Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

CHAPTER 5.4

Pulsed Laser Deposition

Jeffrey T. Cheung

Rockwell

Science

Center

5.4.1

INTRODUCTION

A physical vapor deposition process must have three basic components: a source

material, a substrate, and an energy supply to transport material from the source

to the substrate during film deposition. Of these three components, the form of

energy and its associated mass transfer mechanism are what set one technique

apart from another. In fact, thin-film deposition techniques are often named after

these mechanisms, such as sputtering deposition and E-beam evaporation. Pulsed

Laser Deposition (PLD), as the name suggests, uses a high-energy pulsed laser as

an external power source to ablate the source or target material

[1,2].

Interaction

is short but intense, and induces ablation via a cascade of complex events. Theo-

retical and experimental studies of this phenomenon have always been a subject

of intense interest and have spawned new ideas in laser applications. One idea,

Pulsed Laser Deposition, was first demonstrated by Turner and Smith [3] nearly

three decades ago and has become a widely accepted technology of today.

Despite its early inception and demonstration, PLD development was slow and

sometimes stagnant until the 1980s. The reason is twofold: the lack of a niche for

this technique to compete against the more mature thin-film growth technologies,

and the lack of reliable and affordable high-energy lasers. It has remained an ob-

scure research tool for nearly two decades. Its first explosive growth was triggered

by the successful preparation of high-quality superconducting YBa2Cu307_x films

by PLD [4]. At the same time, breakthroughs were made in laser technology and

$25.00 AH rights of reproduction in any form reserved.

694

5.4.2 Pulsed Laser Deposition System

695

operational costs have declined steadily, especially for

the

UV excimer laser (which

is considered the system of choice for PLD). This technological advancement

played a major role in transforming PLD from a laboratory curiosity to a main-

stream material growth technique.

5.4.2

PULSED LASER DEPOSITION SYSTEM

Among all physical vapor deposition processes, the PLD hardware is the simplest

and most flexible. Figure

shows a schematic representation. The center piece is

a high-energy pulsed laser placed outside the vacuum chamber.

The laser beam is focused onto the target surface. The type of laser used for

PLD has been evolving with time. Ruby [3,5] and Nd:Glass lasers [6] were used

in early studies. They were replaced by the more reliable NdiYAG [7] and TEA-

CO2 lasers [8] in the 1970s. In the 1980s, the trend moved toward using short-

wavelength lasers to produce ablation plumes with higher plasma temperature and

thin films with better crystallinity. Consequently, UV excimer lasers have become

the laser of choice. To date, the majority of PLD systems use either the 248 nm

KrF excimer laser or the 193 nm ArF excimer laser. Over the last 10 years, ex-

Fig.

Reactive gas inlet

Laser beam

Vacuum pump

Schematic representation of

PLD system.

696 Chapter

5.4:

Pulsed Laser Deposition

cimer laser performance has improved drastically. This engineering triumph has

made PLD competitive against other thin-film deposition techniques.

To avoid deep surface pitting caused by repetitive ablation, the focused beam

changes location constantly, by target rotation, by beam rastering, or by a combi-

nation of both. The most popular configuration uses rotating taiget disks mounted

on a carousel multiple holder for multiple-layer growth. Commercial target holder

of this design are available from a number of vendors.

Special substrate holders are used to address the special environmental needs

for growing oxide films. These holders must survive the harsh condition of high

operating temperature (up to 850°C) in an oxidizing environment. The most suc-

cessful designs use high-power halogen quartz lamps or nichrome wire heaters

embedded inside an oxidation-resistant Inconel sheath. The latter has proven to

be particularly compact and durable.

Target/substrate geometry is a key parameter and strongly affects the film qual-

ity. Normally the substrate and target are placed face to face a few centimeters

apart, such that the substrate just touches the tip of the ablation plume, as shown

in Figure

2(a).

Other configurations are also used for specific purposes. Figure 2(b)

shows an "off-center" configuration to improve thickness uniformity. Figure 2(c)

shows an "off-axis" geometry where a rotating substrate is placed along the di-

rection of the ablation plume [9]. Film deposition takes place when radially

dif-

fused ablation species land on the substrate, whereas the heavy particles continue

their trajectories. Figure 2(d) represents an extreme case where the substrate is

facing away from the plume to reduce splashing [10]. Finally, Figure 2(e) shows

the use of a shadow mask placed between the substrate and the target to divert

plume flow and prevent splashed particles from reaching the substrate [11].

