen keine durch den Laser beeinflussten Zonen, weder für ns-Laserpulse noch für ps-Laserpulse.

Anhand der Strukturierung des P3 Schnittes an Perowskit-Solarzellen wird die Anwendbarkeit des zuvor aus den experimentellen Ergebnissen abgeleiteten Ablationsmodell demonstriert wobei eine gute Übereinstimmung des Ablationsmodells mit den beobachteten Ergebnissen erreicht wurde. Basierend auf den Erkenntnissen aus den experimentellen Untersuchun-gen zu den Laser-Material-Wechselwirkungsprozessen wurden durch optimierte P2- und P3-Strukturierung kürzlich Minimodulwirkungsgrade von mehr als 19.4% erreicht.

(9)
(10)

CHAPTER

1

Introduction

Solar cells play a key role in the sustainable generation of electrical power to help combatting the climate change. The generation of electrical power by solar cells is based on the photo-electric effect, first observed in the 19thcentury and later elaborated by Albert Einstein. This effect is based on the absorption of light by a semi-conducting absorber material (solar cell). The absorbed light might generate electron-hole pairs which can be separated and extracted to perform electrical work. The ratio of performed electrical power to the incident light intensity per area is defined as (power conversion) efficiency. For solar cells, the efficiency is an important measure of the electrical performance of the device. Thus, the electrical characterization of solar cells is based on a standardized solar spectrum (IEC60904-3) which is applied for the measurement of the current-voltage curve and the subsequent calculation of its characteristic parameters (i.e. open-circuit voltage, efficiency etc.). The optimization of the solar cell efficiency is still in focus of current research, but beside the improvement of the power conversion processes of the solar cells, also the utilized materials are under continuous improvement. In addition, in order to reduce also the material amount for fabrication the so-called thin-film solar cells were developed. Since the late 70’s thin-film solar cells were applied powering the first solar calculators on the market. The first thin film solar cells were based on silicon, but nowadays a broad variety of semiconductor and compound materials are established. In 2019, the global production of thin film solar cells combined to solar modules was about 5.5 GWp for Cadmium-Telluride (CdTe), around 1.5 GWp for CIGSe (Cu(In,Ga)Se2) or, with regard to the lattice structure, chalcopyrite solar cells), and about 400 MWp for modules based on amorphous silicon solar cells [1]. Within the last decade, the so-called perovskite solar cells have been attracted attention due to their remarkable opto-electronic properties and simultaneously low manufacturing costs. Especially, the CIGSe and perovskite solar cells are most promising candidates for inexpensive and highly efficient solar module fabrication. Moreover, the tunability of their opto-electronic properties makes these materials ideal for

(11)

2

tandem applications. In general, thin-film solar cells are composed of two contact layer and a photo-active layer in between with an overall thickness of the layer stack between some hundreds of nanometers to few micrometers, deposited on a substrate material for mechanical stability. Usually, the voltage of one single solar cell is too low for the majority of technical applications, wherefore many cells need to be interconnected in series and a so-called solar module is built. Uniquely for thin-film solar modules is their pinstripe-like, monolithic cell-to-cell interconnection scheme. Thereby, the interconnection is created by alternating layer deposition and layer patterning steps, wherefore it is called monolithic interconnection. The layer patterning steps are a balancing act between selective material removal and the avoidance of unwanted spatial material alterations under the condition of well-defined scribe line geometries. These demands can be fulfilled by a laser, enabling patterning with highest precision and reproducibility without any tool wear. Usually, the laser emits pulses with a Gaussian intensity distribution, whereby the high peak intensities in the beam center can easily exceed critical thresholds of the material whereas the low-intensity flanks might create material alteration aside the scribe line. Nevertheless, in the case of patterning of solar cells for monolithic series interconnection the laser is the tool of choice. But, special care must be taken to avoid material alterations and damages of the surrounding material. In the case of laser patterning of CIGSe solar cells, in the past decades, several material alterations aside the laser patterned trenches were reported. Thus, in the case of the relatively new perovskite solar cells these observations shall kept in mind for the development of new patterning processes. Although several groups worked on the laser patterning of CIGSe and perovskite based solar cells, by now there are no material-specific models available describing the processes of laser-material interaction so that the patterning is rather empirical. Thus, the experimental investigation of the laser-material interaction processes for patterning of CIGSe and perovskite solar cells and the correlation of the observed results with morphological, compositional, and electronical alterations is the aim of this thesis. Moreover, by means of these material alterations laser-induced damages within the material were derived. The experimental results are summarized and a schematic for the laser-material interaction processes with regard to the utilized laser setup is created. The achieved comprehension of the laser-material interaction processes promotes the further optimization of currently available patterning processes for advancing from single cells to module level.

• Motivation

In the past two decades the CIGSe solar cells have been improved by several research institutes and also industrial manufacturers started their fabrication of solar modules [1]. Thereby, numerous efforts were undertaken to the optimize of the monolithic series interconnection. Thereby, especially the replacement of needle based patterning steps by laser based processes

(12)

Introduction 3

became in focus to decrease the interconnection width for better electrical performance and to enhance the reliability of the patterning processes itself. It was observed that the replacement of the needle based patterning processes is challenging due to the thermal sensitivity of the CIGSe material since in-adequate laser patterning parameters easily lead to compositional and electronic material alterations [2–6]. On the other hand side, i.e. Westin et al. [3] reported for the first time to make use of the thermal sensitivity of the CIGSe by selective evaporation of absorber constituents to transform the CIGSe material in a conductive phase instead of its removal for P2 patterning. Moreover, by means of optical inspection a material alterations in the vicinity of the scribed lines was reported [4,7] and investigations of the material composition and crystallinity indicated a partially transformation of the CIGSe in a Cu-rich phase [2,3,5,8]. All of these experimental findings were carried out on different samples which were fabricated by different preparation methods, thus the comparison and the correlation of the observed results is impossible. This underlines the need to investigate the laser-material interaction processes for CIGSe solar cells; whereby the laser scribed lines and its vicinities were analyzed to reveal by complementary measurements these alteration. The utilization of the same solar cell baseline samples enables the correlation of the individual achieved results to gain a comprehensive understanding of the laser-material interaction processes. The following questions need to be answered.

1. Does laser patterning affect the elemental composition of the materials vicinity? 2. How does the composition change in vicinity of the laser scribed lines?

