The development and calibration of the atom probe field-ion microscope.


Material Information

The development and calibration of the atom probe field-ion microscope.
Physical Description:
xi, 159 leaves. : ill. ; 28 cm.
Stoltz, Daniel Lewis, 1937-
Publication Date:


Subjects / Keywords:
Field ion microscopes   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Bibliography: leaves 155-158.
Statement of Responsibility:
Daniel Lewis Stoltz.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000582700
notis - ADB1078
oclc - 14162293
System ID:

Full Text



Daniel Lewis Stoltz

A Dissertation Presented to the Graduate Council of
the University of Florida
In Partial Fulfill:ent of the Requirements for the
Degree of Doctor of Philosophy



Dedicated to my wife,

Orleen Faye.


The author would like to acknowledge and thank his

advisory committee, Professors J. J. Hren, L. L. Hench,

R. T. DeHoff, J. B. Conklin and E. H. Hadlock. In particu-

lar he would express his gratitude to his chairman, Profes-

sor J. J. Hren, who convinced the author of the importance

of this research and who maintained an open and creative

research environment to accomplish it. Also, the expert

help of J. H. Bottcher, D. A. Jenkins, C. S. Kim, H. R.

King, E. C. Logsdon and U. T. Son in completing the con-

struction and calibration of the atom probe field-ion

microscope was gratefully appreciated.

The author's wife, Orleen Faye, is also acknowledged

for her constant encouragement, inspiration and good humor.

Also, the author's parents are thanked for their lifelong

encouragement and support.

Finally, the financial support of the National Aero-

nautic and Space Administration and Advanced Research

Products Agency was appreciated.










Introduction . 6
Theoretical Background . 8
Experimental Verification .. 24


Introduction . 33
Technical Design and Material
Selection .. 37
Description of Atom Probe .. 40


Introduction . 72
Performance of Individual Components 72
Operational Performance and
Calibration . ... .89


Introduction . .. 115
Description of Microscope 116
Description of Field Emission
Analysis . 126




LIST OF REFERENCES ............... 155

BIOGRAPHICAL SKETCH ............... 159


Figure Page

1 Potential energy diagrams for (a) a
surface atom, (b) a surface ion, and
(c) a surface ion in the presence of
a field . .. .11

2 Three possible field evaporation models:
(a) simple ionic bonding, (b) evapora-
tion over the Schottky barrier after
transition from atomic to ionic state
at Xn, and (c) ionization at xn
followed by immediate evaporation .. 11

3 Potential energy curves illustrating
the various correction terms for both
the image-hump model (a) and the
intersection model (b) .. 15

4 Energy scheme for field evaporation of
a nonmetallic impurity as: (a) a mon-
atomic ion and (b) a molecular ion 23

5 Principle of operation of the atom
probe field-ion microscope ... 34

6 Atom probe field-ion microscope ..... 41

7 Ultra-high vacuum system for atom probe 42

8 Initial atom probe specimen holder 47

9 Physical arrangement of specimen
holder and manipulator in atom probe .. 48

10 Final specimen holder and mechanical
manipulating assembly ... 50

11 Specimen holder and manipulating rod
assembly in biased position .. 53

12 Principle of microchannel-plate image
intensifier . ... .55

LIST OF FIGURES (continued)

Figure Page

13 Microchannel-plate image intensifier
for atom probe . .. 57

14 Atmospheric high voltage feedthrough
for atom probe . ... 62

15 Circuit diagram of preamplifier .... 67

16 Detection circuitry for atom probe 68

17 Specimen manipulation provides observa-
tion of nearly entire surface of tip 77

18 Analog mode circuitry for Spiraltron
detector . .... 83

19 Neon ion current vs. applied potential
to Spiraltron detector in analog mode 85

20 Optimum focusing potentials for micro-
channel-plate and phosphor-coated
screen . ... 94

21 Micrograph of tungsten covered with
gaseous contaminants at 780K ...... 98

22 Random signals detected from triggered
Tektronix 549 Storage Oscilloscope on
500 psec traces . ... 100

23 Tektronix 549 Storage Oscilloscope
20 psec traces of field desorbed species
from tungsten surface, where Vdc =
7.5 kV and Vp = .9 kV ... .101

24 Micrograph of tungsten at 780K imaged
in helium-10 percent neon ... .106

25 Tektronix 549 Storage Oscilloscope, 20
and 10 psec traces for field-evaporated
tungsten at 780K in helium-10 percent
neon . . ... 107

