Buschow K.H.J. (Ed.) Concise Encyclopedia of Magnetic and Superconducting Materials

Подождите немного. Документ загружается.

energy discrimination. The sampling depth of PEEM

correlates with the mean free path of low energy

electrons in the studied material and lies typically

between 2 nm and 5 nm for metals.

The emitted electrons are accelerated in the strong

electric ﬁeld between the sample and the ﬁrst micro-

scope lens (B10

Vmm

1

), and are then imaged with

magniﬁcation by electron optics, consisting of typi-

cally 2–4 electrostatic or magnetic lenses. A back fo-

cal plane aperture limits the angular spread of the

transmitted electrons and also acts as a simple energy

ﬁlter, reducing the chromatic and spherical aberra-

tions of the microscope. Projector lenses magnify and

focus the image onto the electron detector, typically

an electron sensitive phosphor and/or a channelplate

detector, producing a visible image. Deﬂector and

stigmator elements in the microscope steer the elec-

tron beam and correct for eventual mechanical

misalignments. Using synchrotron radiation micro-

scopes, has achieved a spatial resolution close to 20

nm (Anders et al. 1999).

2. Contrast Mechanisms

Contrast in PEEM results from local variations in the

light absorption, the work function, and the topog-

raphy of the sample. Topographic structures such as

edges, bulges, and holes locally distort the electric

ﬁeld close to the sample and modulate the image in-

tensity. Work function contrast arises from inhomo-

geneities in the surface composition and thereby the

work function. This contrast mechanism is particu-

larly strong using UV radiation close to the work

function cut-off.

Chemical, magnetic, and elemental contrast arises

from Near Edge X-ray Absorption Fine Structure

(NEXAFS), which is discussed in more detail in Sur-

face Chemistry: Electron Yield Spectroscopy (Orme-

rod 2001). As an example, an x-ray absorption spec-

trum of a Co/FeMn/NiO multi-layer is shown in Fig.

2(b). Different elements in a material can be identiﬁed

by their characteristic core hole binding energies and

absorption edges. The ﬁne structure close to the ab-

sorption edge—the inset shows the Ni L

2,3

edge—can

be further analyzed, providing information about the

chemical and magnetic state of the material. For ex-

ample, the multiplet structure at the L edges, marked

by arrows in Fig. 2(b), is characteristic for nickel

oxide (Regan et al. 2001).

PEEM microscopes are typically used in three

modes of operation: (i) full-ﬁeld imaging, (ii) local

spectroscopy, and (iii) spectromicroscopy. In the full-

ﬁeld imaging mode images are acquired at ﬁxed pho-

ton energy. In the local spectroscopy mode local

x-ray absorption spectra are acquired in predeﬁned

regions by scanning the photon energy and using the

spatial resolution provided by the PEEM electron

optics. In the spectromicroscopy mode, stacks of

Figure 1

PEEM-2 Photoemission Electron Microscope and x-ray

beamline at the Advanced Light Source, USA. X rays

are monochromatized by a spherical grating and focused

into a 30 30 mm

spot on the sample. The electron

microscope column produces a magniﬁed image of the

local x-ray absorption on a phosphor, which is imaged

by a slow-scan CCD camera.

Figure 2

(a) Energy spectrum of emitted electrons after the

absorption of x-rays. (b) X-ray absorption spectrum

measured in total electron yield of a Co/FeMn/NiO

sample. The absorption ﬁne structure at the Ni L

peak

(inset) results from the presence of nickel oxide.

1230

Thin Film Magnetism: PEEM Studies

PEEM images are acquired as a function of photon

energy, generating a three-dimensional dataset with

two lateral and one energy dimension.

3. X-ray Magnetic Linear and Circular Dichroism

X-ray Magnetic Circular Dichroism (XMCD) and

X-ray Magnetic Linear Dichroism (XMLD) are es-

tablished spectroscopic x-ray techniques, which are

widely used for the investigation of ferromagnetic

and antiferromagnetic surfaces and thin ﬁlms (Thole

et al. 1985, Schu

tz et al. 1987, Alders et al. 1998, see

also Magnetism: Applications of Synchrotron Radia-

tion). Both methods probe the modiﬁcation of the

electronic structure of a material in the presence of

magnetic order, resulting in an intensity variation of

the near edge x-ray absorption ﬁne structure. X-ray

magnetic dichroism has been extensively used at 2p to

3d transitions (L

2,3

edge) in 3d transition metals, ex-

ploiting the strong spin–orbit interaction of 2p core

levels and the strong exchange splitting of 3d valence

levels (Schu

tz et al. 1987). More recently XMLD has

been applied in studies of collinear antiferromagnets

(e.g., Alders et al. 1998). In order to illustrate the

appearance of both dichroism effects, x-ray absorp-

tion spectra are shown in Fig. 3, measured in single

microscopic magnetic domains on a Co/LaFeO

bilayer.

The sample was grown by molecular beam epitaxy

(MBE) on a SrTiO

(001) substrate (Scholl et al. 2000).