3. How does the compositional change affect the electronic properties of the solar cell? 4. Does the compositional change of the material aside affect the local charge carrier lifetime? 5. Which influence does the laser pulse duration have on the material alterations in the

vicinity of the scribe lines?

On the other hand side, perovskite solar cells and their application for solar modules attracted attention due to the remarkable opto-electronic properties and motivated researchers worldwide to study these promising materials for solar cell applications. Beside fundamental laboratory research, with regard to industrial manufacturing several groups started experimentally the patterning of perovskite cells for monolithic series interconnection and reported of successful patterning [9–13]. But at the same time, also several challenges, such as debris removal and perovskite degradation for patterning were reported. The creation of low-ohmic contact resistances for the interconnecting scribe line (P2) for module fabrication is currently one of the most challenging aspects. By means of a systematic study, facing the removal of debris for P2 patterning, it was presumed that remaining debris of the perovskite absorber layer

(13)

4

turns in a barrier phase (possibly PbI2) impeding low series resistances. Up to now, these challenges are solved by increased scribe line widths an multiple patterning to force debris removal to decrease resistive losses [9,10]. The reported experimental findings are based on samples of different perovskite solutions and preparation methods whereby a correlation of the observed features of laser based patterning in not possible. Thus, the exceptional optical and electrical properties of perovskite solar cells and their interesting potential for tandem-applications motivates the investigation of the laser-material interaction processes. Similar to the investigations of the laser-material interaction processes for CIGSe, the perovskite solar cells were patterned by fine laser scribed lines with systematically varied fluence. Subsequently, the vicinities aside the scribed lines were analyzed focusing on morphological, compositional, and electronical material alterations in the vicinity of the scribed lines. To enable a correlation of the individual experimental results, the same type of perovskite baseline samples were utilized for the experiments. For a comprehensive understanding of the laser-induced interaction processes, the following questions will to be answered.

1. Despite the assumption of PbI2as residues at the scribe line bottom, a clear verification is missing; thus, what are the residues at the scribe line bottom made of?

2. Can the remaining residuals be reduced by using temporal shorter laser pulses? 3. Does laser based patterning affect the vicinity of the scribe line?

4. How does the applied laser pulse duration affect the vicinity of the laser scribed line?

Answering the above raised questions and creating specific models for the laser-material interaction enables a correlation of morphological, compositional, and electrical properties of the scribed lines and its vicinities. These results enable an insight into the underlying laser-material interaction processes whereby the current laser based patterning process for perovskite solar cells can be optimized with regard to low-ohmic series interconnections and minimum laser-affected zones aside the scribe line.

• Objective & approach

This thesis investigates the processes which cause material ablation by the interaction of the laser light with the considered material. After these processes are revealed, a sophisticated model for predicting the laser-material interaction processes is created. At the present time, mainly two different types of laser are widely applied for industrial manufacturing of solar cells, on the one hand side the relatively affordable, small and feasible nanosecond laser and on the other hand side picosecond laser which are more expensive and for technical reasons

(14)

Introduction 5

larger. The ns laser pulses and the thousand times shorter ps laser pulses putatively cause different interaction processes within the considered material, thus the application of these laser pulse durations is reasonable for the investigation of the laser-material interaction and its underlying processes. The utilization of ps laser pulses allows an insight on processes based on “instantaneous” and very localized energy deposition with no or only very low thermal impact. In contrast, the application of temporal longer ns laser pulses causes a more gentle treatment with notably higher thermal impact whereby thermal effects on the material can be studied. For the sake of comparison, the utilized ns and the ps laser were operated at the same wavelength. In this work, a commercially available ns laser as well as ps laser were utilized for the experiments on the CIGSe and perovskite solar cells. The investigation of the laser-material interaction processes for both materials is based on a systematic variation of the laser fluence for patterning of fine lines. Subsequently, the scribed lines were analyzed by microscopic inspection of the scribe line morphology, and its compositional and electronical properties before and after laser treatment were characterized at three representative positions: at the scribe line bottom, at the edge of the trenches, and at remote areas. The microscopic inspection was carried out by reflected light (OM), laser-scanning as well as by scanning electron microscopy (SEM). The elemental composition of the utilized materials was additionally determined by energy-dispersive x-ray spectroscopy (EDX). In the case of CIGSe, the local conductivity of the phase transformed scribe line was analyzed by atomic force microscopy in current sensing mode (c-AFM). In order to investigate the electronic properties of the material aside the scribed line, the photoluminescence (PL) was investigated, either by its intensity imaging, or by spectrally or time resolved measurements. The diagram in Fig.1.1depicts the three main parts of this thesis. This cumulative thesis is structured as follows.

Chapter 2:This chapter gives an overview of the theoretical basics, relevant for the comprehen-sion of the experimental results. In the first part the basics of laser and its interaction processes with materials are introduced. The peer-reviewed manuscript "‘Laser Patterning of Thin Films"’ describes the practical aspects of laser patterning for monolithic series interconnection. In the second part, the utilized solar cells are briefly described and an equivalent circuit diagram is presented and the requirements for laser patterned monolithic series interconnections are discussed.

Chapter 3: The utilized laser patterning tool and its characteristic properties are presented in this section. Moreover, the simulation model of the laser-induced heat flow is introduced and an overview of the utilized measurement methods given.

Chapter 4: An overview of each peer-reviewed manuscript utilized in this thesis is given, including its individual intention, used samples and the applied measurements.

(15)

6

Chapter 5:This is the main part of this thesis. Here, the peer-reviewed manuscripts, summariz-ing the experimental results of the laser-material interaction for both investigated kinds of solar cells are presented. Each manuscript is separately motivated and the results are brought into context with the laser-material interaction processes. Furthermore, additional experimental results are included in order to supplement the findings described in the presented manuscripts. At the end of this chapter for both materials a model of the laser-material interaction with regard to the applied laser pulse duration is given.

Chapter 6: The experimental achievements are concluded and a practical application of the created model for perovskite patterning is given. This chapter closes with an outlook considering further investigations.

Figure 1.1: Diagram of the three main parts of this thesis for the investigation of the laser-material interaction processes.