26 Constructed field emission and ion
microscope . . 117


LIST OF FIGURES (continued)

Figure Page

27 Ultra-high vacuum system for microscope 119

28 Sequence involved with specimen exchange
mechanism . . 122

29 Atmospheric high voltage feedthrough
circuitry for microscope .. 127

30 Calibration of Brandenburg high voltage
supply ..... . 128

31 Circuit for point-by-point measurement
of in I vs. 1/V . ... 130

32 Circuit for automatic measurement of
n I vs. 1/V . .. 132

33 Calibration of photomultiplier current
vs. Quantum Radiometer ... .134

34 Field-ion and field emission images of
tungsten obtained from fiber-optic
screen . . 136

35 Field emission data for tungsten ... .138

36 Silicon weight loss vs. time for
different HF-HNO3 solutions ... 141

37 An acceptable silicon tip mounted to
a nickel lead . ... 143

38 Field-ion micrographs of phosphor-
doped silicon imaged at 780K in a
helium-10 percent neon mixture at
5x10-4 Torr . ... 144

39 Computer simulated field evaporation
of the (100) pole of silicon ... 148


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



Daniel Lewis Stoltz

December, 1972

Chairman: John J. Hren
Major Department: Materials Science and Engineering

Since field evaporated surface atoms are identified

with the atom probe field-ion microscope, the theoretical

parameters of the essential field evaporation process are

reviewed. The current status of the image-hump and inter-

section models, field strength, field evaporation rates and

the field evaporation of alloys is described. The experi-

mental verification discussions include the rate sensitivi-

ties, imaging gas effect and pulsed field evaporation. In

addition, current atom probe results are discussed that

indicate further experimental data are needed to provide a

quantitative field evaporation theory.

A detailed description of the technical design, mate-

rial selection, fabrication and construction of the indi-

vidual components for the completed atom probe field-ion

microscope is presented. In particular, the ultra-high

vacuum system, specimen holder, internal microchannel-plate

image intensifier, Bendix Spiraltron detector, impedance

matched of the high voltage pulse to the high voltage feed-

through circuitry, high voltage power supplies and elec-

tronic circuitry are discussed separately.

The operational performance and calibration of the

completed atom probe field-ion microscope are described.

In particular, the specimen holder and mechanical manipu-

lator assembly allowed the tip surface a minimum of 60

rotation providing the tip was precisely positioned within

.3 mm of the optical axes. Voltages were applied up to

24 kV without breakdown. The Spiraltron detector was oper-

ational in both the analog and pulse counting modes and was

found to limit the effective sensitive area of the tip to

30 and 50 A2 for 400 and 800 X tip radii, respectively.

Critical alignment of the Spiraltron detector and the speci-

men tip was achieved by optical, laser and electronic means.

A detrimental defocusing effect was observed and was attrib-

uted to a long time constant and residual charge buildup at

the perimeter of the probe holes in the microchannel-plate

and phosphor-coated screen. This defocusing effect was

corrected by utilizing the microchannel-plate and screen as

electrostatic einzel lenses. Optimum focusing potentials

were determined.

Gas desorption studies of tungsten surfaces indicated

from the 20 Psec traces with the Tektronix 549 Storage

Oscilloscope the following ionic species: H+, N N 0 ,

02, CO2 WN2 and possibly W20. The efficiency in detect-

ing aligned single particles was about 50 percent. Although

the efficiency of the fixed probe hole method was less, the

method was successfully used to monitor the field evapora-

tion of the tungsten.

Field evaporation studies of tungsten at 780K in He-10

percent Ne revealed from the 10 and 20 Usec traces of the

Tektronix 549 Storage Oscilloscope the following species:
CO2, H+, Ne +, N, O0, WO,4. WO23. WH2, WO*-H20- and
2 22'2*H2 and
possibly W205 and WH4 No tungsten-helium and -neon com-

pounds were identified from these traces; however, the over-

lapping of the various isotope forms of tungsten-hydride and

-helide compounds prevented their unambiguous distinction.

The efficiency of detecting aligned surface atoms was less

than 50 percent, but this was improved by aligning the probe

hole on the inside of the ionization.ring.

The practical mass resolution of the atom probe was
.5 AMU (atomic mass unit) for m/n ratios below 100 and

.7 AMU for m/n values above 100. These values were ob-

tained for the assumed pulse factor of two. It was dis-

covered that the determined AMU values did not differ more

than .2 AMU when the electronic dead time was assumed a

maximum of .060 Usec, so it was neglected in the ionic

species identification. Accurate determination of the

pulse factor and electronic dead time can be obtained when

a more favorable statistical representation of detected

species is obtained.



