Thin film materials through the interfacial assembly of inorganic networks


Material Information

Thin film materials through the interfacial assembly of inorganic networks
Physical Description:
xi, 167 leaves : ill. ; 29 cm.
Culp, Jeffrey Thomas
Publication Date:


Subjects / Keywords:
Thin films, Multilayered   ( lcsh )
Organometallic chemistry   ( lcsh )
Supramolecular chemistry   ( lcsh )
Chemistry thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 2002.
Includes bibliographical references.
Statement of Responsibility:
by Jeffrey Thomas Culp.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029642932
oclc - 51947401
System ID:

Full Text







For Garrett


It has been said that no man is an island unto himself, and nowhere is this truer

than in the scientific community. Experiments are proposed, results are discussed,

conclusions are debated, and out of skepticism, truths are extracted and new ideas are

born. Indeed, the life's blood of science is collaboration and I have benefited greatly

from the countless discussions with fellow colleagues throughout the course of this work.

First and foremost, I would like to thank Professor Mark Meisel and Ju-Hyun Park in the

Physics Department at the University of Florida for performing all of the magnetics

measurements presented in this dissertation. Equally as beneficial were the many

discussions the data provoked. I thank you both for your hard work, your knowledge that

you shared with me, and above all for providing a working relationship that was just as

rewarding personally as it was professionally.

The data collected at the Advanced Photon Source (APS) were essential to the

work described in this dissertation and I would like to acknowledge all those who assisted

in the experiments. I would like to thank Professor Randy Duran for getting the

University of Florida involved in the Materials Research Collaborative Access Team

(MRCAT) at the APS and for our fruitful discussions. The beamline support staff at the

MRCAT were a valuable asset and I would like to especially thank Nadia Leyorovska,

Holger Tostmann, and William Lavender for their assistance. I also thank Professor

Pulak Dutta and his group at Northwestern University for their aid in getting our

Langmuir diffraction experiments off the drawing board. To my fellow researchers from

the Duran group at UF, I want to say thanks for making the long hours and hard work

bearable with your assistance and your wit. Lastly, I want to thank Guyanga

Weerasekera and Dr. Mark Davidson who were my left and right hands at the beamline.

I do not know how I could have done it without them.

My day-to-day research was made all the more enjoyable by my friends and co-

workers in the Talham group. I would especially like to thank Gail for taking me under

her wing in my first year and assisting me with the metal phosphonate project. I really

enjoyed our golf outings as well and hope we can hit the links again sometime. Thanks

go to Missy for her professional guidance and personal friendship. She was always there

when my reactions failed, my personal life met a crisis, or when I just needed a trip to the

Salty Dog to regroup. May our paths cross again. I thank Isa for all of her help with the

AFM, SEM, and BAM experiments. To all the rest, past and present, I am lucky to have

had the experience of working with such a professional group and value the personal

relationships I made with all of them.

I want to also thank my neighbors in the Boncella group for their advice, their

chemicals, and all the laughs we shared. I graciously acknowledge Professor Katherine

Williams and Mr. Russell Pierce for their assistance with the AA experiments. Thanks

are also due to Professor Eric Lambers at MAIC for his help with the XPS experiments

and Karren Kelley at the Electron Microscopy Core Laboratory for her assistance with

the SEM experiments. The work performed by everyone in the glass, electronic, and

machine shops, the stockroom, and those in the chemistry department staff is also

gratefully acknowledged.

The professional influences of my fellow co-workers are interwoven throughout

the pages of this dissertation, but perhaps less obvious are the influences of those who

have touched me personally, and in doing so, provided the friendship, guidance and

inspiration which lead me to where I am today. A special thank you is given to my dear

friend Tina Rakes for always being there when I needed her. I thank her for all the times

she watched my son when I needed to work, for listening when I needed to talk, for

caring, and most of all for being a friend. Her presence will be missed. Thanks also go

to my friend Marcia Winter for cheering me up when I was down and for giving me a

place to stay while in transition. I may not have made it through this without you. I also

want to thank my old buddies Jason Doyle and Kirk Thrasher who were always just a

phone call away. True friends indeed stand the test of time.

I also want to express my deepest love and sincerest thanks to my parents for

always being there. None of this would have been possible without their selfless

devotion, their sacrifices, and their unending support. They taught me the value of hard

work and wove the moral fabric of my soul. I am truly lucky to have such wonderful

parents. I also say thanks to my brothers Chad and Luke and my sisters Cyndi and Jodi

for always being there for me. They are not only my family; they are my friends.

Though the course of this work, I have been influenced by many people either

personally or professionally. My advisor, Daniel R. Talham, is one of those rare

individuals who has influenced me in both ways. Professionally, Dr. Talham has

provided me with exceptional guidance with my research and provided an excellent

environment in which to learn. He has given me the direction necessary to achieve the

goal at hand, while at the same time the necessary freedom to develop my own ideas,

pursue my own course, and to learn by my own mistakes. I would also like to thank Dan

personally for the tremendous support he gave me through a most difficult time in my

life. If not for his understanding, his patience, and his inspiration my goals would have

fallen out of reach.

The long hours and mental devotion required for a project of this magnitude are

perhaps felt most by those closest to you. For her sacrifices, I give Stacy my deepest

thanks. The pressures on our relationship were more than anyone could be expected to

bear, and I apologize for not telling her often enough how much she was appreciated.

Her sacrifices did not go unnoticed. I also want to thank my son Garrett. He is my life,

my love, and my inspiration. Through it all, he has been patient beyond his years and I

am very proud of him. I can only hope that one day he will look back on this with

understanding. Until then, we have a lot of catching up to do.


ACK N O W LED G M EN TS .............................................................................................. iii

A B S T R A C T .................................................................................................................... x


1 SUPERMOLECULAR CHEMISTRY AT INTERFACES.......................................... 1
Supermolecular Chemistry.................................................................................I....
Assembling Inorganic Networks at Interfaces......................................................... 4
Thin Film Characterization Techniques .................................................................. 7
Conventional M ethods ....................................................................................... 7
Characterization of Thin Films Using Synchrotron X-ray Radiation................. 13
Grazing Incidence X-ray Diffraction (GIXD).... .............................................. 20
X-ray Absorption Fine Structure (XAFS)......................................................... 24

STRUCTURE AND MAGNETIC PROPERTIES................................................... 34
In tro d u ctio n .......................................................................................................... 34
Experim ental Section............................................................................................ 36
R results and D discussion ......................................................................................... 38
Sam ple Preparations......................................................................................... 38
Magnetic Properties of Mn(O3PC6Hs)H20 and Co(O3PC6H)'H20 .................. 42
Search for Spin Glass or Precursor Phases........................................................ 49
Negative Magnetization in the Cobalt-Rich Samples........................................ 52
C onclu sio n............................................................................................................ 53

D IF F R A C T IO N ....................................................................................................... 54
Introduction ....................................................................................................... 54
Experim ental Section............................................................................................ 55
R results and D discussion ......................................................................................... 56
Manganese Octadecylphosphonate Film........................................................... 56
Azobenzene Derivatized Manganese Phosphonate Film................................... 58
Lanthanum Octadecylphosphonate Film........................................................... 60
C onclu sio ns .......................................................................................................... 63

AT THE AIR W ATER IN TERFACE ....................................................................... 64
Introduction .......................................................................................................... 64
Experim ental Section............................................................................................ 68
Results.................................................................................................................. 72
Langmuir M onolayers and LB Film Transfer................................................... 72
Spectroscopic Analyses.................................................................................... 75
XAFS Analysis ................................................................................................ 77
X-ray Diffraction and GIXD ............................................................................ 79
M agnetic Properties ......................................................................................... 80
Discussion ....................................................................................................... ..... 82
Choice of System and M onolayer Behavior..................................................... 82
Structure of the N network ................................................................................. 84
M agnetism ....................................................................................................... 85
M echanism and Structure Directing Elem ents.................................................. 87
Conclusions .......................................................................................................... 88

FE-CO AND FE-MN SQUARE GRID NETWORKS............................................... 89
Introduction .......................................................................................................... 89
Experim ental Section............................................................................................ 91
Results and Discussion ......................................................................................... 93
Langmuir M onolayers...................................................................................... 93
Infrared Spectroscopy ...................................................................................... 94
Grazing Incidence X-ray Diffraction................................................................ 96
M agnetism ....................................................................................................... 97
Conclusions ........................................................................................................ 100

Introduction ........................................................................................................ 101
Experim ental ..................................................................................................... 103
Results ................................................................................................................ 105
Film Structure.................................................. .......................................... 105
DC M agnetometry ......................................................................................... 106
AC M agnetometry ......................................................................................... 111
Discussion ......................................................... ................... ............................. 115
M agnetic Anisotropy .................................................................................... 115
Spin Glass Behavior....................................................................................... 117
Conclusions ........................................................................................................ 119

PRUSSIAN BLUE FILMS ON TEMPLATED SURFACES .................................. 121
Introduction ........................................................................................................ 121


Experim ental ..................................................................................................... 124
Results and Discussion ....................................................................................... 127
Film Deposition .......................................................................................... 127
M agnetism ..................................................................................................... 133
Conclusions ........................................................................................................ 137

OF LINEAR CHAIN AND 2D HEXAGONAL NETWORKS .............................. 138
Introduction ..... .............. ... ............. .... ....... ........ .. ........ ..................... 138
Experim ental ..................................................................................................... 141
Results and Discussion ...................................................................................... 144
Brewster Angle M icroscopy........................................................................ 144
Infrared Spectroscopy ........................................... ...................................... 146
Grazing Incidence X-ray D iffraction.............................................................. 149
M agnetism ......................................... ................................................. ........... 152
Structures of the Networks............................................................................. 153
Conclusions ................................................ ....................................................... 156

LIST OF REFEREN CES............................................................................................ 158

BIOGRAPHICAL SKETCH ....................................................................................... 167

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Jeffrey Thomas Culp

December, 2002

Chair: Daniel R. Talham
Department: Chemistry

Mixed metal phenylphosphonates of composition MnxCoi-x(03PC6Hs)H20 were

prepared with 0 < x < 1. The mixed metal solid solutions are homogeneous and

isostructural with the single metal parent compounds. The magnetic phase diagram,

down to 2 K, was constructed over the entire composition range. No evidence of spin

glass behavior was observed for any concentration at any temperature.

A series of Mn2' and La3+ organophosphonate Langmuir-Blodgett films have

been structurally characterized by grazing incidence X-ray diffraction using synchrotron

radiation. The organophosphonate amphiphiles include both straight alkyl chain and azo-

benzene groups in the organic tails. The metal oxygen networks within the films are

found to be isostructural to related organic/inorganic layered solids.

Reaction of a Langmuir monolayer of an amphiphilic pentacyanoferrate (3+)

complex with Ni2+, Co2+, and Mn2+ ions from the subphase results in the formation of

two-dimensional cyanide-bridged network at the air-water interface. The networks can

be transferred to various supports to form monolayer or multilayer lamellar films by the

Langmuir-Blodgett (LB) technique. The magnetic properties of these films were

investigated as monolayer, bilayer, and multilayer assemblies. The materials possess

interesting physical properties such as magnetic anisotropy, magnetic ordering, and spin-

glass behavior.

Thin homogenous magnetic films comprised of various Prussian blue analogues

have been prepared through the sequential absorption of the appropriate metal ions and

hexacyano complexes onto hydrophobic surfaces that were first templated with a Fe-CN-

Ni two-dimensional grid network deposited as a Langmuir-Blodgett monolayer. The

films show exceptional surface coverage and magnetic behaviors similar to their solid-

state analogues with ordering temperature ranging from 5 K to 210 K.

The preparation of low dimensional inorganic networks through the reactions of

Langmrnuir monolayers containing low symmetry amphiphilic metal complexes with

aqueous metal cyanides was investigated. The reaction of an amphiphilic

iron(III)terpyridine complex with aqueous Ag(CN)2- resulted in the formation of AgCN

crystallites at the air-water interface as shown by grazing incidence X-ray diffraction.

The reaction of an amphiphilic nickel(II)cyclam complex with aqueous Ni(CN)42- or

Cr(CN)63" yielded cyanide-bridged products as evidenced by infrared spectroscopy;

however, the structures of the products remain uncertain due to a lack of X-ray

diffraction from the materials.


Supermolecular Chemistry

The intricate complexity of biological systems can humble even the most able of

synthetic chemists. Their vast array of structural diversity, from the molecular to the

macroscopic level, is both aesthetically pleasing in its symmetry and awe-inspiring in its

functional efficiency. Countless eons of trial and error have perfected synthetic

processes wherein simple chemical building blocks self-assemble with lock and key

precision into complex superstructures such as proteins, enzymes, DNA, and cell

membranes. Each of these subsystems then works in tandem in intricate processes to

create something so incredibly complex as life. Chemists have long looked to these

natural systems for inspiration, hoping to break down and understand the underlying

mechanisms of this process called self-assembly, with the hope of one day mastering this

same level of synthetic control, where by simply providing the appropriate building

blocks, complex structures with specifically tailored physical properties could be

achieved. This synthetic paradigm of creating complex chemical systems from relatively

simple building blocks has been termed supermolecular chemistry.1 As a general term,

the process includes synthetic techniques such as self-assembly, crystal engineering, and

nanoscale chemistry. While having its roots in natural systems, many of the goals in this

area of chemistry are focused on developing novel materials with molecular recognition,

catalytic, magnetic, electrical, and nonlinear optical properties for use in separations and


nanoscale device applications. As such, it could be said that biological systems provide

the inspiration and functional materials provide the motivation. 2'3

Supramolecular chemistry as a synthetic approach takes advantage of weak to

moderate bonding interactions between complementary molecular components to create a

structure that is greater than the sum of its parts. The inter-molecular forces that have

traditionally been employed are Van der Waals in nature and include pi-pi stacking,

hydrogen bonding, and host-guest interactions. 1.2 These weaker forces allow for an

annealing of the final structure to a thermodynamic rather than kinetic product. To

expand the potential applications of supermolecular chemistry, many research groups

have investigated transition metal coordination geometries as architectural driving

forces.4-13 The coordinate covalent bond has several properties that fit well with the

supermolecular synthetic approach. The bond strength is intermediate between the

relatively weak Van der Waals interactions and stronger covalent bonds and thereby

offers a compromise. The bonding can now be more robust, but labile enough to allow

for the self-annealing process that is so advantageous to self-assembled systems. Also,

coordination complexes have well characterized, and often predictable, geometries and

bonding angles that can aid in predicting apriori the structural motif of the final

assembly. Supramolecular chemistry can, therefore, be reduced in complexity to a

system in which individual linear and angular components combine into one, two, or

three-dimensional arrays.14'15 Aside from their unique bonding and geometrical

properties, transition metals also inherently possess useful physical and chemical

properties such as variable oxidations states which can lead to charge transfer

phenomena, colored materials ranging throughout the visible spectrum, cooperative

magnetic behavior, and catalytic abilities.

Due to the many advantages offered by transition metal complexes, numerous

researchers have developed synthetic strategies incorporating metal ions into various

supermolecular arrays. These strategies often involve various combinations of blocking

ligands, bridging ligands, and complex geometries. Some of the results to date include

"zero-dimensional" clusters,16-21 polygons and polyhedra,14"5"26-29 one-dimensional chains

and ladders,30'35 two-dimensional sheets,36'"44 and three-dimensional networks, 13'45-47

many of which possess interesting chemical and physical properties.

One area of potential application for supermolecular assemblies is in the area of

nanoscale devices. The electronics industry, in particular, is on a continuous quest for

smaller, faster components, and less-expensive fabrication methods. The last few

decades have seen a dramatic decrease in component sizes and increases in computer

speeds, but the trend may be approaching a limit. It is generally agreed that optical

diffraction and the opacity of lens materials or photomask supports will likely make

current photolithographic methods ineffective for fabricating features below 100 nm.48

To truly break into the nanoscale regime, a new "bottom up" approach may be beneficial.

One can envision a process where individual molecular building blocks can be assembled

into nanoscale conductors or switches. It is by this method that the synthetic chemist

may find application in supermolecular assembly processes.

In addition to electronic devices, high-density magnetic storage media are also

attractive endeavors for supermolecular chemists. Molecular magnets offer advantages

such as size homogeneity, solubility, transparency, relatively benign synthetic conditions,

and the potential for multifunctional materials.49'5 To date, magnetic clusters with

significant reversal fields have been realized; however, molecular magnets with

practically useful blocking temperatures remain elusive. 16-25

The application of self-assembly methods to the fabrication of nanoscale devices

is still in its infancy, and the level of control and complexity needed for such

sophisticated materials has yet to be realized. However, one can argue based on the

results reported to date that the potential exists. What is needed now is a better

understanding of the self-assembly process and what tools the chemist can use to direct

the architectures of the final materials. In addition, since many of the aforementioned

applications of supermolecular systems will require the positioning, servicing, and

interfacing of these systems at surfaces,57"65 we have undertaken an investigation into the

structure directing ability of an interface in the assembly process and what conditions

would allow for the positioning of coordinate-covalent networks at a surface.