(16)

CHAPTER

2

Theory

2.1

Laser

The word LASER is an acronym, meaning the technical device which is named after its basic function, the light amplification by stimulated emission of radiation. The operating principle of a laser can be found in the literature of Weber [14] and Eichler [15]. In this thesis, two different laser were utilized for patterning. Thereby, its parameters like wavelength, pulse repetition rate, pulse duration, laser beam diameter and intensity were, with regard to the specific task, adjusted. The influence of the laser parameters on the patterning task is discussed in the attached manuscript “Laser Patterning of Thin Films” (cf. section2.1.3page 2, [16]), except for the laser beam propagation and the laser beam intensity; thus they are introduced in the following.

2.1.1 Parameters

In the following, a laser beam which is emitted by a laser source and propagates in space is considered; its spatial intensity distribution follows a Gaussian function. The laser beam diameter increases continuously, proportional to the distance of the laser source. For distances far away from the laser source, the laser beam diameter increases linearly with the traveled distance. In this case, the diameter increase depends only on the laser wavelength λ described by the divergence angle θ0. The propagation behavior is described by the beam quality factor

M2. M2 has a value of one if the beam propagates without any spatial disturbance; the

utilization of mirrors causes always diffraction, and the divergence angle of the considered laser beam increases resulting in M2 >1. Thus, the laser beam has to be collimated by a lens

(17)

8 2.1 Laser

for guidance to the sample surface. In Fig. 2.1the propagation of the beam behind the laser source is schematically shown, whereby the collimation is achieved by placing a bi-convex lens at the focal point of the utilized lens.

Figure 2.1: Schematic of a divergent laser beam propagation before and after its collimation. Adapted from [15].

Now, the collimated laser beam can be guided through the beam path to the focusing lens where beam becomes focused on the sample surface (cf. Fig. 2.2). The laser beam diameter in focus position 2ω00can be approximated by equation2.1.

Figure 2.2: Illustration of focusing a Gaussian laser beam.

00= 4·λ· f

π·D (2.1)

Here, f refers to the focal length of the utilized lens, λ to the wavelength of the laser beam and D represents in the incoming beam diameter.

(18)

Theory 9

The considered sample surface is hit by the incoming flux of photons; by means of Planck’s constant h, the speed of light c, the wavelength λ of the photons and the photon fluxΦ, the resulting intensity of the laser beam can be calculated (cf. equation2.2).

I = h· c

λ·Φ (2.2)

The incoming laser intensity spreads over a certain cross-section, for a Gaussian intensity distribution the specific intensity can be calculated for each axial beam radius λ(x)along in the direction of propagation and for a specific lateral radius r. The equation2.3describes this relation. I(r,x) = Imax·  ω0 ω(x) 2 ·e−2 r2 ω(x)2 (2.3)

In the special case of r= x=0, the peak intensity Imaxcan be calculated. Usually, the beam intensity in the focal plane x= 0 is of interest, in this case ω0 equals ω(x)and equation2.3 can simplified to equation2.4:

I(r) =I0·e −2r2

ω20 = dP

dA. (2.4)

As a result, from eq.2.4, the power P of a laser beam can be calculated by integrating the intensity over the corresponding area A, described by eq. 2.5:

P= Z IdA=2·π Z ∞ 0 r·I(r)dr= 1 2·I0·π·ω 2 0 (2.5)

Since the temporal evolution of the laser intensity is needed for calculations of the laser-material-interaction (cf. 3.2); an additional term is introduced to describe the applied laser intensity as function of the considered beam radius and calculated time step. Hereby, the temporal evolution of the laser intensity is approximated by a Gaussian curve [14]. By that, equation2.4becomes2.6: I(r,t) = I0·e −2r2 ω20 ·e−4·ln(2) t2 τ2p. (2.6)

Here, τp refers to the full width half maximum (FMHM) pulse duration and t is the calculated current time step. The pre-factor 4ln(2) expresses the width of a Gaussian curve approximating

(19)

10 2.1 Laser

the temporal shape of the laser pulse. After implementation of the temporal-evolution-term, equation2.5 becomes equation2.7:

P(t) = 1 2·I0·π·ω 2 0·e −4ln(2)t2 τ2p. (2.7)

To calculate the temporal evolution of the pulse energy Ep, a special-case approach [17] must be applied, resulting in eq. 2.8[18]:

Ep = Z ∞ −∞P(t)dt=2 Z ∞ 0 P(t)dt= 1 4·I0·ω 2 0·τp· s π3 ln(2). (2.8)

The pulse energy Ep of a single laser pulse can also be derived from the measured averaged power P divided by the pulse repetitions rate PRR of the laser (cf. 2.9), but then the temporal dependence of the pulse energy is lost.

Ep = P

PRR (2.9)

In order to make different laser patterning processes comparable the so-called fluence F is introduced; it is a measure of the optical energy per unit area. It can be calculated by integrating the applied laser intensity distribution over time (pulse duration) resulting in eq. 2.10[18]:

F(r) =2 Z ∞ 0 I(r,t)dt= 1 2 ·I0·τp r π ln(2)·e −2r2 ω20 (2.10)

Since the fluence is always calculated for its maximum value, which is given in the spatial center of the laser beam (r=0), the fluence can calculated by re-arranging equation2.8and substitution of I0in equation2.10, equation 2.11is achieved:

F= 2·Ep

π·ω20. (2.11)

2.1.2 Laser-material interaction

The interaction of light with a surface generally causes absorption, transmission and reflection of light. Their individual shares depend on the specific material properties and the intensity of light. When the laser light hits the material surface, i.e. the light experiences the interface

(20)

Theory 11

between two media of different refractive indices, a defined portion of the incident light is reflected; corresponding to Fresnel law. In the case of perpendicular illumination of the plane, polarization of the light does not matter for the reflectance. In the case of linear optical behavior, the surface reflectance can be calculated by eq. 2.12[19].

R= (n−1)

2+k2

(n+1)2+k2. (2.12)

The degree of reflection R depends on the material specific refractive index n and its extinction coefficient k. After the reflected part of the light is subtracted from the incident light, the remaining portion can be transmitted or absorbed. The length wherein the intensity of the incoming light is reduced by a factor of e is called absorption length. By means of Lambert-Beer’s law (cf.2.13), the portion of light of a certain wavelength (corresponding to the photons energy Eph), which is absorbed within thickness x, can be calculated.

