David Turnbull

DAVID TURNBULL
Autobiography
Youth
I was born (on February 18, 1915) and reared on a family farm in northwest Illinois. Our
community, Elmira, was settled in the 1840's, primarily by immigrants from the border and western
highland regions of Scotland. Later there were migrants from the Appalachian hills who were
attracted by jobs in the nearby soft coal mines and by the relatively cheap land in the region.
Among them was a woman, widowed by a Kentucky feud, who was determined to remove her
several sons from the cycle of violence.
Three of my grandparents were, as children, among the Scottish immigrants and the fourth, my
maternal grandmother, was an orphan, thought to be of Pennsylvania Dutch descent, who had been
adopted into the family of one of my father's great uncles. The Turnbulls were an old border family
who, according to my English friends, were notable for trans-border raids on English cattle herds.
My father's mother was from a highlands family, the McLennans, whose native tongue was Gaelic.
My mother's father, George Murray, presumably descended from the Atholl Murrays, became a
prosperous farmer in the community. My grandmother Murray was the only one of my
grandparents alive when I was born. As a little boy I much enjoyed the trips in a buggy or sleigh,
with my mother, to visit her at the Murray homestead. My only sibling, a brother, Theodore, was
four years younger than I. He became a medical doctor.
My father operated our farm in partnership with his brother, Edwin, and sister, Mary Ellen,
neither of whom ever married. They lived in the house of their parents with a niece, Mary Irene,
fifteen years older than I, whose mother died when she was an infant. An exuberant young woman
and skilled equestrienne, Mary Irene was as a sister to my brother and me. My parents lived in a
nearby house, adjacent to my aunt's, built just after their marriage. Three of my father's brothers
David Turnbull -- Autobiography 2
had farms in the neighborhood, and with their families often visited us on Sunday afternoons.
Some of my recollections go back to the time of World War I. I remember the concern of the
family for my favorite cousin, Willie, a brother of Mary Irene, who was in the Argonne offensive
and my joy when he returned unharmed.
Our farm's extent was 400 acres, only half of which was cultivated. The remainder was quite
hilly and wooded, mainly by oak and walnut trees. It was crossed by a spring-fed stream that
provided water for our livestock. Our principal crops were corn, oats, barley, soybeans, clover, and
alfalfa. We had a steady animal population of about 100 cattle, 80 sheep, 300 pigs, 100 chickens and
30 horses. We farmed entirely with horses until, in my fifteenth year, we obtained a tractor. The
quality of our soil was well below the norm for the region, but through hard work and the skill and
shrewdness of my father and uncle, we lived quite comfortably until the onset of the great
depression. We led quite a cloistered life, rarely venturing more than a few miles from home. We
had no radio until I was age sixteen and I never entered a movie theater until my second year in
college.
My father had a high school education, supplemented by a short accounting course. As a
young man he was quite an outstanding athlete, starring in track and football. He was dynamic and
pragmatic and exceptionally skilled as a mechanic as well as in the other agricultural skills. He was
somewhat impulsive and would fearlessly herd, quite unprotected, the most powerful and potentially
dangerous of the farm animals. He advised me that if I showed it no fear, an animal wouldn't bother
me, but I was never able to face a huge shorthorn bull or a neighbor's German shepherd dog as
unconcernedly as he did. For many years he served as township supervisor, as well as chairman of
the county board of supervisors. Also he was, for a considerable period, president of the county
Farm Bureau Federation. He was a strict but devoted parent with a touch of Gaelic sentimentality.
I was never in a state as forlorn as that of the prodigal son, but had I been I am certain that my
father's response would have been like that of the father in the parable.
After a two-year course at Oberlin College my mother was, until her marriage at age 35, a
country school teacher. She was a quiet and gentle woman who ardently supported the feminist
causes of her day. She did paintings and played the piano quite well. About the time of my
brother's birth, at age 42, she developed an arthritic condition which increased in severity until her
death thirteen years later. Despite her illness, she was always cheerful and managed to perform,
with some help from my brother and me, the varied and onerous duties of a farm woman at that
time. In temperament and intellect I was most like her. In analytical ability I somewhat resembled
my father but, to his considerable disappointment, I had none of his athletic skill.
My family belonged to a strict Presbyterian sect of the type described by Barbara Tuchman in
her "Bible and Sword". The sect placed great emphasis on the Bible and stressed the teachings of
the Old Testament as much, or even more, than those of the New. Only Psalms were sung at the
main religious services and, in view of the uncompromising monotheism of the Old Testament, the
Trinitarian doctrine, while formally accepted, was treated rather gingerly and with a bit of
embarrassment. My father and aunt were active members of the church and I was required to attend
Sunday School and listen to the sermons. At home the Bible was read aloud, one chapter each
morning, serially from Genesis through Revelations. Though formally a member, my mother did
not attend the church and was, I believe, quite skeptical of its dogmas. In guiding my brother and
me she stressed right conduct, with no reference to "right" belief, and often cited the ethical precepts
expressed in the Sermon on the Mount, the Lukan parables, and the Old Testament wisdom
literature. As a little boy I was fascinated by the tales of such Biblical heroes as Gideon and David.
While still quite young I thought much about philosophical and religious questions. Admiring my
father I wanted to accept his beliefs, but I became skeptical of them quite early. While repelled by
the Christian salvation theology, I came in my early teens to admire the exalted ethical ideals of
Jesus and for a time tried, I'm afraid with rather little success, to live up to them. The problem of
Jesus as a historical figure is one that has always intrigued me.