The historical development of the atom probe field-ion

microscope originated when field ionization was used as a

source for mass spectrographic investigations by Inghram

and Gomer, () Beckey(2) and Block.(3) Results obtained

from these studies of H2, D2, N2, 02 and hydrocarbon gases

ionized by a field-ion emitter were quite different from

previous results obtained from conventional ion sources.

Mass spectrographic analysis of the field evaporated ionic

species, however, was difficult experimentally because of

the rapid blunting of the emitter tips. Despite this prob-

lem, Miller and Thomsen(4) observed the copper hydride

products of a hydrogen promoted field evaporated copper

emitter. Also, Vanselow and Schmidt(5) observed several

platinum and platinum oxide species that were field evapo-

rated from platinum tips which had been heated between

1300 and 15000K. Finally, Barofsky and Miiller(6) performed

the first spectrometric analysis of metal emitters cooled

between 21 and 3000K. Field evaporated species of beryllium,

iron, cobalt, nickel and copper were identified as hydrides

by a focusing mass spectrometer which was capable of scan-

ning up to 90 AMU (atomic mass units) in approximately 9

seconds. However, the sensitivity of this instrument was

limited by the significant signal-to-noise ratio. While

concentrating on means to minimize this problem, Mller(7)

conceived the idea of detecting only one single particle

and eliminating the noise discrimination problem by impos-

ing a strict correlation between the initiation of the field

evaporation and the detection event.

In order to satisfy these conditions the developed

instrument, the atom probe field-ion microscope, combined

the atomic resolution of the field-ion microscope with the

single particle detection capability of the time-of-flight

mass spectrometer. By locating a probe hole in the center

of the phosphor-coated screen of the field-ion microscope

a single surface atom could be selected and field evaporated

to the aligned ion detector. In 1967 the first prototype

atom probe was completed by MUller et al.(8,9) This instru-

ment has been modified recently to provide better vacuum

capabilities, specimen manipulation and photographic record-

ing of the images. Brenner and McKinney(10) independently

constructed an atom probe that was similar to MUller's but

featured an internal gimbal system to provide two axes of

rotation for the tip. This instrument was designed also to

apply a negative high voltage pulse to the tip through a

flexible, unshielded wire loop, rather than applying a posi-

tive pulse directly to the tip as done originally by MUller.

The main disadvantage of this method is the reported loss

in mass resolution and detector efficiency of the atom

probe.(11 At this date, these two research groups are the

only ones conducting active research with the atom probe

field-ion microscope.

The initial significant experimental result obtained

by Muller with the atom probe was the detection of three-

and four-fold charged metallic ions and their oxide and

nitride compounds which had been field evaporated from

tungsten, rhenium, tantalum and niobium surfaces.(9,12) In

addition, Brenner and McKinney (13) independently observed

two and three positive charges for iridium and confirmed

the highly charged ions for tungsten. It was also observed

that the amount of these highly charged ions increased as

the field evaporation rate was increased. Another surpris-

ing observation made by MUller with the atom probe was the
adsorption of helium, neon and argon at the metal surfaces.(14)

The field strength required to remove these adsorbed inert

gases was sufficient to evaporate the surface metallic

atoms simultaneously. (14) Brenner and McKinney reported

that the adsorption of helium was observed even when the

helium gas pressure was reduced to 5x109 Torr.(15) The

reported .2 AMU mass resolution capability with MUller's

atom probe allowed the important observation of field de-

sorbed metal-noble gas molecular ions, RhHe+2 and WHe3. (14)

Other experiments in which tantalum was field-evaporated

in helium and neon imaging gases indicated that the surface

binding of helium and neon might be an activated chemi-

sorption with the activation energy supplied by the strong

polarization of the surface atoms exposed to the field. (16)

All of these experimental results are not predicted by the

current field evaporation and ionization theories and must

be incorporated in them to provide a quantitative descrip-


The experimental results obtained by the atom probe

field-ion microscope emphasize the unique sensitivity and

versatile capabilities available with the instrument for

surface studies and microanalysis. Consequently, this power-

ful instrument can supply sophisticated experimental evidence

for many scientific disciplines. For example, in physics

the atom probe can be used to study the following: field

ionization and field evaporation parameters to provide a

quantitative description of these processes, the nature of

"clean" metallic surfaces, the measurement of the polariza-

tion factor and the high electric field penetration of a

metallic surface, the peculiar formation of metal-inert gas

molecules, and the influence of emerging defects on the

surface's structural and electronic properties. Chemists,

on the other hand, might apply the atom probe to the follow-

ing investigations: determination of single particle chemi-

cal composition, mass spectrometry of molecular species

(since the molecule will not be fragmented by this tech-

nique), the formation of the adsorbed gaseous layers,

and possibly catalytic reactions, by the high field.