I(Eph) =I0(Eph)·e−α(Eph)·x Eph = h·c λ (2.13)

The inverse of the absorption length is the absorption coefficient α, it describes the degree of linear absorption by means of the extinction coefficient and the lights wavelength λ, shown by equation2.14.

α= 4·π·k

λ . (2.14)

In the case of non-linear absorption, the linear absorption coefficient must be replaced by an intensity dependent absorption coefficient. The remaining part of light which is not absorbed is transmitted through the material.

Temporal temperature evolution

The laser-material interaction starts with the absorption of a certain portion of light (cf. section

2.1.2) and the subsequent interaction with the materials electrons. In case of metals the laser energy couples to free electrons of the electron gas and in the case of semiconductors the laser light interacts with the electrons of the valence band. Under the condition, that the incoming photons become absorbed by the considered material, different processes can be triggered depending on the dose of irradiation. Weak material excitation is achieved by sufficiently low irradiation dose, causing i.e. photoluminescence or introducing modifications

(21)

12 2.1 Laser

of material properties due to heating and melting. In contrast, higher irradiation dose can cause strong material excitation which can result in ionization or full destruction of the material (ablation) [19]. However, irrespective of the utilized pulse duration and the material properties, the absorbed light will be finally converted into heat. Between excitation and the heating of the material, several processes occur. In Fig. 2.3the temporal evolution of the temperature of the excited electrons Teand of the temperature of the solid-state lattice Tl, for a material that has been excited by a laser pulse, is shown [20–23]. Accordingly, the processes occurring in the material can be summarized as follows: Within the first few femtoseconds the incoming light is absorbed within a small fraction of the material by the electrons, which are excited far in excess of the lattice temperature (step I). After laser excitation two competing processes occur. One is ballistic motion of the excited electrons into the material; the other process is the collision-induced energy exchange between excited and non-excited electrons (step II). Some of these electrons diffuse into deeper regions of the excited material at much lower speed than that of the ballistic motion (step III) and transfer their energy to the lattice by electron-phonon coupling (step IV) and also by diffusive transport to the bulk (step V) [20,24]; at this point all excited electrons transferred their energy to the lattice, its temperature Tl equals at this point the electron temperature Te. Subsequently (step VI), the materials temperature increases and melting, boiling and evaporation set in and initiate subsequent process as deformation, voids and cracks resulting in material ablation. Additionally, by means of Fig. 2.3it can be seen that the absorption of the laser pulse forces processes within the material, such as i.e. collisions and ballistic motions, which can introduce alterations of the materials micro-structure before visible changes of the material surface become apparent [25]. Moreover, since the pulse energy of ps laser pulses is deposited in the material before its temperature raises, it enables non-thermal patterning since the influence of spatial heat conduction and screening of the incident beam are diminished and the energy deposition is very localized [19]; in contrast to ns laser pulses whose energy is deposited in time scales where spatial heat conduction becomes relevant and screening of the incident beam might occur. Hence, the creation of a model of the laser material interaction processes for patterning of CIGSe and perovskite solar cells is based on the utilization of picosecond and nanosecond laser pulses enabling the observation of thermal (ns) and putatively non-thermal (ps) laser-material interaction processes.

Laser ablation processes

Laser ablation refers to laser-induced processes of material destruction or material disintegra-tion by means of rapid changes of the material temperature, its density and ambient pressure which causes subsequently material removal. In the last decades different models were de-veloped to describe the experimentally observed material behavior after laser excitation. In general, immediately after illumination the laser energy is spread within the volume.

(22)

Depend-Theory 13

Figure 2.3: Schematic of the temporal evolution of the electron and lattice temperature of laser-excited solids. Source: [21]

ing on the introduced laser fluence and applied laser pulse duration a characteristic spatial and temporal temperature profile is established, whereby the applied laser pulse duration determines the heating rate of the material. In the following typical laser-induced processes causing material ablation are described.

• Melting

When the temperature of a certain solid material approaches its melting point, nucleation occurs at the material surface. The lower pressure at the surface initiates nucleation and melting of the surface and the melting front propagates towards the bulk material [19]. This process is commonly named heterogeneous nucleation with subsequent melting and occurs typically at low heating rates. Turnbull et al. [26] investigated the behavior of the melting front moving towards the bulk and determined the speed of sound as maximum velocity for the moving melting front within solid materials. The time, within the heated layer becomes molten was also investigated, in the case of a 100 nm thick silicon layer the melting front spread across the material within 250-500 ps corresponding to a velocity of 200-400 m/s [27,28]. In contrast, homogeneous nucleation and melting occurs especially for heating rates in the range of tens of 1012K/s which initiates over-critical heating of the material. Thereby the lattice temperature

(23)

14 2.1 Laser

exceeds by far the melting point and as a result nucleation occurs simultaneously at different depths within the considered volume and homogeneous melting set in [27]. Depending on the strength of excitation and crystallinity of the considered material, homogeneous melting occurs approximately on ps timescale [28].

• Vaporization and boiling

The process of material vaporization is characterized by particles leaving the surface of a boiling liquid phase. Miotello et al. [23] investigated the boiling liquid phase of a heterogeneously nucleated metal surface and determined the rate of leaving particles with approximately 1 atomic layer per nanosecond. It was concluded, that this material removal process is too slow and can be neglected for laser-induced ablation [29]. In contrast, pulsed laser cause over-critical heating of the material which undergoes the step-by-step temperature increase by phase changes [27,28]. In the case of ultrashort laser pulses it is assumed that homogeneous nucleation and phase explosion (also called explosive boiling) processes directly set in [23]. Phase explosion occurs within over-critically heated liquids when homogeneous boiling starts [30]. At this meta-stable state any material inhomogeneity causes instantaneously the creation of gas bubbles propagating immediately to the surface expulsing liquid droplets and vapor [31]. Beside the above described processes, for short (ns), high intense laser pulses further processes might occur. It was shown by Sokolowski-Tinten et al. [32] that short pulses can heat a thin surface layer instantaneously to fluid state with nearly solid density at high temperature which causes as a result mechanical stress in the underlying material in the range of several tens of GPa. The rapid expansion of the pressurized liquid film initiates partial cooling of the material; running in a liquid-gas coexistence regime whereby decomposition into an inhomogeneous phase of gas and liquid at low density takes place [27,32]. The shares of melting, vaporization and boiling depend basically only on the applied pulse intensity; thus, as a result of the Gaussian intensity distribution of the laser pulse all these processes can occur simultaneously [19].