I began elementary school at age 5 1/2. The school, located in Elmira village (pop. ~100), 1 1/2
miles from our farm, had two classrooms, but when I attended, there was but a single teacher and
only one classroom was used. I walked to and from school until I was in grade six; after that I
reached it on ponyback. About thirty students were distributed over the eight grades of the school,
and only two, besides me, were in my class. Following the usual practice in country schools,
lessons were recited to the teacher by each grade in front of her desk, while the students in the other
grades were expected to continue their study.
Apparently I had a mild case of dyslexia, for I was virtually unable to read until my third year
in school. Because of this disability I couldn't read the arithmetic problems, so my learning in all
subjects was impeded. When my difficulty became apparent, my mother proceeded to instruct me
each evening at home, working patiently with me for more than a year. My progress was
discouragingly slow until one evening, when I was faced with what seemed to be a most formidable
lesson, bristling with many strange new words. My mother braced herself for the worst, but much
to her and my amazement, I quickly read the lesson through without prompting or error, and then
went on to read several lessons in advance. Ever after I read well and voraciously. I quickly caught
up with my class in all subjects and also became a flawless speller. This remarkable change in my
reading ability occurred, I think, when I grasped the use of context as a guide.
Our community was not a part of any secondary school district, but my parents managed to
have me admitted to Kewanee High School which, at that time, was considered to be the most
outstanding one, academically, in the area. Kewanee was a manufacturing (steam boilers, tools, and
textiles) town with a population of 17,000 located about eight miles from our farm. It could be
reached only by way of unimproved dirt roads and no transportation was provided. I drove at first a
model "T" and then a model "A" Ford car to the school or, when the roads were impassable,
walked. One or two neighbor children, who also attended the school, rode with me. Upon
beginning high school my age, 12 1/2 years, was below the state specified minimum, 15 years, for
drivers. However, in the rural areas this minimum was not enforced and largely ignored; also there
was, at the time, no driver licensing. Fortunately, I drove to and from school the entire four years
without mishap.
The transition from a country to a strange city school, with 600 students, was quite difficult for
me. The few farm students were looked down upon and sometimes ridiculed for their country ways
by many of the city students. Academically the school entirely came up to my parents'
expectations. It was staffed by a group of outstanding and dedicated teachers, mostly women.
Undoubtedly students at the time were fortunate that most highly qualified women chose the
teaching profession since, as Derek Bok and others have noted, the dictates of the temporal mores
virtually excluded them from most of the other learned professions.
I was especially inspired by the Misses Minnie Trask and Wildred Ewan, who posed
demanding intellectual challenges and encouraged imaginative but rigorous modes of thought.
Minnie Trask, a frail and elderly spinster, taught mathematics with a rather "corny" style, but she
had a profound grasp of her subject and assigned difficult problems. On occasion she would pose
a geometry problem and predict that no one in the class would be able to solve it. Such challenges
were, for me, irresistible and, after strenuous effort, I always solved the problem. To my
considerable embarrassment, Miss Trask would display my solution and tell the class that I had
done a great thing. To that point, which was in my third high school year, my scholastic
performance had been considerably above average, but far from outstanding. After this I excelled in
mathematics, physics, and the social sciences. The latter were taught by Wildred Ewan, a young
woman who encouraged her students to "think the unthinkable" and to examine all viewpoints,
objectively and critically, however unconventional they might appear. It seemed that I learned most
from having to solve difficult problems and rather little from classroom exposition, however
excellent it was. Because of this, the emphasis of my own teaching has been on posing meaningful
and challenging problems. In this I have tried to follow the dictum attributed to Galileo, "You can't
teach a person anything, you can only help them to find it within themself."
Often I spent the high school lunch hour at the well-stocked city library and there discovered
periodicals such as Harper's, the Atlantic, the New Republic, and the Nation, all of which I enjoyed
reading, as well as some of the more popular magazines, such as the Saturday Evening Post. I well
remember the rosy predictions on the economic future in most of these magazines, as well as in the
newspapers, shortly before the onset of the great depression. I completed high school in 1932 at
the depression's nadir and the death of my mother occurred at the same time. After her death, our
aunt Mary Ellen provided, with great devotion, what maternal care my brother and I needed. My
father remarried four years following my mother's death.
My parents began assigning me regular farm chores, which I eagerly undertook, when I was
age six and by age twelve I was doing, when not in school, the work of a grown man. During the
busiest season, from mid-May to mid-September, this meant rising at 4:30 A.M. and working till
7:00 P.M. six days a week, with rather brief intervals for meals. However, I much enjoyed the
woods and fields and most of the varied farming activities, especially working with horses and
herding cattle. Indeed, I would have preferred to spend my life operating our farm, but an asthmatic
condition, which became apparent when I was age two, greatly impaired my ability to perform
certain essential tasks. By the time I completed high school, it was clear that I would have to find
some career other than farming. In the post-Calvinist milieu of my home, it was taken for granted
that any career should be directed toward human betterment, as it was understood. However, I had
only vague notions of what the possibilities were. I definitely did not want to become a minister or
medical doctor, but thought that I might prepare to teach mathematics and history. To this end, I
entered Monmouth (Illinois) College, a small (~400 students) liberal arts institution governed by our
religious sect.
College and Graduate Study
The Presidential campaign of 1932 was in progress at the beginning of my first college year. I
joined about 25% of the other students in voting for the socialist candidate, Norman Thomas, in a
straw poll conducted by the college newspaper. Later I was highly impressed by the policies and
leadership of Franklin Roosevelt which, in my judgment, saved the American family farm for at
least two more generations. Since that time I have always supported the nominees of the
Democratic party for the Presidency. In this, I broke with the political persuasion of my father's
family that, since the Civil War, had always been Republican or Progressive.
I found that to qualify as a teacher I had to complete five or six courses in a subject called
"Education", so I took two of these during my first college year. Never before nor later did I
encounter courses that seemed so trivial and devoid of intellectual content. I resolved, rather than
take any more such courses, to give up my plans to become a teacher. Later I learned that at most
other colleges the program in "Education" was much the same as at Monmouth.
At the end of my first college year economic conditions were still very grim and it appeared
that we would surely lose our farm if the depression continued. I recall, at the beginning of a
"bumper" oat harvest, calculating that the 3 cents/bushel return for our crop could hardly pay the
taxes on the land on which it was grown. I was giving much agonized thought to values and aims
when I came to feel that we are happy to strive for a "good" society, but would be quite bored to live
in a society, if ever attained, with no major problems. Suddenly it seemed that life would be
pointless if my ideals ever were realized. This outlook, which I later found expressed vividly by
Dostoevsky in his "Notes from the Underground", deeply depressed me for a considerable period.
However, I gradually realized that my knowledge and perspective were much too limited for any
conclusive assessment of values and ends. At that point I decided that I would find greatest
satisfaction in endeavoring to advance knowledge and understanding through the practice of
science. I sensed that my intuitive intellect was, perhaps, better suited to the social than to the
physical sciences. However, I decided to take up the physical sciences because I thought them
more effective for enhancing knowledge. From that time, when I was eighteen, to the present, I
have, though often frustrated and discouraged, directed my efforts to the limits of my capabilities, to
the advancement of physical science.
My high school experience inclined me toward physics, but Monmouth's physics offering was
very limited. At the beginning of my first college year an adviser had suggested that I keep my
options open by taking a chemistry course. This I did and found the lectures of the chemistry
professor, W.S. Haldeman, highly interesting and inspiring. They showed me that chemistry was
much more challenging, intellectually, than I had appreciated during my high school course.
Therefore, I decided, at the beginning of my second college year, to concentrate in chemistry and to
take all of the mathematics courses offered. Analytical and physical chemistry were taught by
another outstanding professor, G.W. Thiessen. Actually, I learned more from him, because of his
challenging problem assignments, than from Haldeman and it was his teaching which attracted me
to physical chemistry. I excelled in chemistry and Haldeman urged me to go on to Ph.D. studies.
He recommended me to several graduate schools and I received and accepted a scholarship offer
from the University of Illinois which would cover my tuition and most living expenses.
At the time, Worth H. Rodebush and Thomas E. Phipps were the physical chemistry
professors at Illinois, and I was fortunate in having close association with both. My first year at
Illinois was tense and very difficult. The entering graduate students in chemistry were informed
that the department's research facilities could accommodate only half of them, so the half with the
lowest grades and qualifying test scores at the end of the year would have to leave. As a
scholarship holder, I was required to serve quarter-time as a teaching assistant. In the first term I
was assigned to grade problem sets in a physical chemistry course which I at the same time audited,
since it was considerably more advanced than my Monmouth course. During the second term I
supervised a section of the physical chemistry laboratory and developed a new experiment for it. I
found that my Monmouth training was substantially less advanced than that of first-year students
from Illinois and other research universities. However, well before the end of the year I had caught
up with these students and probably had passed most of them. I was able to do this, I believe,
because at Monmouth I had been taught the simple basic concepts well. In striving to provide their
undergraduates with the most advanced training possible, the research universities have tended,
perhaps, to slight the simpler concepts central to the discipline. Often such neglect becomes
painfully evident during oral qualifying examinations. Perhaps the small liberal arts colleges can
still compete effectively in the training of science undergraduates by concentrating on these simpler
concepts. Also, because of their small classes and the greater commitment of their faculty to
teaching, these colleges can outdo the large universities in individualized laboratory instruction.
Such instruction was, indeed, an invaluable part of my Monmouth education.
Having excelled in courses, including several in organic chemistry and physics, and qualifying
tests during my first year, I was awarded a fellowship in each of the following two years. These
came with no teaching obligations and provided, more than adequately, for my living expenses. Out
of the $1,400 total of scholarship and fellowship money received in my three years at Illinois, I saved
$300.
Rodebush was, at first, my adviser. With Wendell Latimer he is credited with having
introduced the hydrogen bond concept into chemistry. He was considered, rightly I think, a genius
of sorts, but one who did not exert himself unduly. He was curt and sometimes snappish, but
behind his rough facade he was fair and quite caring of his students. His lectures were disjointed
and cryptic and highly unpopular with most of the students. Nevertheless, they usually dealt with
challenging issues and often exhibited profound physical insight. It was as a result of these
lectures that I became fascinated by the problem of nucleation. Rodebush pointed out that to persist
and grow, a nucleus had to reach a critical size such that it would contain hundreds of molecules.
How this size could be reached mystified me. Perhaps I would not have been so mystified had I
grasped the full implications of the principle of microscopic reversibility. Anyway, I resolved to
study nucleation if the opportunity arose, and I listed as a problem in my notebook "the kinetics of
crystal nucleation in liquid Hg".
For my earliest seminar at Illinois, Rodebush asked me to review the paper, then just published
in the Journal of Chemical Physics by Henry Eyring and coworkers, associating crystal melting
with an abrupt appearance of "communal entropy". Upon studying the paper, it seemed to me that
the argument hinged on the assumption, which I thought incorrect, that self-diffusion could not
occur in the crystalline state. In reviewing the paper, I emphasized this criticism, though fearful that
I might have misunderstood the development. Actually, to my considerable relief, Rodebush
approved and liked my critique.
At Rodebush's suggestion I took on what must be almost the ultimate challenge in metastable
structure synthesis, that of producing pure ammonium metal. Rodebush thought that such a
synthesis might be achieved by electrolysis of ammonium salts dissolved in a non-hydrogen
bonded liquid of very low dielectric constant, like dioxane. He was encouraged by the reports of
Kraus and coworkers from Brown University that in similar solvents certain salts did exhibit
appreciable solubility and electrical conductivity. Phipps undertook to instruct me in the techniques
of glass blowing and high vacuum necessary to proceed with the investigation. Our efforts to
synthesize pure ammonium came to naught, but I gained valuable experience from them.
Guided by Phipps, I then began an investigation of the periodic, with field, deviations of the
thermionic emission currents from the Schottky theory. Phipps had discovered these deviations
while checking his techniques preparatory to a planned study of the contact potential of I2 on W.
He and earlier students, Seifert and Clemens, then measured the deviations at intermediate fields.
Phipps thought the behavior quite bizarre and hardly believable. My task was to repeat the
measurements and extend them to higher fields. During my study I benefited greatly from the
tutelage of Ralph Seifert as well as that of Phipps. The attainment of ultra-high vacua was crucial to
experimental success, and the emission behavior during my final measurements indicated vacua
better than 10-13 torr. My results fully confirmed Seifert's and showed further that the deviations
persisted with increasing period and amplitude to still higher fields.
With these results Phipps decided that publication could be risked so two papers were
prepared, the first by Seifert and Phipps and the second by myself and Phipps, and submitted to
Physical Review. Both were accepted, and then published in the autumn of 1939. Later, at Irving
Langmuir's suggestion, Wayne Nottingham of MIT repeated our experiments and obtained the
same results. The deviations were interpreted as a quantum interference effect by H. Mott-Smith
and by Eugene Guth and coworkers.
Learning and working with Phipps was an invaluable experience. I have not known a more
selfless and dedicated scientist than he was, nor any who imposed higher standards of care,
accuracy and repeatability in experimentation on himself and his students.
Having completed my thesis, I received the Ph.D. degree in June, 1939, and then sought an
academic position. Few such positions were open at the time, but following Rodebush's strong
recommendation, I was offered and, as he advised, accepted an instructorship in physical chemistry
at Case (then known as Case School of Applied Science). The practice of visitation, when applying
for a university position, was not then common, and I neither visited Case nor was interviewed by
anyone from there before accepting their offer.
Case School of Applied Science
When I began teaching there, Case was just initiating doctoral programs in chemistry and other
fields, but its major emphasis was still on undergraduate instruction. I found that I was required to
carry a very heavy teaching load, amounting to 24 hours of laboratory and classroom instruction per
week. My duties included teaching: all sections of the physical chemistry laboratory, a required
course in physical chemistry for the metallurgists and another for the physicists, and enough
sections of elementary chemistry to complete the load. There were about 20 to 25 students per
section of the physical chemistry laboratory which meant that 10 to 12 experiments had to be in
operation simultaneously. At Monmouth we set up our own experiments and had 6 to 10 hours to
complete each one. In contrast, the Case schedule allowed a student only 3 hours per experiment.
This meant that every experiment had to be "canned" and I had to do the "canning" and
maintenance with very little assistance.
My first afternoon of instruction in the Case physical chemistry laboratory was my worst
teaching experience. I found that most of the students, though they had taken laboratory courses in
analytical chemistry, were quite inept with little sense of what precautions were essential when using
delicate equipment. All the afternoon I rushed from one crisis to another as ammeters smoked and
glassware broke. Somehow the class and I survived. In time I learned where the troublespots were
and eliminated some by redesign of experiments.
Most of the Case students were quite intelligent and teaching a section of Freshman Chemistry
could be very stimulating. The students were lively and I was kept "on my toes" by their, usually,
penetrating questions. However, as they advanced to upper classes, most became rather docile and
less vibrant intellectually. In many of the engineering courses they were required to write detailed
and lengthy reports and to present solutions to the same problems, with trivial variations, many
times over. This routine left little time for reflection and exercise of the imagination and so, I
believe, stultified intellectual development. Such a mode of education was not unique to Case but
was, I think, prevalent in most engineering schools at the time.
My first term at Case was very strenuous, and soon after it began I developed, at certain
periods of the day, severe stomach pains. I had experienced similar pains from age thirteen, but
they never were serious enough to impede my work or study. However, at Case they became so
nauseating that I could hardly teach, so I had to seek diagnosis and treatment. Fortunately, I was
directed to a brilliant diagnostician, Harley Willims, who immediately attributed my symptoms to a
duodenal ulcer. When his diagnosis was confirmed, I went on a strict diet with medication and the
symptoms quickly disappeared. They recur periodically, but are relieved when I exclude certain
items from my diet for a short period. Our elder son, Lowell, developed similar symptoms, with the
same diagnosis, also at age thirteen.
Beside carrying a heavy instructional load I managed to do some research at Case. This
research included: investigations, in collaboration with Samuel Maron, of the validity of certain
scaling relations for the thermodynamic properties of gases at high pressures; some experimental
studies, which I still think highly of, in the physical-organic field; studies of the kinetics of
oxidation of metals and of the corrosion of certain alloys in non-aqueous media; and studies, in
collaboration with Carl Prutton, of the mechanisms of lubrication under high loads. Also I guided
the Ph.D. research of two students, Delton Frey and Paul Hartman. The results at Case of my
collaborators and me, are described in eighteen papers.
In late 1942 I was offered a position on the Manhattan Project, at the University of Chicago's
"Metallurgical" Laboratory. In my middle teens I had been, as was my mother, an absolute pacifist,
but my pacifism eroded as it became clear that the German dictator really meant to inflict the
horrors promised in his writing and oratory. Thus, I had little moral objection to the project
research, which was clearly motivated by a near hysterical fear that Germany would soon perfect the
"bomb". However, I was reluctant to sever my academic connection, and so decided to remain at
Case where, throughout the war, I was heavily engaged in training the several classes of naval cadets
graduated each year and in research on synthetic rubber and on the improvement of additives for
heavy-duty lubricants.
Various experiences at Case aroused my interest in solid state science, or "materials" science as
we now label it. Since I was teaching the metallurgists physical chemistry, I decided to inform
myself about metallurgy, and found it quite fascinating. My investigation of metal oxidation
kinetics led me to the papers of Carl Wagner, which highly impressed me. The arcane always
appealed to me and, from my chemist's perspective, it seemed remarkable that transport rates in
crystals could, as Wagner's theory indicated, be increased many orders of magnitude by the
presence of minute concentrations of point defects. I wondered about the role of extended
imperfections in such transport, and resolved to investigate it when I could. My admiration of
Wagner was such that I hoped to study with him when the war was over. Actually, he joined the
MIT faculty after the war and I was fortunate to have had some professional association with him. I
felt greatly honored when, in 1952, he tapped me to succeed him as chairman of the Gordon
Conference on the "Physics and Chemistry of Metals"; this conference later fissioned into the
conferences: "Condensed Phase Physics" and "Physical Metallurgy".
My interests in solid state science deepened further during my final years at Case. In this
period I was especially instructed by the books of Hume-Rothery and Seitz on metal physics. I
resolved to conduct my future research in this field, but my Case situation did not seem favorable
for this, so I looked for opportunities elsewhere.
When I decided at age eighteen to go into science, I visualized for myself a monkish type of
life and so thought that I would never marry. However, as time went on, I came to feel more keenly
a void from the absence of family life. In my last year at Case I married Carol Cornell, who has
been my loving and steadfast companion ever since. Two sons, Lowell and Murray, were born to
us and we adopted a daughter, Joyce, when she was one year old.
One of my metallurgy students was Jack Lubahn who, after his undergraduate study, remained
at Case investigating, with George Sachs, the flow and fracture of metals. Early in 1946 he
informed me that he was joining a new metallurgy group being formed at the General Electric
Research Laboratory, in Schenectady, N.Y., by a Captain J.H. Hollomon. He said that Hollomon
expected to hire a physical chemist and asked if I would be interested in the position. Shortly
afterward I met with Hollomon at Lubahn's home. Hollomon, then a trim young man aged 26, was a
most unusual person with quite an overpowering personality. He was brash, intense, completely
self-assured and overflowing with enthusiasm about prospects for the new group. He described the
fascinating, but poorly understood, responses of metals to mechanical and thermal treatments and
his plans to form an interdisciplinary team, with representation from metallurgy, applied mechanics,
chemistry and physics, to attack the problems posed by this behavior. He was certain that these
researches would lead to greatly improved ability to design and synthesize new materials that would
find important technological uses and expressed the view that equipment performance was
becoming more materials than design limited. He assured me that I would be provided with the
freedom and resources for the studies of nucleation and diffusion, which I hoped to make, and also
aroused my interest in the martensite transformation and solid state recrystallization. The position
was offered, at a salary about 15% less than I was receiving at Case, and I accepted it. My career at
Case had been somewhat disappointing, but I learned much from it that would prove useful in my
later research.
General Electric Research Laboratory
I joined G.E. in late 1946, becoming the fourth member of Hollomon's group. John Fisher,
who was trained as an applied mathematician, joined the group not long after I did. The ambience
of G.E.'s research laboratory, then guided by Guy Suits, was pervaded by an aura of excitement and
excellence and of collegiality that was quickly extended to newcomers. I was fortunate to have
joined it when most of its distinguished elder scientists were still active and to have had interactions
with some of them, including Irving Langmuir, Albert Hull and Willis Whitney. At this time Suits
spent most of his time at the laboratory and periodically reviewed the research of individual staff
members at luncheons, also attended by some of the senior scientists. At these meetings I found
Suits' comprehension of the wide variety of topics discussed remarkably acute and his suggestions
for my own research to be generally quite helpful. Unfortunately, Suits' direct interactions with his
research staff dwindled sharply as the laboratory grew and as the demands on his time by the G.E.
executive offices in New York increased.