An important capability of PLD is the deposition of films in a reactive environ-

ment. Gas inlets are usually located just behind the vacuum window through

which the laser beam transmits. This serves the dual purpose of introducing reac-

tive gas and preventing condensation of evaporants on the window, which can re-

duce its optical transmission. Unlike any other physical vapor deposition tech-

niques, PLD can be carried out in any background condition from high vacuum to

atmospheric pressure [12].

The flexibility of PLD has brought out many system modifications. Most of

them are designed to activate reactive gas in order to achieve better stoichiometry

and lower growth temperature. Examples include the use of a discharge ring in

the ablation plume to lower the growth temperature for YBa2Cu307_x (YBCO)

superconducting films [13], a microwave discharge hydrogen source to grow dia-

mond film from a graphite target [14], and a microwave discharge seeded nitro-

gen nozzle beam to synthesize yS-carbon nitride (CNx) films [15].

Another notable example is the combined use of a Pulsed Laser Evaporation

source with Molecular Beam Epitaxy to conduct "Bandgap Engineering" [16].

5.4.2 Pulsed Laser Deposition System

697

Rg.2.

HH Target

1 I Plume

Substrate

I>^^^

Substrate holder

HHI Shadow mask

(a) Conventional configuration

(b) Off-Center configuration

to improve uniformity

L\\V\\\\M

to improve uniformity and

morphology

(d) Reverse face configuration

to improve morphology

(e) Shadow-masked configuration

to improve morphology

Five difterent PLD target/substrate configurations.

Composition depth resolution of two monolayers can be achieved by modulating

the deposition flux density with laser repetition rate.

Experiments using two synchronized lasers [17] were also reported . In these

experiments, one laser is used for target ablation and the other laser with lower

energy density is used to illuminate the as-deposited film to promote nucleation

and enhance crystallinity.

This list of variations of PLD is by no means complete. PLD's flexibility is still

being exploited to further improve its capability.

698

Chapter 5.4: Pulsed Laser Deposition

5.4.3

THE ABLATION MECHANISM

Material ablation induced by high-power laser radiation is a very complex process

and can be described as a series of heterogeneous events. Sequential steps are

shown in Figure

A full cycle takes approximately one microsecond to complete.

At first, the target material absorbs the leading edge of the laser

pulse.

If the en-

ergy density is above the ablation threshold, sufficient photons will be absorbed

Rg.3.

Laser beam

(a) Photon absorption

(b) Molten surface layer

formation

(d) Plasma heating and recoil

force induced splashing

(e) Plume expansion

Molten layer cooling

Schematic representation of

various

stages of laser-solid interaction leading to ablation plume

formation.

5.4.3 The Ablation Mechanism 699

to melt and vaporize surface material. Under typical PLD experimental condi-

tions,

the energy fluence is a few J/cm^ for

"^-second pulses. It is almost an or-

der of magnitude higher than the ablation thresholds of most materials. The peak

power density under this condition exceeds 10^ W/cm^. The electric field inside

an absorbing medium under this condition can be estimated from the following

relation:

£ = (24)/cEoAi)^^^ (1)

where E electric field

<E>

power density

dielectric constant in vacuum

n refractive index of the medium

c speed of light

For a material with n =

1,5

under a radiation power density of

X 10

W/cm ^,

the corresponding electric field inside the bulk reaches 5 X 10^ V/cm^, high

enough to cause local dielectric breakdown. Therefore, the molten surface layer

vaporizes to form a dense plasma. The early stage of plasma flow is hydrodynamic

and expands isoentropically into the vacuum like a supersonic nozzle beam. It

absorbs the remaining portion of the laser pulse to reach a higher temperature.

Plasma temperature has been determined to be in the range of 10,000K by optical

emission spectroscopy.

During the rapid supersonic expansion, a shock wave is generated. Its recoil

pressure exerts a force strong enough to ripple the molten surface layer. The os-

cillatory motion of the ripples propagates along the surface and is interrupted by

the liquid-solid interface at the edge of the molten pool. The continuous propa-

gation crashes against the solid interface and expels liquid droplets into vacuum.

It is the major cause of "splashing."