• Expansion and mechanical stress

The process of material ablation is accompanied by thermal expansion of the excited material. After excitation, the considered material is converted in a highly non-equilibrium state of high temperature and high pressure. The material starts to expand and causes mechanical stress within the neighboring layers. In the case that the mechanical stress exceeds the tensile strength of the material, non-reversible damage and material destruction occurs, followed by ablation [33]. In order to approximate the maximum stress, Meshcheraykov [34] introduced a

(24)

Theory 15

simplified approach based on a solution of the thermo-elastic behavior for a round plate with fixed edges and a constant temperature gradient across the edge, described by equation2.15

σmax=

E·α·∆T

2(1ν). (2.15)

The approximation of the maximum mechanical stress is based on the the materials Young modulus E, the linear thermal expansion coefficient α, the temperature difference∆T and the poisson ratio ν.

2.1.3 Manuscript: Laser patterning of thin-films

Table 2.1: Basic information of the peer-reviewed manuscript: Laser Patterning of Thin Films

Title: Laser Patterning of Thin Films

Authors: Bert Stegemann, Christof Schultz

Book title: Digital Encyclopedia of Applied Physics

Year: 2019

Pages: 1-30

DOI: 10.1002/3527600434.eap830

This work is an invited paper in the “Digital Encyclopedia of Applied Physics”. The article aims to give an overview of the laser-material interaction and practical aspects for laser based patterning of thin- films. In this thesis the article is intended to explain the basics relevant for laser based patterning of thin-films, supported by examples for practical applications, based on the own work.

(25)

1

Laser Patterning of Thin Films Bert Stegemann and Christof Schultz

HTW Berlin – University of Applied Sciences, Berlin, Germany

Abstract

Laser-based patterning of thin films has rapidly revolutionized materials processing over the last years and is increasingly replacing conventional patterning techniques such as photolithography or mechanical patterning. Currently, it is widely used for both fundamental research and practi-cal applications. According to the desired requirements, laser and processing parameters can be specifically selected for selective, precise, reproducible, and nearly damage-free patterning of sin-gle layers and layer stacks of a wide variety of materials. This article summarizes the fundamentals and basic processes of laser ablation and discusses the influence of the essential laser parameters on laser–matter interaction. Moreover, an overview on practical aspects such as different optical setups, process control and approaches for selective layer ablation are given. A variety of current examples for the application of lasers for patterning thin films is introduced with a clear focus on photovoltaic applications. These examples highlight that laser patterning is a versatile, flexible, and reliable technique for processing a wide range of thin-film materials, which will be increasingly implemented in the future.

Keywords laser; thin films; ablation; patterning; solar cells

1 Introduction 1

2 Fundamentals 2

3 Practical Aspects 13

4 Application Examples 19

5 Summary and Outlook 26

Acknowledgments 27

Related Articles 27

References 27

1 Introduction

Laser material processing has become an essential manufacturing technique in thin-film technology. In particular, in thin-thin-film

solar cell processing, the structuring (or scribing) of the functional layers is typi-cally carried out with lasers of different wavelengths, pulse duration, and pulse shapes. The materials to be processed are

Encyclopedia of Applied Physics.

© 2019 Wiley-VCH Verlag GmbH & Co. KGaA. DOI: http://dx.doi.org/10.1002/3527600434.eap830

(26)

manifold, comprising metals, amorphous and crystalline semiconductors, dielectrics, transparent conductive oxides (TCOs), organic and organic–inorganic hybrid mate-rials. But laser patterning is not only found in photovoltaics. For example, the patterning of TCO layers is also performed in the manu-facturing of flat panel displays, touchscreens, or organic light-emitting diodes.

What all thin films to be patterned have in common is that the typical layer thickness is between a few hundred nanometers and less than 10 μm. But often not only single thin layers on a substrate must be processed, but there is a stack of several layers, of which only a single layer is to be patterned, while all other layers should remain unaffected. When patterning, e.g. thin-film solar cells, it is important to process the relevant layers in such a way that the material is removed in very fine lines. In this case, the material must be completely removed without damaging or modifying the edge of the scribed lines or the substrate or the underlying layers. This represents an enormous challenge from the technological as well as from the materials science point of view. Due to this constraint, lasers emitting short and ultrashort pulses (in the nanosecond, picosecond, or fem-tosecond range) are required for thin-film patterning to avoid heat effects, rather than continuous-wave (cw) lasers, which are, e.g. applied for drilling and welding [1] or in laser annealing [2].

The aim of this article is to provide the comprehensible basics for thin-film laser pat-terning and to present illustrative application examples. The focus is put on details of our own and the core of further published work in the utilization of lasers in patterning of thin films in photovoltaics. Accordingly, this article is organized as follows: It begins with a fundamental part that introduces the main properties of laser radiation and laser pulses that are of importance for laser-based mate-rial processing, followed by a description of the principles of laser–material interactions and the basics of thin-film patterning in pho-tovoltaics. Subsequently, practical aspects

are discussed, such as the advantages of laser structuring over conventional patterning methods, the advantages and drawbacks of different optical setups, approaches for selective layer ablation as well as techniques for controlling and analyzing the patterning result. It concludes with a description of application examples and recent develop-ments in the laser patterning of thin films, focusing on materials that are predominantly used in photovoltaics.

2 Fundamentals

2.1 Lasers

Lasers can be distinguished by the state or the physical properties of their laser-active material (e.g. solid state, semiconductor, dye, gas lasers) or by their type of excitation (e.g. optical by means of discharge lamps or diodes, current injection, electron beam, gas discharge). With respect to laser-based material processing, particularly the follow-ing parameters that characterize the laser radiation are relevant:

• Wavelength (𝜆): influences optical absorp-tion, reflecabsorp-tion, transmission, resolution • Pulse duration (𝜏p): influences interaction

time, transient processes

Pulse energy (Ep): influences fluence, abla-tion rate

• Beam profile: determines excess energy and heat energy input

• Beam diameter (2𝜔0): influences spot size, depth of focus, intensity, fluence

• Spatial and temporal coherence.