Most of my associations at G.E. were within Hollomon's group. However, I had many
valuable interactions, besides those with the more senior scientists, with members of other groups,
especially with Bernard Vonnegut, Vincent Schaefer, Fraser Price and with Leroy Asker and several
others in his Semiconductor Studies section.
Hollomon was like no other manager. He was rarely neutral about anything and had very
strong likes and dislikes of people and ideas. These were expressed openly and vehemently and
often changed dramatically from time to time. Those closely associated with him usually were
welcomed to his inner sanctum or consigned to his outer doghouse. Most of us made, I think,
several circuits between the sanctum and the doghouse. Hollomon would advocate an idea or model
vociferously and stubbornly but, if confronted with contrary evidence of a convincing nature, would
quickly and completely reverse his position without the slightest show of embarrassment and then
uphold the contrary view with as much vigor as he did his former one. He adhered to no set rules.
One time he arranged, all for the same date, that he would go to Cleveland to talk about my research,
Fisher would go to Syracuse to talk about Hollomon's research, and I would go to Pittsfield to talk
about Fisher's research. At the time none of us thought these arrangements odd.
Hollomon's leadership was, indeed, dynamic and inspiring. He established a climate highly
conducive to interdisciplinary cooperation and to daring modes of thought and experimentation. In
this ambience we were inclined to choose the most outrageous over the more conventional
explanations for newly discovered phenomena. By adhering to this practice we would, on occasion,
look rather foolish, but we were also led to ideas and insights which proved to be very important.
Hollomon's group grew rapidly and by the early 1950's its achievements were widely
recognized. These achievements and those of several other groups--especially the Shockley group
at Bell Labs, Chicago's Institute for the Study of Metals under Cyril Smith's leadership, and Nevill
Mott's group at the University of Bristol--demonstrated the power of the interdisciplinary approach
to materials research and led to the recognition and labeling of a "Materials Science"
"superdiscipline".
In these and other laboratories there were, at the time, highly effective interactions between
physicists, metallurgists, chemists, and applied mechanicians. Metallurgists were remarkedly
receptive to the ideas of physical scientists because they felt, perhaps, that they had gone as far as it
seemed possible to go with the empirical approach to alloy design and development. In turn,
leading physicists were challenged by metallurgical problems and willingly listened to and learned
from metallurgists. Interdisciplinary exchanges occurred at various meetings, sponsored mainly by
the metallurgical and physical societies. A highly influential forum for such exchanges was
provided by the yearly weekend seminars sponsored by that most applications-oriented of
metallurgical organizations, The American Society for Metals. These seminars were attended by
several hundred metallurgists from industry and the universities. Among the participants were such
physicists as Clarence Zener, Conyers Herring, Percy Bridgman, and Harvey Brooks, and
metallurgists such as Cyril Smith, Larry Darken and Paul Beck, as well as a number of us from
G.E.
Our research attracted visits by many distinguished scientists from other laboratories and
arrangements were made, a rare practice in industrial labs at that time, for some to spend extended
periods with us. Among these was F.C. Frank, whose inspiration and collaborative efforts, in the
summers he spent with us, resulted in several major contributions. One distinguished foreign
scientist, after spending two weeks with us, remarked to Hollomon that never before had he seen
"such first class research done by such second class people". I prefer to think that most of the
people were first class in one respect or another and needed only a climate such as pervaded the
Hollomon group to realize their potential. This, of course, may have been what the visitor really
meant. It is, I believe, fair to say that when Hollomon, later in his career, had charge of large
organizations which he himself had not formed, that his leadership was not as effective as in our
G.E. group. He was brilliant as an innovator but less successful as a renovator.

Thus, I began at G.E. with the freedom and resources and in a climate which seemed close to
ideal for doing the research I most wanted to do. Also during my early years at G.E. I developed a
more mature perspective on the nature and practice of science from reading some of the
philosophical writings of Percy Bridgman and Philip Frank. They did not lead me to accept logical
positivism, but did impress on me the provisional and utilitarian nature of scientific theory. This
impression was, I believe very helpful to my thought.
Broadly, my General Electric research centered on the experimental characterization of, and
theories for, phase changes and atomic transport in condensed systems. The materials investigated,
with many collaborators, included glasses, ceramics, semiconductors, organic compounds, and
polymers, as well as metals. I began by examining the theory of nucleation and with experiments
on the short-circuiting effects of extended imperfections on metal diffusion and on the
undercooling of liquids. My examination of nucleation theory led to consideration of transient
nucleation and a numerical evaluation of the transient for a very simple set of conditions. John
Fisher and I then collaborated in a number of studies in which we adapted the theory of nucleation
kinetics to condensed systems and developed a theory, with Hollomon's collaboration, for
nucleation of the martensite transformation. In these studies Fisher exhibited his remarkable ability
for circumventing the complexities of problems and so exposing their essence.
The investigations of short-circuit effects in diffusion were made in collaboration with Robert
Hoffman. We found that grain boundaries, indeed, serve as highly effective short circuits for self-
diffusion in silver and that edge dislocations, under conditions where they should be little
dissociated, are equally effective. In extracting the short-circuit component we used a simple
analysis made by Fisher, which related the concentration profile of the diffusing species to the
crystal and short-circuit diffusivities.
My research on the undercooling of liquids was, at first, exploratory and somewhat desultory;
then it evolved into a major effort. When I arrived at G.E. the experiments of Schaefer and
Vonnegut, in Langmuir's group, on ice nucleation were underway. Also, Vonnegut was
investigating the solidification behavior of dispersions of oxide coated tin droplets. Keeping
informed on their experiments, I began investigating the undercooling of liquid gallium, which was
convenient for study since its thermodynamic crystallization point, Tm, was only a little above room
temperature. I found that 1 cm3 specimens sealed in evacuated tubes: readily undercooled several
tens of degrees, ΔT°, before nucleation onset; growth following nucleation was very rapid,
recalescence raised the temperature at continuing growth almost to Tm; ΔT° varied considerably
from sample to sample but, for a given sample, generally remained the same through several melting
↔ crystallization cycles; with increasing superheating of the liquid prior to cooling, ΔT° increased
to a point and then remained constant; at large undercooling, nucleation could be initiated by a slight
tap on the wall of the tube containing the specimen. These effects were all very interesting but,
upon examining the literature, I found that, beginning with the observations of Fahrenheit on the
freezing of water in the early eighteenth century, they had all been documented, for one liquid or
another, much earlier. So it seemed hardly surprising that gallium behaved as it did. Then, I believe
in the autumn of 1948, I was told by Hollomon that John Chipman had asked him to recommend a
speaker on "Solidification" for an M.I.T. Metallurgy Department colloquium. Hollomon promptly
recommended me as "a world expert" on solidification, and I found that he had already arranged for
me to speak at the colloquium on a date only two weeks later. I was quite disconcerted and set
about thinking what I could say, without seeming foolish, to an audience most of whom, I
presumed, would be better informed than I about my topic. At this point, the import of the droplet
studies of Vonnegut on tin and of Schaefer on water suddenly became clear to me. I then
concluded that any liquid should exhibit large undercooling at nucleation onset if dispersed into
isolated droplets with a number density much larger than that of heterophase impurity nucleants in
the original mass. To test this idea I formed a dispersion of mercury droplets by shaking mercury
in a test tube containing a solution of sodium oleate in alcohol. The droplets were quite large, their
diameters were probably in the 0.1mm range, but they undercooled 30 to 40° before there was
measurable crystallization, as indicated dilatometrically.
I then went confidently before the M.I.T. audience and explained the droplet and thermal
history effects and described my results on mercury. The presentation seemed to be well received
and almost as an afterthought I expressed the view that even those liquids which crystallize to close-
packed structures might undercool deeply if the heterophase nucleants could be isolated, e.g. by the
droplet technique. I suggested that the common experience that these liquids begin to crystallize at
almost negligible undercooling might reflect only that close-packed "dirt" was far more ubiquitous
than dirt with more exotic structures. My "afterthought" was greeted with polite skepticism by Bert
Warren and others in the audience. This skepticism was based on the then prevalent view that the
atomic short-range order in monatomic liquids was only trivially different from that in close-packed
crystals. Thus it was thought that with slight undercooling the liquid should collapse almost
continuously into the crystal or perhaps crystallize by a grain coarsening mechanism. Such a mode
of crystallization also seemed consistent with the common experience that crystals melt at only
slight superheating. At the time many theorists identified the temperature at which the crystal would
become unstable, due, e.g., to a vanishing shear modulus or increase of vacancy concentration to
some critical level, with that at thermodynamic melting. The implication seemed to be that the
melting of close-packed crystals was, thermodynamically, almost a second order transition.
Actually, as Clarence Zener had pointed out, the shear moduli of most close-packed crystals are
quite high at Tm and falling, with temperature, only slowly on a course which would extrapolate to
zero only at temperatures exceeding 1.5 Tm. Also Mendenhall and Ingersoll had reported in 1908
that droplets of certain metals which formed f.c.c. crystals deeply undercooled before crystallizing.
However, their observations were not followed up and their significance went unappreciated. They
came to my attention only after I had completed most of my undercooling studies.
Soon after the M.I.T. colloquium John Fisher asked Vonnegut and me to discuss our
solidification results at a meeting of the Science of Metals Club. This club had been founded by
Fisher and met one evening each month in an Italian-style Schenectady restaurant. It drew
scientists interested in metals from all laboratories in the Schenectady area. In my talk I again
expressed my prediction that liquids which form close-packed crystals would deeply undercool if
freed of heterophase nucleants. The response was again quite skeptical and, this time, not so polite.
David Harker dissented vehemently, and rashly promised to eat his hat if molten copper could be
undercooled more than a few degrees. This was a challenge I could hardly ignore.
To make the undercooling studies I enlisted the collaboration of Robert Cech who designed
and had constructed a microscope hot stage in which the metal particles could be heated and cooled
while they were observed microscopically. Crystallization was manifested by an abrupt surface
roughening and, for the higher melting metals, by a sudden brightening or "blick" which attended
recalescence. We demonstrated conclusively that molten droplets of a number of high purity
metals, including copper, silver, gold, nickel, palladium, cobalt, and iron, could be deeply
undercooled to temperatures at least 18% below their melting points before crystal nucleation
occurred. Also droplets could be held at 15% undercooling for indefinite periods without
crystallizing. David Harker graciously accepted the validity of our findings and appeared at the
next Metals Club meeting with a headpiece fashioned from Swiss cheese.
Microscopically the high crystal nucleation resistance revealed by our experiments showed that
crystallization hardly can result from mere collapse of the liquid structure, rather some essential
reconstruction of the liquid short-range atomic order (SRO) must occur. F.C. Frank proposed that
an icosahedral type of SRO, which by itself is non-crystallographic, might predominate in the
liquid. Its transformation to close-packed crystalline SRO would require reconstruction. In
support of this proposal Frank showed that, with an interatomic potential of the Lennard-Jones type,
the energy of a 13 atom cluster is lower in an icosahedral than in either close-packed crystalline
configuration.
In terms of classical nucleation theory our results indicated a large crystal-melt interfacial
tension. At that time I thought, as I expect David Harker did, that this tension would be primarily
energetic in origin. Probably Harker understood, as I then didn't, that it should be possible to form
a crystal-melt interface in a monatomic liquid at negligible energy cost, provided there is no density
deficit. However, the considerations of Frank, as well as those arising later from the dense random
packing models for amorphous structure, indicated that the density deficit could be eliminated only
by marked localization of positions on the liquid side of the interface. Such
localization, effectively a layering in liquid structure, would be attended by substantial negentropy.
Thus while the energy or negentropy can vanish separately they cannot do so simultaneously.
Later, at Harvard, Frans Spaepen and Robert Meyer showed that the upper limiting entropy
decrease associated with this interface localization would be more than adequate to account for the
magnitude of the crystal-melt tension indicated by the nucleation experiments.
In retrospect, I feel greatly indebted to David Harker for the stimulus which impelled me to
investigate the undercooling of liquids such as Cu and Ni. Also I benefited from his occasional
sharp criticisms of my ideas about certain other problems. He was an incisive thinker who was
quick to detect a flawed argument.
After the microscopic studies I investigated, by dilatometric and calorimetric methods, the
kinetics of nucleation in various Hg droplet dispersions. Those dispersions constituted by droplets,
3-5 microns in diameter, coated with Hg laurate or Hg benzoate films exhibited very large onset
undercoolings, about 0.35 Tm, when cooled at 1°/minute. The kinetic studies revealed that
isothermal nucleation in these dispersions was stochastic in both volume and time and so described
by a single valued frequency/volume. These were, I believe, the first experiments in which such
single valued frequencies of solidification were found and measured. The large undercooling at
nucleation onset, ΔT°, exhibited by Hg suggested that the other metals might show even larger ΔT°
than we found initially. However, I didn't find time to investigate this possibility.
The results of the undercooling experiments naturally suggested that liquid metals might form
glasses if quenched rapidly enough. At first I discounted this possibility since extrapolation of the
melt viscosities into the undercooled regime, on an Arrhenius course set at Tm, indicated that they
would approach solidity level values only at temperatures approaching 0°K. This was still my view
in the autumn of 1957 when I began a seven-month leave from G.E. spent mainly at the Cavendish
Laboratory of Cambridge University, where I shared an office with Morrel Cohen, then on leave
from the University of Chicago. In this period I visited several European laboratories, including
that of J.A. Prins at Delft. During my visit Prins pointed out--as I could have known had I read
certain of the papers cited in Kauzman's 1948 review--that some liquids composed of small
compact molecules and bound only by van der Waals forces, exhibit viscosities which, though small
at Tm, deviate positively from an Arrhenius course as temperature falls and rise sharply, but
continuously, to the solidity level at temperatures far above those indicated by the Arrhenius
extrapolation from Tm. These observations suggested that the liquid structure becomes locked by
repulsive forces at some point during its collapse with cooling.
Upon returning from Cambridge I discussed what I had learned from Prins with Morrel
Cohen. In these and later discussions we developed the ideas about the glass transition and
conditions for glass formation which were described in several papers published over the following
few years. A major feature of our development was the association of configurational locking of
the liquid during its compaction, by cooling or pressurizing, with a vanishing "free volume". We
supposed that the elementary step in liquid flow would require a local density attenuation of
amplitude greater than some critical value. Glass formation was associated with a transition in the
atomic movements from a predominantly translatory to oscillatory character with increasing
compaction. These considerations and the evidence that crystallization was reconstructive led us to
predict that all fluids, including metals and other monatomic ones, would form glasses, when
sufficiently compacted, by undercooling or pressure, if crystallization did not intervene.