Besides applying the atom probe to specific field-ion

microscopic problems, the material scientist has a very wide

scope of possible investigations for which the atom probe

can supply invaluable information. For instance, order-

disorder transformations, spinodal decomposition, solute

atom distribution, interfacial segregation, oxidation and

corrosion, surface diffusion, surface binding energy deter-

minations, nucleation of vapor-deposited films, heterogene-

ous nucleation of defects and microanalysis.

The present research will describe in detail (a) the

theoretical foundation of the phenomena embodied in field-

ion mass spectrometry, (b) the design associated with pro-

ducing such an instrument, and (c) data relative to the use

of the field-ion and field emission microscopic features

and the single particle identification capability of the

calibrated atom probe device.




The attainment of a "clean" solid surface involves

sophisticated techniques and ultra-high vacuums below 1010

Torr. Under these conditions reasonable periods of time

are realized to conduct experiments concerning the nature

and characterization of the "clean" surface prior to its

coverage with gaseous adsorbates. One of the most unique

methods used to create "clean" surfaces is associated with

field-ion microscopy. In field-ion microscopy a nearly

hemispherical, atomically smooth surface must be produced

to develop radially projected images with 106 magnification
and 2 to 3 A resolution. To produce this "clean" surface

from a cryogenically cooled specimen the electrostatic field

at the tip is increased until the surface contaminates are

desorbed and the atomic asperities are evaporated as ions.

This technique was designated "field evaporation" by its in-

novator, MUller,(i7,181n 1941. At that time, experimental

techniques were adequate to produce the necessary 5 to 8 V/A

potential at the tip and to observe the stripping of the

surface atoms with a helium image. In addition to producing

"clean" surfaces, field evaporation can be repeated on a

specimen to achieve an atomic layer by atomic layer serial

sectioning of the bulk structure and to observe any defects

in the bulk as they emerge at the surface.

Since field evaporation is essential to field-ion

microscopy, it is important to understand'how the field

evaporated surface is formed and its influence on the re-

sulting field-ion image's interpretation. Ideally, the

field evaporation theory should determine initially the

evaporation of individual atoms with respect to temperature,

field strength and chemical properties. Next, the theory

should consider the surrounding structure's influence on the

evaporating species. Such parameters would include crystal

structure and orientation, electronic structure, specimen

size, solute and impurity atoms, and the presence of residual

gases in the microscope body. From these factors one should

be able to predict the final shape or "end form" of the

evaporated specimen, plus the necessary parameters to con-

trol the evaporation process. Due to the complexities in-

volved with this process, the current theory provides only

a qualitative description and is being modified constantly

to better explain sparse amounts of available experimental

data. Consequently, it is the purpose of this section to

review briefly the current status of the field evaporation

theory and discuss the pertinent experimental measurements.

Theoretical Background

Activation Energy

The primary theoretical problem involves determining

the activation energy, Q, necessary to field evaporate

surface atoms as a function of the applied potential, V.

Millerr(1) envisioned this process as a special case of a

thermionic cycle and expressed the activation energy needed

to remove a surface atom without an applied field as

Qo = A + E In nW (1)

where A is the sublimation energy needed to free the surface

atom from its potential well at the surface, as shown in

Figure l(a). The terms E I represent the additional energy
n n
required to ionize the sublimated or gaseous atom with a

charge of ne; i.e., the term In corresponds to the nth ioni-

zation potential. When the metallic atom is ionized, the

ejected electrons from the atom return to the metal surface

and contribute a potential energy, no, to the metal. Conse-

quently, P in relation (1) represents the total, not abso-

lute, work function because it corresponds to the potential

energy difference of an electron located at the Fermi sur-
face and at infinity in a vacuum.(19) Consequently, the

activation energy, Qo, is illustrated in Figure l(b) as the

energy increment between the bottom of the atomic potential

well and the zero level of the ionic potential. Note that

the zero levels of the ionic and atomic potential energies

are separated by the amount E I nO.
n n
When a positive potential is applied at the specimen

tip the ion experiences a repelling potential equivalent to

-neFx, where F is the electric field, X is the distance

between the ion and the metal surface, and ne is the number

of electronic charges. Correspondingly, an attractive

image-force term is formed by the rearrangement of surface

charge on the metal due to the ion's presence and is equal

to -(ne) 2/4x. In addition, Brandon(21) proposed that

another short range repulsive term Vrn(X) must be included.