Assembling Inorganic Networks at Interfaces

From the numerous interfaces available for study, we chose the air-water interface

as our medium in which to work. The choice was obvious for several reasons. The air-

water interface is readily available and inexpensive; aqueous monolayers systems

(otherwise known as Langmuir monolayers) have been well studied by our group and

countless others over the past century; many experimental techniques are available and

well understood for characterizing Langmuir monolayers; the final products can often be

transferred to solid supports by the Langmuir-Blodgett technique for further

characterizations of their structure and physical properties.66,67

The general strategy is outlined in Scheme 1-1. To form extended inorganic

networks at the water surface, an amphiphilic ligand system or transition metal

Air-water interface su yeu
> ~~supermolecular y v_

__O D, /C 1 -0-

Scheme 1-1. A strategy for assembling coordinate covalent networks at the air-water
interface. The geometry of the networks can be controlled through appropriate choice of
amphiphile, bridging species, and solute building blocks.

complex is spread in sufficient quantity to form a molecular monolayer on the water

surface. The subphase will be a solution of another metal complex. Either the

amphiphilic complex or the dissolved complex will contain cyanide ligands that can act

to bridge the two complexes together resulting in a coordination polymer at the air-water

interface. To achieve the desired geometrical product, either the complexes themselves

will be designed with appropriate blocking ligands attached, or the air-water interface

will direct the two-dimensional motif by limiting reactivity in the third direction.

The process for characterizing molecular monolayers on water surfaces by surface

pressure vs. area isotherms was first described by Langmuir in 1917,68 and is outlined in

Scheme 1-2. A volatile solution of an amphiphilic molecule is spread on a water

subphase contained in a Teflon trough. By compressing the moveable barriers, the

amphiphiles Wilhelmy balance moveable

Iba rrer
II- 6 -PIP '


im i \\


mean molecular area
Scheme 1-2. A schematic depiction of the control and characterization of molecular
order in aqueous Langmuir monolayers is shown. As the film is compressed,
organization of the amphiphiles occurs resulting in a change is surface tension. The
variation in surface tension is monitored by a Wilhelmy balance and is converted to a
surface pressure vs. mean molecular area isotherm.

effective surface area per molecule decreases and the molecules are forced to close-pack
into a condensed state. During the compression, the surface tension as measured by the

Wilhelmy plate varies with the degree of order in the monolayers. Extrapolation of the
steepest part of the resulting isotherm to the x-axis gives the mean molecular area per
molecule. In addition, a flattening of the isotherm at high pressure and smaller area is
attributed to the collapse of the film, or the limiting mean molecular area. The behavior

Hydro obic


/* P

Tmnrfemned biar

Scheme 1-3. A schematic depiction of the Langmuir-Blodgett technique for preparing
thin films for aqueous Langmuir monolayers is shown. In addition to the method shown,
transfer can also result by starting with a hydrophilic surface submerged in the subphase
followed with a single up-stroke. The method thus provides a unique level of control in
preparing monolayer to multilayer thin films.

of the isotherm can give indications as to whether the amphiphiles have cross-linked into

a polymeric network on the water surface. Changes in the onset area, slope, and limiting

mean molecular area between isotherms on pure water and isotherms run on metal

containing subphases give definitive evidence for an interaction between the amphiphile

and the subphase metal ions.

Thin Film Characterization Techniques

Conventional Methods

Characterization of reaction products formed at an air/water interface presents a

significant challenge since the quantity of material is on the order of micrograms. To

obtain further evidence of reaction it becomes necessary to collect the film onto a solid

support. The Langmuir-Blodgett technique, as shown in Scheme 1-3, allows for the

transfer of a monolayer from the water surface onto a hydrophilic or hydrophobic surface.

The Langmuir-Blodgett technique is a method for preparing multilayer assemblies by a

step-wise deposition procedure.6667 The film is compressed to the desired surface

pressure and the substrate is dipped through the film into the subphase. The alkyl chains

interact with the substrate on the down stroke and transfer to the solid support. The

surface at the end of the dip is now terminated with the hydrophilic end of the

amphiphiles and, upon removing the substrate, another layer of amphiphiles is transferred

on the upstroke. The result is a bilayer of the amphiphiles held together through dipolar

or covalent interactions between the hydrophilic head groups of the amphiphiles. The

method also allows for the transfer of a single monolayer if the solid substrate is

hydrophilic and the deposition procedure starts with the substrate submerged in the


As alluded to earlier, the structural characterization of the sub-microgram

quantities present in thin films presents a significant challenge. The transferred film can

be subjected to analysis by FTIR, UV-Vis, X-ray photoelectron spectroscopy (XPS),

atomic force microscopy (AFM), scanning electron microscopy (SEM), and X-ray

diffraction. To use UV-Vis and FTIR spectroscopies to characterize small quantities of

materials like those obtained in monolayer reactions at a water surface, intense absorption

processes in the product are needed. With transition metal complexes, d-d absorption

bands are typically weak and the presence of other more intense absorption mechanisms,

such as ligand-to-metal charge transfer, or ligand pi-pi* transitions is needed in order to

be a useful characterization method in thin films.

FT-IR spectroscopy is usually quite useful in the study of thin films. The

sensitivity of the method to submonolayer amounts of material is made possible through

attenuated total internal reflectance crystals, hereafter named ATR crystals. These

substrates are single crystal silicon or germanium and can be used clean as hydrophilic

substrates or made hydrophobic by application of a layer of an alkyl silane. The

geometry of the crystal is such that the incident IR radiation is internally reflected

resulting in an evanescent wave that travels parallel to the sample surface. Vibrational

modes with moderate to high absorbtivities, such as C-H, C-N, and C-O are readily

observed in monolayers quantities. This technique is used repeatedly throughout the

experiments described hereafter since the majority of the complexes involved contain the

cyanide ligand, and cyanide-stretching vibrations are quite intense.

Cyanide complexes can be well characterized by FT-IR spectroscopy by

monitoring the C-N stretching frequency. These vibrations vary in frequency with

changes in metal oxidation state and coordination environment.69 This behavior is

rationalized by considering the orbital interactions that are involved in transition metals

completed to cyanide. The cyanide anion lone pair is predominantly located on the

carbon and nearly all of the monomeric cyano-complexes known are coordinated through

carbon. This interaction has a strong sigma bonding interaction and effectively lowers

electron density in the C-N antibonding orbital. As such, coordination of C-N to a metal

center results in an increase in the C-N stretching frequency compared to the free ion.

The C-N antibonding orbitals are low enough in energy to have significant pi-bonding

interactions with the metal center as well. The metal donates electrons to this

antibonding orbital through a process call back-bonding. The more efficient the back-

bonding, the weaker the C-N bond, and the lower the C-N stretching frequency becomes.

As a consequence of these two mechanisms, a general trend is observed as one progresses

across the first row transition metals. For trivalent metal ion complexes, the higher

nuclear charge from chromium(III) to cobalt(III) is expected to increase the sigma

donation and strengthen the C-N bond; however, this is nearly balanced by an increased

ability to back-bond to the ligand, which is working to weaken the C-N bond. The result

is that the C-N stretching frequencies remain nearly constant for the hexacyanides of

chromium (III) through cobalt (III), occurring around 2130 cm'1. Another trend observed

in metal-cyanides is that the C-N stretching frequency is lower for the divalent complex

since the lower cationic charge decreases cyanide to metal sigma bonding and increases

metal to cyanide back-bonding. Thus, K3Fe(CN)6 shows a C-N stretch at 2130 cm"1,

whereas the same vibration occurs at 2060 cmn1 in K4Fe(CN)6. A final consequence of

the bonding interactions, and perhaps the most useful effect for characterizing polymeric

metal cyanides, is the shift of the C-N stretching frequency when both the C and N atoms

are coordinated. Coordination at the nitrogen end is similar in nature to the sigma

donation of the C end, but with minimal back bonding abilities. The result, again, is that

coordination through nitrogen removes electron density from the C-N antibonding orbital

and strengthens the C-N bond.70 This, combined with kinematic effects resulting from the

increased rigidity of the structure, works to shift the C-N stretch to higher frequencies

when cyanide assumes a linear bridging mode.71-73 Therefore, this shift will be a key

signature of the formation of cyanide bridges in the structures prepared at the air-water


Elemental composition of thin films can be accomplished through X-ray

photoelectron spectroscopy (XPS). X-ray photoelectron spectroscopy is a surface

sensitive technique that distinguishes elements by the binding energies of their valence

electrons. The sample is prepared on a conducting substrate such as silicon, and

subjected to an electron beam of varying energy. An energy sensitive detector analyzes

the energy spectrum of the electrons ejected from the top few layers of the sample. Peaks

will occur in the spectrum at energies corresponding to the binding energies of the

electrons. Since these energies are unique to the specific elements of interest, they

provide a method for the identification of the elements present in the sample. Further

information is obtained by integrating the peak area in combination with calculations for

the photo-quantum yield of the transition observed. The resulting ratio of integrated peak

areas is proportional to the ratio of elements in the sample.

Surface morphology in thin films can be investigated by microscopic techniques

such as SEM and AFM. Scanning electron microscopy is useful when lateral resolution

on the sub-micron level is sufficient. For more detailed analysis where higher lateral

resolution or accurate depth measurements are necessary, one must resort to AFM. With

the appropriate choice of tips and scanning heads, AFM can give detailed surface

morphology measurements with lateral resolution on the order of nanometers and depth

resolution on the order of angstroms over sample areas varying from tens of nanometers

to a hundred microns. Indeed, with certain systems where tunneling tips are applicable,

lateral resolution can approach the angstrom level, giving a true atomic picture of the

surface composition. This technique is usually limited in its application for LB films

since the surface layer is typically comprised of methyl terminated alkyl chains, which

effectively mask the inner bilayer structure of interest.

Conventional X-ray sources can be used to characterize the lamellar organization

in LB films. For well-organized LB films, the system mimics a powder diffraction

pattern of a layered system under conditions of extreme preferred orientation. The plate-

like domains are randomly arranged in the xy-plane, but have their z-axes aligned

parallel. As such, the diffraction pattern will consist of a series of (001) peaks which can

be indexed to yield interlayer lattice spacings (d) by the Bragg equation: nX = 2dsin(0),

where 0 is one-half the angle (20) where the diffraction peak occurs. In well-organized

LB films, the (001) peaks can be very intense and observable with a film thickness on the

order of 10 bilayers. Diffraction arising from the in-plane structure is typically several

orders of magnitude less intense since there is no preferred orientation of the in-plane

scattering vectors and acquisition times are usually too long on a conventional

diffractometer to be of use. In addition, the random scattering from the substrate is very

intense relative to the scattering due to the thin film sample, and a special technique

called grazing incidence X-ray diffraction (GIXD) is required to improve the signal to

noise ratio. This technique is best performed using high brilliance X-ray sources such as

synchrotrons. The GIXD experiment and another thin film characterizations using

synchrotron radiation called X-ray absorbance fine structure (XAFS) will be described in

detail in the following section.

Characterization of Thin Films Using Synchrotron X-ray Radiation

Synchrotron radiation74 is produced in large particle accelerators called storage

rings. A storage ring is a large, evacuated circular ring that may be a kilometer or more

in circumference. It is lined with steering magnets (bending magnets) and rf generators,

which work in tandem to keep packets of positrons or electrons circulating at energies of

several gigaelectron volts. At these energies, the particles travel at velocities very near

that of light. A consequence of this circular trajectory is a constant transverse

acceleration of the packets of charged particles and a continuous emission of white

radiation. The energy of this radiation is dependent on the energy of the particle beam

and spans the region from ultraviolet to hard X-rays. The X-ray radiation is many orders

of magnitude more intense than that from a typical diffractometer. Relativistic effects

also work to force a very narrow collimation of the resulting radiation. This collimation

makes a synchrotron source comparable to an optical laser with regard to beam

divergence. This broad band of radiation leaves tangentially from the "bending magnets"

in the storage ring at several different locations called beamlines. These beamlines can

be equipped with optics and/or crystal monochrometers to select single wavelength

radiation as necessary for individual experiments. These types of synchrotrons

employing bending magnets are termed "second generation" synchrotrons.

The research requiring synchrotron radiation reported in this dissertation was

performed at the Advanced Photon Source (APS), Argonne National Laboratory,

Argonne, IL.75 An aerial photograph and schematic of the facility are shown in

Figure 1-1. Production of the beam begins with an electron gun that emits 100 keV

electrons into a linear accelerator linacc). Here the electrons are ramped in energy by a

series of accelerating structures to 200 MeV before striking a water-cooled 7 mm thick


.97 :ci -:X^ *

S' ",MO"L ES

Figure 1-1. An aerial photograph and schematic depicting the synchrotron facility at the
Advanced Photon Source, Argonne, IL (taken from reference 75).

tungsten disk. This interaction produces electron-positron pairs. The APS can operate

using either electrons or positrons, but positrons are normally selected since their positive

charge minimizes interactions with residual gas ions in the ring. The positrons are sent

through a second stage linac that further accelerates them to 450 MeV. From here, the

positrons are sent to a 368 m circumference booster ring where four 5-cell rf cavities

increase the positron energies at a rate of 32 keV per turn. In 0.25 sec, the positrons orbit

the ring 200,000 times as their energy climbs to 7 GeV. At this energy, the positrons

have a velocity approximately equal to the speed of light. They are then injected into the

1104 m circumference main storage ring that is configured as a set of curves connected

by straight sections. As the positrons change trajectory, their velocity changes, and

radiation is emitted tangentially to the arc of the particle beam. Insertion devices, either


Figure 1-2. A schematic of a typical (undulator or wiggler) insertion device showing the
rows of alternating pole magnets. Interaction of the radiation with the magnetic field
results in an enhanced intensity of the X-rays through constructive interference effects.
Synchrotrons such as the Advanced Photon Source, which employ insertion devices, are
known as third-generation facilities (image taken from ref 75).

wigglers or undulators, are incorporated into the straight sections and greatly enhance the

synchrotron radiation.

Facilities, such as the APS, which incorporate insertion devices are known as a

"third generation" synchrotrons. These devices consist of a series of magnets oriented

horizontally wigglerss) or vertically (undulators) to the main beam path (Figure 1-2). The

magnetic poles alternate in polarity and introduce a sinusoidal motion to the main beam.

This acceleration results in the emission of radiation. Undulators use a series of small

deviations to produce an interference effect that results in nearly monochromatic

radiation with vastly enhanced intensity. Wigglers, on the other hand, introduce a

stronger magnetic field that destroys the sinusoidal motion via relativistic effects and

1020 .....- .... ...
%. .... .... ,..... (s rm

,. J .. ., . .'
0 "'A LS I


E Dang W"RU m'bu

t O .. .. J HSL ..>.-. i . ....J . .
10 Cis -4 0i0

0 10 owricl
0 0 -2 19- 1904

to,4- is' ue w
Photon Energy (ceV)

Figure 1-3. The X-ray brilliance of the Advanced Photon Source (APS) compared to
other X-ray sources (taken from reference 75). Facilities such as the APS that
incorporate insertion devices are known as a "third generation" synchrotrons.

produces equal intensity radiation over a broad range of energies. Compared to a

bending magnet source employing a single magnetic pole, the radiation from a wiggler is

enhanced relative to the number of magnetic poles. The insertion devices can produce a

power density higher than that found on the surface of the sun and a combination of flux

and brilliance 104 106 greater than conventional X-ray sources (Figure 1-3). Third-

generation storage rings maximize those x-ray beam qualities, flux and brilliance, that are

needed for frontier experimentation. Flux and brilliance are benchmarks of x-ray beam

quality. Both are based on a measure of the number of photons per second in a narrow

energy bandwidth and in a unit of solid angle in the horizontal and vertical directions.

Flux is the number of photons per second passing through a defined area, and is the

appropriate measure for experiments that use the entire, un-focused x-ray beam.

Brilliance is a measure of the intensity and directionality of an x-ray beam. It determines

the smallest spot onto which an x-ray beam can be focused.

Thirty-five "sectors" are marked on the experiment hall floor. Each of these

sectors comprises two beamlines, one originating at a bending magnet in the storage ring

lattice, the other at an insertion device. These sectors are the domain of the APS users

and are the locations where a wide range of experiments take place. The beamline in

which the University of Florida has an interest is maintained by a multi-institutional

consortium call the Materials Research Collaborative Access Team, or MRCAT, and is

located at sector 10I OD.76

The MR-CAT undulator line is fully operational. As described earlier, an

undulator insertion device provides enhanced X-ray intensity over a narrow energy range.