As the reactive gas pressure increases,

does the frequency of collisions among

various species. Multiple collisions take place and promote condensation among

molecular and atomic species to form clusters ranging from dimers to submicron

clusters. Their density and distribution are sensitive to background gas pressure

and the distance to target surface. In general, the cluster concentration in the

plume increases with background gas pressure and the distance from the target.

This description is an oversimplified version of a complex process. There have

been many experiments, using an array of in situ diagnostic tools such as optical

emission spectroscopy, laser-induced fluorescence, CCD camera, ion probes, and

time-of-flight mass spectroscopy (TOFMS) to monitor and understand this pro-

cess [18]. However, even with such powerful diagnostic instrumentation, the

physics of laser ablation is still far from being completely understood. Despite the

simple experimental setup of PLD, its ablation mechanism is no doubt the most

sophisticated among all physical vapor deposition techniques.

700 Chapter 5.4: Pulsed Laser Deposition

5.4.4

ADVANTAGES AND LIMITATIONS

The unique approach of PLD has its advantages and limitations. The advantages

have stimulated the rapid growth, and the disadvantages have kept it from becom-

ing a large-scale production tool in the technological mainstream.

5.4.4.1 Advantages

When a multiple-component target is heated on its way to vaporization, individ-

ual constituent elements segregate in the solid (or melt) because of their different

volatility. Elements with the highest vapor pressure will be vaporized, first fol-

lowed by the less volatile ones. In other words, the source material does not melt

and vaporize "congruently," and a thin film deposited under this condition does

not maintain the target stoichiometry. Unfortunately, many materials behave this

way, ranging from the semiconductor GaAs to the superconductor YBa2Cu307_x.

To grow thin films of these materials by simple thermal evaporation, individual

sources must be used to compensate for different individual vapor pressures. In

PLD,

the short thermal cycle induced by laser pulse provides a simple solution to

this problem. The volume heated by each pulse is characterized by the laser spot

size multiplied by the penetration depth, L =2(D • r)

^''^,

where D is the thermal

dififusivity and r is the laser pulse duration. If L is less than the thickness of ma-

terial ablated per pulse, then the stoichiometry of the plume will be the same as

the target, enabling the use of a single source. This criterion is satisfied by many

materials.

CRYSTALLINITY ENHANCEMENT

Ablation species are energetic particles with kinetic energy up to a few keV's and

potential energy in the form of high plasma temperature (i.e., electronic excita-

tion).

In the presence of background gas for reactive deposition, kinetic energy is

retarded by collisions to just a few electron volts, which is ideal for enhancing

film crystallinity. A comparison of the crystallinity of ZnO thin films deposited

under two different target-substrate configurations gives convincing evidence of

this effect. If the substrate faces the target, as in Figure 2(a), where it receives the

full benefit of impingement of the energetic particles in the ablation plume, the

(0002) X-ray diffraction peak of the thin film has a FWHM width of 0.21°. If

the substrate is placed parallel to the plume, as in Figure 2(c), where it completely

avoids direct exposure to the plume, the (0002) diffraction peak FWHM is 0.43°.

On the other hand, the role of potential energy in the form of electronic excita-

5.4.4 Advantages and Limitations 701

tion is not as clear. Its signiiacance lies in activating the reactive background gas

via collision-induced energy transfer.

REACTIVE DEPOSITION

The high plasma temperature of the ablation plume acts as an energy reservoir for

activating the reactive background gas. Therefore, PLD bears some resemblance

to magnetron sputtering. However, in magnetron sputtering deposition of oxides,

film quality is very sensitive to oxygen partial pressure in the Ar/02 sputtering gas

mixture. Excessive oxygen causes resputtering, whereas insufficient oxygen re-

sults in films with high oxygen deficiency. The optimal window of oxygen partial

pressure is narrow. PLD does not have this shortcoming. In addition to oxides,

PLD has also been effective in growing nitride films in a nitrogen ambient [19].

MULTILAYER GROWTH

Because of stoichiometry preservation by PLD, only a single source is needed

to grow films with chemical complexity. Therefore, multiple-layered structures,

such as superconductor [20] and ferroelectric [21] memory devices can be grown

easily by ablating multiple sources.

FAST TURNAROUND TIME

PLD only needs a very small amount of target material as long as the target is

larger than the laser spot. Therefore, it is an ideal way to grow thin films of exotic

materials when only a small quantity is available. Because of the experimental

simplicity, PLD is also known for its fast turnaround time when growing a thin

film of a new material starting from its powder form. This technique gives mate-

rial scientists a quick peek at new material properties with minimal effort and

time.