The relative in-plane orientation of the photon polar-

ization and the sample magnetization are shown on

the right. The size of the XMCD effect in ferromag-

netic Co, which appears as a variation of the L

and

resonance intensities, depending on the relative

orientation of the sample magnetization M and the

circular x-ray polarization direction P, is a measure

for the element specific, atomic, spin, and orbital

moment and the angle enclosed by the magnetization

and the polarization vector: DIB M

cos+ðM; PÞ

(Fig. 3(a)). Alternatively, XMLD is particularly use-

ful for the investigation of collinear antiferromagnet-

ism because it probes an anisotropic electronic

structure. Ideal antiferromagnets do not show XMCD

because they are magnetically compensated. The ab-

sorption spectrum of the antiferromagnetically or-

dered Fe atoms in LaFeO

shows a change in the

relative intensity of the L

and L

multiplet peaks due

to XMLD, depending on the relative orientation of

the LaFeO

magnetic axis and the linear x-ray polar-

ization (Fig. 3(b)). The size of the effect is again a

measure for the element specific, atomic, magnetic

moment M, and the angle enclosed by the spin axis A

and the linear polarization vector E: DIB M



½1  2cos

+ ðA; EÞ:

XMCD was ﬁrst applied to ferromagnetic domain

imaging in a pioneering work in 1993 (Sto

hr et al.

1993). More recently, microscopic antiferromagnetic

domains have been imaged for the ﬁrst time using

XMLD (Scholl et al. 2000, Nolting et al. 2000). For

the study of ferromagnetic domains in transition

metal ferromagnets, e.g., Co, images are acquired

with the photon energy tuned to the peak of the L

and L

resonances. The ratio image acquired with

circular polarization is the XMCD image (Fig. 4(a)).

Alternatively, images acquired with opposite polari-

zation can be divided. Dividing PEEM images ac-

quired with linear polarization at the magnetically

sensitive peaks A and B at the L

edge of the anti-

ferromagnet LaFeO

generates the XMLD image,

representing the antiferromagnetic domain structure

of a material (Fig. 4(b)).

Different colors in the XMCD and XMLD image

correspond to different directions of the magnetic

moment. In Co (top) we distinguish three classes of

domains, which have magnetizations pointing up

(shown here as the darkest shade), down (lightest

shade), and left or right (medium shade, not distin-

guished in this measurement geometry). The XMLD

image of LaFeO

shows two classes of domains

with the in-plane projection of the atomic magnetic

Figure 3

(a) X-ray Magnetic Circular Dichroism (XMCD) and

(b) Linear Dichroism (XMLD) spectra of 1.2 nm Co/40

nm LaFeO

/SrTiO

(001). The LaFeO

spectrum was

acquired at the Fe edge. The spectra show opposite

dichroism effects (a) at the L

and L

resonance in the

ferromagnet Co (XMCD) and (b) at the L

multiplet

peaks A and B in the antiferromagnet LaFeO

(XMLD).

1231

Thin Film Magnetism: PEEM Studie s

moments pointing up down (darker shade) and left-

right (lighter shade). Note that the shown images

were acquired at identical sample positions, making

use of the element speciﬁcity of x-ray PEEM. The

correspondence between the two domain structures

indicates a uniaxial interface exchange coupling of

sufﬁcient strength, aligning the magnetization in the

ferromagnet with the in-plane component of the

magnetic axis in the antiferromagnet.

4. Imaging of Local Exchange Bias

One important application of domain imaging using

the PEEM technique is the study of exchange biased

thin ﬁlm systems. Exchange bias at a ferromagnet/

antiferromagnet interface leads to a unidirectional

anisotropy of the ferromagnetic layer, pinning its

magnetization into a preferred direction (Meiklejohn

and Bean 1956); see also Magnetic Films: Anisotropy.

This pinning is employed in magneto-electronic de-

vices for stabilization of the magnetization in a mag-

netic reference layer. The major application of this

effect is in magnetic hard disk heads containing a

read element based on the giant magnetoresistance

phenomenon (see Giant Magnetoresistance and Mag-

netic Recording Technologies: Overview). The unique

combination of elemental speciﬁcity, surface sensitiv-

ity, and ferro- and antiferromagnetic contrast of

PEEM allows studying the magnetic structure direct-

ly at the interface between two coupled layers, pro-

viding crucial insight into the microscopic coupling

mechanism (Ohldag et al. 2001). The unidirectional

pinning of the magnetization in a ferromagnet in

contact with an antiferromagnet can be made visible

by PEEM as shown in Fig. 5.

XMCD images of a ferromagnetic Co layer on La-

FeO

were acquired in remanence as function of the

applied magnetic ﬁeld. Figure 5 shows a subset of a

ﬁeld series, containing about 30 images. The strong

uniaxial interface coupling to the AFM prevents a

permanent rotation of the horizontally oriented mag-

netic domains (gray), which have a magnetization ori-

ented perpendicular to the ﬁeld direction. Vertically

Figure 5

(a) Magnetization reversal of Co layer pinned by

LaFeO

underlayer. The images were acquired in

remanence after applying magnetic ﬁeld pulses of

increasing strength. The correspondence of

magnetization direction to domain brightness is

illustrated on the right. (b) Local hysteresis loops were

calculated in single domains, showing exchange bias of

opposite sign (triangles/squares). The average loop

(crosses) was acquired by averaging the XMCD

intensity over the whole PEEM ﬁeld of view.

Figure 4

X-ray magnetic dichroism images of (a) the

ferromagnetic domain structure in Co and (b) the

antiferromagnetic domain structure in LaFeO

in a

Co/LaFeO

/SrTiO

(001) multilayer sample. Different

shades refer to different domain orientations as

indicated in the boxes. The images demonstrate a

parallel orientation of the in-plane component of the Fe

and Co moment due to interfacial exchange coupling.