2.1.1 Gaussian Beams

Usually, the emitted radiation in the funda-mental TEM00 mode can be approximated by a Gaussian profile. Thus, the beam in the direction of propagation possesses a radially symmetric intensity distribution, which allows easy and stable focusing as it is essential for defined and controlled laser patterning. Such Gaussian beams possess a caustic as shown in Figure 1. In this figure, r is the distance to the radially symmetric axis,

(27)

Laser Patterning of Thin Films 3 ω(z) ω0 +z0 –z0 2zR r θ z

Figure 1 Gaussian laser beam: width𝜔(z) as a function of the distance z along the beam, 𝜔0, beam waist; zR,

Rayleigh range; 2zR, depth of focus; Θ, total angular spread. zis a point in the propagation direction, 𝜔 is the beam radius,𝜔0is the beam radius at

z =0, and Θ is the total angular spread. At the beam waist, i.e. at z = 0, the beam radius has its smallest value.

A prominent feature is the diffraction-limited beam diameter 2𝜔0, which diverges rapidly as it propagates away from the focus. At the Rayleigh length zR, the diameter of the beam is increased by a factor of√2 with respect to the focus for a given wavelength𝜆, as described by

zR= π𝜔2

0

𝜆 (1)

Apparently, the Rayleigh length behaves inversely proportional to the laser beam diameter. Thus, a large beam diameter is necessary for a minimal (diffraction-limited) focus diameter. Both beam waist and Rayleigh length not only depend on initial beam diameter but also on focusing optics.

The spatial Gaussian intensity distribution and its characteristic features are shown in Figure 2. Considering the spatial dependence of the intensity, it approaches zero for r→ ∞. From this fact, several definitions of the position of the beam radius have emerged. Widely used definitions are where the peak intensity is decreased to a value of 1/e or, alternatively, to a value of 1/e2times of the maximum intensity value.

The intensity I of the Gaussian beam at the focal plane as a function of the radius r results

from the integration of laser power P over the area A as follows:

I(r) = dP

dA=I0•e −2r2

𝜔20 (2)

In certain cases, laser diodes might be used for thin-film patterning. Their differ-ent parallel and perpendicular divergence angles result in elliptical beam spots, instead of a circularly symmetric Gaussian profile. Circularizing a collimated laser diode beam requires special optics, such as cylindrical lenses or spatial filters.

2.1.2 Pulse Duration

The laser pulse duration 𝜏p is defined as the full temporal width of the pulse at half maximum (FWHM). In pulsed laser material processing, the energy of the laser pulse is delivered to the free electrons of the material, which transfer their energy into the lattice.

Depending on the duration of the laser pulses, different physical processes can dom-inate the ablation behavior, see Figure 3a,b. In principle, there are two cases: (i) If the laser pulse is shorter than die duration of the inter-action between the electrons and the lattice, the energy transfer occurs after the end of the pulse, leading eventually to material removal or plasma formation (i.e. nonthermal or cold ablation). (ii) If the laser pulse is longer than die duration of the interaction between the electrons and the lattice, energy is still induced after the electrons start transferring

(28)

36.7% 13.5% I0/e I0/e2 0 ω0 r(ω) 2ω0 I(ω) I0

Figure 2 Intensity profile of a Gaussian laser

beam (TEM00mode).

Nanosecond laser pulses (ns) Ultrashort laser pulses (short ps, fs)

Plasma shielding

(a) (b)

Thermal penetration Optical penetration Ablation plasma

Figure 3 Laser–matter interaction for short (ns) (a) and ultrashort (ps, fs) (b) pulses.

their energy into the lattice. This excess of energy diffuses into the material in the form of lattice vibrations, i.e. phonons, and causes an increase in the material temperature, the formation of melt and vapor, and eventually material removal.

Typical collision times between electrons and phonons are several hundred fs. How-ever, to fully transfer the energy of an excited electron to the lattice, multiple collisions and phonon emission processes are needed [3–5]. The corresponding electron–phonon relaxation time is therefore in the range of a few 10 ps, e.g. in semiconductors in the order of 1–20 ps [6]. At this time, thermal equilib-rium between the electrons and the lattice is

obtained, i.e. the electron temperature equals the lattice temperature: Tel=Tl.

If a laser pulse is longer than this mate-rial specific relaxation time (i.e. above

𝜏p ≈ 10 ps), various physical phenomena might occur that make precise material pro-cessing challenging. A plasma builds up of free charge carriers caused by the interaction of the laser beam with the material surface. The growing plasma plume attenuates the laser pulses and shields the material from incident laser pulses [7, 8]. In addition, a heat-affected zone is formed, leading to melt formation. In the processing of compounds or multicomponent systems, stoichiometric changes of the material can occur due to different melting and evaporation enthalpies.

(29)

Laser Patterning of Thin Films 5

By cooling and condensation of the material, the material can change depending on the temperature gradient into a crystalline or amorphous structure. The flaking of the ablated particles or excessive heating can thus lead to a shock-affected zone and to undesired cracks in the material.

Material processing with ultrashort laser pulses (𝜏p < 10 ps) might have several advantages. In this case, the laser energy is deposited into the electrons of the material on a timescale short compared with the relax-ation time (cf. Figure 3b). Thus, the electrons and the lattice are not in thermal equilib-rium. The electron and lattice system are treated as two subsystems with temperatures

Teland Tlin the so-called two-temperature model [9]. The system is driven into a highly nonequilibrium state, and the process will have a nonthermal nature. Material removal begins when the pulse is already over [4, 5, 10]. No plasma plume can develop, and plasma shielding is reduced or even avoided, that the entire radiation energy, except for the reflected portion, can be used for mate-rial removal. The thermal penetration depth

Lth,which determines the heat-affected zone, can be estimated as follows [7]:

Lth=2√D𝜏

p (3)

D is the diffusion coefficient and 𝜏p the pulse duration. Before the electrons deliver their previously absorbed energy to the lattice, they travel a certain distance in the material. The average distance Le is determined by the following equation [7]:

Le= √

De•𝜏e (4)

Deis the diffusion constant of the electrons and𝜏eis the time in which the electron gas has transferred the energy to the lattice. Since

𝜏eis much smaller than𝜏p,it follows Le≪ Lth. Accordingly, the heat input decreases in prin-ciple with shorter pulse durations. By using shorter pulse durations, it is also possible to ablate finer structures, because the energy is introduced more locally and lateral heat flux is avoided. Thus, the ablation efficiency

is improved, the threshold fluence decreases [11]. Since heat conduction into the mate-rial is negligible, the heat-affected zone is reduced. Also shock and thermal stress are reduced, since no melt is present. However, the higher intensity of these ultrashort pulses generates nonlinear effects within the mate-rial like avalanche breakdown or nonlinear absorption whereby the kinetic energy of the excited electrons is already sufficient to modify the material at scales smaller than the optical penetration depth or the diffusion length [7, 12]. Moreover, it was found that even for such short pulses the pulse energy can be accumulated within the material depending on the thermal properties of the material itself, resulting in heat-induced material modifications [12].