Upon returning to G.E. I set out to experimentally test our prediction that all liquids should
form glasses if crystallization could be bypassed. I had noted the extremely deep eutectic in the
Au-Si system and decided that the eutectic alloy melt might be the best candidate for glass
formation since the undercooling required to reach the glass temperature from its low liquidus
should be minimal. I was preparing to test for glass formation by this alloy, using Cech's drop tube
method, when the report of Duwez and his students, that this alloy did form an amorphous solid
during rapid melt quenching, appeared in Nature. Morrel Cohen and I then pointed out in a paper,
also published by Nature, that our theory indicated that deep eutectic compositions should be
especially favorable for glass formation. Professor Duwez later noted that he was guided by our
deep eutectic criterion in his further search, following the Au-Si discovery, for alloy glass formers.
In retrospect, it is clear that the cooling rates realizable in the Cech drop tube would not have been
adequate to quench the Au-Si alloy to a glass. Its glass temperature is exceptionally low so that its
glass formation requires imposition of the very high thermal gradients achieved in the rapid
quenching techniques pioneered by Duwez and his students. However, the drop tube method
would have been sufficient to quench droplets of the Pd rich Pd-Si eutectic alloy to a glass, as was
demonstrated by the later experiments of my student, Alvin Drehman, at Harvard. This alloy proved
to have a glass temperature relatively much higher than that of the Au-Si eutectic.
Following our studies of the undercooling of liquid metals my colleagues and I applied the
dispersion technique to characterize the nucleation of various other transformations. These
included investigation, in collaboration with Robert Cormia, of the crystallization of dispersed
droplets of various n-alkane liquids, with chain lengths ranging from 16 to 32 carbon atoms. We
found that the undercooling, ΔT°, at measurable nucleation onset was surprisingly small, only of
order 0.04 Tm, compared with the metal values of 0.2 to 0.35 Tm. Our kinetic studies showed
conclusively that isothermal nucleation in each of the high purity liquids, n-hepta- and n-octa
decane, was stochastic in both volume and time. Further, the extremely high measured value of the
temperature coefficient of nucleation frequency/volume, a factor of seven to eight thousand per

degree, was in excellent agreement with the predictions of classical nucleation theory. The
abnormally low undercooling at nucleation onset may reflect considerable local alignment of the
chain segments in the melt so that reconstruction of the SRO necessary for crystallization is
substantially less than for other liquids. Later Fraser Price, Robert Cormia and I measured the
frequency of crystal nucleation in polyethylene by the droplet method. We found that ΔT° was
about 0.15 Tm which is 5 times larger than that typically exhibited by bulk specimens. The high
nucleation resistance of the polymer relative to that of the C16-C32 n-alkanes was attributed to
resistance to chain folding, which we supposed necessary for nucleation.
Having found that crystallization in liquids usually is initiated by heterophase nucleants,
present inadvertently, I became actively interested in the problem of heterogeneous nucleation in
solids. It was well known that nucleation of incoherent phases in solids often is initiated at
extended imperfections but that particles of coherent phases may form copiously within the crystal.
What was not known were the departures from equilibrium required for the onset of measurable
nucleation of new phases within the perfect crystal. J.H. van der Merwe and Harvey Brooks,
independently, had adapted the Read-Shockley theory for grain boundaries to the calculation of the
interfacial tension, σαβ, of incoherent interphase boundaries. Their calculations indicated that with
increasing crystallographic misfit, δ, σαβ should at first increase rapidly and then become
approximately level at δ of only a few percent. Putting these values of σαβ into the expression for
the classical nucleation rate Brooks showed that the homogeneous nucleation frequency of those
incoherent phases having the larger misfits should never reach measurable levels.
The austenite-martensite interface in iron alloys is thought to be incoherent. To assess the role
of heterogeneities in martensite nucleation Cech and I investigated the transformation behavior of
small, 50-100 µm diameter, single crystal spheres of Fe-Ni alloy formed during the fall of the
molten alloy droplets in a tubular furnace. We found that the martensite start temperature in some
of the spheres was at least 175° below that in the bulk alloy. From this behavior we concluded that
martensite formation in these alloys must be initiated at extended imperfections rather than by
homogeneous nucleation, in accord with Brooks' calculation.
The interfaces between face-centered cubic lead and tetragonal tin also are expected to be
incoherent. Thus it would seem that the diffusive motion of Sn in Pb(Sn) alloy solutions should be
virtually frozen before an undercooling sufficient to nucleate Sn homogeneously could be reached.
Borelius and his students had made careful resistometric studies of the precipitation of Sn from
these alloys. In interpreting their results they supposed that the nucleation was homogeneous,
starting from homophase concentration fluctuations, but ignored any contribution of interfacial
tension to the nucleation barrier. Also they supposed that precipitate particle growth occurred by
volume diffusion. Their suppositions seemed quite unreasonable to me. My calculations indicated
that the growth rate of their Sn particles must have been several orders of magnitude greater than the
maximum permitted by the accepted volume diffusion rates. Harry Treaftis and I then undertook
experiments on these alloys in which we thoroughly characterized the isothermal time dependence
of, and the effects of, thermal cycling on the precipitation rates.
Our experiments indicated that in any isothermal experiment all the operative nuclei were
present from time zero; there was no measurable homogeneous nucleation. Also the activation
energy for precipitate growth proved to be only 0.4 of that for volume diffusion of Sn in Pb. To
account for these results I supposed that precipitation actually was effected by sweeping short
circuits which deposited rods or plates behind them. This model was fully confirmed by our later
microscopic observations which showed that precipitation occurred by the deposition of colonies of
more or less parallel tin lamellae from moving grain boundaries, which served as short circuits for
the transport of Sn to the leading edges of the lamellae. To test the model further Warren DeSorbo
and I followed the precipitation calorimetrically as it occurred in a dispersion of small Pb(Sn) alloy
spheres, with diameters less than 40 microns. After thorough homogenization most of the alloys
were single crystals. We found that the total extent of precipitation was only 0.0025 to 0.10 of that
exhibited, under the same conditions, by bulk specimens, thus, confirming that extended
imperfections are essential for the nucleation of precipitate. I learned later that our observed Pb(Sn)
precipitate morphology was similar to that sometimes found in other alloys and that its evolution,
known as "discontinuous" precipitation, had been explained phenomenologically by Cyril Smith.
Our results thus confirmed Brooks' prediction that the resistance to homogeneous nucleation
of incoherent phases should be extremely high. They also demonstrated the power of the
dispersion technique in reducing or eliminating transformation from heterogeneous nuclei. Left
open was the question of the resistance to homogeneous nucleation of coherent phases.
The definitive theory for the initiation of coherent precipitation was developed by John Cahn
beginning with his treatment, in collaboration with John Hilliard, when both were members of our
G.E. group, of spinodal decomposition in strain free (δ = 0) systems. A central parameter of the
theory is the gradient energy, which determines the interfacial tension between the coherent and
parent phases and imposes a minimum on the wavelength of concentration fluctuations which can
be amplified in the spinodal regime. Following the appearance of the Cahn theory the kinetics and
morphology of precipitation beginning in the spinodal range came to be extensively investigated.
However, there were no measurements of the frequency of coherent nucleation in the temperature
ranges above the spinodal points. Shortly after I moved to Harvard, I.S. Servi, then at the Kennecott
Copper Research Laboratory, invited me to consult with him on his resistometric study of the
precipitation of cobalt from its dilute solution in copper; in this system δ is quite small and the
initial precipitate is coherent. Servi's measurements were highly accurate and we decided that they
could be adapted to determine the nucleation frequency of coherent cobalt. The essential
experimental requirements were: that the specimens be brought directly from the homogenization
to the precipitation temperature without intermediate quenching and reheating; and, of course, that
precipitation temperatures be chosen which were well above the spinodal limits. We found, for
initial cobalt concentrations ranging from 1 to 2.7 at.%, that undercoolings of 60 to 90°,
corresponding to supersaturation ratios about 2.0, were required for the onset of copious
nucleation. We deduced the isothermal nucleation frequencies and found that their temperature and
concentration dependences were in reasonable agreement with the predictions of classical nucleation
theory. Subsequent determinations by other investigators have been in fair accord with our results.
The results on liquid undercooling also raised the question: why do crystals generally melt at
slight superheating? In classical nucleation theory the nucleation resistance is due, primarily, to the
work of forming the crystal-melt interface. This same resistance should oppose the formation of
such interfaces within the perfect crystal.
The explanation proposed by Frenkel and others for the slight superheating actually observed
was that, at capillary equilibrium, the external surface of a crystal, at its melting point, would be
coated with a thin molten film. As heat is supplied this film migrates inward, but if the interfacial
rearrangement frequency ki is very high the interfacial superheating will be hardly detectable when
thermal gradients of the usual magnitude are imposed. Under these conditions little can be learned
of the ultimate superheatability of the crystal interior. However, there are materials, usually
covalently bonded, in which ki is so small that the interface temperature is virtually identical with the
ambient when only small thermal gradients are imposed. It was known that some of these materials,
e.g. albite, would withstand large superheating without internal melting. The kinetics and
morphology of melting of two materials of this type, quartz and P2O5 crystals, were investigated by
J.D. Mackenzie, Norman Ainslie, Robert Cormia, and myself. We measured the dependence of the
rate of melt-in from the external surface on superheating and found no evidence of interior melting
to the highest temperatures imposed, which for quartz corresponded to a superheating exceeding
350°.
Early in the 1950's Suits approved the initiation of a high pressure-high temperature project
which was to have as its main object the synthesis of diamond. Hollomon had a prominent part in
the deliberations leading to this decision. The history of the high pressure project, culminating in
the synthesis of diamond by the team of Bundy, Strong, Hall, Bovenkerk, and Wentorf, has been
thoroughly aired from opposing viewpoints, elsewhere. In my judgment credit for the synthesis
belongs to all the individuals on the team and to the G.E. Laboratory management for taking the
decision to proceed with the project and for providing the considerable resources which were crucial
for its success.
There was a long history of efforts to synthesize diamond in a metastable state at one
atmosphere pressure. Since my research had shown that up to moderate departures, at least, from
equilibrium crystal nucleation in liquids and in many solid state transformations is effected entirely
at heterogeneities I became interested in the possibility that diamond might grow in the metastable
regime providing the heterogeneities which presumably nucleated graphite could be separated from
them. I was further attracted to this possibility by Bridgman's remarks to Zener at the authors'
breakfast preceding an A.S.M. weekend seminar, that he took seriously evidence that diamonds
found in certain meteorites formed under metastable conditions. He mentioned a failed attempt at
M.I.T., motivated by Karl Compton, to grow diamond crystals by exposing them to hydrocarbon
vapor at an elevated temperature.
Sometime later I initiated, with the collaboration of Gerald Sears, William Rocco, and Richard
Oriani and with Hollomon's enthusiastic support, a project to grow diamonds metastably.
Essentially, our strategy was to avoid graphite nucleation by using diamond seed crystals so small
and perfect that they, hopefully, would not include graphite nucleating heterogeneities. Our
calculations indicated that, under these conditions, the crystals should grow as diamond when
exposed, at sufficiently elevated temperature, to gaseous or condensed solutions supersaturated with
carbon. However, our attempts were unsuccessful. Sears and Rocco noted that, under some
conditions, the surface morphology of crystals exposed to carbon vapor changed in a way
suggestive that there had been some growth, but the changes were small and the crystals actually
had lost a little weight in the process. When the high pressure synthesis was achieved we
discontinued our efforts. In retrospect, I think that we gave too little attention to the state of the
carbon in the nutrients and to possible growth inhibition by tightly bound contaminants which may
have been present on the surfaces of our seeds. In recent years other groups apparently have
succeeded in growing diamond in the metastable low pressure regime, but I am not familiar with the
detailed procedures used.
I joined G.E. at a time when the Laboratory and, indeed, the company as a whole were rather
loosely organized. The lines of authority often were indistinct and the leadership was quite
informal. The president of the company, Charles E. Wilson, was a self-made man, with little formal
education, who had advanced, from a lowly position, through the ranks. He was a huge man with a
striking presence and spoke with a booming voice in an evangelical and rather "corny" style. He
held extremely conservative economic and political views. However, he was a humane person,
passionately devoted to his company and very proud of its technological achievements. He would, I
think, rather have G.E. renowned for "lighting the world" than for 15% after tax profits. He was
proud of the Research Laboratory and was reported to have said that the Company, having benefited
so much from science, had a special obligation to support it.
While he was president, there were summer meetings of groups of selected professional and
managerial personnel at the company-owned "Association" island in Lake Ontario. The groups met
serially, each for three days, and were constituted by a mix--engineers, scientists, managers,
salesmen, etc.--of about 300 people drawn from all parts of the company. Each day there were
meetings at which the group was regaled with rosy accounts of performances in selected enterprises
and company components. Liberal amounts of time were allotted to recreation, organized by
assigning each attendee to one or another of four teams--"red", "green", "orange" or "blue"--which
then competed in various sporting events. Near the end of the meeting a winning team was
announced and duly recognized. The competition seemed to free pent-up juvenile instincts in most
attendees. Sumptuous meals were provided but hardly could be enjoyed because at every mealtime
the dining hall was turned into a virtual bedlam by shouted slogans and almost continuous
boisterous cheering and table pounding by members of one or another of the teams. At the end of
the first day there was held what was meant to be a solemn ceremony, under a huge and aged elm
tree, in which the first-time attendees were inducted into the G.E. professional elite. The climax of
the meeting came on the afternoon of the third day with the arrival of Mr. Wilson. His arrival, by
seaplane, was announced by the firing of blanks from a cannon and, as he was launched ashore, the
entire group lined up in the dock area while the band played martial airs. Upon landing he shook
hands with everyone present and when we reassembled in the meeting hall he delivered an
impassioned oration which profoundly affected most of his audience.
For me the affair was rather distasteful but, in some ways, very funny. Certainly it was highly
vulnerable to satire and ridicule. Indeed, Kurt Vonnegut, who was present at one of the sessions I
attended, apparently found it a rich source of material for his first novel. At that time Vonnegut was
the Research Laboratory's publicist. He wrote two news releases on the research of my colleagues
and myself. I thought that he, as well as all other science writers I've known, overdramatized their
news stories. However, he understood much better than the great majority of them what he was
writing about. In retrospect, I suppose that the Island meetings did serve to knit the diverse and
scattered G.E. professional community together and boost the esprit de corps of most of its
members.
In the early 1950's Mr. Wilson was succeeded, as company chief executive, by Ralph Cordiner,
who had been trained as an accountant. Cordiner's approach to management was, in sharp contrast
with Mr. Wilson's, cold and analytical and he seemed utterly obsessed with the "bottom line".
Apparently he thought the loose organization of the company somewhat scandalous and set about
imposing a rigid managerial hierarchy with the responsibilities of all managers, from the lowest to
the highest levels, precisely defined. Actually, before he assumed full responsibility there was
already a trend, motivated by the rapid company growth, toward better defined organization.
The Research Laboratory also grew rapidly in the immediate postwar period and it was
reorganized in 1950. William Ruder, a gentle low-key person, was head of the Metallurgy
Department at the time of this reorganization. However, the reorganization of Metallurgy was
designed primarily by Hollomon, who became department head at Ruder's retirement. The
department was divided into four sections. Hollomon asked John Fisher to manage the Physical
Metallurgy Section and me to manage the so-called Chemical Metallurgy Section. This was a type
of responsibility that I never asked for nor expected to have, but I accepted it, as it was understood
that I would continue to participate directly in research. The broad objective of the department was
to develop new materials and processes, hopefully technologically important, through improving
the understanding of: structure and its evolution and control, properties, and materials-property
relations. The Chemical Metallurgy Section was to concentrate mainly on the characterization and
understanding of structure, from the atomic level up, and its evolution. Its steady population was 20-
25 staff scientists and about 10 technicians. The scientists, even those who were new Ph.D.'s, were
free to choose, after due consideration of the advice of others, their problems and associations.
Thus my responsibility was not to direct the research of 20 scientists, rather it was to: stimulate
interactions between them, represent them before higher management, allocate resources, and bring
important new research opportunities to their attention. Such opportunities were approached
through voluntary decisions of staff members and by hiring new personnel who were able and
interested in the new area.
At one time or another, 35 scientists and engineers were members of the Section during the
period (1950-1958) that I headed it. About two-thirds of them began their post-university careers
in the Section. Most achieved high professional distinction. Among the major lines of
investigation initiated and carried forward by the Section were: the development of the theory of
spinodal decomposition by J.W. Cahn and J.E. Hilliard; studies of the formation and mechanical
properties of metal whiskers by S.S. Brenner and G.W. Sears; J.S. Kasper's determinations of the
complex crystal structures of certain intermetallic phases and his and F.C. Frank's prescription of
the rules for the structures now known as Frank-Kasper phases; other structural studies by W.L.
Roth, B.W. Roberts, B.F. Decker, L. Guttman, M.B. Webb, E.H. Jacobsen, and C.W. Tucker;
studies of physi- and chemi-sorption on clean and structurally well characterized surfaces by G.
Ehrlich, T.W. Hickmott and E.G. Brock; D.A. Vermilyea's research on anodic oxidation and
electrocrystallization; research on the kinetics and morphology of crystal growth by G.W. Sears,
R.L. Fullman, J.W. Cahn, W.B. Hillig, J.B. Newkirk, and J.L. Walker; characterization and
interpretation of thermal properties by W. DeSorbo and R.A. Oriani; atomic transport by R.E.
Hoffman, E.W. Hart, and R.H. Doremus; kinetics and morphology of solid state transformations
by J.W. Cahn, R.E. Cech, J.E. May, A.H. Geisler, H. Rosenbaum, and J.B. Newkirk. I am proud
of the achievements of these scientists as members of the Section and in their later careers. The
scientific opportunities were, on the whole, well chosen and the overall program was, I think, quite
well balanced. In retrospect, I think that one serious lack was a strong effort on inorganic materials
synthesis.
As section head I sometimes was called upon to help resolve various personal and
interpersonal problems of staff members. These experiences, I'm sure, led me to a much better
understanding of the range and variety of human personality. A yearly ritual, in which section
heads were heavily involved, was the rating of staff members for the purpose of salary action. To
begin the process each section manager ranked their staff members from first to last on the basis of
their perceived actual and potential achievements. A young scientist who might have accomplished
little might be ranked ahead of a productive older scientist if he was thought to show great promise.
The next step in the procedure was interleaving the section rankings within a department. Then the
department rankings were interleaved, with much argument, to reach a laboratory-wide ranking from
one through the number, of order 350 , of the laboratory staff. There was even considerable
controversy over the precise order in the middle of the ranking. After the final order was agreed to
salary changes were approved such that the salary of each individual would be brought into near
conformity with the salary-age curve corresponding to the decile which included their rank. The
dispersion of these curves, especially at the middle deciles, was far from sufficient to justify the
effort spent in fine tuning the order. After the salary changes were decided the section head met
with each section member to inform them of the salary action and assess their performance. In
these meetings there was the temptation, resisted by the more experienced and perceptive ones, for
managers to assume an air of technical omnipotence and moral superiority. The evaluations were,
of course, quite subjective and it is my experience that moral excellence has rarely been a prime
criterion in the selection of executives, department heads, managers, etc. Perhaps the most
surprising reaction I encountered in an evaluation meeting was that of a young staff member who,
when informed that his salary was to be substantially raised, berated me severely and sincerely
because he thought the increase was much more than his performance deserved. He told me that, as
a company stockholder, he feared that such "irresponsible" salary action would jeopardize his
investment. I was pleased to learn that there existed someone with such depth of conviction.
The G.E. salary review process seemed overly elaborate and complex, but it apparently was
typical of those of many other industrial laboratories at the time and even of some now. It had the
merit that every staff member was at least noticed and his efforts reviewed at least once each year.
While the size of the sections remained about constant the administrative responsibilities of
section heads were gradually increased to the extent that they began to interfere seriously with my
direct research activity. I preferred to concentrate on science rather than administration and
relinquished my managerial position after my return from Cambridge in 1958.
In 1954 I accepted Frederick Seitz's invitation to join him in editing a set of volumes on Solid
State Physics. Seitz's treatise on "The Modern Theory of Solids", published in 1940, had a
profound influence on the definition and development of the discipline of solid state physics. Due
to the tremendous growth of the field following World War II he felt a pressing need to revise and
expand the treatise, but the task seemed too great for one author and the essential subject matter too
voluminous for containment in a single volume. The plan developed was to publish volumes
serially each containing several comprehensive, in depth treatments of basic topics by experts in the
fields. The coverage was to be very wide, ranging over the entire fields now labeled "Condensed
Matter Physics" and "Materials Science". I was to represent the materials science end of the
spectrum. Dr. Kurt Jacoby, a scholarly man who was then vice president of Academic Press,
enthusiastically supported the project and arranged to have the books published by his company.
The volumes, now about 40 in number, became the Solid State Physics series. Henry Ehrenreich
joined us as a co-editor in 1967. Seitz had a remarkable capacity to inform himself of and to digest
significant developments in all areas covered in the series. I know of no one more scrupulous than
he was in answering his correspondence, apparently all of it, promptly and in detail. He seemed
able to read and write anytime and anywhere, quite oblivious to surrounding distractions.
During most of my G.E. career I was also an adjunct professor of metallurgy at Rensselaer
Polytechnic Institute (R.P.I.). This activity was a part of a cooperative program between G.E. and
R.P.I. whereby young G.E. staff members could do research at G.E., supervised by an adjunct
professor, which could constitute an R.P.I. Ph.D. thesis. The G.E. students had to fulfill the R.P.I.
course and examination requirements for the degree. In this program I guided the Ph.D. research
of David Vermilyea, Sidney Brenner, Herman Rosenbaum, and Norman Ainslie. They all worked
on problems which were of Ph.D. caliber and fitted well with the G.E. research program. As an
adjunct professor I also gave, on occasion, a graduate course in kinetics which met after hours at the
laboratory and was taken by R.P.I., as well as by G.E., based students.
By the late 1950's it appeared that G.E.'s management had become somewhat disenchanted
with their liberal policy, especially as it had been exercised in the Hollomon group, of permitting
scientists more or less free rein in their choice of problems. Mr. Charles Wilson's sentiments
notwithstanding, any industrial organization expects substantial visible returns from support of its
own basic research. It is a truism that, for its part, a research laboratory must keep a close check on
the pulse of the company's technology to insure that the important needs of that technology are
clearly appreciated by the laboratory scientists. Also, there is greater likelihood that research results
will be utilized when a continuum of activity, from basic research through manufacturing operations,
is maintained. Such a continuum assures overlapping interests and effective communication all
along the line. Insuring that it exists is the joint responsibility of laboratory and company
management. For its part, company management must promote a climate in the manufacturing
laboratories which will attract and hold outstanding engineers and applied scientists. When I joined
G.E. such a climate seemed to exist in most of these laboratories as shown by the ability of their
members to understand and critically evaluate the Research Laboratory's activities. It is my
judgment, with which many may disagree, that later G.E. management policies led to major shifts of
emphases in the applied laboratories toward short-term goals and "fire fighting". As a result the
central laboratory had to take much greater responsibility for "technology transfer" and it
redistributed its efforts accordingly. Another reason that industries often become disenchanted with
their basic research stems from the view, which seems prevalent throughout industry, as it was at
G.E., that the time between a scientific advance and its technological application, if any, should take
no more than 5-6 years. Actually, there are numerous examples in the materials field, G.E.'s
Lucalox development is one, where this time has taken more than 15 years. In my opinion, the
difficulties that industries have had in benefiting from their research are due in part to their
preoccupation with short-term goals and in part to the often too cavalier attitudes of their
laboratories toward the company's technological needs.
Anyway, because of the seeming shift in the laboratory's emphasis toward more programmatic
and applied research and since I was approaching mid-career I began in 1960 to consider more
seriously the academic positions which were offered to me from time to time. Other reasons for my
changing attitude were: my leave at the Cavendish Laboratory reminded me how pleasant and
stimulating a university environment can be; the retirement policy of G.E., as well as in other
companies, was quite rigid, requiring retirement at age 65 or even before--as one of my colleagues
put it, "one can't grow old gracefully in this setting"; perhaps the activity at G.E. which I found most
pleasant was in helping to train new staff members for research. About 15 of these--including the 4
whose Ph.D. research I guided in the R.P.I.-G.E. program--began their professional careers in
collaborative research with me. Naturally I was attracted by the prospect of devoting full time to
this activity; finally, my experimental designs were always quite simple and fell far short of fully
utilizing the magnificent facilities for sophisticated experimentation provided by G.E.