1232

Thin Film Magnetism: PEEM Studies

oriented domains, which have an easy axis along the

magnetic ﬁeld, switch from black to white at about

100 Oe, indicating the reversal of the magnetization

parallel to the applied ﬁeld. Remanent hysteresis loops

of two individual domains are shown at the bottom of

Fig. 5. The two representative loops show a relative

shift of the plus/minus and minus/plus switching ﬁelds

of about 50 Oe. This corresponds to a positive (neg-

ative) exchange bias ﬁeld in the triangles (squares)

domain of þ25 (25) Oe. Averaging over a large

number of domains yields the average loop (black).

5. Conclusion

In comparison to other domain imaging techniques

x-ray PEEM stands out by its ability to separately

study the chemical, electronic, and magnetic proper-

ties of different materials in a layered or alloyed ma-

terial. The moderate surface sensitivity of the

technique permits the investigation of ultrathin mul-

tilayers but at the same time PEEM is sufﬁciently

sensitive to detect sub-monolayer amounts of mag-

netic moments. The spatial resolution of PEEM can

be further improved by aberration correction. Imple-

mentation of correction techniques promises an im-

provement of the spatial resolution close to the

physical limit of 1–2 nm. Another future application

of PEEM is the investigation of dynamical magnetic

processes, such as spin precession and magnetization

reversal, utilizing the pulsed structure of synchrotron

sources.

Bibliography

Alders D, Tjeng L H, Voogt F C, Hibma T, Sawatzky G A,

Chen C T, Vogel J, Sacchi M, Iacobucci S 1998 Temperature

and thickness dependence of magnetic moments in NiO epit-

axial ﬁlms. Phys. Rev. B 57, 11623–31

Anders S, Padmore H A, Duarte R M, Renner T, Stammler Th,

Scholl A, Scheinfein M R, Sto

hr J, Se

ve L, Sinkovic B 1999

Photoemission electron microscope for the study of magnetic

materials. Rev. Sci. Instrum. 70, 3973–81

Bauer E 1994 Low-energy electron microscopy. Rep. Prog.

Phys. 57, 895–938

Duden T, Bauer E 1998 Spin-polarized low-energy electron

microscopy. Sur. Rev. Lett. 5, 1213–9

Meiklejohn W H, Bean C P 1956 New magnetic anisotropy.

Phys. Rev. 102, 1413–4

Nolting F, Scholl A, Sto

hr J, Seo J W, Fompeyrine J, Siegwart

H, Locquet J-P, Anders S, Lu

ning J, Fullerton E E, Toney M

F, Scheinfein M R, Padmore H A 2000 Direct observation of

the alignment of ferromagnetic spins by antiferromagnetic

spins. Nature 405, 767–9

Ohldag H, Regan T J, Sto

hr J, Scholl A, Nolting F, Lu

ning J,

Stamm C, Anders S, White R L 2001 Spectroscopic identi-

ﬁcation and direct imaging of interfacial magnetic spins.

Phys. Rev. Lett. 87, 247201/1–4

Ormerod R M 2001 Surface Chemistry: Electron Yield

Spectroscopy. In Buschow K H J, Cahn R W, Flemings

M C, Ilschner B, Kramer E J, Mahajan S (eds.) Encyclopedia

of Materials: Science and Technology, Elsevier, Oxford, Vol.

9, pp. 9006–8

Regan T J, Ohldag H, Stamm C, Nolting F, Lu

ning J, Sto

hr J,

White R L 2001 Chemical effects at metal/oxide interfaces

studied by x-ray-absorption spectroscopy. Phys. Rev. B 64,

214422/1–11

Scholl A, Ohldag H, Nolting F, Sto

hr J, Padmore H A 2002

X-ray photoemission electron microscopy, a tool for the in-

vestigation of complex magnetic structures. Rev. Sci. Instrum.