Experimenters have control over the undulator and can tune the energy range of the

output by adjusting the "gap" between the magnetic poles. The larger the gap, the wider

the energy range and lower the intensity of the resulting radiation. Typically the gap is

set to 100-200 eV to compromise intensity and beam stability. Final selection of

monochromatic radiation over an energy range of 4.8-30 keV is accomplished by a cryo-

cooled Si (111) double crystal monochromator designed by the IIT Center for

Synchrotron Radiation Research. The optimum energy range is between approximately 6

keV and 13 keV, since lower energy radiation is strongly scattered by air and higher

energies require operation with crystal harmonics. As such, operations outside of the

optimum range, while possible, are less trivial and require greater attention to

experimental parameters. The second crystal has a piezoelectric tuning actuator with a.c.

feedback and a Bragg-normal motion that permits some degree of fixed-offset operation.

The monochromator assembly is housed in a separate hutch (called the A-hutch) that is

upstream from the main experimental station.

The term monochromator is somewhat of a misnomer since the radiation coming

off the crystals contains harmonics due to higher order Bragg reflections. This is

compensated for inside the main experimental hutch (called the B-hutch) by reflecting the

beam off a flat, 60 cm long harmonic rejection mirror with Pt and Rh coatings. By

adjusting the tilt of the mirror, the lower energy radiation can be selectively reflected

while the higher energy harmonics are unaffected. This angle can be found by scanning

the mirror tilt and monitoring the intensity of the reflected beam. The intensity will

remain relatively constant until the angle passes a critical angle where the intensity drops

rapidly. By setting the mirror tilt at a value just prior to the cut off angle, higher energy

radiation is effectively discriminated since the angle at this position is above the critical

angle for the higher energy components. The profile of the beam coming off the mirror is

ovular with a width of- 5mm and a height of- 3 mm. The intensity of the beam varies

over this area, with the most intense section in the center of the beam. A homogeneous

section is selected by the use of motorized slits with a motor resolution on the order of


The presence of a vertically deflecting mirror at the front of the experimental

station requires all optical components to be mounted on an incline in order to be inline

with the X-ray beam. This is accomplished by use of an X95 rail system, which also

performs the function of standardizing all component mounts to one of two rail-to-beam

distances. A second float glass mirror is available for use as a steering mirror for liquid

scattering experiments. This allows for a more facile transition between experimental


The X-rail terminates down stream at an 8-circle Huber goniometer that is

mounted on a large positioning table. The table can be moved vertically and horizontally

perpendicular to the beam to synchronize the goniometer center to the beam position.

Two of the 8 circles that control the detector position have encoded motors that permit

continuous scanning and data acquisition using the multichannel scaler described below.

Several detector types are available for use in MR-CAT sector. High flux

measurements, such as those required for incident beam monitoring, are accomplished

through Daresbury design spectroscopy ion chambers for use on the main X95

spectroscopy rail or with smaller Cornell-type ion chambers that may be mounted on the

spectroscopy rail or on the Huber goniometer detector arm. Lytle-type fluorescence

detectors are also available and typically used in XAFS spectroscopy. The goniometer

detector arm can be fitted with sensitive Nal scintillation detectors for high-resolution

diffraction experiments. Data collection is accomplished through an instrument chain of

Keithley electrometers, V-F converters and a 32 channel multi-channel scaler. The multi-

channel scaler permits the simultaneous monitoring of the monochromator energy,

detector outputs, and the goniometer detector motor positions for use in slew scan

operations. Motor commands and data acquisition are handled by the MX system,

written by William Lavender.77

mirror defin flux scintillation
_______ E=_ mirrstor silts monitor counter
X-rays from Im rl 0_.
monochomator 1 U |---

Figure 1-4. Schematic showing the experimental set-up for GIXD experiments. The
sample surface can be either a solid support or a water surface. The steering mirror is
float glass coated with a platinum strip and controls the incident angle to the water
surface. For solid samples, the steering mirror is unnecessary since the incident angle
can be controlled by the sample position.

Grazing Incidence X-ray Diffraction (GIXD)

To obtain X-ray diffraction patterns of monolayers requires a specialized

technique developed over the past two decades (Figure 1-4). The technique is called

grazing incidence X-ray diffraction (GIXD) and requires intense X-rays from synchrotron

sources. The theory behind the technique has been described in detail and only a brief

outline will be presented here.78

The theory behind the technique is based on well-known laws of optics that

govern the interaction of electromagnetic radiation with interfaces consisting of a change

in index of refraction (n). A wave incident on a flat interface at some angle ai is both

refracted into the second medium and reflected off the surface. As Ca approaches some

critical angle, c,, the angle of the refracted wave with respect to the surface approaches

zero and the incident wave is totally reflected. At incident angles just below the critical

angle, the incident wave vector factorizes into two components, one horizontal and one

vertical. This horizontally propagating (evanescent) wave has an exponentially

dampened amplitude along the surface normal and a resulting penetration depth that

varies as a function of the incident angle. At X-ray wavelengths, the penetration depth in

water when xi = cx is approximately 100 nm, but decays rapidly to approximately 5 run

when ai = ('2)aX. The critical angle for water, at X-ray energies on the order of 10 keV,

is 2.4 mRad. At this angle, the incident radiation has become surface sensitive. At

oi = oc, the evanescent and incident waves are in phase and the amplitude of the

evanescent wave is effectively doubled. Since intensity is the square of the amplitude,

the evanescent intensity is quadrupled at the critical angle. As the incident angle

decreases below the critical angle, the incident and evanescent waves become more and

more out of phase and the intensity of the evanescent wave decreases. In a typical GIXD

experiment, the X-rays are incident on the water surface at 0.85%c, which is an effective

compromise between keeping the X-rays surface sensitive and keeping the evanescent

intensity high. This same approach applies to LB films transferred onto solid supports as


This experimental technique has been readily applied to the structural

characterization of aqueous Langmuir monolayers and LB films.78'80 In all cases of

crystalline films, the materials are found to behave as 2D powders. That is, there is no

preferential orientation of Bragg planes and the crystals are randomly dispersed through

rotational disorder about the z-axis perpendicular to the plane of the film. While these

two-dimensional powder patterns do not provide the detail available in traditional single

crystal diffraction experiments, knowledge of the limited structural motifs possible for

the packing of alkyl chains combined with lattice-energy calculations and the information

derived through GIXD experiments can provide a structural picture of the LB film in

surprising detail. 78-80

In the simplest model, a domain of the film is treated as a two-dimensional crystal

consisting of uniformly oriented rigid molecules. The scattering pattern is then governed

by the structure factor reflecting translational order of the molecular centers in the plane

of the monolayer and the form factor of the individual molecules. The translational order

within the xy-plane will give rise to a diffraction peak at 20 when the (h k) lattice planes

make an angle Ohk with the evanescent beam fulfilling the Bragg condition X = 2dsin(9hk).

Unlike a three-dimensional crystal, there is no restriction on the z-component of the

scattered beam. The Bragg scattered beam may go into the water or exit the surface at an

angle Oz. As a result, the 2D lattice confines the scattering vectors to Bragg rods instead

of points. The Bragg rod will have an intensity profile, due to interference effects

between the scattered wave and the wave reflected from the surface. The two waves will

be in-phase when the scattered angle, Oz, matches the critical angle, oc, and a maximum

will occur in the Bragg rod profile. At higher angles, the two waves become out of phase

and the intensity decays. Coupled with this phenomenon, is the interaction of the

molecular form factor with the Bragg rod profile. The structure factor for a rod like

molecule is large only on a plane perpendicular to the rod long axis. The intersection of

this plane with the Bragg rods will give rise to diffraction maxima.

Analysis of the GIXD patterns for Langmuir monolayers has been described in

intricate detail78-80 and the following brief description of the method is heavily borrowed

from these references. Close-packed alkyl chains, assuming rotational symmetry, can

basically adopt three packing motifs: hexagonal, centered-rectangular, and oblique. If the

molecules are all standing perpendicular to the film plane and the packing has hexagonal

symmetry; in rectangular notation, the three lowest order peaks (0 2), (1 1) and (1 -1) are

degenerate (the (1 0) and (0 1) reflections are absent due to symmetry), and the Bragg rod

will have its maximum at 0z = 0. If the molecules are all standing perpendicular to the

film plane, and the unit cell stretches or shrinks towards a nearest neighbor, the cell is

centered-rectangular. The (0 2) degeneracy is removed and two peaks with an

approximately 2:1 intensity ratio will be observed in the xy-plane. If the unit cell

stretches towards a nearest neighbor, the (0 2) peak will be at a larger angle than the (1 1)

(1 -1) degenerate peak. If the cell shrinks, the opposite inequality is observed.

Symmetry can also be removed by tilting of the alkyl chains. If the alkyl chains tilt in a

nearest neighbor direction, the cell is distorted to centered-rectangular and one of the

degeneracies is removed. The result again is that two peaks with an approximately 2:1

intensity ratio will be observed in the xy-plane. The Bragg rods for the two peaks will

have different intensity profiles with the non-degenerate peak having its maximum at 0z =

0, and the degenerate peak will have its maximum at 0z > 0 with 0z dependent on the

magnitude of the molecular tilt. To calculate the tilt angle, it's more convenient to plot

the diffraction data relative to the incident wavevector "K"' where Kxy = (4x/X)sin0xy and

K = (27k/,)sinO,. Now the tilt angle 4 can be calculated via the relationship

tan = K& / [(Kdy)2-[(/2)(Knxy)]2]1' where Kd is the Kz value for the degenerate

reflection, Kdxy is the Kxy value for the degenerate reflection, and Kny is the Kxy value for

the non-degenerate reflection. Similar geometric arguments can be applied to calculate

tilt angles when the alkyl chains tilt towards a next-nearest-neighbor, or in an

intermediate direction.79

For the systems described throughout this dissertation, the analysis of the

diffraction data is concerned primarily with the structure of a two-dimensional inorganic

network contained within the Langmuir monolayer or Langmuir-Blodgett film. The

structure of the organic chains are of secondary importance since the rigid nature of the

inorganic lattice will most often be incommensurate with the alkyl chain packing. The

result of such a mismatch will be either a disordered arrangement of the organic chains or

the formation of small aggregated domains, neither of which are well suited to fulfill the

conditions for high quality diffraction. The interpretation of the diffraction data can also

be more complex for inorganic systems, since the number of packing arrangements and

possible lattices are much larger than in simple rod-like hydrocarbon chains. The Bragg

rod profiles for planar inorganic networks will have their maximum intensity at the

horizon due to the finite thickness of the quasi-two-dimensional system and as such yield

little structural information. Interpretation of the in-plane diffraction pattern is made

somewhat simpler by the reduction in Miller indices to two values for a planar lattice.

Additional information can also be obtained through systematic absences present in the

indexed diffraction peaks that result from crystal symmetries. However, a complete

solution to the structure is still difficult if not impossible due to the infinite possible

arrangements of the constituent atoms. The best approach is to use comparisons to

analogous solid-state structures in combination with a chemical intuition of typical

bonding arrangements and complex geometries to arrive at a structure that agrees well

with the diffraction data.

X-ray Absorption Fine Structure (XAFS)

Complementary to x-ray diffraction is another technique useful in the structural

characterization of materials. The technique is x-ray absorption fine structure

(XAFS).81'82 When a collimated beam of monochromatic radiation travels through matter

of thickness x, its intensity decays according to I / Io = eX, where Io and I are the


, A..................................................

Figure 1-5. The absorbance coefficient for a typical substance at X-ray wavelengths
spanning an adsorption edge energy (Eo). The fine structure is the result of interference
effects between the propagating and backscattered photoelectron waves. The fine
structure in the post edge region is the phenomenon known as XAFS (figure taken from
reference 82).

incident and transmitted intensities and gt is the linear absorption coefficient. As the

energy of the photons is increased, p generally decreases until a critical energy is reached

where the absorption suddenly increases several-fold. This discontinuity is referred to as

an absorption edge and relates to the ejection of a core electron from an atom to a

continuum state. Further increasing the energy will cause a further decrease in A until

another absorption edge is encountered. These absorption processes are designated as K,

L, M, etc. according to the orbital state from which they originate. There is one

absorption edge for the K shell, three for the L shell, five for the M shell, and so on.

Closer inspection of the X-ray absorption spectrum for materials often reveals a

fine structure in the gax vs. E plot in the pre-edge and post-edge regions, Figure 1-5.

Peaks in the pre-edge region arise from absorption processes involving the excitation of

core electrons to bound states and can provide insight into bonding information such as

energetic of virtual orbitals, electron configuration, and site symmetry. Oscillations in

the region 40-1000 eV beyond the absorption edge arise from final state interference

effects involving scattering of the outgoing photoelectron from the neighboring atoms.

These oscillations are the XAFS for X-ray absorption fine structure (also called EXAFS

by some authors). In between the pre-edge and post-edge regions is the X-ray absorption

near edge structure (XANES) that arises from effects such as many-body interactions,

multiple scattering effects, distortion of the excited state wavefunction by the coulomb

field, band structures, etc.

Since XAFS is an effect arising from the interference of the outgoing

photoelectron wave from the absorbing atom with an incoming wave backscattered from

a neighboring atom, then a detailed analysis of this interference effect could provide

details about the local structure surrounding the absorbing atom. Indeed, this is the value

of XAFS spectroscopy. As a local probe, XAFS can complement X-ray diffraction data

since diffraction arises from a periodic arrangement of atoms over a large lattice and

XAFS focuses on the coordination shell within 10 A of the absorbing element. Of

course, XAFS cannot compete with the amount of information available through single

crystal diffraction where complete structure solutions are derived. But with samples

where single crystals are unavailable, XAFS can offer a unique probe into local

environments that may not be obtainable by other methods. For example, XAFS is

applicable to gasses, liquids, solutions, and crystalline or amorphous solids. In addition,

since XAFS results from an absorption process, it is element specific. Although the

XAFS phenomenon and its basic explanation in terms of quantum mechanical

interference effects have been known since the 1930, the phenomenon did not become a

practical experimental tool until Sayers, Stern, and Lytle distilled the essential physics of

the process into the standard XAFS equation and proposed a simple method of data

analysis.83 This achievement, coupled with the availability of tunable, high flux, high

energy-resolution synchrotron facilities, has led to an exponential growth in the number

of XAFS experiments performed since 1970.

The interpretation of an XAFS spectrum, X(E), acquired as a function of

energy, first involves normalization to the background absorption (Po) by

X(E) = [I(E) Eo(E)] / io(E) [1-1]

followed by the conversion from E space to k space via the relationship

k = [(8t2m I h2)(E Eo)] / [1-2]

from which structural parameters can then be determined by application of the standard

XAFS equation

X(k) = ZNjSi(k)Fj(k)e-2 k2 e 2r I.(k) sin(2kri + 4k)
ikI2 k[1-3]

Here Fj(k) is the backscattering amplitude from each of the Nj neighboring atoms of the

jth type with a Debye-Waller factor ofo cj (to account for thermal vibration and static

disorder). The total phase shift of the photoelectron is 3jj(k) and contains contributions

for the absorbing atom i and the backscattering atom]. The term e2r1 ,) takes into

account inelastic losses in the scattering process with X1 being the electron mean free

path, and the amplitude reduction factor, Si(k), takes into account inelastic losses due to

multiple excitation. Thus, each XAFS wave is determined by the backscattering

amplitude (NjFj(k)) modified by the reduction factors Si(k), e'-2^, and e2r' (k), the 1 / kr2

distance dependence, and the sinusoidal oscillation which is a function of interatomic

distances (2krj) and the phase shift (1jj(k)).

