High precision stress measurements in semiconductor structures by Raman microscopy [Elektronische Ressource] / presented by Benjamin Uhlig

Benjamin Uhlig

Sujets

• Physik

Informations

• Publié par: technische_universitat_dresden
• Publié le: 01 janvier 2009
• Nombre de lectures: 66
• Langue: English
• Poids de l'ouvrage: 22 Mo

Extrait

High Precision Stress Measurements
in Semiconductor Structures
by Raman Microscopy
A dissertation of the
TU Dresden, Germany
for the degree of
Doctor of Natural Sciences
(Dr. rer. nat.)
presented by
BENJAMIN UHLIG
Accepted on the recommendation of:
Prof. L. M. Eng, TU Dresden
Prof. A. Michaelis, IKTS Fraunhofer Dresden
2009c
Abstract
Stress in silicon structures plays an essential role in modern semiconductor tech-
nology. This stress has to be measured and due to the ongoing miniaturization
in today’s semiconductor industry, the measuring method has to meet certain
requirements.
The present thesis deals with the question how Raman spectroscopy can be used
to measure the state of stress in semiconductor structures. In the first chapter the
relation between Raman peakshift and stress in the material is explained. It is
shown that detailed stress maps with a spatial resolution close to the diffraction
limit can be obtained in structured semiconductor samples. Furthermore a novel
procedure, the so called Stokes-AntiStokes-Difference method is introduced. With
this method, topography, tool or drift effects can be distinguished from stress
related influences in the sample.
In the next chapter Tip-enhanced Raman Scattering (TERS) and its application for
an improvement in lateral resolution is discussed. For this, a study is presented,
which shows the influence of metal particles on the intensity and localization of
the Raman signal. A method to attach metal particles to scannable tips is success-
fully applied. First TERS scans are shown and their impact on and challenges for
high resolution stress measurements on semiconductor structures is explained.d
Kurzzusammenfassung
Spannungen in Siliziumstrukturen spielen eine entscheidende Rolle fur¨ die
¨moderne Halbleitertechnologie. Diese mechanischen Verspannungen mussen
gemessen werden und die fortlaufende Miniaturisierung in der Halbleiterindus-
trie stellt besondere Anforderungen an die benutzte Messmethode.
Diese Arbeit beschaftigt¨ sich mit dem Thema, inwieweit Raman Spektroskopie
zur Spannungsmessung in Halbleiterstrukturen geeignet ist. Im ersten Kapi-
tel wird der Zusammenhang zwischen Raman Peakverschiebung und mecha-
nischer Spannung erlautert.¨ Es wird gezeigt wie man detaillierte stress maps
in strukturierten Halbleiterproben erhalt¨ mit einer Auflosung¨ nah am Diffrak-
tionslimit. Daruber¨ hinaus wird ein neuartiges Verfahren, die sogenannte
Stokes-AntiStokes-Differenz Methode vorgestellt mit deren Hilfe man Einflusse¨
von Topographie, Gerateef¨ fekten und Drift von den zu messenden Span-
nungszustanden¨ in der Probe unterscheiden kann.
Im nachsten¨ Kapitel wird diskutiert, inwiefern der Ansatz von Tip-enhanced Ra-
man Scattering (TERS), also spitzenverstarkter¨ Raman Streuung genutzt werden
kann um die laterale Auflosung¨ bei Raman Spannungsmessungen zu erhohen.¨
Hierzu wird eine Studie prasentiert,¨ die zeigt, welchen Einfluss Metallpartikel
auf Erhohung¨ und Lokalisierung des Ramansignals haben. Eine Methode um
Metallpartikel an scannbare Spitzen anzubringen wird erfolgreich angewendet.
Erste TERS-Scans werden gezeigt und deren Bedeutung und Herausforderun-
gen bei der hochaufgelosten¨ Messung von Spannungen in Halbleiterstrukturen
wird erlautert.¨CONTENTS e
Contents
Titel a
Abstract b
Contents e
1. Introduction 1
2. Stress Measurements 5
2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1. Raman history . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.2. The Raman effect . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3. Measuring stress with Raman . . . . . . . . . . . . . . . . . 8
2.2. Theoretical considerations . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.1. Raman scattering . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2. Influence of strain on phonons . . . . . . . . . . . . . . . . 10
2.2.3. Uniaxial stress . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3. Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.1. Raman Microscope . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.2. Calibration issues . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.3. Sample preparation . . . . . . . . . . . . . . . . . . . . . . . 20