73, 1362–8

Scholl A, Sto

hr J, Lu

ning J, Seo J W, Fompeyrine J, Siegwart

H, Locquet J-P, Nolting F, Anders S, Fullerton E E,

Scheinfein M R, Padmore H A 2000 Observation of anti-

ferromagnetic domains in epitaxial thin ﬁlms. Science 287,

1014–6

Schu

tz G, Wagner W, Wilhelm W, Kienle P, Zeller R, Frahm

R, Materlik G 1987 Absorption of circularly polarized x rays

in iron. Phys. Rev. Lett. 58, 737–40

Sto

hr J, Padmore H A, Anders S, Stammler T, Scheinfein M R

1998 Principles of x-ray magnetic dichroism spectromicro-

scopy. Surf. Rev. Lett. 5, 1297–308

Sto

hr J, Wu Y, Hermsmeier B D, Samant M G, Harp G R,

Koranda S, Dunham D, Tonner B P 1993 Element specific

magnetic microscopy with circularly polarized x rays. Science

259, 658–61

Thole B T, Laan G, van der Sawatzky G A 1985 Strong mag-

netic dichroism predicted in the M

4,5

x-ray absorption spectra

of magnetic rare-earth materials. Phys. Rev. Lett. 55, 2086–8

Tonner B P, Dunham D, Droubay T, Kikuma J, Denlinger J,

Rotenberg E, Warwick A 1995 The development of electron

spectromicroscopy. J. Electron Spectrosc. Rel. Phenom. 75,

309–32

A. Scholl

Lawrence Berkeley National Laboratory, Berkeley

California, USA

Thin Films, Multilayers and Devices,

Superconducting

Thin ﬁlms are an essential part of all electronic tech-

nologies. This is particularly true for superconducting

instrumentation, sensors and electronics, in which the

active devices themselves are made from deposited

ﬁlms, often a sequence of ﬁlms made as a multilayer

structure. This is in contrast to some semiconductor

technologies, in which devices are often created by

doping of bulk single crystals, say silicon or GaAs,

and are then linked together with interconnects and to

passive components (capacitors, resistors) that are

made from deposited ﬁlms. In the most highly devel-

oped superconducting technology, namely Josephson

junction circuits operating at 4.2 K, the entire circuit

of junctions, resistors, capacitors, insulators and in-

terconnects is made from ﬁlms that are deposited at

room temperature (a significant difference from semi-

conductor processing, which is carried out at high

temperatures). Thus, the control of thin-ﬁlm processes

1233

Thin Films, Multilayers and Devices, Superconducting

is especially critical to the success of superconducting

electronics technologies. The important ﬁlm synthesis

methods, the properties of superconducting ﬁlms and

the devices made from them are the subjects of this

article, which therefore forms a link between articles

that are concerned predominantly with different ma-

terials or physical phenomena and articles that deal

with low-current applications.

The very significant differences between low-tem-

perature superconducting (LTS) materials and high-

temperature superconducting (HTS) materials have

already resulted in some divergence in the approaches

being used to make, understand and use LTS and

HTS ﬁlms. In each section of this article LTS ﬁlms

will be described ﬁrst, then the differences in the HTS

cases will be identiﬁed. It is important to note that

LTS ﬁlm studies and technology date from the late

1950s and are, hence, well established, whereas HTS

ﬁlms are very recent and the ﬁeld is subject to rapid

change.

1. Thin Films

1.1 Thin-ﬁlm Synthesis

The various deposition methods that are available to

make thin ﬁlms of all kinds can be readily used for

superconductors. These methods are thermal and

electron-beam evaporation and sputtering. Many el-

emental superconductors that have been used for

both science and applications studies (e.g., lead, tin,

indium, aluminum, zinc) can be made as ﬁlms by the

simplest methods, which is thermal evaporation in a

moderate vacuum (10

4

Pa or 10

6

torr) onto a subst-

rate at room temperature. The nature of the substrate

is not critical, glass is quite adequate. Alloys of these

elements can be made by evaporation from two

sources or sometimes (e.g., PbBi) from a single

source.

Films of greater technological interest (e.g., nio-

bium, tantalum and their alloys with other transition

metals) are not easy to evaporate thermally, but elec-

tron-beam evaporation and sputtering are straight-

forward for such materials. Magnetron sputtering

sources have become the method of choice for the

niobium ﬁlms used in Josephson circuits. Alloys can

be made from a sputtering target of the appropriate

composition. The superconducting properties of tran-

sition metals are more sensitive to impurities, such as

oxygen, than the s–p metals and a better vacuum

system must be used, say p10

6

Pa (10

8

torr). The

transition metals make excellent getters, so placing a

liquid-nitrogen-cooled surface near the substrate very

effectively lowers the system pressure. The substrate

temperature can be 300 K for niobium ﬁlm deposi-

tion, but the resistance ratio of clean ﬁlms made at

this temperature will be limited by the grain size.

Higher substrate temperatures are needed for some

transition metals (e.g., tantalum) and to increase the

grain size of niobium. In particular, high-quality

epitaxial ﬁlms with long electronic mean free paths,

such as those used in SIS particle detectors, and epi-

taxial superlattices with mean free paths extending

over many layers (see later) are grown at tempera-

tures as high as 800 1C.

To form ﬁlms of compounds based on niobium

which have higher critical temperatures, such as

NbN, NbCN and the family of A15 compounds

Sn, Nb

Ge, Nbj

Al (see Superconducting Materi-

als; Types of ) somewhat more complicated methods

are used. The nitrides and carbonitrides, which are

important materials for Josephson circuits operating

at 8–10 K, are made by sputtering niobium targets in

argon plus nitrogen or nitrogen plus methane onto

substrates at 100–1000 1C (Braginski and Talvacchio

1990). The A15 compounds have generally been

made by coevaporation from electron-beam sources,

say niobium and tin, but cosputtering from multiple

targets or sputtering from a single mixed target can

also be used. The geometry of the evaporation sourc-

es (see Fig. 1) creates a spread of compositions across

an array of substrates (Hammond 1975). This allows

measurements to explore the effects of stoichiometry

on the physical properties of the ﬁlms. To form the

crystalline structure of the compound, a heated

substrate has to be used. Typical optimum temper-

atures are 880 1C for Nb

Sn, 900 1C for Nb

Ge.

These high deposition temperatures have limited the

interest in A15 compounds for electronics applica-

tions, although values over 700 1C are now accepted

as necessary for HTS ﬁlms.

Figure 1

Arrangement of sources and substrates to achieve a

spread of compositions: (a) array of substrates heated to

800 1C; (b) niobium deposition-rate monitor; (c) tin

deposition-rate monitor; (d) niobium electron-beam

source; and (e) tin electron-beam source.