A theoretical discussion on the origin of these parameters and their relevant

effects on XAFS spectra is beyond the scope of this introduction;81'82 however, a few

general points can be made. Since the sinusoidal XAFS oscillation results from the

interference sin(2kr) term, with a frequency 2r in k space, the more separated the

absorbing atom and backscattering atom (larger r) the higher the frequency of the

oscillation. The intensity of a spherically propagated wave decreases as 1/r2, so the

XAFS amplitude decays rapidly with distance. While the amplitude function Fj(k)

depends mainly on the type ofbackscatterer, the phase function contain contributions

from both the propagating atom and the backscattering atom. It should also be mentioned

that the standard XAFS equation is an approximation based on the assumption that the

process results from a single-scattering event. This is normally a valid assumption since

multiple-scattering pathways are merely a sum of all the scattering pathways that

originate and terminate at the central (absorbing) atom. As such, the path lengths are

typically quite large resulting in significant attenuation and high frequency oscillations

that tend to destructively interfere. Multiple-scattering can become important when

atoms are aligned in a collinear array. In these types of systems, the propagated wave is

strongly forward-scattered by the intervening atom, resulting in significant amplitude

enhancement. This process is most evident in systems where the bond angles are 180 +

30. In these cases, modifications to the standard XAFS equation are necessary to take

into account the multiple scattering processes.

harmonic fluorescence
synchrotron rejection detectorM
suu1Ce mirror Z I filter/slits

,, m -- ---! I^ r -'-'l i
X-ry s.-..J IH/ monior
XI. sample reference
monochronmtor monitor foil

Figure 1-6. Schematic of an experimental setup for collection XAFS spectra in
fluorescence mode.

The discussion to this point has centered on XAFS taken in absorbance mode, but

the technique is also applicable in fluorescence mode. In fact, for thin film samples, such

as those obtained by LB methods, the absorbance due to the sample is negligible and data

can only be collected as fluorescence. A schematic of a typical experimental

arrangement for collecting fluorescence XAFS spectra is shown in Figure 1-6. The

source radiation is scanned over the required energy range by movement of the

monochromator and reflected off a harmonic rejection mirror located inside the

experimental hutch. The incident flux is monitored by a ion chamber positioned

upstream from the sample-fluorescence detector assembly. The sample is then oriented

45 relative to the incident beam. The fluorescence detector (Lytle detector) is then

placed at 45 relative to the sample with a Z-1 filter placed between the sample and Lytle

detector. There is also a set of collimators oriented between the sample and Lytle

detector to minimize randomly scattered X-rays from reaching the detector. A thin foil of

the same element to be measured in the sample is position behind the sample housing and

in front of a second ion chamber so that a reference absorbance edge can be measured in

situ. This reference foil provides a convenient method of energy calibration since the

absorbance edge is known to within a fraction of an eV for most elements. The measured

fluorescence spectrum can then be converted to an experimental spectrum p.(E) where the

y-axis is the total linear absorption coefficient and the x-axis is energy. For fluorescence

experiments, g(E) is calculated by 4(E) = F / Io where F is the measured fluorescence

intensity and Io is the measured incident X-ray intensity. Prior to the experiment, the

linearity of the detector responses should be verified. This can be accomplished by

monitoring the ratio F / Io with full Io and after attenuation with appropriate filters to

(1/2)Io. If the detectors are linear, the ratio ofF / Io will not vary by more than a few

percent. If the response is not linear, then the sensitivity of the detectors should be

adjusted by varying the detector gasses. To minimize background scattering from the

sample support, the sample should be prepared on an "X-ray transparent" support such as


An XAFS scan normally involves scanning the energy from ~-150 eV before the

edge to -1000 eV past the edge. The scan step size can be varied to give a higher

resolution in the vicinity of the edge (- I eV / step) and a slightly lower resolution after

the edge (- 3-5 eV / step). In addition, multiple scans can be taken and averaged for a

better signal to noise ratio.

Once the data has been collected, the process of extracting the important

structural information from the XAFS spectrum can begin. This process is greatly

simplified by application of an appropriate software package such a WinXAS84 that has

been designed specifically for XAFS data reduction. In addition, the WinXAS program

has been designed to accept inputs from modeling programs such as FEFF785 which can

calculate theoretical XAFS parameters based on atomic coordinates prepared via the

input program ATOMS.86 Using these three programs in tandem, one can efficiently

perform the background subtractions, Fourier transformations, and curve fitting routines

necessary for interpretation of the XAFS data.

The first step in the process of data reduction is the removal of a "raw

background" which is normally present as a smoothly varying, low-order polynomial

evident in the pre-edge and post-edge data curve and the normalization of the data to the

edge step. In the WinXAS program, these steps can be preformed in one step by simply

fitting the pre-edge (typically linear) and post-edge data (typically a second order

polynomial) to two separate functions and subsequently subtracting them off The next

step is to convert the g.(E) from E space to k space. In order to do this, the edge energy

(Eo) must be determined. With WinXAS, this is done by finding the inflection in the

edge step using a second derivate curve. Once (Eo) has been determined, the conversion

to k space can proceed by simply selecting the conversion step (with the proper (Eo)

input) with the WinXAS program. The data at this stage still contains other "background

factors" such as spectrometer baseline, beam harmonics, elastic scatterings, etc., which

will are removed by fitting to a cubic spline function. The power and number of nodes in

the spline function can be varied to get the best fit. The fit window is varied to provide

the best result and is typically in the range 2.5 < k < 11. Quality of fit can most easily be

judged by monitoring the Fourier transformed "radial plot" which varies in real time in

WinXas as the fitting window is varied. The main objective is to minimize peaks below

one angstrom in the radial plot and to ensure that the radial plot remains relatively

constant as the fitting window is changed. It is best to try several different cubic splines

to determine which gives the best fit.

Once the background has been subtracted; the final stage is the Fourier transform.

This is best done in combination with a Bessel window function that minimizes the high

frequency ripples that result from the finite size of the transform data. It is also

advisable to truncate the data window at values that give the smoothest continuation from

one end of the data window to the next. Once the window has been chosen, the Fourier

transform proceeds by opening a separate window where the results are displayed for

four different values ofk weighting. Typically for transition metals, it is best to use a k3

weighting factor which amplifies the lower intensity high k data, giving better results.

The Fourier transform now yields the "radial plot" which consists of a series of

peaks corresponding qualitatively to different coordination shells. The peak positions are

not absolute and are offset by a phase shift. To fit the radial data, one must construct a

model of the coordination environment using the program ATOMS. The important data

here are atom type, coordination number, bond angles, and bond lengths. These values

are read by the input program FEFF and imported into WinXas as the starting points for

the fitting routine. The major problem with XAFS fitting is now apparent, as there are

several fitting parameters in the XAFS equation to vary. The theoretical values

calculated by FEFF are quite accurate, but if at all possible, model compounds should be

run with known structures and similar bonding interactions to extract expected variables

such as the Debye Waller factor and edge energy shifts. For the FeNi grid network,

values were obtained from FeCo Prussian blues reported in the literature. With each

variable having a starting value and expected range, the fitting can begin. In the

beginning, hold different variables constant, and vary each individually to see the effects

of each variable, then steadily progress including more variables until a good fit is


obtained. With a large number of variables, the fit is not absolute, but if the modeled

cluster gives a good quality fit with reasonable values of the non-structure variables then

the model cluster is well supported.



Even before Clearfield's elucidation of the structure of the prototype

a-Zr(HP04)2H20,87 layered metal phosphates were extensively studied primarily because

of their ion exchange capabilities.88 This initial interest has been extended to metal

phosphonates where similar architectures are found89"93 and now includes organic

networks that can be varied to further modify the properties of the layered solids.94"99

Recently, layered metal phosphates and phosphonates have been shown to exhibit

interesting magnetic phenomena, including magnetic ordering, canted antiferro-

magnetism 00107 and antiferromagnetic resonance,108 and they have been studied as

models for two-dimensional (2D) magnetism. Our group has also extended these studies

from the solid-state, 109 to monolayer110-112 and multilayer thin films, 113"119 where similar

properties have been observed.

As part of our interest in 2D magnetism in metal phosphonate solids and thin

films, we have investigated a series of mixed-metal Mn2+/Co2 and Mn2+/Zn2+

phenylphosphonates. Two possibilities exist if mixed metal phases form, each giving rise

to different magnetic behavior. If ions of a different spin state organize in an ordered

fashion, then a new superstructure is formed giving rise to the possibility of

ferrimagnetism if the spin state of the two ions is different. Alternatively, if the ions

distribute randomly, then a solid solution results. Historically, mixed metal solid

solutions have been extensively studied120 because they exhibit altered magnetic behavior

and provide an opportunity to study the details of magnetic ordering mechanisms.

Systems based on Mn2/Co2 have been popular choices, as the materials cover a range of

dimensions, from quasi-lD121-123 to quasi-2D124 to 3D, 125 and in most cases an isotropic

(Heisenberg-type) interaction describes the coupling between S = 5/2 spins of Mn(II)

ions, while an anisotropic (Ising-type) interaction describes the coupling between

"effective" S = 1/2 spins of Co(H). Consequently, upon dilution these materials

experience an interesting blend of competing spin and lattice dimensions. Despite

previous studies on mixed-metal solids, there are still some unanswered questions. For

example, in some cases, but not all, the combination of random mixing and magnetic

frustration leads to spin glass behavior. 126 In addition, some ferrimagnetic systems have

exhibited the interesting effect of negative magnetization.127,128 New examples of mixed-

metal magnetic systems, either structurally ordered or as solid solutions, can provide the

opportunity to further study some of these phenomena.

The divalent metal phenylphosphonates form an isostructural series

(Figure 2-1),91 and we find that the mixed-metal analogs form as solid solutions of

formula MnxCoI-(O3PC6H)H20O or MnxZn.l-x(O3PC6H)'H20O. At low temperature, the

pure Mn and pure Co2' phenylphosphonates experience long-range antiferromagnetic

order at TN ; 12 K and 4 K, respectively. Upon dilution, the ordering temperatures are

reduced compared to the values found for the pure compounds, and the resulting

magnetic phase diagrams are reported here. For diamagnetic Zn2 doping, i.e. MnxZnl-,,

the reduction of TN follows the prediction of mean field theory for x > 0.6 and this

Figure 2-1. In-plane and cross-sectional view of Mn(03PC6H5)H20. Crystallographic
data are taken from reference 5. Key: oxygen, small open circles; manganese,
crosshatched circles; phosphorus, diagonal-hatched circles (phosphorus atoms above and
below the plane are distinguished by hatches with different directions).

magnetic phase diagram was reported previously. 129 However, for the MnxCol-x

compounds, the reduction of TN with doping concentration is weaker than expected on

the basis of mean field theory. For MnxCol-x at low temperatures, the magnetization of

the Mn-rich specimens, i.e. x > 0.25, is characterized by canted antiferromagnetic

behavior. On the other hand, the magnetization of the Co-rich specimens, i.e. x < 0.25,

exhibits a very small negative magnetization behavior when the zero-field cooled and

field cooled data are compared. The magnetic phase diagram for MnxCo.-

x,(O3PC6Hs)'H20 is reported here.

Experimental Section
Materials used. Reagent grade Mn(NO3)2'4H20, CoCl26H20 and

phenylphosphonic acid (C6H5PO3H2, 95%) were purchased from Aldrich (Milwaukee,

WI) and used without further purification. The water used in all reactions was purified

with a Barnstead NANOpure purification system that produced water with an average

resistivity of 18 MNQ cm. Mn(O3PC6Hs)H20 and Co(O3PC6Hs)H20 were synthesized by

mixing equimolar amounts of the appropriate metal ion solution with a solution of

phenylphosphonic acid (pH adjusted to 5-6 with 0.1 M KOH) both heated to 60C prior

to mixing. The solutions were allowed to stir for two hours at this temperature. For each

sample, the precipitate was filtered, washed with water and subsequently with acetone,

and then dried under vacuum.

Preparation of Mn1Col-x(O3PC6Hs)H20O compounds. The mixed-metal

phenylphosphonates MnxCoi-x(O3PC6H5)H20O were prepared in a manner similar to the

pure metal phenylphosphonates but with slight modification. In each case, aqueous

solutions of the metal salts in the desired molar ratios were heated to 60C and added to a

solution containing a slight excess of phenylphosphonic acid at pH 5-6. The resultant

solutions were stirred for only 10 minutes before filtering the precipitate. The products

were washed with water and acetone, and finally dried under vacuum. In all cases, the

final Mn:Co ratios (determined by atomic absorption) of the solid-state materials were

similar, i.e. within 10 %, to those of the starting metal salt solutions.

Instrumentation. Atomic absorption (AA) measurements were performed on a

Perkin-Elmer Model 3100 atomic absorption spectrometer with a photomultiplier tube

detector. For AA analysis, the solid-state samples were dissolved in a 1.0 M HC1

solution. X-ray diffiraction was done with a step scan (0.02 20/step, 2 sec/step) using a

Phillips APD 3720 X-ray powder diffractometer with the Cu Ka line as the source.

Electron paramagnetic resonance (EPR) spectra were recorded on a Bruker (Billerica,

MA) ER 200D spectrometer modified with a digital signal channel and a digital field

controller. Data were collected using a U.S. EPR (Clarksville, MD) SPEC300 data

acquisition program and converted to ASCII format using a U.S. EPR (Clarksville, MD)

EPRDAP data analysis program. Magnetization and AC susceptibility measurements

were performed using a Quantum Design MPMS SQUID magnetometer. The DC

measurements were made with a measuring field of 100 G or 1.0 kG when sweeping the

temperature, or were made at 2 K while sweeping the field up to 50 kG. The AC

susceptibility measurements used frequencies ranging from 17 Hz to 1.5 kHz and an AC

field amplitude of 4.0 G. Additional low frequency (19 Hz) AC susceptibility

measurements were performed with a homemade mutual inductance coil of a standard

design. 130 High frequency (14 MHz) studies were conducted in a homemade tank-circuit

biased with tunnel diode. 130,131 For all of the magnetic studies, powder samples were

contained in gelcaps or plastic vials, with the exception of the work performed at 14 MHz

when the sample was loaded directly into the housing of the coil. The background

signals arising from the gelcaps and vials were independently measured and were either

negligible or subtracted from the data.

Results and Discussion

Sample Preparations

In order to encourage homogeneous solid solutions of composition

MnxCol1,(O3PC6H5)H20O, and to prevent any annealing into a multi-phased system,

samples were quickly precipitated and collected immediately. This procedure resulted in

a decreased crystallinity of the solid solutions, relative to what is possible with the pure

phases, although it is sufficient for powder XRD analyses and does not appear to

influence the magnetic properties. All attempts to prepare mixed-metal samples of high

crystallinity by slow growth techniques resulted in the formation of physical mixtures

and/or multi-phase materials.

Table 2-1. Concentration of manganese in MnxCol-x(03PC6H5).H20 determined from
AA spectroscopy and unit cell parameters determined from the 110, 011 and 030 hkl
reflections in the corresponding powder XRD patterns.

Mol % Mn a+0.01 b0.01 c0.01

100 5.73 14.34 4.94
95 5.73 14.34 4.94
82 5.70 14.34 4.92
68 5.68 14.34 4.90
55 5.66 14.34 4.88
35 5.65 14.34 4.88
30 5.64 14.34 4.87
21 5.62 14.34 4.86
19 5.63 14.34 4.85
11 5.61 14.34 4.85
10 5.61 14.34 4.85
0 5.60 14.34 4.83

Structural Characterizations
The relative percentage of manganese and cobalt in the solid solutions was

determined from AA analyses (Table 2-1). Although AA spectroscopy gives an average

stochiometry, it cannot provide information about the structural homogeneity of the

samples. Therefore, X-ray diffraction was used to determine if the final product consists

of single or multiple phases. The structures of the pure manganese and cobalt

phenylphosphonate compounds consist of layers of quasi-two-dimensional metal-

phosphorus-oxygen sheets that define the ac plane, while the organic moieties project

between the layers thus defining the b-axis (Figure 2-1).90'91 These materials are known

to crystallize in the same space group, Pmn21, with slight modifications of the ac basal

plane spacings.90'9' However, the inter-plane distances are almost identical because both

compounds contain the same organic phenyl group.




.0 b
I lA I. -M^^JL
16 20 24 28 32 36
20 (degrees)


o A

1'6 17 1'8 1 2'0
2 (degrees)

Figure 2-2. A. X-ray powder diffraction patterns of a: Co(03PC6H5-)H20,b:
Mno.35Coo.65(03PC6H5)H20, C: Mn(O3PC6-I)'H20. B. Expansion showing, from left,
110, 030, 011 hkM reflections for a: Co(O3PC65)'H20, b: Mno.35Co0.65(03PC6H5)H20,c:

Due to the symmetry of the Pmn21 space group, the 100 and 001 reflections are

systematically absent, so the highest order reflections containing in-plane structural

information are the 110 and 011. Although both of these reflections contain an inter-

plane contribution, this distance remains essentially constant for all compositions. The

110, 030, and 011 reflections conveniently occur consecutively over a small range of

20 in the X-ray diffractogram, making them a practical series for monitoring variations in

the ac lattice spacings. The position of the 030 reflection in all samples is an internal

reference that confirms that the inter-plane distances do not change as a function of

doping, allowing the 20 values for the 110 and 011 reflections in the doped materials to

be used to determine the in-plane lattice spacings.