2.4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4.1. Instrument characterization . . . . . . . . . . . . . . . . . . 21
2.4.2. Capability of the tool and diversity of measurements . . . 36
2.4.3. Industrial chip structures . . . . . . . . . . . . . . . . . . . 41
2.4.4. Stokes Anti-Stokes method . . . . . . . . . . . . . . . . . . 56
2.5. Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.5.1. Defocussing . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.5.2. Linescan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5.3. Finite element stress simulation . . . . . . . . . . . . . . . . 69
2.6. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71f CONTENTS
3. Enhancing the Raman Signal 73
3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.2. Theoretical basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2.1. The diffraction limit . . . . . . . . . . . . . . . . . . . . . . 77
3.2.2. Electrical field enhancement . . . . . . . . . . . . . . . . . . 78
3.3. Experimental issues . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.3.1. Application of metal nanoparticles . . . . . . . . . . . . . . 81
3.3.2. Scanning probe . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.3.3. Tip picking . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.3.4. Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.4. Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.4.1. SERS experiments with metal nanoparticles . . . . . . . . 86
3.4.2. TERS results . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.5. Remarks & Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.5.1. SERS studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.5.2. TERS experiments . . . . . . . . . . . . . . . . . . . . . . . 102
3.5.3. Improved resolution and artifacts . . . . . . . . . . . . . . 104
4. Summary & Outlook 107
A. Appendix 111
A.1. Peak fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
A.2. Mathcad calculations . . . . . . . . . . . . . . . . . . . . . . . . . . 113
A.3. FE simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
B. Bibliography i
List of Figures ix
Curriculum Vitae xiii
Publications xv
Acknowledgement xvii
Erklarung¨ xixIntroduction 1
1. Introduction
Stress or strain is one of the most important material properties in present mi-
croelectronic devices. Film stresses due to thermal mismatch of different layers
can lead to film delamination, formation of voids or cracks and electromigration
induced damage [1]. On the other hand there is process induced stress which
can enhance the mobility of the electric carriers in the transistor gate channel
and thus improve the device characteristics [2, 3]. It is therefor crucial to find a
way to measure the amount and kind of stress in the device.
A well known and widely used method for stress measurements in semiconduc-
tor industry is obtaining the wafer curvature or bow and calculating the stress
situation in the film. However this method gives only an integral value over the
whole wafer. Going one step further one can use interferometric measurements
to gain excess to local wafer curvature. Measuring the lattice constant by the
means of x-ray spectroscopy increases the spatial resolution even more. Though
being very sensitive to small stress variations one is still restrained by the size of
2the measuring spot in the range of 0.5 mm .
Convergent beam electron diffraction (CBED) with the help of transmission
electron microscopy (TEM) lamellas enables measurement in the nanometer
regime. However destructive sample preparation, time consumption and exten-
sive modeling limits this method for use in today’s semiconductor metrology.
A suitable method for measuring stress in semiconductor structures is Raman
microscopy. It is a fast, non destructive technique which offers good stress
sensitivity and spatial resolution. Unfortunately the resolution is limited to the
laser spot size, around 500 nm.
Following Moore’s Law [4], the semiconductor industry is moving to smaller
and smaller structures which results in more complex elements and new kind of
materials like strained silicon or silicon-germanium [5, 6]. Thus, new challenges
arise when physical properties have to be measured. In order to meet the present
demands in transistor dimensions (< 100 nm) one has to use UV laser or to go
beyond the diffraction limit and find new techniques. Tip enhanced Raman
spectroscopy (TERS) is such an approach.2 Introduction
Figure 4.4 shortly summarizes the motivation of this thesis. Starting in the
upper left corner, unwanted stress after film deposition or structuring as well
as intentionally applied stress, for example i