In fact, PLD is sometimes perceived as the material scientist's equivalent of

what "rapid prototyping" has been to mechanical engineers.

5.4.4.2 Limitations and Solutions

PLD has some limitations. Following is a list of the most common ones and their

solutions:

LARGE-AREA UNIFORMITY

The difficulty of achieving large-area uniformity has always been PLD's main

drawback. There are several reasons for this. First, the laser ablation plume is

702 Chapter 5.4: Pulsed Laser Deposition

non-Lambertian. Its spatial profile follows a (cos^)" distribution where the expo-

nent n varies from n = 3 to /i = 11 and higher. In addition to the narrow angular

profile of the plume, separation between the target and substrate is only a few

centimeters in order to take full advantage of the laser-induced plasma. Both lim-

itations pose severe problems in scaling up to large area.

There have been many attempts to meet this challenge. Solutions include an

off-center substrate mount, an off-axis substrate mount, and a rastering laser spot

over a large target [22]. Verdicts are still out to decide as which approach is the

most practical and appealling.

ROUGH SURFACE MORPHOLOGY DUE TO "SPLASHING"

A major concern of PLD is the presence of micron-size particles on deposited films

due to "splashing," which is the consequence of intense laser-target interaction.

There are three major causes of splashing.

Exfoliation. During laser ablation, materials are removed in a preferential

orientation along the direction of the incident laser beam. The departed material

leaves deep channels and thin columns on the surface. Such microstructures are

mechanically weak and are easily broken by thermal shocks [23]. Debris lands on

the substrate surface, resulting in rough morphology. The signature of this type of

splashing is the appearance of irregular-shaped micron-sized particles.

Subsurface boiling. Subsurface boiling was originally referred to as "true

splashing" by Ready [24]. It occurs if the time required to transfer laser energy

into heat is shorter than the time to evaporate a surface layer with a thickness on

the order of absorption depth. Under this condition, the subsurface layer is super-

heated before the surface material is completely vaporized. Boiling subsurface

material explodes from the surface and ejects micron-size molten globules. Tour-

tellote [6] estimated a threshold power density for subsurface boiling of a metal-

lic target to be 5 X 10^ W/cm^ and higher for dielectric materials. This power

density range is higher than typically found in PLD. Therefore, subsurface boil-

ing has a negligible significance except for a few materials containing very vola-

tile constituents such as Hgi_xCdxTe (high Hg pressure) [25].

Shock-Wave-Induced Droplet Expulsion. Unlike subsurface boiling, in which

the force that expels liquid droplets comes from within the bulk, shock-wave-

induced droplet expulsion is caused by the recoil pressure exerted by the shock

wave of the departing plume. The signature of this type of splashing is also con-

densed molten globules, similar to subsurface boiling, but much smaller, typically

in the submicron range. In contrast to subsurface boiling, the threshold power

density for this mechanism to occur is low, nearly as low as the ablation thresh-

old. Therefore, there is little chance of eliminating it completely. For some com-

pounds, such as CdTe, ZnTe, and ZnSe, PLD-grown films are smooth without any

5.4.4 Advantages and Limitations

703

trace of splashing. This is consistent with the recoil pressure/surface melt inter-

action model, because these materials do not have a melt phase. The vaporization

of these materials is dominated by sublimation.

SOLUTIONS TO AVOID SPLASHING

Since the inception of PLD, there has been a continuous struggle to eliminate

splashing. Many experiments have been attempted. Some of them can be consid-

ered as the brute force approach, while the others rely on more subtle principles.

Here, we only list the most conmion remedies to this problem.

Target Surface Conditioning. The use of dense targets, mechanically pol-

ished surface, or ablation with low-fluence laser pulses prior to film growth can

reduce exfoliation.

Particle Filter. The particle filter approach relies on the fact that particles in

the ablation plume travel much more slowly (about 10'^ cm/sec) than atomic and

molecular species (10^ cm/sec and higher). Particles can therefore be removed by

a synchronous mechanical shutter

[26],

rotary vane filter [27] (shown in Figure 4),

or multiple-disk velocity filter [28]. These methods are effective in removing par-

Fig.

Substrate

(zz:

Laser beam

A rotary vane particle

filter

[27].