Powder XRD patterns for the pure manganese and pure cobalt phenyl-

phosphonates, as well as that of the Mno.35Coo0.65 sample are shown in Figure 2-2. The

similarity of the patterns in Figure 2-2A make it clear that the mixed metal systems are

isostructural with the parent compounds. An expansion of the region between 20 values

of 16-20 in Figure 2-2B shows the 110, 030, and 011 reflections for the same three

compounds. For the mixed-metal example, discrete 110 and 011 reflections are observed

at 20 values between those of the pure Mn21 or Co21 phases, while the 030 reflection

remains the same for all three samples. These observations are consistent with the

formation of a single homogeneous solid solution. Similar results were seen for all

compositions, and Table 2-1 lists the corresponding a, b, c cell parameters for the pure

and doped materials as calculated from the 110, 030, and 011 reflections. The cell edge

lengths systematically shift in value as a function of x. The absence of any reflections

corresponding to the pure single ion phenylphosphonates in the XRD patterns of the

mixed-metal phenylphosphonates, combined with the observation that the detected

reflections have 20 values between those of the two pure compounds, provide convincing

evidence that single phase solid solutions have been formed.

Electron Paramagnetic Resonance

Evidence for microscopic homogeneity of the solid solutions comes from EPR.

The cobalt phosphonate is EPR silent at X-band, while the manganese analog gives a




2000 4000 6000
Magnetic Field (G)

Figure 2-3. Room temperature EPR signals for polycrystalline a: Co(03PC6H5) H20,b:
Mno.g4Co0.A6(03PC6H5)H20, c: Mn(03PC6H5)'H20.

broad line that is structureless as a result of dipolar interactions (Figure 2-3). 132 The

anisotropy of the EPR line width has previously been used to demonstrate the two-

dimensional exchange pathways in the layered manganese phosphonates. 109,132 As the

percentage of Co2+ in the solid solution increases, the Mn2+ signal broadens (Figure 2-3)

reflecting the randomization of the identity of the Mn21 ions nearest neighbors. In the

solid solution, there is no signal due to crystallites of pure Mn(O3PC6HIs)-H20.

Magnetic Properties of Mn(03PC6H5)'H20O and Co(03PC6Hs)H20

The magnetic properties of several manganese and cobalt organophosphonates

have been studied previously. 101,102,104,105,108 The manganese phosphonates undergo a

long-range ordering transition to a canted antiferromagnetic state at temperatures ranging

from 12 K to 18 K, depending on the identity of the organophosphonate. Pure

Mn(03PC6H5)H20 orders at TN ; 12 K. 108 The cobalt phosphonates also order

antiferromagnetically, and for Co(O3PC6Hs)H20, we observe TN 4 K, as we describe

later in this section.

0 50 100 150 200 250 300
T (K)

T (K)

Figure 2-4. A. The temperature dependence of the DC magnetic susceptibility for
Mn(03PC6H5)H20 after zero field cooling the specimen to 2 K and then measuring in a
field of 1 kG. The results of a fit using a S = 5/2 Heisenberg high temperature expansion
for T > 20 K are shown by the solid line with the result J = -2.27 + 0.02 K, as described
in the text. B. The temperature dependence of the DC magnetic susceptibility for
Co(O3PC6Hs)H20 after zero-field cooling the sample to 2 K and then measuring in a
field of 1 kG. The results of a fit using an S = 1/2 Ising high temperature expansion for T
> 9.5 K are shown by the solid line with the result J = -2.43 + 0.05 K, as described in the

The data in Figure 2-4 show the temperature dependence of the static magnetic

susceptibility of the pure Mn and Co materials, acquired by cooling the samples in zero

magnetic field and measuring in a DC field of 1 kG. The broad maximum in the

susceptibility, X-, is characteristic of low dimensional antiferromagnetic interactions

when short-range order correlations become greater than the thermal fluctuations of the

spins. These short-range correlations are established by magnetic exchange interactions,

J, which are typically considered to be limited to nearest neighbor spins. In other words,

the Hamiltonian may be written as

H -= -g s,-Sj [2-1]

where Y runs over all pairs of nearest neighbor spins S, and Sj. The susceptibility data

for the manganese phosphonate may be fit with a 2D high temperature series

expansion133 for a quadratic layer ofHeisenberg S = 5/2 spins based on eq. [2-1], and the

solid line in Figure 2-4A shows the best fit to the data with J= -2.27 0.02 K. The fit

was restricted to T > 20 K since at lower temperatures the fitting procedure is not valid.

In the case of the pure cobalt phenylphosphonate, Eq. [2-1] still describes the simplest

interactions for the case of this 2D, S = 1/2 Ising system when the spin operators are

restricted to their z-components.134 The solid line in Figure 2-4B is a fit, for T > 9.5 K, to

a 2D, S =1/2 Ising high temperature series expansion,135 using an exchange constant of J

= -2.43 0.05 K. It is noteworthy that the magnetic exchange parameters are very

similar in spite of the significantly different spin values and spin dimension.

Previous studies'08 have identified the ordering in Mn(03PC6H5)H20 as a

transition to a canted antiferromagnetic state in analogy to other manganese

organophosphonates.'0' The magnetic moments assume a non-collinear orientation that

produces a weak ferromagnetic moment that lies within the plane of the manganese ions.

This moment, and hence the transition from the paramagnetic to the canted

antiferromagnetic state, can be observed in a difference plot of the magnetization as a

7 -..******..
7 A
6 MO
6" -
3 M -

2 n -

I '^ .-r, , i , I .


0 A&""& AA A BA AA

5 10 15 20

T (K)

Figure 2-5. A. Field cooled (FC) and zero-field cooled (ZFC) magnetization data of
manganese phenylphosphonate are shown as a function of temperature. Both data sets
were acquired with a 100 G measuring field. B. The difference between field cooled and
zero-field cooled magnetization versus temperature for Mn(O3PC6H5)H20 measured in a
field of 100 G.

function of temperature for experiments performed in field-cooled (fc) and zero-field-

cooled (zfc) conditions, AAMfcfc (Figure 2-5). The ordering temperature, TN, may be

identified in the Mc data as the temperature where the magnetization begins to deviate

from its high temperature paramagnetic behavior, and from Figure 2-5, TN = 11.7 K for

Mn(03PC6Hs)H20. Another parameter, TN*, is defined as the temperature at which

AMfc-zfc differs significantly from zero. These two temperatures, TN = 11.7 K and

D Mn o.95Co0.5

10 []
0" 15

'e 10


0- -n an -l-ri i- - i-

4 6 8 10 12 14
T (K)

Figure 2-6. The difference between the field cooled and zero field-cooled magnetization,
AM, is shown as a function of temperature. Typical data from the Mn-rich (i.e. x > 0.25)
samples are shown when the magnetic field, for measuring and field cooling, was 100 G.

TN* = 11.5 K, are identifiable in Figure 2-5. The value of TN* changes as a function of

the magnitude of the applied measuring field and as a result, for small values of TN, it is

best to acquire data with a smaller measuring field, typically 50 100 G. Due to this

dependence upon measuring field, it is important to realize that TN* will always be lower

than TN, but TN* is nevertheless evidence of an ordered state with a weak ferromagnetic

moment. In contrast to the weakly ferromagnetic manganese compound, the pure cobalt






10 B



0 2 4 6 8 10 12 14 16
T (K)

Figure 2-7. A. Field cooled and zero-field cooled DC magnetization for x =0.1. Both
data sets were acquired with a 100 G measuring field, and on this scale, the difference
between the two data sets is not visually detectable. B. The difference between field
cooled and zero-field cooled magnetization from A is shown as a function of temperature.
The onset of a negative magnetization occurs at TN, and this signature is characteristic for
all the Co-rich (i.e. x < 0.25) samples.

compound is antiferromagnetic with TN = 3.9 K, as determined from both DC

magnetization and AC susceptibility measurements.

Magnetic Properties of the Solid Solutions

Typical magnetization plots for the solid solutions MnxCoI-x(O3PC6Hs)H20 with

0.25 < x < 1.00 are shown in Figure 2-6. The ordering temperatures identified in the

magnetization vs. temperature plots are consistent, within experimental resolution, with


10 -


S / Tetracritical Point

0.0 0.2 0.4 0.6 0.8 1.0

Figure 2-8. The magnetic phase diagram of MnxCo-.(O3PC6H5)H2O indicating the
ordering temperature vs. Mn2 concentration. The phase diagram has a tetracritical point
at x = 0.25, as described in the text. The present work was restricted to T > 2 K. The
lines are guides for the eyes and are described in the text.

the temperatures of anomalies in the ac susceptibility studies. Like the pure

Mn(O3PC6Hs)H20, the solid solutions with x > 0.25 form canted antiferromagnets in the

low temperature state. For x < 0.25, the AMfc.zfc vs. temperature plots still reveal the

ordering temperature, but the magnitude of AMfc-zfc is much smaller and negative,

Figure 2-7. This point is discussed further, later in this section. Nonetheless, the

ordering temperatures were confirmed with AC susceptibility measurements, and they are

included on Figure 2-8.

The mixed Mn/Co phenylphosphonates can be thought of as magnetically doped

pure manganese or pure cobalt lattices with the other metal ion as impurity.

Consequently, a reduction of TN from the pure systems is anticipated. Since the magnetic

exchange interactions and lattices are similar, the primary differences are the spin values

and the dimension of the spins (i.e., Ising-like or Heisenberg-like). Therefore, the

reduction of TN is not expected to be as strong as it is for the case of doping with

diamagnetic spins, and these general tendencies are reflected in the phase diagram in

Figure 2-8. In the Mn-rich regime, TN(x) closely follows a linear function with an x = 0

intercept (solid line) at the TN value obtained for the pure Co material. For 0.25 < x <

0.60, the perturbation of the magnetic correlations is stronger as the percolation threshold

is approached and the reduction of TN follows a trend qualitatively represented by the

dotted line. For the Co-rich samples, there is not sufficient resolution in the identification

of TN to allow a specific x dependence to be identified, so the general trend is sketched

by the dashed line. The prediction of a tetracritical point at x = 0.25 agrees well with a

face-centered square planar lattice containing four nearest neighbors where one spin

species dominates the magnetic exchange. In our case, the Mn2' spins dominate the local

magnetic environment. Tetracritical points120'136 have been observed previously in other

doped magnetic systems containing competing magnetic anisotropies. 137

Search for Spin Glass or Precursor Phases

The assignment of the pure Mn2+ material as a canted antiferromagnetic S = 5/2

Heisenberg-like system and the pure Co2, system as a quantum antiferromagnetic S = 1/2

Ising-like system opens the possibility of forming a spin-frustrated state in a randomly

mixed Mn/Co system. In molecular magnetism, similar studies on layered materials have

been reported.138'39 Thus, bimetallic oxalato layered mixed-metal compounds containing

competing ferro- and antiferromagnetic interactions have been magnetically characterized

and in some cases spin glass behavior has been observed. 140 Spin-frustrated systems

displaying magnetic properties characteristic of spin glasses have also been observed in

doped magnetic materials possessing tetracritical points in their magnetic phase

diagrams. 121,123,141

Time dependent thermal remnant magnetization studies were performed with two

samples, x = 0.30 and 0.68. In one set of experiments, the samples were zero-field

cooled from 300 K to 5, 7, and 12 K in three separate runs. The process of cooling from

300 K to the low temperature fixed point required approximately 80 minutes. After

equilibrium was established, a field of 1 kG was applied, and the magnetization was

monitored for nominally 40 minutes. During this time, the magnetization was observed

to relax towards an equilibrium value, and this process was easily fit by a simple

exponential function, yielding time constants ranging from 700 to 1100 seconds. In a

different measurement, the magnetization relaxation rates of the sample holders were

studied and were determined to be negligible. The total change of the signal during the

measurement after achieving the equilibrium state, as defined by the thermometer of the

instrument, was about 1%. Although these results may be suggestive of behavior

associated with a spin glass state, we consider them to be related to the process of cooling

the powder samples. A simple cooling model142 provides a plausible explanation for the

measured relaxation rates. It is noteworthy that the same type of behavior was observed

for both samples and at all three of the temperatures that were studied. In other words,

the experiments covered several of the magnetic phases shown in Figure 2-8, and in

every instance, the behavior was always the same.


S. HDC =1 kG


0 5 10 15

Figure 2-9. The real component of the AC susceptibility for Mno.i8Coo.82(03PC6H5)H20
at 17 Hz and an amplitude of 4 G. The AC susceptibility was studied in an applied field
of zero and 1 kG, corresponding to the open and filled circles, respectively. The
identification of TN is consistent with the values determined by DC magnetization
techniques. The AC response is understood as arising from the dynamics of the magnetic
domains and the powder nature of the specimens, as described in the text.

In a second set of studies, the AC susceptibility of samples was investigated. The

temperature dependence of the real component of the AC susceptibility in applied

magnetic fields of zero and 1 kG are shown in Figure 2-9 for x = 0.82. Our AC studies of

all x reproduced, to within experimental resolution, the ordering temperatures seen in the

DC magnetization data. However, when no external DC magnetic field was present, new

peaks were observed in the AC susceptibility signals, and these features were not present

in the DC magnetization data. Upon application of a 1 kG field, these features were

suppressed and, therefore, can be attributed to the dynamics of the magnetic domains and

the powder nature of the specimens. The temperatures of the transitions as measured by

AC susceptibility did not appear to be frequency dependent from 17 Hz to 1.5 kHz.

Therefore, no spin glass behavior was observed in any of the samples at any temperature


In summary, no evidence of a spin glass state was obtained in our measurements.

It is important to note that a spin-flop transition has been observed in pure

Mn(03PC6H5)H20 in magnetization vs. field studies performed at 2 K,108 and a similar

spin-flop transition is seen for the solid solution with x = 0.84. However, no spin-flop

signatures were observed for samples with x < 0.84, where an increasing intrinsic

background arising from competing magnetic spins may have masked the spin-flop

transitions. Furthermore, we were particularly curious about the possibility of precursor

behavior in the region near the tetracritical point, i.e. a region bounded by the solid,

dotted, and broken lines in Figure 2-8. However, as discussed at the beginning of this

section, no magnetic glassy behavior was observed in this region. Finally, we note that

for x = 0.30 and 0.35, our studies down to 2 K did not reveal any anomalies indicative of

crossing into an "intermediate" phase.120 Naturally, specific heat studies may provide

additional information concerning the existence of and the identification of such a phase.

Negative Magnetization in the Cobalt-Rich Samples

For x < 0.25, ordering is observed, but the value of AMf-zfc is small and negative,

Figure 2-7. Features in the ac susceptibility are observed at the same temperatures, so we

associate these temperatures with the transition to long-range antiferromagnetic order.

The negative magnetization shifts observed for the Co-rich specimens contrasts with the

positive magnetization shifts detected for the Mn-rich materials. The phenomenon of

negative magnetization has been identified previously in a variety of ferrimagnetic

materials. 127,128 Naturally with the doped Co-rich specimens, similar arguments may be

made if small regions of ferrimagnetic ordered phase are present. However, negative

magnetization is also observed in the pure Co material, although it is even weaker than

observed in the data shown in Figure 2-7B. Negative magnetization has previously been

observed in the canted antiferromagnet,143 and the same phenomenon may be responsible

for the behavior observed for the cobalt phase (x < 0.25).


A new series of mixed metal phenylphosphonate solid solutions,

MnxCo-(O3C6H5)H20O, have been prepared and their magnetic properties investigated.

Each composition undergoes long-range magnetic ordering to a canted antiferromagnetic

state at temperatures, TN < 12 K, and a magnetic phase diagram has been constructed

based on individual DC and AC susceptibility measurements. For both the Mn2 and

Co2' high concentration limits, TN decreases relative to the pure single ion phosphonates,

consistent with what is expected for magnetic ion impurity doping. The phase diagram

includes four phases with a tetracritical point at x = 0.25 K, indicating a competition

between the Heisenberg-like Mn2 and the Ising-like Co2+ spins, with the S = 5/2 Mn2

dominating the local environment. While prior studies on mixed-metal systems

possessing competing spin types have shown evidence for spin-glass behavior, no such

state is observed in the MnxCoI-,(O3C6Hs)H20 solid solutions.