1234

Thin Films, Multilayers and Devices, Supercondu cting

To prepare ﬁlms of the HTS materials, many of the

techniques described in this section are being used

(see High-temperature Superconductors: Thin Films

and Multilayers). The presence of three elements plus

oxygen in the HTS materials leads to increased com-

plexity of the deposition systems, particularly those

for cosputtering, sequential sputtering (a fraction of a

monolayer of each material is deposited sequentially

and repeated many times), coevaporation (MBE) and

sequential evaporation. Two comparatively simple

methods have quickly become popular and repro-

ducibly yield very-high-quality ﬁlms. These are off-

axis sputtering (Eom et al. 1989) and laser ablation

(Inam et al. 1988). As in the case of A15 compounds,

HTS ﬁlms are deposited at high temperatures (600–

750 1C) with the best ﬁlms being made at X700 1C.

Unlike the transition metals, HTS ﬁlms are deposited

in high oxygen pressures (sometimes 250 Pa) and,

hence, the base pressure of the vacuum system is less

relevant. One ﬁlm process that is widespread for

semiconductors, chemical vapor deposition (Li et al.

1991), is being tried for HTSs with some success. It

has the potential to be a commercial process for ﬁlms

of large area. At present typical ﬁlm substrate sizes

are 4 in or 5 in diameter wafers for niobium circuits,

with plans in one laboratory to move to 8 in and 2 in

diameter for HTS devices.

A very significant difference between LTS and

HTS ﬁlms is in the choice of substrates. Epitaxy be-

tween the substrate and ﬁlm is essential for HTS

materials, hence, a very much more limited choice of

substrates is available. Because of the high deposition

temperatures, substrates for both HTS and A15 ﬁlms

must not react chemically with the ﬁlm at tempera-

tures up to around 800 1C. For most LTS ﬁlms this is

not a consideration, but for those applications re-

quiring epitaxial ﬁlms with long electronic mean free

paths, the substrate again must be chosen to have a

good lattice match and to be nonreactive.

1.2 Film characterization

Given the possibility that impurities introduced into

ﬁlms during deposition will change their properties

from those exhibited by bulk samples of the same

material, characterization is an important part of

thin-ﬁlm research. Important measurements include:

the resistivity of the ﬁlm at 300 K, which will be in-

creased by impurities and disorder; the resistivity ra-

tio between 300 K and (say) 10 K, which is strongly

reduced by impurities and electron scattering from

the ﬁlm surfaces; the transition temperature T

, which

is generally decreased by impurities but can some-

times be increased by disorder; the critical current

density J

, which can be most easily measured by

making a constriction in the ﬁlm; the stoichiometry,

which can be determined by Rutherford backscatter-

ing, inductively coupled plasma spectroscopy or en-

ergy/wavelength dispersive x-ray analysis; and the

crystalline perfection, which is studied by x-ray scat-

tering or channelling techniques.

Of these characterization techniques, one of the

simplest and most informative is the measurement of

resistivity and resistivity ratio ðr

300K

10K

Þ. For ex-

ample, r

300K

C15 mO cm for niobium implies an elec-

tronic mean free path of 3 nm. As phonon scattering

is reduced with decreasing temperature, scattering

from grain boundaries and lattice defects (or the ﬁlm

surface) becomes dominant, for ﬁlms deposited in a

poor vacuum system scattering by included impuri-

ties becomes important. Thus, a resistivity ratio of 7

implies a mean free path at 10 K of 21 nm, whereas 50

implies 150 nm. The ratios of 7 and 50 are typical of

niobium ﬁlms made on room-temperature and heated

substrates. If the ﬁlm thicknesses are larger than the

mean free path, the resistivity ratio is an immediate

indicator of ﬁlm quality. In some epitaxial niobium

ﬁlms, resistivity ratios as large as 300–400 in 2 mm

thick ﬁlms have been reported. For many materials

(e.g., silver, niobium, molybdenum, tin, gallium) pre-

cision deposition techniques enable ﬁlms to be de-

posited with mean free paths that are so long at low

temperatures that the resistance ratio is generally

limited by surface scattering, unless the ﬁlms become

very thick.

These measurements are commonly used for both

LTS and HTS superconducting ﬁlms. In the HTS

case, an important additional parameter of interest is

the surface impedance R

at frequencies in the range

from 1 GHz to 100 GHz. The importance of micro-

wave applications of HTS ﬁlms has created the need

to measure R

from 4.2 K to at least 77 K (see

Superconducting Microwave Applications: Filters). As

for good HTS ﬁlms is at least an order of mag-

nitude below that of copper, a copper cavity cannot

be used for the measurements at 77 K. A niobium

cavity can be used at 4.2 K. No standard measure-

ment method exists, but a technique which deter-

mines the Q of a resonator made of two HTS ﬁlms

facing each other is becoming popular (Taber 1990).

1.3 Physical Properties of Films

Many of the properties of superconducting ﬁlms are

discussed in other articles, in that they are identical to

those of the bulk material. However, thin ﬁlms are

often used in preference to single crystals or poly-

crystalline samples, for three reasons. First, they al-

low certain experiments to be carried out more easily.

Examples are high-frequency properties, where a ﬁlm

can be prepared with a smooth clean surface more

easily than the bulk; nonequilibrium properties,

where the sample volume can be made small; Jose-

phson effects in weak links (see Electrodynamics of

Superconductors: Weakly Coupled ), which are most

easily deﬁned in ﬁlms; tunnelling spectroscopy for

1235

Thin Films, Multilayers and Devices, Superconducting

which tunnel junctions are prepared by oxidizing

clean ﬁlm surfaces; and proximity effect studies,

in that ﬁlms can easily be made with thickness

comparable to a coherence length (at least in LTS

materials).