2.1.3 Wavelength

The physical basis of the laser material inter-action is absorption, i.e. the transfer of as much of the incident laser energy as possi-ble to the sample to be processed. Besides the pulse length, the amount of energy absorbed depends on the laser wavelength

𝜆 and on the absorption properties of the

material.

The emitted laser wavelength is deter-mined by the laser transition in the specific active medium and the resonator. By using nonlinear processes, doubling, tripling, or chirping, the emitted frequency nearly every wavelength can be generated and specifically selected for the materials to be processed. Currently, the available range extends from X-ray (∼3 nm) till the far infrared (∼1 mm) [13–15]. According to Planck’s equation, the wavelength 𝜆 is correlated to the photon energy EPh:

Eph=hc

𝜆 (5)

When the laser light hits the material sur-face, i.e. the light experiences the interface between two media of different refractive indices, a defined portion of the incident light is reflected, corresponding to Fres-nel laws. The remaining part is absorbed

(30)

or transmitted by the material, depend-ing on the optical and geometric sample properties.

The absorbed radiation has different opti-cal penetration depths. To remove material by laser ablation, sufficient absorption of the laser radiation by the material is necessary. Since the absorption depends on the laser intensity, two characteristic cases are consid-ered, i.e. material interaction with low and high laser intensities. For weak intensities, according to Beer’s law, the intensity I at a depth x in the material is given by

I(x) = I0e(−𝛼x) (6)

Accordingly, the intensity of the radiation decreases from the initial intensity at the sample surface I0 with increasing penetra-tion depth, as determined by the material-and wavelength-dependent absorption coefficient𝛼. In semiconductors, low inten-sity light can only be absorbed if Ephexceeds the bandgap energy Eg(i.e. Eph< Eg). Nonlin-ear absorption is neglectable. In the case of high laser intensities, additional non-linear effects such as multiphoton absorption and avalanche ionization become more likely and increase the absorption of the mate-rial. Thereby, photons with energies smaller than the bandgap energy (Eph < Eg) might be absorbed. This happens especially with transparent materials at laser wavelengths in the near-infrared spectral range and depends on the photon density and its interaction cross section [7].

The optical penetration depth l𝛼 of the laser radiation and thereby the volume in which the laser energy is deposited are given by

l𝛼=𝛼−1 (7)

Accordingly, l𝛼depends on the absorption coefficient and thereby on the laser wave-length. Thus, depending on the individual process demands, the utilized laser wave-length must be chosen carefully and further aspects such as sample geometry, desired quality of the patterned structures, and the excited volume must be considered.

2.1.4 Fluence

For the representative and comparable description of the material processing by laser, the laser fluence F is usually specified. The fluence is defined as the incident energy per effective area, which is dependent on the Gaussian profile of the laser pulse.

A distinction is made between the peak fluence F0and the threshold fluence Fth. The peak fluence indicates the maximum fluence of the laser spot, while the threshold fluence indicates – in the present case – the onset of the material removal. The fluence F results from temporal integration of the intensity (cf. Eq. (6))

F(r) = F0•e −2r2

𝜔20 (8) This relation between the radius of the ablated area on the surface and the peak fluence F0 was established by Liu [16]. For determination of the threshold fluence Fth, the peak fluence is systematically varied and the diameters of the corresponding ablated areas are determined. From Eq. (8) it follows that

d2= (2r)2=2𝜔2

0•[ln(F0) −ln(Fth)] (9) Accordingly, from a so-called Liu plot of the squared measured ablation diame-ters against the logarithmic peak fluence, the threshold fluence can be determined by a fitting line, see Figure 4. Moreover, from the slope of the linear fit, 𝜔0 can be deduced.

Since the pulse energy is (i) obtained by integration of the peak fluence over the area and (ii) related to the pulse repetition rate fp and the average power Pm, the peak fluence can be expressed as follows:

F0= 2•Pm

fp𝜋𝜔20 (10) Thus, the peak fluence can be easily cho-sen in the experiment by adjusting the aver-age power and the pulse repetition rate.

For the patterning of photovoltaic mate-rials, often a certain width of the scribe must be achieved to ensure proper electrical isolation or to provide a sufficiently large

(31)

Laser Patterning of Thin Films 7

Figure 4 Liu plot for the determination of the

threshold fluence (ablation threshold).

Threshold fluence

Fth

Peak fluence F0 (log scale)

Square diameter

d

2

distance for low contact resistances. In the case of thin-film solar cell patterning, the diameter of the laser spot is required to be in the range of about 25 μm. To estimate the laser beam diameter at the focal plane, a simplification of a parallel laser beam of a wavelength𝜆 with a radius of 𝜔pcan be used, as long as the beam radius is far beyond the dimension of the wavelength:

𝜔′ 0= 𝜆f 𝜋k𝜔 p (11) where f is the focal length of the lens and k the beam propagation factor, which is calcu-lated using the relation k = 1/M2. Moreover,

in order to calculate the length of the waist of the laser beam at the focus position z

R, Eq. (12) can be used, which is similar to Eq. (1) but it also takes the beam quality into account:

z′ R=𝜔20

k𝜋

𝜆 (12)

Both beam radius and fluence describe the laser beam properties; thus they are indepen-dent from the material properties.

2.2 Principles of Thin-Film Laser Ablation

2.2.1 Laser–Matter Interaction

In pulsed laser material processing, the energy of the laser pulse is delivered to the free electrons of the material, which transfer their energy into the lattice. In principle,

there are two cases: (i) If the laser pulse is longer than die duration of the interaction between the electrons and the lattice, energy is induced after the electrons start transfer-ring their energy into the lattice. This excess of energy diffuses into the material in the form of lattice vibrations, i.e. phonons, and causes an increase in the material temper-ature and melt and vapor formation. (ii) If the laser pulse is shorter than die duration of the interaction between the electrons and the lattice, the energy transfer occurs after the end of the pulse, leading to material removal in gas phase or plasma phases (i.e. nonthermal or cold ablation).