We have recently described the preparation of a series of metal phosphonate

containing Langmuir-Blodgett films. 109-1.1,113-119.132,144,145 These films, modeled after

known layered organic/inorganic solids, demonstrate that it is feasible to incorporate an

inorganic extended solid-state network into the hydrophilic region of a Langmuir-

Blodgett bilayer assembly. Examples prepared to date include several

divalent" 1,113,114,119,132,145 and trivalent metal phosphonate networks113'14'117 formed with

straight chain alkylphosphonates as well as with organophosphonate amphiphiles

containing azobenzenel17 and tetrathiafulvalene1"9 functional groups. The inorganic

lattice greatly enhances the stability of the resulting LB films"18 and enables introduction

of traditionally solid-state properties such as magnetism into these thin film

materials.109'132'145 In addition, by employing functional phosphonic acids, LB films that

combine properties of organic and inorganic assemblies have been prepared. 17,119

To date, direct structural characterization of the inorganic component of these

metal phosphonate films is lacking. Analysis has generally relied on comparisons of

spectroscopic and magnetic properties of the films with those observed for similar solid-

state analogs. These comparisons have offered convincing, but nevertheless indirect,

evidence that the inorganic lattices formed in the LB films are isostructural with the

known solids. We report here direct structural characterization of the inorganic lattices in

a series of these metal phosphonate containing LB films using grazing incidence X-ray

diffraction (GIXD). The results confirm the earlier conclusions that the inorganic lattice

that forms in these LB films is isostructural with the solid-state analogues.

Experimental Section

The 16-bilayer manganese octadecylphosphonate145 (MnOPA), 16-bilayer

lanthanum octadecylphosphonate114 (LaOPA), and 15-bilayer manganese (4-(4'-

tetradecyloxyphenyldiazenyl)phenyl)butylphosphonate117 (MnA4) LB films on glass

slides were prepared as previously described. A "bilayer" is comprised of head-to-head

layers of the phosphonate amphiphile sandwiching one metal ion layer.

The grazing incident X-ray diffraction experiments were conducted at the

Materials Research Collaborative Access Team (MRCAT) beamline at sector 10 of the

Advanced Photon Source, Argonne National Laboratory, Argonne, Illinois.76 The

beamline is equipped with an undulater insertion device, a double silicon crystal

monochromator, harmonic rejection mirror, and a Nal scintillation counter mounted on

an eight-circle Huber goniometer. 146 The sample was mounted in the center of the

goniometer and aligned to make the X = 1.254 A X-rays incident on the sample at an

angle of 1.8 mrad. The evanescent wave produced at this low incidence angle allows for

enhanced surface sensitivity. The beam was 2 mm wide by 0.2 mm high and irradiated

an approximately 50 mm strip along the sample surface. The scattered X-rays were

detected by scanning the scintillation counter through a plane parallel to the sample

surface. Bragg rod profiles were obtained by scanning over the peaks of interest in the xy

plane with successive steps in the z direction (perpendicular to the sample surface). The

Figure 3-1 In-plane and cross-sectional view of Mn(03PC6H5)H20. Crystallographic
data are taken from reference 16. Key: oxygen, small open circles; manganese,
crosshatched circles; phosphorus, diagonal-hatched circles (phosphorus atoms above and
below the plane are distinguished by hatches with different directions).

scattered X-rays were collimated through a set of Soller slits prior to detection giving an

instrumental resolution in Qy of 0.0015 A'1, and Qz of 0.01 A"1' where

Qxy= (47r/X)sin,0xy and Qz = (21r/X)sinOz.

Results and Discussion

Manganese Octadecylphosphonate Film

The structural prototype of the Mn(O3PR)H20 manganese phosphonates is the

phenyl analog, Mn(O3PC6Hs)LH20, the structure of which was determined by Cao et al

and is shown in Figure 3-1.91 The solids crystallize in the Pmn21 space group, and the

phenylphosphonate has unit cell parameters a = 5.734, b = 14.33, and c = 4.945. The

lamellar material consists of two-dimensional manganese-oxygen networks in the ac

plane separated in the b direction by the phenyl groups of the phenylphosphonate ligands.

If the in-plane manganese-oxygen network is considered alone, it can be described with a



5.5 5.0 4.5 4.0 3.5 3.0 2.5

Figure 2. The grazing incidence X-ray diffraction pattern obtained on a 16-bilayer
sample of MnOPA.

face-centered rectangle unit cell with edges a = 5.734 A and b = 4.945 A (in the two-

dimensional cell, the b axis corresponds to the c axis in the Pmn21 space group).

The background corrected GIXD pattern obtained from the MnOPA LB film is

shown in Figure 3-2. The pattern consists of four distinct diffraction peaks corresponding

to lattice spacings of 4.27, 3.71, 2.88, and 2.44 A. The latter three peaks are due to

scattering from the manganese-oxygen network of the inorganic lattice and can be

indexed to a face-centered rectangular cell with Miller indices of (1 1), (2 0), and (0 2),

respectively, corresponding to unit cell parameters of a = 5.76 A and b = 4.88 A. The

(1 0) and (0 1) reflections are absent due to the centered cell. The indexed diffraction

pattern of the LB film agrees very well with the unit cell for the solid-state manganese

phenylphosphonate verifying that the inorganic network formed in the LB film is

isostructural with the analogous solids. Analysis of the peak width for the isolated

(0 2) peak by application of the Scherrer equation147 yields an average structural

coherence length of-180 A.

The intense broad peak in the diffraction pattern at a spacing of 4.27 A is typical

of inter-alkyl chain distances.78'79 The presence of a single diffraction peak is usually

interpreted as hexagonal close packing of freestanding chains. This packing motif is

unlikely in MnOPA due to the structural constraints of the underlying inorganic network

that yields an average cross-sectional area per alkyl chain of 28 A2. Interlayer spacings

determined from previous X-ray diffraction data indicate a tilt angle of approximately

30, and Bragg rod scans of the GIXD peak confirm this assessment. The absence of the

additional diffraction peaks that are expected from a lower symmetry cell is likely due to

a small coherence length of the alkyl packing, resulting from significant disorder in the

organic network induced by the lattice mismatch between the rigid inorganic layer and

the alkyl chain packing. A similar situation has been observed for alkyl chains tethered

to a rigid polymer backbone.148

Azobenzene Derivatized Manganese Phosphonate Film

The GIXD pattern for a 15-bilayer film of MnA4 is shown in Figure 3-2. As in

MnOPA, the pattern shows the expected peaks corresponding to the (2 0) and (0 2)

Bragg planes for the manganese phosphonate lattice at spacings of 2.88 A and 2.44 A.

The (11) peak expected at 3.71 A is obscured in MnA4 by the strong scattering from the

organic groups. The spacings for the inorganic lattice yield the same centered

rectangular cell, a = 5.76 A and b = 4.88 A, observed for the MnOPA film, verifying that

the same inorganic network forms in each case.

0 6-


5.5 5.0 4.5 4.0 3.5 3.0 2.5

Figure 3-3. The grazing incidence X-ray diffraction pattern obtained on a 15-bilayer
sample of MnA4.

As expected, the packing of the organic groups is different in the two manganese

phosphonate films. In contrast to the intense reflection at 4.27 A in MnOPA, the MnA4

film has a strong reflection at 3.79 A and a weaker reflection at 4.59 A. Previous

investigations into the structure of azobenzene-containing alkane thiol monolayers on

gold by Caldwell et al. suggested that the azobenzene groups, when tethered to the

surface with flexible alkyl segments, can aggregate into tightly packed islands forming a

herringbone motif.94 Thus, the azobenzene unit, although not completely independent of

the Au( 11) surface, strongly influences the overall packing arrangement.94 In the case of

the SAMs on gold, the azobenzene groups were proposed to pack with hexagonal

symmetry with an intermolecular separation of 4.5 A, which gave rise to a single

diffraction peak at 3.9 A. The presence of two diffiraction peaks at 4.59 A and 3.79 A in

MnA4 suggests a deviation from hexagonal symmetry and therefore a different packing

arrangement of the azobenzene groups than was observed for the SAMs on gold. This

variation likely results from a combination of the rectangular symmetry of the underlying

manganese phosphonate lattice and the shorter alkyl tether in MnA4 preventing a high

degree of aggregation. A unit cell for the organic network can be assigned by assuming

the azobenzenes tilt by 22 along the direction parallel to the manganese face diagonal.

This arrangement would result in an oblique unit cell of dimensions a = 7.0 A and b = 5.8

A and y = 140 for which the reflections at 4.59 A, and 3.79 A can then be assigned to the

(1 0)org and (0 l)org Bragg planes, respectively (the subscript org is used to differentiate

the organic network when it is considered independently from the inorganic network).

The (2 -2)org reflection for the organic lattice is predicted at 2.88 A, commensurate with

the (2 0) reflection of the parent inorganic network. This peak may contain intensity

from both networks. Support for this assignment comes from the peak width, which is

larger than either the related 2.44 A peak that arises from the inorganic network or the

corresponding peak from Figure 3-1 for MnOPA. This arrangement of the azobenzene

groups would also lead to significant K-interactions perpendicular to the (0 l)og Bragg

plane and 7t-stacking is supported by UV-Vis spectroscopy, reported previously. 17

Lanthanum Octadecyiphosphonate Film

Trivalent lanthanum phosphonates are also known to form layered structures in

the solid state, now consisting of two-dimensional oxygen-bridged La3' networks

separated by the organic substituents of the phosphonate group. The GXDCD pattern for

the LaOPA LB film is compared to the diffraction pattern obtained for a powdered

sample of lanthanum butylphosphonate in Figure 3-4. The similarities in the diffraction

patterns of the two materials indicate that the inorganic network formed in the LB film


5.5 5.0 4.5 4.0 3.5 3.0

Figure 3-4. The grazing incidence X-ray diffraction pattern obtained on (top) a 16-
bilayer sample of LaOPA compared to a X-ray powder diffraction pattern of (bottom)
lanthanum butylphosphonate.

is isostructural with the shorter chain solid-state analog. The trivalent lanthanum ion

requires two phosphonates (one monobasic and one dibasic) to preserve electric

neutrality, resulting in a higher density of alkyl chain packing than is observed for the

manganese phosphonates. Bragg rod scans of the GIXD peaks and interlayer spacings

obtained from conventional X-ray diffraction indicate an absence of any significant tilt

angle in the alkyl chains. As a result, the organic and inorganic networks lie on the same

two-dimensional lattice, and the observed diffraction peaks are a composite of scattering

from the inorganic extended network and the alkyl chain molecular network. The entire

pattern can be indexed to an oblique cell with a = 12.05 A, b = 10.55 A, and y = 72. The

assigned unit cell is essentially a 1.5*a, 2*b super cell of the in-plane cell derived from

X-ray powder diffraction studies of lanthanum methylphosphonate.149 More complete

structures of lanthanum benzylphosphonate and lanthanum phenylphosphonate

Table 1. The calculated and observed lattice (d) spacings, in A, for the proposed
lanthanum octadecylphosphonate unit cell, a = 12.05 A, b = 10.55 A, and y = 72.

(h,k) calc obs

(0,2) 5.02 5.0

(2,2) 4.53 4.57

(3,0) 3.82 3.85

(3,2) 3.63 3.68

(3,-2) 2.67 2.67

(2,-3) 2.56 2.56

(-2,4) 2.07 2.11

(4,-3) 1.90 1.98

have been determined;150 however, the methyl derivative was chosen as the structural

analogue to the LB film since both are alkylphosphonates. The methylphosphonate

diffracts in a triclinic space group with a = 5.398, b = 8.168, c = 10.162, a = 73.76,

3 = 83.89, and y = 73.5. 149 The in-plane lanthanum-oxygen network of an individual

layer can be assigned two-dimensional parameters based on this structure of a = 5.398 A,

b = 8.168 A, y = 73.5 with two lanthanum ions per unit cell. The larger unit cell for the

LB film is necessary to make the organic and inorganic sublattices commensurate. The

calculated lattice spacings are compared to the experimental spacings in Table 3-1. This

cell contains three lanthanum ions and six phosphonate groups and provides an average

molecular cross-sectional area per phosphonate group of approximately 20 A2, consistent

with a close-packed, upright organization of the alkyl chains.


The GIXD experiments on a series of Mn2+ and La3+ metal phosphonate LB films

prove that the inorganic networks in these films are isostructural with their known solid-

state analogs, confirming earlier assignments that were based on spectroscopic data. The

three examples described here make up an interesting series. The LaODP film provides

an example where the inorganic and organic networks can be described with the same

two-dimensional unit cells. In the MnA4 film, the two networks are commensurate, but

best described with independent cells. And finally, the packing of the organic network in

the MnODP film appears to be incommensurate with the inorganic network. These

observations reinforce the idea that the metal phosphonate extended lattice will form

regardless of the organic groups, as long as the space requirements of the metal ion lattice

can be met. The larger energy associated with the metal/ligand interactions determines

the structure and the area available to the organic groups.



Many advances in the pursuit of nanoscale objects make use of supermolecular

assembly, the synthesis of larger structures from molecular building blocks.' Inspired by

biological self-assembly, much supermolecular chemistry holds structures together with

directed, non-covalent interactions such as hydrogen bonding, van der Waals,

electrostatic, and 7t-stacking forces.2'4 However, bonding is not restricted to weak

interactions, and the directional properties of coordinate covalent bonding have also lead

to many interesting structures.2'413

Among the motivations for the pursuit of nanometer scale objects is the need

for electronics architectures that are beyond the scope of present-day lithographic

technologies. 151,152 Such architectures will require both nanometer scale device

components and infrastructures such as wires, insulation, and shielding that can service or

interface with these devices at the nanometer scale.2 In addition to electronics and

information storage, other applications of nanoscale architectures include catalysis and

separations, while nanoscale objects also have tremendous potential as molecular level

probes and transducers for chemical recognition sensing. 152

Many of these applications are likely to require positioning the structures at

surfaces. For example, the electronics architectures mentioned above will have to be

fabricated onto a support. Two-dimensional (2D) grid structures have been proposed as

separations media, 153156, which will require their positioning at an interface between

phases. Interfaces may also play a role in the "manufacture" of supermolecular

structures, providing a way of directing interactions by orienting molecules at the surface

for subsequent reaction. Therefore, there is a significant need to investigate the

application of supermolecular assembly processes at interfaces.57-65 With a growing

understanding of how to synthesize supermolecular objects, we can now begin to study

how the requirements of such assembly processes can be adapted for fabrication at

interfaces. Included in this goal is the need to investigate ways to use the interface itself

as a structure-directing feature in the assembly of supermolecular architectures.

Air/liquid interfaces are often used to direct assembly processes, and a careful

understanding of these processes is now possible largely as a result of surface sensitive

characterization methods, including grazing incidence X-ray diffraction.80 Traditional

Langmuir monolayers can form two-dimensional molecular crystals,66'7980'157 or can

selectively bind molecules or ions from the subphase to produce multicomponent

assemblies.80"'158'160 Langmuir monolayers are also used to induce the heterogeneous

nucleation of three-dimensional crystals, where chemical or stereochemical features of

the monolayer can direct the morphology, orientation, or chemical identity of the product

crystals. 161-163 Supermolecular objects have also been prepared in situ at the air/water

interface.80 For example, the 2x2 and 3x3 metal ion molecular grids first described by

Lehn et al.'164-166 been formed at the air/water interface by reaction of Langmuir

monolayers of the linear multidentate ligands with aqueous metal ions.62'63 These and

other examples are included in a recent comprehensive review by Kuzmenko et al.80

Less common are extended two-dimensional covalent grid networks. The best-described

examples are those of Michl and coworkers who use the principles of modular chemistry

to prepare surface-anchored two-dimensional covalent networks.153155'167'168

As part of our investigation of these issues, this article describes the assembly of a

two-dimensional nickel-iron-cyanide grid network at the air/water interface. Numerous

solid-state compounds based on bridging cyanides are known'9'21'22'33'38'51'52'169'172 with

the prototype being Prussian blue. Recent interest in these compounds stems from the

cyanide ligand's ability to efficiently mediate magnetic exchange, and many new mixed-

metal Prussian blue-like structures have been developed with fascinating magnetic

properties.51,52'169'173'174 Structures with one- and two-dimensional coordinate covalent

networks are also known,33'170'171 and metal cyanide complexes have been used as

building blocks in the preparation of "zero-dimensional" clusters.19,21,22'172 In all cases,

the structure directing elements are the well-defined bond angles of the transition metal

complexes and the linear bridging cyanide ligands.

Our approach for assembly at an interface is outlined in Scheme 4-1. The target is

a square grid nickel-iron-cyanide network that arises from the 90 bond angles around the

starting iron cyanide complex. The product is a single monolayer of a two-dimensional

square grid because the amphiphilic dialkylaminopyridine ligand confines the iron

complex to the interface, which then directs the condensation reaction within the plane of

the water surface. In the absence of the interface, the pentacyanoferrate (3+) starting

complex is capable of forming bridges that lead to geometries other than a square grid,

and when the reaction is carried out in solution, only amorphous products are observed.