So, the ionic potential in the presence of a field, as

illustrated in Figure l(c), is equivalent to

Vi(X,F) = Vr X) neFX X (2)

Generally, the short range repulsion term Vrn(X) is neglec-

ted(4) because the ionization process is assumed to occur

at a sufficiently great distance from the atomic surface.

Consequently, expression (2) is approximated by

Sn(ne) 2
Vi(x,F)= neFX- n- (3)

Differentiation of expression (3) to obtain the maximum

energy, referred to as the image-hump or Schottky barrier,

results as follows,

S1/2 3 6n) 1/2
Xs= ) or ) (4)
0 0
where Xs is A and F is V/A. This position Xs is illustrated

in Figure l(c).

If the ionic potential curve of Figure l(c) is super-

imposed on the atomic potential curve of Figure l(a), three

possible models for field evaporation are devised. Under

the condition that E I -nc is small, the ionic and atomic
n n
potential curves do not intersect and the resulting model

represents simple ionic bonding, as shown "in Figure 2(a).

The second and third possibilities assume the usual condi-

tions for metals, i.e., the E I -n$ term is large, which
n n
allows intersection of the two potential energy curves. In

Figure 2(b) the atomic state is stable at the surface and

the two curves intersect left of the Schottky barrier. This

means that the atom has sufficient energy to be ionized at

the intersection distance, Xn. However, to be evaporated

the ion must have additional kinetic energy to overcome the

Schottky maximum or it will be forced back into the stable

interior of the metal. This evaporation process was intro-

duced by Miller (1718) and is referred to as the image-hump

or image-force model. The required activation energy for

this model is given by

Qn Q neFXn ne)2 Q (n3e3F)1/2, (5)

where Q is given by expression (1).

The third possibility represents the atomic potential

curve intersecting the ionic curve to the right of the

Schottky barrier. This occurs when the atom has sufficient

energy to ionize at the intersection point Xm and is

~ II I 1X

1 (_".F) "Ie-n<
O I .I ,l

(a) (b) (c)
Figure 1. Potential energy diagrams for (a) a surface atom,
(b) a surface ion, and (c) a surface ion in the
presence of a field.(20) (Note the difference
in the zero energy level for the atom and ion.)

Figure 2. Three possible field evaporation models:
(a) simple ionic bonding, (b) evaporation over
the Schottky barrier after transition from
atomic to ionic state at xn, and (c) ioniza-
tion at xn followed by immediate evaporation.(20)

spontaneously repelled from the surface. Gomer(22) intro-

duced this intersection model of the evaporation process

and expressed its required activation energy as

Qn = Qo eFXm (e) (6)

where Xm is the distance between the metal surface and the

ionization region; i.e., the point of intersection of the

two potential energy curves. Brandon(23) also noted that

the activation energy of the image-hump model could be

linearly dependent on the field, if a significant short-

range repulsive potential Vrn(X) existed for the ion. This

would effectively reduce or possibly eliminate the Schottky

barrier and invalidate expression (3).

Several corrections to the activation energy expressions

(5) and (6) have been reported for both models. Millerr(2)

and Gomer and Swanson 25) have added an important term,
1 2
y(a-a"i)F to both relations (5) and (6). This term repre-
sents the difference in the dipole polarizability of the

metal atom, aa, and the singly charged ion, ai, and dis-

regards the reported higher order polarization terms. 26

The effect of this polarization difference is equivalent to

assuming larger values of Q for a given field. Typically,

polarization corrections amount to 1 to 3 eV. In addition,

when the different types of surface binding were considered

by Gomer and Swanson(25) with the intersection model, broad-

ening of energy levels and energy shifting became significant

for covalent bonding. Thus, the activation energy expressed

in relation (6) was appropriately corrected, as follows:

(ne) n r
Q = A + E X neFX AE (7)

where AE is the shift of energy level and r is the energy

level broadening value which is given as the ratio of

Planck's constant h and the electronic transition time T.