The second reason for using ﬁlms for scientific

studies is that they can be made to exhibit properties

that are different from those of the same bulk ma-

terial. These differences can arise because of dimen-

sional effects, as ﬁlms can be grown very thin in the

dimension normal to the substrate and can be pat-

terned to be small in the plane. When the thickness of

the ﬁlm is less than the coherence length, it becomes

two dimensional, in that its properties cannot vary in

the direction normal to its surfaces. These dimen-

sional (including dimensional crossover effects) ef-

fects will be discussed in Sect. 2.

Another difference between ﬁlm and bulk materials

can be created because the structure and electronic

properties of the ﬁlm are changed by the deposition

process. This is generally a result of crystalline dis-

order (see Superconducting Materials: Irradiation

Effects). If the ﬁlm is deposited on a substrate that

is too cold to allow ordering to occur, a very-small-

grained or even amorphous ﬁlm can result (see Amor-

phous Superconductors). An early example was the

nonsuperconducting semimetal bismuth, which be-

came superconducting up to 6 K when deposited on a

substrate held at 4.2 K. The T

of gallium is similarly

raised from 1.1 K to 8.5 K by such ‘‘quench conden-

sation.’’ Such ﬁlms were the ﬁrst amorphous metals

(Brandt and Ginzburg 1960, Buckel Gey 1963). Al-

uminum ﬁlms can have a T

up to 6 K compared to

the bulk 1.2 K, the T

depending largely on the con-

ditions of deposition and the strain induced by the

substrate. Such changes in T

are due to modiﬁcation

of both the phonon spectrum and the density of elec-

tronic states of the material. In crystalline transition

metals, with their well known dependence of T

electron-to-atom ratio (see Superconducting Materi-

als: Types of ), a remarkably different dependence is

observed when ﬁlms are deposited at low tempera-

tures to make them amorphous. Such quench-con-

densed ﬁlms have high resistivities and small

resistance ratios (close to 1).

A third reason for using ﬁlms is for superconduct-

ing applications in devices and electronics, as de-

scribed later in this article and in other more specific

contributions on particular application areas.

The same reasons for interest in the physical prop-

erties of LTS ﬁlms apply to HTS materials. However,

the radically different properties of the HTS oxides

do create some different results. For example, the

oxides have anisotropic and very short coherence

lengths. Most HTS ﬁlms are grown with the c-axis

normal to the substrate, with the copper oxide sheets

in the plane of the substrate. The c-axis coherence

length is only a few tenths of nanometers. Thus, the

surfaces of HTS ﬁlms, unlike their LTS counterparts,

are most unlikely to have properties representative of

the bulk material (even those properties along the c

axis). This is because any change in stoichiometry, for

example loss of oxygen, or disorder, including that

due to surface reconstruction of the lattice, can affect

a surface layer whose thickness is comparable to, or

greater than, a coherence length. Thus, any experi-

ments that are surface sensitive, such as tunnelling,

proximity effects and photoemission studies, are un-

likely to measure bulk properties of HTS materials in

measurents on c-axis ﬁlms. Recently, ﬁlms with other

orientations have been deposited (see High-tempera-

ture Superconductors: Thin Films and Multilayers);

measurements on these ﬁlms are likely to reveal more

of the bulk properties of HTS materials. In LTS ma-

terials the coherence length is generally much longer

than the thickness of any surface layer, although

niobium and A15 ﬁlms can exhibit problems of sur-

face degradation unless care is taken.

Effects of dimensionality are also different in HTS

materials, in that even in the bulk they are strongly

anisotropic and can be considered pseudo two di-

mensional. It is also not possible to produce ﬁlms

thinner than the coherence length along the c axis.

The coherence length is only 4 nm in the a–b plane of

YBCO, so even a-axis ﬁlms are difﬁcult to make thin

compared to this length. The effect on physical prop-

erties of making the ﬁlms thin is most easily studied

in superlattices (see Sect. 2) but remarkably thin HTS

ﬁlms have been made that remain superconducting to

quite high temperatures.

The properties of both ﬁlms and bulk HTS mate-

rials can differ significantly from optimum values,

particularly due to loss of oxygen, to crystalline dis-

order and to doping effects. The trend of these effects

is to decrease T

and increase resistivity until the

material becomes insulating at low temperatures

(Greene and Bagley 1990).

There is one parameter that is very different in

ﬁlms compared to bulk single crystals, especially in

the case of YBCO. This is the value of critical current

. All good HTS ﬁlms with T

close to 90 K, inde-

pendent of their method of preparation, have J

val-

ues at all temperatures oT

much higher than those

for single crystals. It appears that a pinning center is

rather readily produced in thin ﬁlms that is difﬁcult

to reproduce in crystals and that this defect is com-

mon to all ‘‘high quality’’ YBCO ﬁlms to date.

2. Superlattices

As an object of scientific study, multilayers or super-

lattices are an interesting class of ‘‘engineered’’ ma-

terials in the ﬁelds of semiconductors, magnetism (see

Thin Film Magnetism: PEEM Studies), optics, x-ray

optics and superconductivity (Shinjo and Takeda

1987). They are made by sequentially depositing two

or more materials many times (e.g., ABABABy).

1236

Thin Films, Multilayers and Devices, Supercondu cting

The individual ﬁlm thicknesses of A and B approach

one atomic layer in some cases. This repetition of the

ﬁlm sequence distinguishes them from the multilayer

device structures to be discussed later.