I soon received quite a number of highly attractive academic offers. After visiting several
universities and spending the 1962 spring semester there I was most attracted to the offer of a
professorship by Harvard University. The reasons that Harvard most appealed to me were: it was
the only university I visited where there were no organizational barriers between solid state physics,
metallurgy, and applied mechanics; Harvey Brooks, Bruce Chalmers, Nico Bloembergen and
Bernard Budiansky were all in a single division. This arrangement, I thought, should be especially
conducive to the interdisciplinary associations to which I was accustomed at G.E.; the Harvard
graduate students, while relatively few in number, were of the highest quality; Brooks, who was the
Division head, made it clear that I would not be expected to form a large research operation, it was
hoped, in accord with my own objectives, that I would become an educator rather than an "empire
builder". On the negative side I sensed what one of my most distinguished academic friends
warned me of: that Harvard's sociology, with its heavy emphasis on the more formal and theoretical
side of science, and learning in general, would not be conducive to understanding of and sympathy
for the discipline of materials science. However, I noted that Bruce Chalmers was successful and
happy at Harvard and I was confident that Harvey Brooks fully understood and appreciated the
importance of materials science in both the educational and scientific frameworks. My wife and our
children, who were in their early teens, were all happy in Schenectady and were no more than
lukewarm towards the prospect of moving to the Boston area. Because of their feelings and my
long and happy association with the Research Laboratory I wavered for some time over the Harvard
offer. Finally I did accept it after two of my best friends, John Fisher and Fraser Price, on whom I
could always rely for frank advice, convinced me that it was in the best interests of my family and
myself to do so.
In retrospect, my G.E. experience was a very exciting one of professional growth and
maturation and it provided invaluable perspective and preparation for my educational and research
efforts at Harvard.
Harvard University
In the autumn of 1962 I began my Harvard career as Gordon McKay Professor of Applied
Physics at a salary about 40% below my last one at G.E. However, over the years at G.E. I had
been generously compensated and saved enough so that my family lived comfortably on the lesser
income.
Every year at Harvard, excepting one when I was on sabbatical at Stanford, I taught a graduate
course on "Kinetics of Condensed Phase Processes". It was presented in four parts: (1) basic
reaction rate theory, (2) atom movements, (3) theory of capillarity, and (4) kinetics of phase
changes. During the 23 years it was taught, more than 200 persons from Harvard and M.I.T. either
took it for credit or faithfully audited it. They were graduate students, postdoctoral fellows, and
young faculty members drawn from the fields of physics, metallurgy, chemistry, geology, ceramics
and applied mechanics. Various of my students urged me to write a book based on my lectures, but
I never found the time and energy to do so.
As part of the course I gave several laboratory demonstrations. In one, showing the
superheatability of water, the procedure, adapted from Wakeshima and Takata [Journal of the
Physical Society of Japan, Vol. 13, p. 1398 (1958)], was to inject water droplets (a few tenths mm in
diameter) into the lower end of a column of liquid along which a thermal gradient was imposed,
such that the temperature ranged from 30° to 300°C between the foot and top of the column. The
essential properties of the column liquid were: virtual immiscibility with water; density greater than
that of water; and normal boiling point in excess of 300°C. As the water droplets rose in the
column they became superheated upon passing the 100° level. When cavitation, from the first
vapor nucleus, occurred, the droplet exploded in quite a conspicuous manner. The superheating
was inferred by noting the level at which the explosion occurred. In some demonstrations droplets
reached the 225°C level before exploding. One year Robert Apfel, a student specializing in
acoustics and fluid mechanics, was fascinated by the demonstration and decided to develop it into a
Ph.D. thesis study, which F.V. ("Ted") Hunt, the acoustics professor, and I agreed to co-direct.
Apfel ingeniously modified the procedure so that water droplets could be "trapped" acoustically,
and then observed, at various levels of the column. In this way he characterized the temperature and
pressure dependence of the cavitation and demonstrated that water droplets could be heated to
279°C, corresponding to 179° superheating, before cavitation onset.
The other formal courses I taught at Harvard were taken, mainly, by undergraduates and were
in the areas of thermal and introductory solid state physics. I developed thermodynamics from its
experiential basis and strongly emphasized problems whose solutions demanded deep physical
insight. My problems and mode of instruction in thermodynamics were not, I'm afraid, overly
popular with the upperclass students; most were applied mathematicians who would have been more
comfortable with the more formal approach to the subject. For a number of years I taught one
section of a course in applied physics in which students learned physics by carrying through major
laboratory projects and solving problems. Each student was required to complete four projects
during the year, distributed over the areas: mechanics, electricity and magnetism, optics, and thermal
physics. I was responsible for the thermal physics section in which students carried out projects
such as: measurement of the Joule-Thomson coefficient; determination of the equation of state of
rubber bands; the tensile strength of liquids; construction and characterization of a fuel cell; study
of monolayer behavior using a Langmuir trough. Students, 8-14 in each section, mostly set up their
own experiments and were called upon to acquire an in-depth understanding of the physics
underlying their projects. I was pleasantly surprised to learn how well first year students grasped,
e.g., the thermodynamics of the Joule-Thomson effect or of rubber elasticity. The course was quite
popular, attracting very good students who welcomed a science course in which the experiments
were not "canned". However, it was highly consumptive of faculty time and for that reason was
finally discontinued.
Upon arriving at Harvard I started research on the formation and properties of amorphous
solids of all types and on molecular transport and structural changes in polymeric and other organic
systems. Later, my students and I also did research on: "fast" diffusion in metals and its
ramifications; crystal growth and phase separation in covalent systems; morphology and kinetics of
precipitation from solid solutions; and crystal nucleation in liquids and glasses. During my
Harvard career I have guided the doctoral research of 40 excellent students. Also 10 postdoctoral
fellows, who had not been my Ph.D. students, have been associated with my research. Their
disciplinary backgrounds, students and postdoctorals, were quite varied: about 50% physics or
applied physics; 25% materials science; 10% chemistry; and the rest electrical or mechanical
engineering.
Much of our research was directed at understanding the formation and nature of metallic
glasses. That the melt ↔ glass transition may occur in a metal was first demonstrated by Ho-Sou
Chen and myself. We first looked for the thermal manifestation of the transition by the alloy
Au4Si, which Duwez and students had melt-quenched to an amorphous solid. However, upon
reheating, the amorphous alloy crystallized without exhibiting the transition. We then thought that
we might retard the crystallization, which was to the metastable γ phase, by replacing a substantial
part of the silicon with germanium. Indeed, the ternary alloy so formed did exhibit the transition,
both thermally and rheologically, and it could be cycled several times between its undercooled melt
and glass states without crystallization which, however, did occur at a temperature, Tkc, only a few
degrees above the glass temperature, Tg. That the rheological transition occurred in the range of the
thermal transition was shown by partial spheroidization of alloy films at T > Tg and by measured
shear viscosities, 1010 to 1013 poise, of the fully relaxed alloy, well within the fluid range. The
changing isothermal viscosity as the alloy approached its relaxed state also was measured. Later
Chen and I found that the binary alloy Pd82Si18, as well as certain ternary alloys, formed by some
replacement of Pd or Si in the binary, also exhibited the glass transition, both thermally and
viscometrically. In these alloys the displacements of Tkc above Tg were generally greater than in the
Au-Si-Ge alloy. The displacement, Tkc-Tg, is, of course, one index of the resistance of the alloy
glass to crystallization, the larger it is the less should be the cooling rate needed for quenching the
alloy melt to a glass. Chen and I found that Tkc-Tg could be increased considerably, by selected,
but not indiscriminate, substitution of other elements for part of the Pd or Si. Later, at Allied
Chemical, Chen in collaboration with Donald Polk, also one of my former students, designed
selective alloying procedures whereby the glass formability of transition metal based alloys was
increased sufficiently for them to be processed cheaply into large masses of glassy metal strip.
Our later research at Harvard indicated that, in accord with the theory for crystallization
kinetics, Tkc-Tg increases with the glass temperature scaled by the liquidus temperature, Tl, i.e., Trg
= Tg/Tl. Impurity additions may increase Trg either by lowering Tl or raising Tg. It appears that
the liquid lowering effect is, as Cohen and I proposed, the predominant one, but the impurity effect
on Tg is, in some alloys, quite appreciable. Recently H.W. Kui, A.L. Greer, A.J. Drehman and I
found that, when heterophase nucleants were removed, the alloy Ni40Pd40P20, which has one of the
highest known Trg (~0.67) for metals, could be melt-quenched to a glass in virtual bulk form
(thickness ~1 cm) at rates of only 1 deg/sec. Then Kui and I showed, again taking care to eliminate
nucleants, that the alloy could be reheated from its glass to its liquidus temperature at rates as low as
2.5 deg/sec without crystallizing.
Chen and I also obtained the first evidence, by transmission electron microscopy, that upon
reheating certain metallic glasses, they were ternary Pd-based alloys, phase separate before
crystallizing. Later, my students C-P.P. Chou and M.A. Marcus characterized the occurrence and
course of the phase separation in certain of these alloys more quantitatively by small angle x-ray
scattering studies. Thus the thermal, viscometric, relaxation, and phase separation behaviors of
metallic glasses were all shown to be remarkably parallel to those of the common nonmetallic
glasses.
Another major line of our Harvard research was the characterization and modelling of the
structure of metallic glass. My students, Brian Bagley, Charles H. Bennett, Slade Cargill, Donald
Polk, and Frans Spaepen, all made important contributions to this research. In the late 1950s J.D.
Bernal showed that the structure (DRP) formed by the densest random packing of hard spheres,
constrained by curved surfaces only, was amorphous and apparently stable internally. Detailed
examination of the local configurations showed that most were tetrahedral and only a very small
fraction could be described as half octahedral. Bernal and others thought that the structure
modelled that of monatomic liquids, but in 1964 Morrel Cohen and I noted that the structure
apparently was "locked" like that of a glass, and proposed it as prototypical of a monatomic glass
structure. Bennett formed a DRP structure by a computer simulated condensation of hard spheres
in which the spheres were placed at surface binding sites, selected in a various ways, with no
subsequent reconstruction allowed. The Bennett structures were somewhat less dense than
Bernal's, but tetrahedral configurations were predominant while half-octahedral ones were scarce.
Bagley and Cargill carefully characterized the diffraction of x-rays by electrodeposited Ni-rich
amorphous Ni-P alloys. Cargill then showed that the radial distribution function (RDF) of Ni
atoms deduced from these results and the RDF's of other glassy metals then available, were in good
agreement with the DRP structures of Bennett and with that constructed by Finney following
Bernal's prescription. Later Polk suggested that the metalloid atoms were sited in the larger holes
of a DRP structure formed by the metal atoms. This second "Polk" model is now thought to be
oversimplified, but it implied a highly developed short range order such that each metalloid atom is
virtually surrounded by metal atoms. Such compositional ordering in amorphous Au-Si alloys had
been indicated by Chen's thermal measurements and its existence in other metal-metalloid alloys
was shown by measured partial distribution functions.
Bagley, one of my earlier students, and I thought that properly selected amorphous materials
might exhibit ferromagnetism. We looked first for a liquid ferromagnet. It appeared that certain
molten cobalt-based alloys might exhibit ferromagnetism if sufficiently undercooled. Bagley
prepared some of these alloys, but found no convincing evidence, using the rather crude probes at
our disposal, that any of the molten phases were ferromagnetic. Bagley then turned his attention to
the properties of amorphous films formed by condensation techniques and he and I were among the
first to demonstrate the occurrence of ferromagnetism in amorphous solids; it was exhibited by a
Co-P alloy formed by flash evaporation and condensation. However, apart from investigations by
J. DeLau, who spent a leave from the Philips-Eindhoven Laboratory with us, and Dawning Pan, we
had little subsequent effort on ferromagnetism. Other electronic property studies were made by my
students, C.H. Lin and M.A. Marcus, who characterized, using sensitive differential techniques, the
effects of structural relaxation on the electrical resistivity of some Pd-based alloy glasses.
My students and I became interested in the low temperature mechanical properties of alloy
glasses after being informed by Pol Duwez that some Pd-based glasses exhibited remarkable room
temperature ductility. This subject also interested Michael Ashby, then on our faculty. To account
for the highly localized mode of the low temperature plastic deformation, Polk and I proposed a
model in which compositional disordering produces flow softening. Later, Spaepen and I
suggested a complementary explanation for flow softening based on the free volume model. Ashby
and his students measured the elastic moduli of amorphous Co-P alloys and confirmed that the
shear modulus was low compared to that expected for crystalline materials. Denis Weaire, then a
postdoctoral fellow at Harvard, in collaboration with Ashby and his students, accounted for the
relatively low modulus, using the DRP model for glass structure. Ashby and his students also
found that the creep behavior of amorphous Co-P alloy, at temperatures well below Tg, exhibited
substantial relaxation effects. Later, Alan Taub and Spaepen characterized the creep of Pd-based
alloy glasses, in the temperature regime just below Tg, precisely and thoroughly. They found that,
depending on the extent of structural relaxation, the isothermal shear viscosity varied by at least four
orders of magnitude and they were the first to measure the temperature dependence of the viscosity
of the glass in an isoconfigurational state.
The relation between atomic diffusivity and viscosity of glasses is a problem that always
intrigued me, but it was very difficult to determine experimentally because of the extremely low
values of glass state diffusivities. One of my students, Martin Rosenblum, decided to try to
measure these diffusivities by monitoring the decay in amplitude of concentration differences in