For most solid inorganic materials, electron-phonon relaxation time is in the range of 100 fs–10 ps. If a laser pulse is longer than this material specific time, elec-tron and lattice temperatures have not to be discussed separately.

2.2.2 Film-Side and Glass-Side Ablation

Laser patterning of thin films is usually based on ablation and etching, driven by thermal, physical, mechanical, or photochemical pro-cesses [8, 17, 18]. Usually, laser radiation is absorbed and converted into heat so that subsequently melting and evaporation occur. Here, patterning is discussed in terms of local material removal of thin solid layers by means of short and ultrashort laser pulses. In the case of short pulses (ns), the interaction of the beam with the evaporated material

(32)

plays a significant role in the patterning process whereby it can be neglected for ultrashort (i.e. short ps, fs) pulses [8]. There are at least two distinguished cases delivering the laser beam to the sample, from the front side or film side (i.e. direct ablation), which is the most used regime, and from the back side or glass side (i.e. induced ablation) of the sample whereby a transparent substrate is necessary. In the following sections, both approaches are discussed. A mixture of both basic approaches, i.e. direct induced ablation, is also possible.

2.2.3 Direct Ablation

Direct ablation can be considered the

clas-sical laser patterning approach. Typically, it is driven by absorption, heat transport, melting, and evaporation of the material [19]. In more detail, the process steps are illustrated in Figure 5 and can be explained as follows: First, the laser beam irradiates the sample surface where it is absorbed. The absorbed radiation is converted into heat, which diffuses into deeper regions (step 1). The second step is dominated by further expansion of heat, leading to melting and evaporation and eventually to local strain. Depending on the respective pulse length, interaction with the rising vapor or plasma might occur. For short pulses, which are over before any material is removed from the surface, mainly a further expansion of heat into deeper regions occurs. For longer pulses, screening of the laser beam might set in. As a result, excited species (due to evaporation and ionization) leave the sam-ple surface, and the still lasting laser pulses are attenuated by the plume (step 3) and Plasma is formed, which further grows till it collapses [7]. For very high intensities and long pulses, the plasma can become process limiting. The third step describes further heat expansion within the material, which leads to liquid or even gaseous expulsion, which is commonly named disintegration or ablation.

A well-known example for direct ablation laser patterning is the edge isolation process

of Si wafer–based solar cells. There, a laser scribes a trench along the outer edge of the wafer in order to cut the emitter layer of the Si wafer. This process is utilized to improve the shunt resistance, thereby avoid-ing performance losses of the solar cell by preventing leakage currents along the wafer edges from the emitter to the base contact [20]. A further example is the P2 patterning step in the monolithic serial interconnection of chalcopyrite thin-film absorber using Cu(In,Ga)Se2(or CIGSe) as absorber layers (cf. Section 2.3.1) [21].

2.2.4 Induced Ablation

If the substrate is transparent, e.g. glass, laser patterning can be performed through the substrate, and one can use the so-called induced ablation. Figure 6 illustrates the three characteristic steps. The laser beam passes through the transparent substrate and is absorbed at the interface of the material to be processed (step 1). The temperature at the interface rises and fast thermal expansion or internal plasma formation leads to melting and evaporation of the material, which in turn produces local strain (step 2). If the energy input is sufficiently large, fracturing and also further melting and evaporation might occur and results in ablation, ejection, and expulsion of the material under laser impact (step 3). Thus, this process is often referred to as laser lift-off [22]. For the sake of completeness, it should be added that, conversely, lift-off processes can be adopted to remove dielectric films from opaque sub-strate, such as Si wafers, by single-shot ps and fs laser pulses [23].

Several experiments have demonstrated that laser patterning by induced ablation has several advantages. The required laser pulse energy is generally lower than for direct ablation, since not the entire volume to be removed has to be evaporated [19]. Moreover, screening by the plume is avoided. Further advantages are patterns of higher quality, i.e. sharp-edged patterns, and smaller heat-affected zones [18, 24, 25].

(33)

Laser Patterning of Thin Films 9 Absorbing layer Substrate Expansion of heat Vapor/plasma plume Step 1 Step 2

Depending on pulse duration Interaction with plume Shielding Melting, vaporization Step 3 Expansion of heat Vapor/plasma plume Liquid/gaseous expulsion (a) (b) Laser beam Laser beam

Figure 5 Schematic representation of the characteristic steps of direct ablation. Source: Bovatsek et al. [18].

Reproduced with permission of Elsevier.

The physical processes enabling induced ablation of thin films were elucidated by time-resolved microscopy and considered laser pulse penetration depth and film thick-ness [22, 26]. Two limiting cases of laser

lift-off processes were identified: For pene-tration depths larger than or approximately equal to the layer thickness, the layer is completely melted and drips off. This pro-cess is called laser-induced forward transfer,

Referenzen

ÄHNLICHE DOKUMENTE

In summary, the comparison of PSCs using PCBM and PPCBM layer reveals that diffusion of PCBM molecules into grain boundaries of a polycrystalline perovskite film takes place and

It turned out during experimental analysis that pulse length has a significant influence on quality of material processing especially for use of ultra-short pulse laser beam

With varying degrees, their surfaces were irregular and rough compared to those from PZE osteotomy (Figure 5.6, 5.8, 5.9, and 5.10). This finding agrees with what Panduric et al.

6–8 Based on these results, magnetic patterning of perpendicular multilayer films has been exploited using fo- cused ion beams 9 or ion irradiation through a mask 6–8,10 to create

The resulting catalyst film was structured by an interference pattern by means of multiple nanosecond laser pulses, using the second harmonic of a Nd :YAG

For shorter pulse durations in the low nanosecond scale, the laser intensity and therefore the evaporation velocity and the recoil pressure is very high, whereas the melt time is

The behavior of the threshold fluence can be explained by varying absorption (due to changes in the surface reflectivity), chemical changes of the surface (e.g. due to

According to the equation (4) this is a consequence of the strongly increasing threshold fluence and decreasing penetration depth with increasing pulse duration (cf. Figure 5a):