In semiconductors the long electron wavelength

(B10 nm) implies that the bulk electronic band struc-

ture can be radically changed in the direction normal

to the layers of a superlattice, once the individual

layers have thickness comparable to this wavelength.

This regime is relatively straightforward to reach with

current growth capabilities such as MBE. Thus, semi-

conductor superlattices have become a ﬁeld of intense

activity, for both their research interest and device

potential. In metals, the short electron wavelength

(a few tenths of nanometers) implies that such band

structure changes cannot be observed. When metals

become superconducting, however, the electron wave-

length is replaced by the coherence length as the im-

portant length parameter. In LTS materials, it is easy

to make superlattices with layer thickness much less

than coherence lengths (typically tens of nanometers).

Hence, there is scientific interest in superconducting

superlattices, in which the components A and B can

be superconductor (A) with another superconductor,

metal, semiconductor or insulator (B).

It is important to note that in HTS superlattices, at

least those with c axis normal to the substrate, it is

again not possible to make the layer thickness thinner

than the coherence length, which is only a few tenths

of nanometers. If the HTS materials were strictly two

dimensional, layering them into a superlattice should

have no effect on T

unless structural effects occurred

(see later).

Unlike semiconductor superlattices, at present

there are no important devices made from supercon-

ducting superlattices (in contrast to multilayer device

structures), although the Nb/Al–Al oxide–Nb trilayer

process that is now the metallization of choice for

Josephson circuits was a direct result of studies of

tunnelling in niobium–aluminum superlattices (Geerk

et al. 1982).

2.1 Superlattice Synthesis

The vacuum equipment that is commercially availa-

ble makes it relatively easy to prepare superconduct-

ing superlattices. A substrate is exposed to alternating

sources of materials, for example by moving shutters

in front of electron-beam sources, or by rotating or

oscillating the substrate in front of two sputtering

sources. Magnetron sputtering sources are particu-

larly suitable because of the stability of their depo-

sition rate. It is often important to heat the substrate,

as the resulting ﬁlms (e.g., niobium, tantalum) will

then have larger grain size, longer electron mean free

path and longer coherence lengths. Superlattices of

HTS materials have been made by the same methods.

Obviously a heated substrate (B700 1C) is always

necessary. In addition, laser deposition lends itself

ideally to the synthesis of HTS superlattices. Two

targets are moved alternately into the focus of the

laser beam and the layer thickness is controlled by the

number of laser shots allowed to impinge on each

target (Lowndes et al. 1990).

2.2 Superlattice Characterization

As a superlattice has an additional periodicity of the

structure in the growth direction, which is superim-

posed on the crystal lattice spacing, an important

measurement is the standard y–2y x-ray scan. An ex-

ample is shown in Fig. 2, where the subsidiary maxi-

ma are due to x-ray reﬂections from the superlattice

(Somekh et al. 1989). The position of these maxima

can be used to determine the superlattice wavelength

(the sum of the A and B layer thicknesses) and their

intensities are related to the interfacial roughness be-

tween layers.

Another important measurement is simply the net

resistivity of the superlattice (in the case of two met-

als). If the electron mean free path is long, that is, the

electrons move freely through the interfaces between

A and B, then the resistivity will have a value between

the values of A and B in their bulk form. However, if

the electrons are scattered at the interfaces, or if the

ﬁlms themselves become structurally disordered due

to the layering process or to strain, then the resistivity

will be higher than the values of A or B. Even in

superlattices of superconductors with other materials

(insulators, semiconductors), or in HTS superlattices,

the resistivity and resistance ratio are very important

parameters to measure.

Transmission electron microscopy can yield ele-

gant pictures of superlattices and can be used to give

the orientation of the individual layers or grains

Figure 2

CuK alpha x-ray trace of niobium–tantalum epitaxial

multilayer (4.9 nm wavelength) grown on A-plane

sapphire to give (110) niobium–tantalum epitaxy; from

the superlattice intensities the interface roughness can be

estimated at 0.22 nm (after Somekh et al. 1989).

1237

Thin Films, Multilayers and Devices, Superconducting

within the layers. In the HTS materials, the layered

nature of the oxide structure gives even more striking

results. One example shows that within the one layer

there is a very abrupt change from YBCO to PrBCO,

presumably the result of ledge growth of the layers of

unit cell dimensions, that is, the layer was growing as

one unit cell of YBCO say when the source was

changed to PrBCO, which continued the growth to

complete the layer. This ledge growth is also clearly

evident in scanning tunnelling microscope proﬁles of

the surface of YBCO ﬁlms (not superlattices), in

which the layers of one unit cell in height are shown

around screw growth dislocations.

2.3 Physical Properties of Superlattices

The superconducting property of LTS superlattices

that has received by far the greatest attention is the

upper critical ﬁeld H

, measured with the ﬁeld ap-

plied parallel to the layers. This is because the di-

mensionality of the superconducting state is readily

determined from the form of the dependence of H

on temperature T.

As an example, consider the specific case of a su-

perlattice of niobium–germanium (germanium ﬁlms

are insulators at low temperatures) (Ruggiero et al.

1982). If the niobium layers are thick (tbx), then the

superlattice behaves like bulk niobium, that is, it is

three dimensional. If the niobium layers are thin (t5x),

and the germanium is thick, then the niobium layers

are isolated from each other and are two dimensional.