compositionally modulated films--a technique used by Hilliard and his students at Northwestern to
measure very small interdiffusivities in crystalline materials. Preparing compositionally modulated
films, a sputter deposition technique was used, was experimentally very difficult, but Rosenblum
persevered and eventually produced good compositionally modulated films with wavelengths ~25Å.
With Spaepen and myself, he measured metal atom interdiffusivities, D, as low as 10-22cm2sec-1
corresponding, as one wit put it, to an average atomic jump rate of "one per semester". We found
that the isothermal D was sharply decreased by structural relaxation, but were able to determine its
temperature dependence, which matched that of the viscosity, for the glass in an isoconfigurational
state. Later A.L. Greer, who joined us as an S.R.C. postdoctoral fellow and later became a faculty
member, and his student, R.C. Cammarata, improved the deposition techniques and then determined
D for Pd- and Fe-based glasses over a wavelength range from 2 to 70Å. Again, they found that the
isothermal D depended strongly on the relaxation state.
Before beginning his research on metallic glass, Ho-Sou Chen had prepared specimens of
amorphous germanium (a-Ge) by electro- and vapor-deposition and was investigating their
structure and thermal properties. He and I made the first determination of the heat of crystallization
of a-Ge; our result has now been confirmed by a number of other investigators. Much later Chen
and Bagley, when at Bell Labs, and Spaepen and I independently used this data to show that there
must be a metastable equilibrium between the amorphous and liquid metal phases of Ge, as well as
of Si, at a point about 200-250° below the equilibrium melting point of the crystal. Our prediction
has been confirmed, at least qualitatively, by recent studies of the melting and solidification behavior
of amorphous overlays on the crystal following short high-energy laser pulses.
Following a custom started by Bruce Chalmers, our group met for coffee at 10:30 each
morning and 3:30 each afternoon. At these meetings my students and I often discussed structural
models for a-Ge. One day John DeNeufville, I believe, brought in some interconnectible tetrahedral
units. He, Polk, Meyer, and some of the others explored what types of four-connected structures
could be formed from these units and it was quickly realized that a fully-connected "random"
network was one such structure. About that time Simon Moss and his student, Joseph Graczyk
had, at MIT, carefully determined, by scanning electron diffraction, the RDF of amorphous silicon
films vapor deposited in high vacuum. Moss described these results to us and Polk compared them
with the RDF of a random network model he had hand-built from the tetrahedral units. The match
of the model to the experiment, including the average distortions of the tetrahedral angles from their
ideal value, was excellent and so originated the first "Polk" model. About this time interest in
amorphous semiconductors, partly stirred by S.R. Ovshinsky's claims that they had exceptional
electronic properties, became widespread and William Paul and his students at Harvard began
experimental studies of their structure and other properties. Also, Henry Ehrenreich became
interested in the theory of amorphous semiconductors and other disordered systems and, in this
period, our three groups--Paul's, Ehrenreich's, and mine--often held joint seminars and other
meetings to discuss amorphous solids. Earlier I had useful interactions with Arthur Bienenstock,
then on our faculty, on the structure of nonmetallic glasses. He and my student, Brian Bagley,
made an important analysis of the upper limiting intensity of small angle x-ray scattering which
could arise from microcrystalline solids.
The Ph.D. studies of my students on polymeric and other organic systems included: the
liquid-glass transition of a van der Waals bonded material, o-terphenyl, by Richard Greet, my first
Harvard research student; the kinetics and morphology of crystallization of linear polymers by
Joanne Burns; point defects and diffusion in molecular crystals by Ray Baughman; viscosity and
kinetics and morphology of crystallization of amorphous selenium by postdoctoral fellows Kee S.
Kim and Richard B. Stephens, and student Geraldine Gross; and effects of magnetic and electric
fields on the structure of liquid crystals by Robert B. Meyer. I had interested Meyer in the problem
of the kinetic resistance to crystallization of liquid crystal phases. We thought that this resistance
might be significantly lower than that of isotropic liquids. Actually, Meyer found that, in its nematic
state, p-azoxyanisole sustained a rather large undercooling, about 70-80°, before the onset of
measurable crystal nucleation. Following my suggestion, he read F.C. Frank's paper, in the
Faraday Society Discussions, on the structure of liquid crystals. Immediately, he comprehended
the paper fully, and, inspired by it, began his investigations, carried through with minimal guidance,
of the magnetic and electric field effects on liquid crystal structure. Perhaps 12-15 of my other
research students transformed, as Meyer did, their original research project into quite a different
one, usually better or anyway more suited to the student's interests and abilities. Fortunately, the
funding agencies, usually the Office of Naval Research or the National Science Foundation, were
quite tolerant of these changes.
Meyer's results aroused the interests of other Harvard faculty in liquid crystals and, I believe it
is fair to say that, largely due to his stimulus, there developed an interdisciplinary program of liquid
crystal research in which faculty from chemistry, physics, and the applied sciences participated.
Meyer had a central role in this program, first as a postdoctoral fellow and later as a faculty
member. Other faculty who were centrally involved included Peter Pershan from Applied Physics
and David Dolphin from Chemistry. The program was considered a model of the type most worthy
of support by the NSF-sponsored Materials Research Laboratories.
While I was at G.E., F.C. Frank and I developed our "dissociative" mechanism for diffusion
with which we interpreted the transport behavior of copper in germanium. In this mechanism, the
impurity is primarily in a substitutional state, but after dissociating from its reciprocal vacancy, it
migrates interstitially until trapped by another vacancy. This mechanism is thought to be
responsible for the "fast" diffusion of many noble and transition metal impurities in four-
coordinated semiconductors. It was long known that gold diffused with abnormal rapidity in lead
and the studies of Ascoli and Kidson showed that the abnormality could not be due to short
circuiting by grain boundaries or dislocations. Many thought that the transport occurred by the
"tumbling" of tightly bound gold-vacancy (Au-v) pairs, formed by association of a vacancy with a
substitutional Au atom, but I thought that, rather, it might be effected by the dissociative mechanism.
I interested my student, Thomas R. Anthony, in the problem, and he pointed out that the tumbling
motion should produce a large and readily measurable enhancement of the lead self-diffusivity,
DPb. The critical experiment was to measure how DPb was affected by Au. While preparing for
the experiment, we discovered that Seith and Keil had already, in the late 1930s, carried it out with a
null result and had concluded, therefore, that Au must migrate in Pb mainly by an interstitial
mechanism. Then, Anthony, Brian F. Dyson, a postdoctoral fellow, and I, confirmed Seith and
Keil's results on the effect of added Au on DPb and in addition established that: Cu and Ag are fast
diffusers in Pb; and that the increase in DPb with added Cu was orders of magnitude less than
demanded by a Cu-v tumbling mechanism.
Later another of my students, J.W. Miller, showed that DPb actually is increased measurably
by added Au, but by an amount very much less than it would be if the tumbling mechanism
predominated. Thus, their transport behavior indicated that the noble metals must, contrary to
Häggs' empirical rules, dissolve, in part at least, interstitially in Pb. We proposed that such
interstitial solution behavior might be expected provided: the host ion cores were separated widely
enough that they would not be overlapped by interstitially sited solute; and the valence of the solute
atom was relatively small. These criteria suggested that fast diffusion in metals by interstitial
mechanisms should be more widespread than had been thought. Our predictions were confirmed
by the experiments of Anthony, Dyson, and myself showing that tin, indium and thallium were
hosts for fast diffusion and by experiments in several other laboratories which indicated the
occurrence of fast diffusion in many other systems fulfilling the specified conditions. Other
Harvard studies of interstitial solution behavior and its ramifications were made by my students:
Wesley Akutagawa, Ann Campbell, Barney M. Cohen, J.W. Miller, William K. Warburton, and
Ali-Reza Yavari, and by my postdoctoral fellows: Gregory R. Gruzalski and Curtis W. Owen.
Professor Stephan Berko of Brandeis University collaborated in some of our studies and
Akutagawa's thesis was co-guided by Professor James W. Mayer of Caltech.
Other main areas of my Harvard research and those collaborating in my investigation of them
were: (1) kinetics and morphology of precipitation from solid solution, with my students King-
Ning Tu and Alexandre N. Rossolimo; (2) kinetics and morphology of phase separation and
viscosity in covalently bound systems with students T.P. Seward, John P. DeNeufville, Charles
H. Drummond and postdoctoral fellow Donald R. Uhlmann; (3) the pressure enhancement of
crystal growth and atom transport rates in covalently bonded materials with Professor J.F. Hays of
the Harvard Geology Department, Donald R. Uhlmann, and students Vincent J. Fratello, Michael J.
Aziz and Eric Nygren. Dr. John M. Poate and associates of Bell Labs joined Aziz, Nygren and
myself in our studies of the pressure dependence of crystal growth and atom transport rates in
silicon. We found that the kinetic coefficients of certain of the processes in four-coordinated
systems were increased 1 to 2 orders of magnitude by application of 30-40 kilobars of pressure,
indicating that they are attended by large negative activation volumes; (4) nucleation and growth of
crystals in metal liquids and glasses with Frans Spaepen and Lindsay Greer and students,
Genevieve Devaud, Hin-Wing Kui, and Carl V. Thompson; (5) the heats of crystallization and
crystallization rates of ion-amorphized Ge and Si with Poate and his associates of Bell Laboratories,
Spaepen and student Edward P. Donovan; (6) morphology and properties of superconducting
alloys formed by rapid solidification processing with Professors Joze Bevk and Michael Tinkham,
postdoctoral fellow Fawwaz Habbal, and students Chen-Hsi Lin and Gregory Clemente.
Interacting with my Harvard students and postdoctoral fellows has been one of the most
rewarding and pleasant activities of my career. In general their intellectual power and imagination
has exceeded even my initial expectations. However, at the beginning of their research many
students were quite inexperienced in the laboratory and their initial progress sometimes was
painfully slow. The beginner tends to idealize the research process and seeks to achieve step by
step perfection using the most sophisticated techniques that can be devised. This approach often
leads to impasse and frustration. The research strategy which I commend to my students, having
found that it worked best for me, is an opportunistic one in conjunction with what, in common
parlance, is known as the "quick and dirty" approach; i.e. to first do a variety of easy experiments
with the crudest workable procedures. Such experiments may not lead to publishable results, but
often they indicate quickly in which direction further effort should be focused. If promising results
are obtained they can then be tested and quantified by more refined techniques.
In this day of complex and expensive methods, many wonder if a small experimental research
group, constituted mainly by graduate students and with modest financial resources, can any longer
have a significant impact on science. However, while the development of advanced experimental
techniques is indispensable to scientific progress, the developers often are specialists who are more
"technique" than "problem" oriented. It is still crucial to recognize the significant problem, whose
solution often demands a concert of techniques. To perform the "critical" experiment, it may often
be necessary for the university scientist to seek out and attract the collaboration of those, wherever
they may be found, with the needed experimental facilities. I emphasize to my students that while
we can hardly achieve distinction with our hardware, we may do so with our wits.
As my students gained experience, their pace quickened to the point that they controlled their
research and often took productive initiatives that I had not anticipated. My role as thesis adviser
has been to acquaint students with significant problem areas and to help them acquire the
intellectual discipline and the appreciation of the standards of achievement essential for the
successful practice of science. When very good students have their research well underway, the
best policy for a research adviser is to "stay out of their way". At Harvard I missed the wide
spectrum of mature and expert advice to which I was accustomed at G.E., but my thought has been
continually renewed and invigorated by my associations with the many brilliant and intellectually
uninhibited Harvard students I've known.
Perhaps the most unpleasant period in my Harvard experience was in the era of student
militancy, roughly between the years 1966 and 1972. This militancy, which developed about the
same time in many universities throughout the world, was in the U.S. at first directed against the
American involvement in Vietnam. Later it was broadened to attacks on the academic, industrial,
and governmental "establishments". By 1968, the overwhelming majority of the Harvard faculty
and administrators probably considered the U.S. Vietnam policy to be seriously mistaken.
Opposition to the policy was generally expressed by faculty as individuals in the broad political
arena. Excepting on issues concerning academic freedom, the University or faculty as a whole
rarely took positions on public policy matters.
Because of the openness and freedom of campus discourse and the important role of the
faculty in the determination of University policies, it was generally thought that Harvard might be
immune to disruption by student radicals. However, this expectation was dashed by the occupation
of Harvard's University Hall by about 200 students in the spring of 1969. The immediate issue was
whether or not to terminate the military officer's training program (ROTC) on campus, but in view
of the ever more threatening tone of the radical rhetoric, it seemed that some issue precipitating a
major clash of the militants with the University was bound to arise. The not too gentle clearing of
the Hall by outside police called in by the University generated much sympathy for the radicals and
their causes among moderate students and faculty. For about the next three years there was much
unrest in the University, marked by shouting down of campus speakers and occasionally of
professors in their classrooms, student strikes, and blockades. This type of student activism
continued into 1972 when, at Harvard and elsewhere, it seemed to abate quite suddenly. After that
time most students became preoccupied with academic performance and career preparation and
were rather indifferent to activist causes. The reasons for the abatement still are not clear, though
some associate it with the ending of the military draft.
As did other faculty, I favored some of the positions and social reforms advocated by the
activists, but was completely opposed to their strong-arm attempts to impose them. Reforms and
other changes in public policy in the U.S. can only be effected, I believe, through the long and often
frustrating political process of convincing the general public that they should be made. However,
the student activists expended most of their effort, not in the public arena, but in trying to force
various policy changes on the universities. With their faculties keenly sensitive to the logical and
moral ambiguities of most positions, universities were the easy targets. The militants were like
puritans in their self-righteousness but, partly debilitated by affluence, were, unlike puritans, quite
deficient in self-discipline.
The militancy had little direct effect on my activities, but the ambience of the University in that
period was sullen and depressing. In 1968 I was elected to the U.S. National Academy of
Sciences. It was a recognition that I had no hint of and never expected. Before my election, I
imagined that the Academy had a central and exalted role in forming national science policy.
However, I was disappointed to learn that membership is more honorary than operative; the main
duty of most members being to elect new members. Also, the election processes seemed rather less
comprehensive and objective than I had thought. Election requires that some Academy member
take the initiative in seeking out detailed information on the candidate's career and making a case for
their election, which will be intelligible and convincing to the other members, drawn from many
diverse fields, who will vote on it.
At Harvard, "materials science" occupies the middle part of a condensed phase science
spectrum which extends from applied mechanics through condensed matter physics. For the
graduate student, it is one of the options within this spectrum. Often in committee and panel
meetings, I have listened to hours of argument, frequently heated, about the definition of "materials
science". Perhaps I am naive, but the definition has always seemed quite clear and simple to me.
Broadly, I believe that it might be stated--as in my von Hippel lecture and prefatory chapter in Vol.
13 of Annual Review of Materials Science--as "the characterization, understanding, and control of
the structure of matter at the ultramolecular (some prefer "supermolecular") level and the relating of
this structure to properties (mechanical, magnetic, electrical, etc.)". However, in professional and
educational practice, materials science is viewed, rather more narrowly, as the characterization,
origin, and property effects of structure as delineated by the extended imperfections (interphase and
intercrystalline boundaries, dislocations, etc.). This area is the special forte of those who label
themselves "materials scientists" on the university campuses as well as in Washington, D.C., and it
constitutes what is most unique in the curriculum offerings of materials science. It is, as my
colleague James Rice put it, the "core component" of materials science. It should hardly need
stressing that this component has been and will continue to be an indispensable element of science
and technology, and any university which expects to have significant representation among leaders
of the field must provide effective instruction in it.
At Harvard we offer just three graduate level half-courses in this core area; they are: "Structure
of Solids", "Kinetic Processes in Condensed Systems", and "Deformation of Solids". Through the
default of other disciplines it has also been necessary for us to provide, in one way or another,
instruction on the thermodynamics of capillarity and of phase equilibria. During my time at
Harvard the research and instructional program in materials science was conducted by just two to
four faculty, and more often two than four. At first, Chalmers and I were the only "permanent"
faculty. Later, Ashby accepted a tenured position, but with the understanding that, in view of his
strong family and other ties to England and Cambridge, he would be strongly attracted to an
appropriate chair, if offered, at Cambridge University. He accepted the offer of such a chair by the
Cambridge Engineering Department in 1974, after a few years on our faculty. Following Chalmers'
retirement, I was for the next five or six years the only tenured materials science professor at
Harvard; then Frans Spaepen was offered and accepted tenure. While I was the single permanent
faculty member, a substantial part of the instructional burden in materials was borne by Joze Bevk,
Frans Spaepen, and A. Lindsay Greer, who were, at different periods, on our junior faculty.
Many wonder how the Harvard materials effort, having only one or two permanent faculty and
no autonomy, can be viable. In most other leading universities, the materials science faculties are
quite large and autonomous. For example, at present, the MIT materials faculty numbers more than
40, and that at Cornell about 15. To survive and flourish at Harvard, "core" materials science must
maintain close interactions with the much larger efforts in condensed matter physics or applied
mechanics or both. Such interactions are facilitated at Harvard by the absence of organizational or
geographic barriers between these fields. During my Harvard career I have had substantial and
meaningful interactions with perhaps 8-10 faculty from fields, mostly condensed matter physics,
other than my own. I was instructed and much stimulated by my associations with these faculty
and I believe it fair to say that they, in turn, benefited much from their interactions with me and other
materials science faculty. Also, in the Harvard milieu the materials science professors have
excellent opportunities to attract graduate students originally enrolled in other disciplines to their
research programs. Attesting to this, only three of my first twenty research students were initially
my advisees.
On the negative side the small size and lack of autonomy of the Harvard materials faculty
places it in an especially vulnerable position when decisions on appointments and other matters
affecting its health and survival are made. All permanent positions on the Harvard faculty of arts
and sciences are, in effect, chairs, each of which is supported by an endowment sufficient to cover,
at least, the academic year salary of the professor. This policy severely restricts the number of
permanent appointments and there is no commitment of even a single chair to the core area of
materials science. When any chair is to be filled there is considerable competition for it among a
faculty with widely distributed, but often quite parochial, interests most of whom have little
understanding or knowledge of core materials science and its history. It is in this situation that the
Harvard sociology poses, as my distinguished friend warned that it might, a serious problem for
materials science.
The Harvard procedure for making a permanent faculty appointment is quite complex and
drawn out. The object is to identify, with extensive advice from outside experts, a world leader
(hopefully "the" world leader) in the field and then attract him/her to Harvard. There are perhaps
five or six stages in this process culminating in a meeting of an ad hoc committee formed by the
university president, after receiving, with the faculty dean's endorsement, a department's
recommendation. Typically, the committee includes the president, the dean of the faculty, several
faculty from departments other than the one requesting the appointment, and three outside experts.
Meeting for about a day, it reviews (or at least has available) all documents pertaining to the
appointment, interviews some faculty who favor, and possibly some who oppose, the department's
proposal, and then recommends a course of action to the president, which he may accept or reject.
Later the university corporation approves pro forma the president's recommendation, if any.
No Harvard junior faculty position is "tenure track". Even though a junior professor (J.P.) has
an excellent performance record, he/she will not be tenured unless perceived at, and outside of,
Harvard as a world leader in a field in which a chair is to be filled. Due to this stringent policy, only
about 1 in 10 J.P.'s is offered tenure. In justification of the policy, it is argued that the J.P., after
his/her Harvard experience, almost certainly will receive an attractive tenure offer from another
major university. However, for a young experimental scientist, denial of tenure means uprooting
and disrupting a research program in the middle of what is likely to be one of the most creative
phases of their career. Also, the policy makes it difficult to recruit outstanding young faculty
trained at universities other than Harvard. The appointment policies of most major universities are
de jure similar to Harvard's, but de facto they are much more relaxed, so that a J.P. is likely to be
tenured if his/her performance is perceived by the university to be merely outstanding. In my
judgment, the Harvard appointments policy is overly pretentious and rather less objective than it is
intended to be. Clearly, much depends on whose advice is sought and which advice is listened to by
the university administrators. However, the policy assures that the permanent appointments
generally will be of high quality.
Some Thoughts on Directions in Materials Science
Some thoughts on the historical development and future course of materials science were
expressed in my von Hippel Award lecture ["Laser and Electron Beam Processing of Materials"
(eds. C.W. White and P.S. Peercy), pp. 1-5, Academic Press, N.Y. (1980)], my Campbell Memorial
lecture [Metallurgical Transactions 12A, pp. 695-708 (1981)], and my prefatory chapter in Annual
Reviews of Materials Science [13, pp. 1-7 (1983)]. Here I will only recapitulate and update them.
Of course, materials technology, empirically based, is thousands of years old, but the
development of materials science, as we know it, did not begin until about the 18th century.
Nevertheless, it is interesting that the general atomic and molecular concepts underlying materials
science were advanced in the ancient world. The Ionian philosophers, Leucippus and Democritus,
seem to have been led to an atomic theory of matter through realization that the identity and quantity
of a material persist through a succession of physical transformations (e.g. evaporation ↔
condensation, melting ↔ freezing). Later, Epicurus suggested that just as a vast vocabulary can be
constructed from only a few symbols so the tremendous variety of material forms might result from
differing combinations of only a few elementary units.
In the modern era, solid state science began, after discovery of the basic principles of
chemistry, with an initial emphasis on the characterization and understanding of ideally structured
crystals and the coexistence of phases. Following notable advances in the understanding of
interatomic cohesion in the early part of this century, it became clear that the rates of atomic and
charge transport in many solids were orders of magnitude greater than predicted by ideal crystal
models. To resolve these disparities point defect models for transport were devised by Frenkel,
Schottky, Wagner and others. Later, the low resistance, relative to that predicted for perfect crystals,
to plastic flow was accounted for by the dislocation models of Orowan, Taylor, and others. Still
later, F.C. Frank showed that the easily measurable growth of polyhedrally shaped crystals at small
departures from equilibrium could be effected by screw dislocations emergent in the crystal
surfaces. Much of the materials science effort in the 1940's and 1950's was centered, as it was at
G.E. in this period, on the development, refinement, and testing of these defective crystal models of
solid behavior. These efforts culminated in vindicating, to a remarkable degree, the model
predictions on the nature, density, and actions of the imperfections. Especially impressive were the
microscopic studies which showed that dislocations exist and act in real crystals in the ways that
had been forecast. In contrast with point defects, extended imperfections increase entropy so
minutely that they should be practically absent from finite crystals in thermodynamic equilibrium to
quite high temperature. Therefore, it was presumed that they must be introduced into actual crystals
by various growth accidents and surface damage. These presumptions were confirmed
experimentally by several investigators and, perhaps more importantly, it was shown that crystals
(e.g. metal "whiskers") could be grown virtually dislocation-free and that such crystals did exhibit
the mechanical strength and growth resistance nearing the levels predicted by the ideal crystal
models.
From my chemist's perspective, the amazing feature of these developments was that the
properties of the crystals were altered so dramatically by the presence of such minute fractions (e.g.
1 in 106) of displaced atoms. These small deviations were aptly called "imperfections in nearly
perfect crystals".
While studies of nearly perfect crystals continued, by 1960 there was a growing research
emphasis on the formation and properties of materials in frozen metastable states, including
amorphous and other highly disordered solids. This new and still continuing emphasis was
stimulated in part by the challenge to theory posed by disordered systems and in part by advances
in experimental techniques adaptable to metastable structure synthesis. Among the principal
techniques used in these syntheses are: ultra-rapid melt quenching, various condensation
procedures (vapor-, sputter-, electro-), ion implantation, ion beam mixing, and laser pulsing. They
have in common the feature that they can effect extraordinarily high rates of energy flow into and
out of thin material sections. Exploitation of these techniques has, during the past three decades, led
to the synthesis of a wide variety of metastable structures. Among these are: amorphous solid
forms of semiconductors and metallic alloys, which have already been alluded to; "quasicrystals",
new and unexpected structures which are, strictly, aperiodic, but which exhibit sharp diffraction
patterns with non-crystallographic (e.g. icosahedral) point group symmetries; solid solutions with
much higher solute concentrations than known heretofore, indeed, it now seems possible, in
principle, by ion implantation or deposition processes, to form homogeneous solid solutions of