R = (CHF2),CH3
1 N
N sc I^ .cSN ______ ,A
N Ni(H20)6(NO3) ]6 N
N ,^,e. N ---- Y ^S --e .
II ^Fe- 6N -Ni-
N I C N subphase -_F I'/JNi / Fe


Langmuir monolayer 2D grid network

Scheme 4-1. Assembly of a two-dimensional square grid network at the air/water

The interface facilitates bridging in the equatorial plane of the amphiphilic complex, and

therefore plays an important role in controlling the final structure.

A potential obstacle to confining reactants to an interface is that reactivity can be

limited by restricted diffusion.156 A gas/liquid interface minimizes this problem, allowing

studies to focus on the structure-directing elements of the reactants and surface. In

addition, structures formed at the air/water interface can be transferred from the water

surface to solid supports using standard Langmuir-Blodgett film methods. The

transferred films allow for a more thorough measurement of the structural and physical

properties of the interface-formed networks. The condensation reaction outlined in

Scheme 4-1 is followed at the air/water interface with surface pressure measurements and

with Brewster angle microscopy (BAM), and the structure of the resulting nickel-iron-

cyanide network is confirmed in transferred films with optical and infrared spectroscopy,

X-ray absorption fine structure (XAFS), grazing incidence X-ray diffraction (GIXD) and

magnetization measurements.

Experimental Section


Materials. Unless otherwise indicated, all reagents were purchased from Aldrich

(Milwaukee, WI) or Fisher Scientific (Pittsburgh, PA) and used without further

purification. The 4-aminopyridine was recrystallized from water prior to use.

Instrumentation. All NMR spectra were obtained on a Varian VXR-300

spectrometer. The characteristic solvent peaks were used as reference values. Elemental

analyses and mass spectrometry analyses were performed by the University of Florida

Spectroscopic Services laboratory, where high-resolution mass spectra were collected on

a MAT 95Q, Finnigan MAT (San Jose, CA). Melting points were obtained on a Thomas-

Hoover Capillary melting point apparatus and are uncorrected. UV-Vis spectra were

obtained on a Hewlett-Packard 8452A diode array spectrophotometer. IR spectra as KBr

pellets were recorded on a Mattson Instruments (Madison, WI) Research Series-1 FTIR

spectrometer with a deuterated triglycine sulfate (DTGS) detector.

N-methyl-4-didodecylaminopyridinium iodide (1). A solution of 3.54 g (0.015

mol) N-methyl-4-amininopyridinium iodide175 and 9.37 g (0.0375 mol) of 1-

bromododecane in acetonitrile (75 mL) was refluxed over 5.5 g K2C03 (0.04 mol) for

three days. The acetonitrile was removed and the organic materials dissolved in

chloroform and filtered. The chloroform was removed under reduced pressure and 20

mL of diethyl ether was added to the orange oil that remained. Addition of the ether

solution to 150 mL of pentane with vigorous stirring precipitated the product. The solid

was filtered and washed well with diethyl ether and dried under vacuum (7.8 g, 91%) IH

NMR(CD3CI), ppm: 8.45, d, 2H; 6.79, d, 2H; 4.11, s, 3H; 3.37, t, 4H; 1.54, m, 4H; 1.19-

1.25, m, 36H; 0.80, t, 6H. Calcd for C3oH57N2l: C, 62.92; H, 10.03; N, 4.89. Found: C,

63.25; H, 10.64; N, 4.89. Melting point: 99-101 C. MS (445).

4-didodecylaminopyridine (2). Demethylation of 1 was accomplished in a

manner analogous to a previously reported procedure for the demethylation of pyridinium

salts.176 A stirred mixture of 7.5 g of 1 and 40 g of pyridine hydrochloride were refluxed

under nitrogen in the absence of solvent. After 24 hours the mixture was cooled and 75

mL of water was added to dissolve the excess pyridine hydrochloride. The crude product

was filtered off and redissolved in 100 mL chloroform. The chloroform solution was

extracted three times with 50 mL portions of concentrated ammonium hydroxide and

dried over anhydrous MgSO4 before removal of the solvent under reduced pressure.

Acetonitrile (100 mL) was added to the oil that remained and the mixture was vigorously

stirred in an ice bath. The precipitated solid was redisolved in diethyl ether (100 mL),

treated with 200 mg of activated carbon and filtered through Celite. The ether filtrate

was mixed with 80 mL of acetonitrile and concentrated under a stream of N2 to

precipitate the pure product as beige solid. The solid was washed with acetonitrile and

dried under vacuum (3.9 g, 70%). 'H NMR (CD3CI), ppm: 8.16, d, 2H; 6.41, d, 2H; 3.25,

t, 4H; 1.57, m, 4H; 1.26-1.31, m, 36H; 0.88, t, 6H.. Calcd for C29H54N2: C, 80.86; H,

12.64; N, 6.50. Found: C, 81.14; H, 12.28; N, 6.58. mp 54-56 C MS (431 (+IT)).

Bis(tetramethylammonioum) pentacyano(4-didodecylaminopyridine)-

ferrate(I) 6H20 (3). The preparation of the amphiphilic pentacyanoferrate complex

was adapted from a previously reported procedure for the preparation ofdisodium

pentacyano(4-octadecylamino-pyridine)ferrate(III). 177 To a solution of 1.9 g (0.0044 mol)

of 2 in methanol (50 mL) at 40 C was added 0.40 g (0.0015 mol)

Na3[Fe(CN)sNH3]xH20. The suspension was stirred for 12 hours in air yielding a dark

purple solution. The methanol was concentrated at room temperature under reduced

pressure to a volume of 10 mL and 40 mL of chloroform added. The insoluble iron salts

were filtered off through Celite and the solvents removed. The product was dissolved in

50 mL methanol and precipitated by the addition of AgBF4 (0.003 mol) in 25 mL

methanol. The solid was filtered, washed with methanol and ether, and transferred to a

methanol solution oftetramethyl ammonium bromide (0.003 mol). The mixture was

stirred vigorously for four hours and then filtered to remove the AgBr. The violet filtrate

was concentrated at room temperature to a few milliliters and added to 75 mL

acetonitrile. Concentration of the acetonitrile under a stream of nitrogen, followed by the

addition of several volumes of acetone precipitated the complex as a violet powder,

which turns blue upon hydration. The solid was dried under vacuum in a desiccator over

P205 (0.412 g, 35 %). IR (KBr pellet): vc-N(cm1) 2126, 2116. Calcd for C42H9oN9O6Fe:

C, 57.78; H, 10.4; N, 14.44. Found: C, 57.30; H, 10.76; N, 14.68.


Materials. Unless noted, all reagents were used as received.

Substrate preparation. Single-crystal (100) silicon wafers, purchased from

Semiconductor Processing Co. (Boston, MA), were used as deposition substrates for X-

ray photoelectron spectroscopy (XPS). X-ray diffraction, FT-IR, UV-Vis, and GIXD

samples were prepared on petrographic slides that were purchased from Buehler Ltd

(Lake Bluff, IL). Samples for SQUID and XAFS investigations were prepared on Mylar

(Dupont) substrates cleaned prior to use with absolute ethanol. The silicon, glass, and

quartz substrates were cleaned using the RCA procedure178 and dried under nitrogen.

All substrate surfaces were made hydrophobic by deposition of a monolayer of

OTS. 179,180

Instrumentation. The LB films were prepared by using a KSV Instruments 5000

trough modified to operate with double barriers. The surface pressure was measured with

a filter paper Wilhelmy plate suspended from a KSV microbalance. Subphase solutions

were prepared from 17.8-18.1 MQ cm water delivered with a Barnstead Epure system.

The XPS spectra were obtained on a Perkin-Elmer (Eden Prairie, MN) PHI 5000 series

spectrometer using the Mg Ku line source at 1253.6 eV. Typical operating pressure was

4 x 10-l0 bar. X-ray diffraction was performed with a Philips APD 3720 X-ray powder

diffractometer with the Cu Ka line, X = 1.54 A. Magnetization measurements were

performed on a Quantum Design MPMS SQUID magnetometer.

GIXD and XAFS experiments using synchrotron radiation were performed at the

Advanced Photon Source, Argonne, IL, at the Materials Research Collaborative Access

Team beamline (sector 10). The XAFS spectra of LB films transferred to Mylar were

recorded in fluorescence mode using a Lytle detector. The sample film was oriented at

45 degrees to the incident beam and the detector at 90 degrees relative to the incident

beam. Energy calibration was accomplished by simultaneously recording transmission

XAFS spectra through the appropriate metal foil positioned behind the LB film sample.

Transmission intensity was measured using an ion chamber detector charged with the

appropriate air/N2 ratio to give a linear response over the scanned energy range. The

sample used for XAFS was 100 bilayers. XAFS scans were taken over both the Fe and

Ni edges from 150 eV before the edge step to 1 keV beyond the edge in separate runs for

each edge. Data analysis was performed using the Winxas program.84 Background

subtraction and normalization to the edge step was done with a linear fit to the pre-edge

region and a second order fit to the post edge region. The atomic absorption correction

was done on the k3 weighted data using a cubic spline function. Fourier transforms of the

k3-weighted data were done in combination with a Bessel window function. No

smoothing functions or Fourier filters were applied to the data (for an overview of the

data analysis see Chapter 1) The GIXD scans were performed on LB films transferred to

glass slides. The sample was positioned in the center of an 8-circle Huber goniometer

and oriented at an angle of 0.13 degrees relative to the incident beam. The incident

beam was collimated to 200 microns high by 1500 microns wide and tuned to a

wavelength of 1.254 A. Diffracted intensity in the xy plane was measured using a Nal

scintillation counter mounted on the Huber goniometer. The diffracted signal was

collimated prior to the detector using Soller slits giving an experimental resolution on the

order of 0.015 A-'.

Film preparation. The amphiphilic iron complex 3 was spread onto the water

surface from a chloroform solution. All multilayer films of the nickel-iron-cyanide

network were transferred as Y-type films onto hydrophobic substrates at a surface

pressure of 25 mN/m over a subphase 1 g/L in Ni(N03) at ambient temperature. The

average transfer ratios for both the upstroke and downstroke were 0.85.

Langmuir Monolayers and LB Film Transfer

The amphiphilic iron complex 3 forms a well-behaved monolayer on the water

surface. Brewster angle microscopy indicates the amphiphile is in a liquid expanded state

at zero surface pressure at room temperature. Compression of the monolayer at room

temperature produces the surface pressure versus area isotherm shown in Figure 4-1.



S 0
u 30 o
3 0
U)* 0
20 0
41 100

25 50 75 100 125
Mean Molecular Area (A2)

Figure 4-1. Room temperature surface pressure vs. mean molecular area isotherms for
complex 3 over pure water (open circles) and over a I g[L Ni(N03)2 (filled circles).

The complex is slightly soluble on pure water and a creep of approximately 15

,2/molecule/hour is seen at a surface pressure of 1 mN/m, Figure 4-2.

The behavior of 3 over a subphase containing Ni2 is markedly different from that

over pure water. Brewster angle microscopy indicates that the monolayer is in a

condensed phase at zero surface pressure when the subphase contains Ni2>. Evidence for

a condensed phase at zero surface pressure is also give by the isotherm shown in

Figure 4-1. The mean molecular area of 52 A2 at the onset of pressure for the complex

over a Ni2 subphase is nearly identical to the mean molecular area at collapse of the

complex over pure water. Additionally, the slope of the isotherm is steeper when 3 is

compressed over a Ni2 subphase. As will be demonstrated, this behavior results from a

condensation reaction between the ferric amphiphile and the aqueous nickel ions.


90 .- -. ,

sI = 1 mN/m
< 80 \

a 70 -

o :

o 60

0 10 20 30 40
Time (min)

Figure 4-2. The change in mean molecular area versus time at a surface pressure (p) of
I mN/m for complex 3 over pure water (open circles), and over pure water with
subsequent injection of a Ni(N03)2 solution (at the time indicated by the arrow) under
the monolayer (filled circles). Condensation of the monolayer occurs immediately after
injection of the Ni2+ solution.

The reaction at the air/water interface was also detected by monitoring the change

in mean molecular area (MMA) versus time at a constant surface pressure as shown in

Figure 4-2. A monolayer of 3 was first compressed over pure water to a pressure of 1

mN/m. A slow and linear creep is seen in the film due to the slight solubility of 3. When

10 mL of a solution of Ni(N03)26H20 (at a concentration to give a final subphase

concentration of ImM Ni2+) is then injected under the monolayer, a rapid drop in the

MMA is seen. The MMA eventually stabilizes at a value of 50 A2, which after correction

for the initial creep of ca. 2-3 A2, agrees well with the 52 A2 seen at 1 mN/m in the

compression isotherm performed over Ni2>. shown in Figure 4-2.

To further characterize the product of the condensation reaction occurring on the

water surface, the networks were transferred to various supports by the Langmuir-

Blodgett (LB) technique. The nickel-iron-cyanide network transferred well as Y-type

films from a subphase containing 1 g/L Ni(N03)2 at a surface pressure of 25 mN/m

giving an average transfer ratio of 85% (due to the rigid nature of the film) on both the up

strokes and the down strokes on all substrates used. In contrast, in the absence of Ni2,

amphiphile 3 transfers on the down stroke, but washes off on the up stroke.

Spectroscopic Analyses

Pentacyanoferrate(III) complexes coordinated to a 4-aminopyridine ligand display

an intense ligand to metal charge transfer band between 500 and 700 nm, depending on

the identity of the 4-aminopyridine derivative and the nature of the solvent.181,182 This

charge transfer band is also observed in the transferred films containing the nickel-iron-

cyanide network. The intensity of the charge transfer band increases linearly with the

number of transferred bilayer and demonstrates a reproducible transfer of the network

from one bilayer to the next. In addition, the presence of this band in the LB films

confirms that the integrity of the iron complex is preserved over the course of film


Evidence that nickel is incorporated in the transferred film is found by XPS. A

survey scan of a monolayer transferred on the upstroke onto a clean silicon wafer shows

both Fe (2pi and 2p3) and Ni (2p, and 2p3) peaks. Analysis of the integrated areas of the

XPS multiplex scans using a take-off angle of 80 degrees and taking into account

differences in photoelectron escape depths183'185 for both sets of peaks yields an Fe:Ni

ratio of 1:1 +/- 10%. This ratio is expected if the Ni2' ions are incorporated into a face-

centered square grid assembly as depicted in Scheme 4-1.


Ji V (a)

I ,(b)

2300 2200 2100 2000 1900
Wavenumber (cm- )

Figure 4-3. FT-IR absorption spectra of the C-N stretching region for (a) a monolayer
film of the nickel-iron-cyanide grid network transferred to a silicon ATR crystal, and (b)
a KBr pellet of pure 3.

Confirmation of cyanide bridging in the monolayer networks is found by

comparing the C-N IR stretches in the LB film and a KBr pellet of pure 3 (Figure 4-3).

In 3, the cyanide-stretching region shows a strong band at 2111 cmn' and a shoulder at

2126 cm"1. These bands are in agreement with the pseudo C4v symmetry of the complex.

The FT-IR spectrum of a monolayer of 3 reacted with Ni2 and transferred to a silicon

ATR crystal shows a dominant cyanide stretching band at 2162 cm'1 and a weaker band

at 2118 cm'1. This shift to higher energy is typical when cyanide assumes a bridging


0 2 4 6 8

Figure 4-4. Magnitude of the Fourier transforms (FT) of the (a) nickel edge and (b) iron
edge k3-weighted XAFS for a 100-bilayer sample of the nickel-iron-cyanide grid network
transferred to Mylar. The "R-axis" has not been corrected for phase shifts. No
amplification factors were applied to either trace.