Note that expression (7) also indicates a linear relation-

ship between Qn and F. Recently, Gadzukl27) reported a

detailed analysis that included calculated energy level

broadening and shift effects for cesium on tungsten and

potassium on platinum. The broadening value was approxi-

mately 1 eV and the energy shift was about 0.3 eV.

Another important factor to consider is the electric

field penetration of the metal surface. By assuming the

metal surface to be an abrupt termination of a periodic

array of atoms, it will be void of conduction electrons and

can be penetrated by the electric field to a finite screen-

ing distance, X. This effect reduces the ionic potential

curve by an amount neFX. In addition, if the intersection

model assumes that the surface mirror plane of the metal

coincides with its assumed surface-vacuum interface, the

surface atoms would experience this electric field penetra-

tion and its term must be added to relation (7). Although

the assumed metal surface does not correspond to the real

surface, the corrected expression (7) provides reasonable

agreement between the calculated and experimental activation

energies for desorbed inert gas atoms and highly charged

ions of tungsten. 26)

These corrections are illustrated schematically in

Figures3(a) and (b) for the respective image-hump and inter-

section models. Considering the various assumptions made

for each model, the image-hump model would appear more valid

if a Schottky hump really existed and is located beyond Xn

in Figure 3(b). However, values of Xs, calculated from rela-

tion (4), for typical experimental parameters were less than

the radius of most ions for both singly and doubly charged

ionic species.(11)

Quantum mechanical treatment of metal surface poten-

tials was undertaken initially by Bardeen(28) and later

refined by Juretschke(29) and Loucks and Cutler. 30) These

calculations indicate that the classical image potential

remains valid at a large distance from the metal surface,

whereas quantum mechanical exchange and correlation forces

predominate near the surface. Consequently, these calcula-

tions also indicate there is no theoretical basis for the

Schottky barrier to exist for singly and doubly charged

ions. In addition, its existence with triply charged ions

is doubtful. However, calculations for higher ionic charges

suggest its possible existence. 26)


n a
T- -- --- -* --- -

xI Vi(x)



.---^Z^E1' /

0 x
I \ V""i (x)-/ E-

Sin-n. -AE-.

V (x)

(a) and the intersection model (b).(9)

Evaporation Field Strength

At the cryogenic temperatures used in field-ion micros-

copy the activation energy, Qn, is assumed negligible.

With this assumption the evaporating field, Fen, can be

determined from the respective activation energy relations

for both the image-hump and intersection models. For the

image-hump model the evaporating field is determined from

relation (5) and is expressed as

(A + I nc)2
F = (8)
e 3 3
n ne

providing the polarization and field penetration terms are

neglected. The evaporation field for the intersection model

was determined from relation (7) and is equivalent to

A + I no 3.6n2/r0
en= nr V/ (9)
e nr
n o

Both A/2 and AE were neglected in expression (7) because

their contributions are small. Qn was assumed zero and Xn

was assigned the single bond atomic radius, ro, of the metal

atom. As mentioned previously, the existence of the Schottky

barrier is difficult to predict theoretically. Consequently,

it is not possible to state if the atom will tunnel or be

thermally excited over the Schottky hump by either of these

two evaporation models. Despite this uncertainty, useful

data(24,31) comparing the evaporation fields for various

metals were determined from the image-hump model by Miller(24

and Brandon.(31) Brandon's(31) calculations also indicated

that the doubly charged ions had lower evaporation fields

than the singly charged ions. Predicting the evaporation

field required to evaporate triply and higher charged ions

is limited because their ionization potentials and sublima-

tion energies are unknown.

Other factors contribute to the strength of the evapo-

rating field which are not included directly in the above

relations (8) and (9). For instance, the degree of protru-

sion of the atom, the location of an atom in the crystallo-

graphic plane and the tip geometry should be included. The

complex interaction of the various field evaporating vari-

ables led Brandon(31) to introduce the following relation

F = 1 a2 Fa (10)

where 81 is the field enhancement factor related to the

local radius of curvature, 82 is the field enhancement

factor due to local lattice geometry, particularly the

lattice step height, and Fa is the average field on a tip.
Gomer ( equated the average field strength F as follows