virtually arbitrary composition; interphase dispersions, prepared by deposition or processing of in
situ composites, with interfacial area densities approaching the ultimate limit.
There are other methods of structure synthesis which I expect may become more widely
exploited in the future. One of these, developed by W.L. Johnson, R.B. Schwarz, and others, is the
formation of metastable structures by solid state interdiffusion of selected materials. There is much
to learn about the thermodynamics and kinetics of these processes and their potential for producing
new materials is probably far from full realization. A somewhat neglected approach in metastable
synthesis is that of eliminating heterophase impurities, e.g. by the isolation or fluxing techniques
alluded to earlier, which may nucleate the more stable state during deenergization before a
metastable one forms. I believe that wider use of this approach would prove quite productive.
Another important area that seems to deserve increased attention is that of interface
modification, by, e.g., irradiation or chemical reaction, in order to increase interfacial strength. Such
processing can be crucial for forming composites with acceptable mechanical strength.
As I have noted elsewhere, the historical emphasis in materials research might be viewed as
having progressed toward the more disordered (and non-equilibrium) states of matter. Indeed, it
appears that we can now fabricate solids in which the ultimate degree of disorder is approached, but
the understanding of the structure and properties of such materials is still in a highly primitive state.
In my judgment, the opportunities for the discovery and study of novel metastable structures are
still so great that they will heavily engage material scientists for some time to come. The present
situation in the synthesis of such structures seems somewhat akin to that in organic chemistry
following establishment of the principles of chemical synthesis; the possibilities for new and
interesting combinations of structural elements seem almost limitless.
The application of the new understanding of materials to ceramics has not progressed as
rapidly as it has to semiconductors and metals. The slower pace of ceramic science has been due
partly to the greater difficulty of processing (e.g. by melting and shaping) ceramics which makes
experimentation on them much harder. I expect that we will see growing emphasis on this branch
of materials science and wider application of the newer techniques to the synthesis of ceramics in
novel metastable forms.
Also, I believe that we will see greatly increased effort in the materials science of condensed
organic systems, especially polymers. Progress in this field has been hampered, I believe, by a lack
of communication between organic chemists and materials scientists. However, there is, I think,
great potential for achieving new understanding and for discovery of still more new polymeric
materials with exceptional electrical, mechanical, and thermal properties.
Some General Observations
As may be clear from what I have written, I am confident that the enlightenment and vastly
improved human capabilities resulting from scientific progress is, on the whole, good for society.
Of course, it is a truism that some of the novel capabilities deriving from physical science can be
and have been put to very bad uses, such as: the creation of terribly destructive weapons; the
development of processes which can lead to serious, and possibly catastrophic, environmental
damage; and the over-indulgence of society with consumer goods. Concerning consumer over-
indulgence, I find myself in general agreement with the viewpoints expressed by J.K. Galbraith in
his "The Affluent Society".
Certainly it is a responsibility of scientists to do their utmost to inform society of the
destructive potential of new weapons and of the great damage that certain practices, e.g. profligate
consumption of fossil fuels, may inflict on the environment. The public response to such
information will be determined by the outcome of political processes. My preference has been to
participate in these processes as an individual citizen rather than as a scientist since, in my
experience, scientific ability does not necessarily assure civic sagacity. Thus, I believe that the
political choices of a well informed and educated public are likely to be as wise as those of
scientists as a group.
Looking back, I am fortunate for having been able, during most of my career, to work toward
the objective I set for myself at age 18--that of trying to advance understanding through the practice
of physical science. I have tried to do this by my own research and by training students to do
research. This activity has been for me intellectually challenging and highly fulfilling.
I am impressed ever more deeply by the natural order, which seems remarkable and awesome.
It can be described by science, but its existence and origin are still a great mystery which seems
beyond our capabilities to resolve. For me it is not necessary to accept some metaphysical solution
of this mystery; it is enough that the order exists and is esthetically pleasing. Thus, I have sought a
maximum commitment to life and humanity with a minimum commitment to dogma. In my view,
commitment to humanity arises naturally from a wide ranging and deeply sensitive human
consciousness which leads one human to feel sharply within one’s self the aspirations or pains of
another. I believe it is such consciousness which has led people to imagine or feel that there exists
a just and humane basis for ultimate reality.