XAFS Analysis

The non-phase-shift corrected Fourier transforms of the k3 weighted XAFS of the

nickel-iron-cyanide network transferred onto Mylar are shown in Figure 4-4 for the iron

and nickel edges. Both sets of data show a similar pattern with three dominant peaks

attributed to the first three coordination shells. For the iron edge, the first two peaks

correspond to the C and N of the cyanide ligand, respectively, and the third peak to the

nickel ion coordinated to the nitrogen end of the cyanide bridge. For the nickel edge

transform, the peak assignments can be made with the first peak corresponding to the

cyanide nitrogen and most likely coordinated water, the second peak to the cyanide

carbon, and the third peak to the iron atom. The significant intensity of the peaks at

0.10 I


.- 0.05


0 1 2 3 4

Figure 4-5. Fit (solid line) to the first two coordination shells of the Fourier transformed
nickel edge XAFS (circles) of the nickel-iron-cyanide grid network on Mylar. The fit
was calculated using FEFF7 codes for a model Ni2+ cluster coordinated in-plane by the
nitrogen end of four cyanides and axially by two oxygen ligands.

approximately 4.5 A in both radial plots has been explained in other cyanide-bridged

systems as resulting from the focusing effect of the linear cyanide bridge.'86'187

The results of a fit to the first two coordination shells in the nickel edge k3

weighted Fourier transformed XAFS data are shown in Figure 4-5. The fit was

accomplished using the program Winxas with inputs from theoretical XAFS parameters

generated from FEFF 7.0 codes85'86 for a model nickel cluster composed of two axial

oxygen atoms and four equatorial nitrogen-bound, iron-terminated cyanide ligands. The

coordination number for Ni was fixed and both shells were fit simultaneously using an

intrinsic reduction factor (So2) of 0.52 for each and an edge energy shift (AEo) of 1. 0 and

6.0 for the first and second shells, respectively. The So2 and AEo values used in the fit

were similar in magnitude to those reported in the XAFS analysis of a similar metal-

cyanide system. 188 The bond distances extracted from the fitting procedure were (in A):

Ni-N, 2.09; Ni-O, 2.11; Ni-C, 3.22; and C-N, 1.13; and are reasonable if compared to

similar compounds.37 3"'40 Fits to the Ni edge XAFS were limited to the first two

coordination shells due to complexities arising from the large number of multiple

scattering pathways contributing to the third coordination shell.

X-ray Diffraction and GIXD

The lamellar order in the multilayer films of the nickel-iron-cyanide network was

confirmed by X-ray diffraction from a 30-bilayer sample. An intense diffraction peak at

2.5 degrees 20 and a weaker harmonic at 5 degrees 20 can be assigned to the (001) and

(002) Bragg reflections and yield an inter-bilayer spacing of 35 A. This inter-bilayer

spacing is reasonable for the size of the amphiphile deposited as Y-type bilayers.

Grazing incidence X-ray diffraction was used to verify the presence of any long-

range in-plane structural correlations in the film. The diffraction pattern obtained for a

39-bilayer sample of the iron-cyanide-nickel network transferred to glass is shown in

Figure 4-6. The counts are normalized to the most intense peak and plotted versus the in-

plane scattering vector Qxy = (47/,)(sin0,y). The three intense peaks can be assigned to

the (2,0), (2,2), and (4,0) Bragg reflections at d spacings of 5.10 A, 3.61 A, and 2.56 A,

respectively, from a face-centered square cell with an edge of a = 10.2 A. The broad

background centered at ca. Qxy = 1.41 A' and the shoulder at 1.58 A -' are likely due to

the poorly organized alkyl chains.78'79 The isolated (4,0) peak was fit to a Lorentzian

function and yielded a full width at half maximum (Qxy,.) of 0.1 A'. Insertion of this

value into the Scherrer equation,'47 = [(1.87r) / (Qxyf,,)], yields an average crystalline


1.0 (2,0) (2,2)
+ ++ (4,0)
4+ +
,, ,+ 41.
+ +

Q ++ *- +
+ + +P + ++

+0.5 + + +
+ + + +
S+ +
r +~

0.0 ,. .
1.0 1.5 2.0 2.5 3.0
Q. (A')

Figure 4-6. The difference between the field cooled and zero field-cooled magnetization,
DM, is shown as a function of temperature. Typical data from the Mn-rich (i.e. x > 0.25)
samples are shown when the magnetic field, for measuring and field cooling, was 100 G.

coherence length () of-60 A, or 5 unit cell lengths, indicating that the 2D networks

cover an average area of approximately 3600 A2.

Magnetic Properties

The magnetic properties of a 10 cm2 sample containing 300 bilayers (150 bilayers

per side) of the nickel-iron-cyanide network transferred to Mylar were investigated by

SQUID magnetometry. Two measurements were performed, one with the sample

surface oriented parallel to the magnetic field and one with the sample surface oriented

perpendicular to the magnetic field. The background corrected field-cooled

magnetization versus temperature obtained in a field of 20 G is shown in Figure 4-7. The

rise in magnetization below Tc = 8 K observed in both orientations is attributed to the

onset of ferromagnetic order. The magnetic behavior is clearly anisotropic, with the

sample displaying a stronger magnetic response when the surface is oriented parallel to


0 3

2 2
1 o
11 .. ..I .... I..: :...
0 ~" 0*** 0 01**
0 5 10 15 20 25 30

Figure 4-7. The temperature dependence of the magnetization after field cooling in 20 G
with the sample surface aligned parallel (filled circles) and perpendicular (open circles),
to the magnetic field. The measuring field was 20 G. The break at Tc = 8 K is indicative
of long-range ferromagnetic order between the Fe3+ (S = '/) and Ni2+ (S = 1) centers.

the magnetic field. The presence of a ferromagnetic state at low temperature is further

supported by the magnetization vs. field data taken at 2 K. The sample shows a rapid

increase in magnetization at low field followed by a gradual approach toward saturation

at higher fields. Cycling the magnetic field at 2 K results in the hysteresis loops

(corrected for the diamagnetic background) shown in Figure 4-8. The plots are

normalized to the magnetization at 5 T. Again, there is clear anisotropy in the magnetic

behavior between the two orientations of the sample with respect to the field. When the

field is parallel to the sample surface, the magnetization increases more rapidly with

respect to the field and the remnant magnetization is 35% versus 8% in the perpendicular

orientation. The coercive field is also slightly anisotropic, being 140 G in the parallel

orientation and 110 G in the perpendicular orientation.

1. . . . . . .

0.5 00
0880 6
0. oo
--. o --- ^ ^---
-o 880 %
-0.5 -oO 1 :040

-1.0 .* . . . . . . . .
-1.0 -0.5 0.0 0.5 1.0
H (kG)

Figure 4-8. Hysteresis loops measured at 2 K with the sample surface aligned parallel to
(filled circles) and perpendicular to (open circles) the applied magnetic field. The
magnetization is normalized to the saturation magnetization.


Choice of System and Monolayer Behavior

Designing a system to result in the formation of a coordinate-covalent network at

the air-water interface requires the appropriate transition metal complex building blocks.

Octahedral transition metal ions possessing linear bridging ligands are well suited to the

assembly of square-grid networks since the required 90 bond angles are inherently

present. By substituting one position with a hydrophobic ligand, the building block can

be made amphiphilic and thus tailored for assembly reactions at the air-water interface.

Condensation of the amphiphilic building block can then be accomplished by reaction

with a suitable aqueous transition metal ion contained in the subphase. Confinement of

the reacting system to the water surface directs the resulting structure to a two-

dimensional motif

The numerous examples of pentacyanoferrate complexes and the substitutional

inertness of the cyanide ligand make this class of compounds well suited to our assembly

strategy. Multilayer films of the single chain derivative of 3, (4-octadecylamino-

pyridine)pentacyanoferrate(III) (4) has previously been described. 189 The sodium salt of

4 was reported to be too soluble for film preparations, but when prepared as a mixed film

with hexadecyltrimethylammonium counterions, stable Langmuir monolayers resulted.

No difficulties with hydration of the single-chain complex were reported.

We decided to modify the pentacyanoferrate complex to the dual chain derivative

to match more closely the size of the amphiphilic ligand with that of the metal complex

head group. The addition of a second alkyl chain also decreased the solubility of the

complex and eliminated the need for long chain alkyl ammonium counterions. It was

found to be beneficial to exchange the sodium counterions for tetramethyl-ammonium

ions to decrease the hygroscopic nature of the complex and to aid its dissolution in

chloroform. The resulting complex forms a Langmuir monolayer that creeps slowly on

water (Figure 4-2), but forms a highly stable film after reaction with aqueous Ni2 ions to

form an insoluble polymeric network.

Evidence for the condensation reaction is seen in-situ at the air/water interface. In

the absence ofNi2, 3 forms a liquid expanded phase upon compression. This behavior is

reasonable, as amphiphiles with twelve-carbon alkyl tails do not normally form

condensed phases at the air-water interface at room temperature.190 Upon addition of

Ni2, a condensed phase is seen in the pressure vs. area isotherm and in Brewster angle

microscopy. The mean molecular area of 52 A2 at the onset of pressure correlates with

the limiting area per molecule of the complex over pure water and suggests that the film

is highly condensed at zero pressure over the Ni2 subphase.

In a mixed-metal cyanide square grid network (Scheme 4-1), a centered unit cell

will have two iron amphiphiles per unit cell. Doubling the area per molecule, determined

from the pressure vs. area isotherm, gives a cell area of 104 A2, which then corresponds

to a cell edge distance of 10.2 A. This value is in agreement with the 10.2 A2 determined

for the cell edge by GIXD and suggests that the mean molecular area is determined by the

lattice spacing of the inorganic two-dimensional grid network and not by the Van der

Waals interactions of the organic chains.

Furthermore, BAM and surface pressure data indicate that the nickel-iron-cyanide

network forms with or without reorganization of the monolayer. That is, the MMA

obtained by compression of the iron amphiphile over a Ni2 subphase is very close to the

MMA obtained after injection of a Ni2 solution under an organized monolayer of the

iron complex. A condensed film is formed at zero pressure over the Ni2 subphase, and

subsequent compression of the film only acts to push together domains that have already

assembled at the interface.

Structure of the Network

The formation of an extended two dimensional array is dependent on the

exclusive bridging of the four in-plane cyanide ligands, as bridging of the trans cyanide

ligand would effectively terminate the structure and result in an amorphous inorganic

polymer. Evidence for a well-organized network from GIXD, XAFS, and FT-IR suggests

that while coordination of the axial cyanide is possible, this mode is most likely labile in

the absence of the added stability brought on by extended bridging interactions.

The results of the GIXD clearly show the presence of a structurally coherent

inorganic network in the mixed metal film. The three peaks shown in Figure 4-6 fit very

well to the expected (h k) pattern for a face-centered square grid network. The unit cell

edge length of 10.2 A deduced from the diffraction data is very similar to that reported in

cubic Prussian blue derivatives. 191 The high background scattering near 1.4 A"k' and 1.6

A-' is in the range of Qxy normally seen for alkyl chain packing and suggests that the

alkyl chains are loosely organized. This observation would be expected in light of the

large area per alkyl chain in the condensed network.

The XAFS data complement the conclusions of the GIXD experiments. The

Ni-N, C-N, and Ni-C distances of 2.09 A, 1.13 A, and 3.22 A, respectively, were

obtained from modeling the Ni edge XAFS. Combining these bond lengths with the

average Fe-C bond length of 1.95 A reported for other Fe-CN-Ni bridged systems,38'40

leads to a Ni-Fe separation of 5.17 A. This value is close to the 5.10 A separation

deduced from the GIXD. The quality of the XAFS fit supports the modeled nickel

coordination environment in which the octahedral nickel ions are coordinated in-plane by

the nitrogen terminus of the cyanide bridge and the axial sites by oxygen, most likely

from coordinated water.


The formation of a structurally coherent inorganic network at the air-water

interface is confirmed by the transition to a ferromagnetic state below 8 K in the

multilayer film containing 150 bilayers per side. The ability of the cyanide ligand to

mediate magnetic exchange between two paramagnetic metal ions is well known and has

been extensively explored in cubic Prussian blue derivatives. In particular, the Fe3/Ni2

Prussian blue'92193 was found to be ferromagnetic with a Tc of 23 K. In addition,

ferromagnetic exchange has also been reported in a series of two-dimensional cyanide-

bridged iron-nickel compounds with T.'s on the order of 10 K.38'194 The ferromagnetic

behavior of these materials is rationalized'"95 by realizing that for octahedral metal

centers, the magnetic orbitals are the Fe3+ (S = '/2) t2g and the Ni2 (S = 1) eg sets, and that

the cyanide orbitals that overlap with each of them are orthogonal.

For the Fe3/Ni2' LB film system, the ordering temperature of 8 K is lower than

the T, of 23 K observed in the cubic analogue, and is more similar in magnitude to the

ordering temperature reported in other low dimensional Fe-CN-Ni networks. 196 Lower

ordering temperatures for the 2D systems relative to the cubic analogues is expected as

the number of exchange pathways per magnetic ion is reduced. Further evidence for a

two dimensional network is obtained from the anisotropic magnetic behavior seen in the

film. The stronger magnetic response of the sample when oriented with the surface

parallel to the magnetic field suggests a magnetic easy axis within the plane of the

network. A strict analysis of a magnetic vector in the film is limited though due to

uncertainties in how the microscopic surface roughness of the substrate affects variations

in the orientation of network sheets relative to the plane defined by the macroscopic

substrate. The high anisotropy of the magnetization does discriminate against the

magnetic behavior arising from cubic Prussian blue-like particles and is highly suggestive

of a low dimensional system.197 More detailed studies on the magnetic properties of the

LB film networks are ensuing since the unique structural features of these monolayer

networks may provide experimental probes of the exchange coupling interactions in

metal cyanide networks and the how the issue of dimensionality influences ordering in

mixed-spin 2D systems. 198199

Mechanism and Structure Directing Elements

The two-dimensional nickel-iron-cyanide network forms at the air/water interface,

but does not require pre-organization of the amphiphiles. The condensation reaction

proceeds in the absence of applied surface pressure when the amphiphile 3 is spread on

the Ni2+ containing subphase, in which case subsequent reduction of the surface area

simply compresses the preformed domains. Compression of the film thus appears to do

little to extend the in-plane order of the networks, and instead, only works to increase the

density of the domains allowing for better transfer of the networks to solid supports. Pre-

organization of the amphiphile, followed by injection of Ni2 ions into the subphase

results in the same network, with no significant difference in domain organization.

Control of the reaction to form the square grid network results from the

orientational constraints of the octahedral metal complex with linear cyanide bridging in

combination with the interface as a structure-directing element. This view is supported

by attempts to form the same networks from solution. The analogous reaction of 3 with

Ni2 in methanol yields an insoluble precipitate, which is shown by X-ray diffraction to

be amorphous. When compared to the homogeneous reaction, the air-water interface not

only directs the structure of the final material, but also acts to enhance the structural

coherence length as well.

It is interesting to compare the mechanism of formation of the metal cyanide two-

dimensional networks to other examples of Langmuir-Blodgett films that contain

inorganic networks. For example, there are now several examples of metal phosphonate

based LB films, where the inorganic extended solid networks determine the in-plane

structures. 109,117,145 The difference is that for the metal phosphonates, the LB films form

with the same structure that forms in the solid-state. The structure is determined by the

inorganic lattice energy. With the metal phosphonates, the LB film processing directs

where the structure will form and affords control of the fabrication to one layer at a time,

but the air/water interface does not act as a structure-directing element. In contrast, the

iron-cyanide-nickel network described here does not form from compound 3 in the

absence of the interface. The interface directs where the reaction will take place and

limits the reaction to one layer at a time, but importantly, it also directs the structure.

An amphiphilic octahedral iron complex containing linear cyanide ligands was

designed and synthesized as a building block for the assembly of two dimensional square

grid nickel-iron-cyanide networks at the air-water interface. The reaction of Langmuir

monolayers of this complex with aqueous nickel ions contained in the subphase results in

the formation of coordinate covalent networks. Characterization of these networks by

various techniques indicates that the structure is two-dimensional and coherent over an

average domain size of 3600 A2. Magnetic measurements indicate a ferromagnetically

ordered state below 8 K with the magnetic behavior highly dependent on the orientation

of the sample with respect to the field. This synthetic method demonstrates that the air-

water interface can function as a structure-directing element in the assembly of new

supermolecular network solids and, in addition, provides a means for transferring these

materials in a controlled fashion to solid supports. This assembly strategy may aid in the

future development of nanoscale materials and with interfacing them at a surface.



Coordination chemistry routes to finite and infinite networks make use of the

predictable directional characteristics of coordinate covalent bonds. 14.15 .28.29.4547,200-202

Tunable variables like stochiometry, template additives, secondary structure building

blocks, or kinetic control are used to determine the final network structure, and several

examples are included in the current issue. Potential applications of inorganic finite and

infinite networks include recognition and sensing, catalysis, electronic and optical

functions, and magnetic effects related to information storage.":2>ou '' It is interesting to

note that several of these applications are likely to involve positioning at surfaces, and

routes to locate the finite or infinite networks at interfaces will be needed. 57-bu62-6s.2r' 204

One approach is to involve the interface directly in the assembly, to carry out the network

fabrication where it will be located. In this case, the interface can play a role in

determining the network structure. Examples of assembly at liquid interfaces have been

published, including some two-dimensional infinite networks.62'63'80'154'155'167'168 The

surface of a liquid retains the structure directing character of an interface, but at the same

time is fluid and can facilitate diffusion of reactants. Careful understanding of these

processes is now possible largely as a result of surface sensitive characterization

methods, including grazing incidence X-ray diffraction as detailed in a recent review.80