However, if the niobium layers are again, thin, and the

germanium layers are thin enough to allow appreciable

coupling by tunnelling between the niobium layers,

then bulk three-dimensional behavior is again recov-

ered, as the superlattice appears again as uniform ma-

terial over the length scale of x. Interesting crossover

effects have been observed as a function of tempera-

ture, in that the coherence length is much longer near

than it is at lower temperatures. Hence, it is possible

to ﬁnd two regimes, one at higher temperatures where

x4t and three-dimensional behavior is observed and

another at low temperatures with xot that exhibits

two-dimensional behavior. As these two regimes have

different temperature dependences of H

,namely

T for three dimensions and H

C(T

T)

1/2

for two dimensions, the crossover between them is ob-

served in the plot of H

vs T (see Fig. 3). These re-

gimes can be thought of as the situations where the

cores of the ﬂux vortices either extend across multiple

layers as Abrikosov vortices (three dimensional) or sit

between the niobium layers in the germanium layers as

Josephson vortices (two dimensional).

A second type of superlattice utilizes superconduct-

ing and normal metals (e.g., SNSNSNy). The cou-

pling between the S layers is by the proximity effect

induced in the N layer. A widely studied example is

niobium–copper superlattices, in which one unusual

effect occurs for small periods (t2 nm). Both the

niobium and copper layers appear to become disor-

dered, with a strong increase in resistivity, a decrease

in resistance ratio and decrease in the T

of the nio-

bium. Dimensional crossover effects can also be ob-

served in SN superlattices as the N layer becomes

longer than its effective coherence length, which in

copper in niobium–copper superlattices was demon-

strated to be about 20 nm (Banerjee et al. 1983).

More recently, a further novel transition was pre-

dicted (Takahashi and Tachiki 1986) and observed

(Karkut et al. 1988) in superlattices of two super-

conductors with very similar T

but with very differ-

ent mean free paths, for example, niobium and a

niobium–titanium alloy. In this case two crossover

transitions were observed. Near T

the vortices are

located in the niobium–titanium but are large enough

to extend across the niobium (three dimensional,

C(T

T)). At a lower temperature x

becomes

less that t

and the vortices relocate into the nio-

bium layers and are conﬁned within them (two di-

mensional, H

C(T

T

1/2

)) (see Fig. 4). At even

lower temperatures, x

NbTi

becomes ot

NbTi

and the

vortices become two dimensional within the nio-

bium–titanium layers, which have a very high H

Work on HTS superlattices is naturally at an ear-

lier stage, although the layered nature of the oxide

Figure 3

Upper critical ﬁelds near T

for Nb–Ge multilayers.

Systematically varying the germanium thickness

between the two-dimensional niobium layers effects a

progression from anisotropic three-dimensional

behavior ½H

c28

ðTÞBðT

 TÞ to ‘‘crossover’’ behavior

and ﬁnally to decoupled two-dimensional behavior

½H

c28

ðTÞBðT

 TÞ

1=2

 (after Ruggiero et al. 1980).

1238

Thin Films, Multilayers and Devices, Supercondu cting

materials has allowed superlattices of very impressive

perfection to be grown. Perhaps the most interesting

question to be answered to date is the value of T

the individual HTS layers are made thinner and as

the coupling across an intermediate layer is made

weaker. The intermediate layer is, for example,

PrBCO which is a lattice-matched oxide that is a

Mott insulator that exhibits high electrical resistivity

at low temperatures. The results of Fig. 5 show that

even layers of YBCO that are one unit cell thick have

a T

of 20 K when widely separated (Lowndes et al.

1990). The T

increases as the coupling increases,

reaching 56 K when the PrBCO is one unit cell thick.

It is evident that HTS materials with a pronounced

perovskite layer structure behave in many respects

as intrinsic atomic-scale multilayers, the alternating

layers being the copper oxide planes (S) and the redox

spacer layers (N or I) (see Superconducting Materials:

Types of ). In the more complex HTS materials

the doping levels on the two types of layer may be

varied independently, their character changing from

ySNSNSNy to ySISISISy depending on the

preparation conditions. The interpretation and mod-

elling of HTS materials as intrinsic multilayer struc-

tures is currently an active area of considerable

importance both for their basic physical properties

and for their behavior in magnetic ﬁelds.

3. Superconducting Device Structures

Superconducting devices are rather simple structures

compared to those that are commonplace in semi-

conductor technology. This perhaps reﬂects the orders

Figure 4

(a) Theoretical temperature dependence of the parallel

critical ﬁelds for a multilayer SS

y . The dashed

and dash-dotted curves indicate the upper critical ﬁelds

when the superconducting order parameter nucleates in

the S and S

layers respectively. The observable H

c28

the higher one of the two critical ﬁelds at each

temperature. The point (T

, H

) is a multicritical

crossover point, d is the layer thickness and D is the

diffusivity (after Takahashi and Tachiki 1986). (b)

Resistive measurements of H

c28

vs tð¼ T=T

Þ for

Nb/NbTi superlattices with layers of equal thickness

showing behavior at t

for samples with different

wavelength L (after Karkut et al. 1988).

Figure 5

Critical temperature of (YBCO)

/(PrBCO)

superconducting multilayers (m ¼1 to 8 unit cells thick)

vs the separating PrBCO layer thickness: D, eight-cell

YBCO; &, four-cell YBCO; m, three-cell YBCO; ’,

two-cell YBCO; *, single-cell YBCO (after Lowndes

et al. 1990).

1239

Thin Films, Multilayers and Devices, Superconducting