David Turnbull

Uploaded by

Copyright:

Available Formats

David Turnbull

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

David Turnbull

Uploaded by

Copyright:

Available Formats

DAVID TURNBULL

several sons from the cycle of violence.

four years younger than I. He became a medical doctor.

and my joy when he returned unharmed.

My father had a high school education, supplemented by a short accounting course. As a

Jesus as a historical figure is one that has always intrigued me.

grades were expected to continue their study.

virtually excluded them from most of the other learned professions.

father remarried four years following my mother's death.

College and Graduate Study

satisfaction in endeavoring to advance knowledge and understanding through the practice of

the advancement of physical science.

to physical chemistry. I excelled in chemistry and Haldeman urged me to go on to Ph.D. studies.

Such instruction was, indeed, an invaluable part of my Monmouth education.

crystal nucleation in liquid Hg".

I might have misunderstood the development. Actually, to my considerable relief, Rodebush

approved and liked my critique.

synthesis might be achieved by electrolysis of ammonium salts dissolved in a non-hydrogen

persisted with increasing period and amplitude to still higher fields.

and by Eugene Guth and coworkers.

accuracy and repeatability in experimentation on himself and his students.

recommendation, I was offered and, as he advised, accepted an instructorship in physical chemistry

anyone from there before accepting their offer.

Case School of Applied Science

maintenance with very little assistance.

and eliminated some by redesign of experiments.

was, I think, prevalent in most engineering schools at the time.

directed to a brilliant diagnostician, Harley Willims, who immediately attributed my symptoms to a

same diagnosis, also at age thirteen.

collaborators and me, are described in eighteen papers.

imperfections in such transport, and resolved to investigate it when I could. My admiration of

conferences: "Condensed Phase Physics" and "Physical Metallurgy".

for this, so I looked for opportunities elsewhere.

General Electric Research Laboratory

executive offices in New York increased.

others in his Semiconductor Studies section.

conducive to interdisciplinary cooperation and to daring modes of thought and experimentation. In

physicists, metallurgists, chemists, and applied mechanicians. Metallurgists were remarkedly

from metallurgists. Interdisciplinary exchanges occurred at various meetings, sponsored mainly by

provided by the yearly weekend seminars sponsored by that most applications-oriented of

G.E. group. He was brilliant as an innovator but less successful as a renovator.

impression was, I believe very helpful to my thought.

on the short-circuiting effects of extended imperfections on metal diffusion and on the

undercooling of liquids. My examination of nucleation theory led to consideration of transient

for circumventing the complexities of problems and so exposing their essence.

crystal and short-circuit diffusivities.

speaker on "Solidification" for an M.I.T. Metallurgy Department colloquium. Hollomon promptly

measurable crystallization, as indicated dilatometrically.

melting of close-packed crystals was, thermodynamically, almost a second order transition.

came to my attention only after I had completed most of my undercooling studies.

recalescence. We demonstrated conclusively that molten droplets of a number of high purity

an icosahedral type of SRO, which by itself is non-crystallographic, might predominate in the

liquid. Its transformation to close-packed crystalline SRO would require reconstruction. In

by marked localization of positions on the liquid side of the interface. Such

localization, effectively a layering in liquid structure, would be attended by substantial negentropy.

magnitude of the crystal-melt tension indicated by the nucleation experiments.

quick to detect a flawed argument.

kinetics of nucleation in various Hg droplet dispersions. Those dispersions constituted by droplets,

repulsive forces at some point during its collapse with cooling.