1 V
Fa = F (11)

where k is a semi-constant (normally 5), V is the applied

tip potential and R is the tip radius. When the product

8182 in relation (10) is practically constant, the stable
tip will exhibit a surface with flat, close-packed planar

regions due to their large lattice step heights. In

achieving this stable condition the weakly bound atomic

sites at the edges of the close-packed planes were field

evaporated initially. Even when the tip surface is stable,

the field strengths over the different regions of the sur-

face are not uniform due to the slight changes in radii of


Field Evaporation Rates

The field evaporation rate ken for both the image-hump

and intersection models assumes that the process is ther-

mally activated and obeys the Arrhenius rate equation of the


ke = Vn exp{-Qn/kT} (12)

where vn is the vibrational frequency of the surface atom,

Qn is the activation energy, k is the Boltzmann constant
and T is the absolute temperature. Further considerations

indicate that quantum mechanical tunneling predominates at

low temperatures.(22'25'32'33) Consequently, the rate con-

stant is approximated by

ke v n exp{-an n3/2} (13)

where an is dependent on the atomic mass as well as the

shape of the potential barrier and is a slow function of the

electric field F. At intermediate temperatures, optimum

rates are combinations of the thermal activation and tunnel-

ing processes. Gomer and Swanson(25) and Tsong 33) extended

their field desorption process to determine the rate of

field evaporation for either lattice atoms or adatoms with

masses similar to the substrate mass, where hv << kT. The

resulting rate appeared as follows

-Qn 4
ke pv exp{- -- + 3} (14)
S27an2 (kT)

where p is the electronic transition probability, v is the

surface atom vibrational frequency and an is the constant

described for relation (13). This expression is similar to

the rate equation for ionic bonding and indicates that the

field evaporation involves three steps: thermal excitation

to higher vibrational levels, electronic transition and

atomic or ionic tunneling. (11) Gomer and Swanson determined

values for the pre-exponential factor, p, which indicated

that a two-electron transition was more favorable than a

single electron transition by two to four orders of magni-


Another field evaporation rate expression was intro-

duced by Tsong26) and Muller(34) which describes the abso-

lute rate of the desorption of single adatoms as a function

of the electric field. This rate is

1 -1
ke(F) = sec (15)

where N is the number of pulses of width T and field

strength F required to evaporate an adatom. Since the

evaporation rate, ke, increases eight orders of magnitude

by raising the field by approximately 10 percent,(26'34)

it is very important to accurately determine the field

strength experienced by the individual adatoms. Consequently,

expression (15) becomes very important when an evaporating

pulse of known strength and duration is utilized with the

atom probe field-ion microscope26) to identify individual

adatoms or lattice atoms. This method of field evaporation

is very versatile because it provides an accurate measure of

the field strength required for evaporating individual atoms.

For instance, Plummer and Rhodin(35) reported that the re-

quired field intensity decreased about 10 percent when scan-

ning from the center to the edge of the tungsten (110) plane.

Field Evaporation of Alloys

Theoretical considerations pertaining to the field

evaporation of dilute alloys differ from the previously

described pure metal case. Observed field-ion images of

dilute alloys reveal a lack of surface regularity and sharply

resolved high index planes. This suggests that the presence

of the solute atoms in the solvent lattice either creates

preferred field evaporation or the solute atoms are merely

invisible. Brandon(36) considered this effect in terms of

thermodynamic quantities. These include AHs (heat of

sublimation of pure solute), Hs (heat of solution), Ed

(dissociation energy of diatomic gas rather than the mon-

atomic solute ion), and HA (heat of adsorption). From

expression (1), which represents the activation energy re-

quired to remove an ion of charge n from the metal lattice,

Brandon determined from the image-hump model that Q =

(n e 3Fn /2, providing Qn was assumed a minimum and the

polarization and penetration corrections were ignored. F

is the field strength required to evaporate the ion at 0K.

The sublimation energies for the monatomic solute ion and

the diatomic gases are approximated by relations (16a, b

and c) respectively.

A A ART (16a)

A A0 + H (16b)

A E Ed + HA (16c)

By substituting these calculated sublimation energies and

the predicted ion charges (n 3/2 = n is a minimum) into

expression (1), the evaporation field, Fn, can be determined.

With the available values of Hs and HA the general condition,

HA > -AHF > Hs, was noted where -AHF is the selected com-

pound's heat of formation. This indicates that if the solute

atom has a lower free energy at the surface as compared to

the bulk, the difference in energy must appear as a reduced

average binding energy for nearest neighboring solvent atoms

at the surface when the migrating solutes reach the surface.

Therefore, as field evaporation occurs, the solvent atoms

will evaporate preferentially to expose the underlying im-

purity atom. This effect was observed by al.(37)

To provide more physical insight into this process,

Brandon(36) selected iron as the solvent and calculated the

field strengths expected to evaporate the various solutes

normally present in iron. The results predicted that either

the solutes evaporated with a preferred ionic charge or

remained in the lattice. More important, however, his cal-

culations indicated that nonmetallic solutes, such as C, 02

and N2, required extremely high evaporation fields and with-

out another evaporation mechanism would remain on the iron

surface. One of the proposed processes considered the field

evaporation of a molecular ion composed of the metal atom

and the nonmetallic solute atom. Brandon 36) presented the

following condition for such an event:

Am (Ed+HA < d (17)
n g n
n n

where Am is the sublimation energy of the molecule from the

metal lattice, E I is the total ionization potential of the
n g
nonmetallic impurity, and Z I is the total ionization poten-
n d
tial of the metal-impurity molecule. The appropriate poten-

tial energy scheme based on the assumption that the metal-

impurity complex will have the identical charge as the metal

ion is shown in Figure 4. Presently, limited data for these

thermodynamic quantities prevent a detailed prediction of

whether molecular ions are either field evaporated or dis-

sociated. However, experimental results from the analyses

of field evaporated species have produced evidence of the

existence of molecular ions. For instance, multiply charged



MsA nf -




Ms+A ads





Ms+A ads



I *I


Figure 4. Energy scheme for field evaporation of a
nonmetallic impurity as: (a) a monatomic
ion and (b) a molecular ion. (36)


oxide and nitride molecular ions of metals, CuO NiO+,

Fe0 FeN2 WO2 and WN were observed by either magnet-

ically focused mass spectrometers(6) or by atom-probe field-

ion analyses. 9)

Experimental Verification

Rate Sensitivities

Examination of the Arrhenius rate equation, ken

Vn exp{-Qn/kT}, and the activation energy for the image-hump

model including the polarization factor

Qn = (n3e3F)12 + 1 (aa-ai) F2 (18)

reveals two means to vary the rate of evaporation. An in-

crease in temperature with both the electric field and acti-

vation energy constant will increase the evaporation rate.

Also, as the field is increased, the evaporation rate

will increase providing the temperature remains constant

and the activation energy decreases. The sensitivity of

the evaporation rate to the field at a constant temperature

is designated field sensitivity,(32) while the sensitivity

of the evaporation rate to the temperature at a constant

field determines the temperature sensitivity. Both terms

are classified as rate sensitivities. Another experimental

approach introduced by Brandon(32) utilizes the variation

of both temperature and field to keep a constant evapora-

tion rate. Mathematically, this method is

related directly to the first two sensitivities and the
experimental value is designated the rate sensitivity.

The Arrhenius relation is differentiated to determine
the theoretical values for these various rate sensitivities
to allow comparison with the experimental work. The field
sensitivity, ST, is determined as follows(20,38):

dd kn(ken) F d n(ken) -FdQ1
T = d an(F) T dF T -T F
The temperature sensitivity, SF, is obtained by differen-
tiating the Arrhenius equation with respect to an(T) and is
shown to be equivalent to

d n(ken) Qn n (
SF = d n(T) F T [ =n (20)

The third sensitivity, Sk, maintains a constant evaporation

rate and is defined as

=S -d n(F) (21)
Sk -d n(T) kn (21)

All three of these sensitivities apply to both the image-
hump and intersection models by merely choosing the appro-
priate field evaporation activation energy relation.
Tsong ) assumed that the pre-exponential term v in
the Arrhenius relation was a slowly varying function of
field and temperature as compared to the exponential depen-
dence of ken for these parameters. Thus, by approximating
vn as a constant one can obtain

Qn = k T Fn en-
n Een

d Q = k T Ln Vn d kn(T) k(22)
Kn T en

Rearranging and solving for kn(vn/ken) results in

An = -ST Sk (23)

In principle, the experimental determination of both

the rate sensitivities, ST and Sk, from relations (19) and

(23), respectively, will allow Q and vn to be measured.

Brandon 32 followed such an approach and calculated a v

value of 5x1010 layers/sec for tungsten from experimental

ST and Sk values. This value was the same order of magni-
tude as the atomic vibrational frequency (1012-1013/sec) if
one uses MUller and Tsong's39) suggestion that the number

of atoms on the surface exposed to an evaporating field may

be as much as three orders of magnitude less than the total

number of surface atoms. MUller and Tsong(39) used this

same technique with tungsten specimens and obtained values

of 6x10 layers/sec. The discrepancy between the two values

indicates the magnitude of experimental error and emphasizes

the need for more experimental and theoretical work.

Another means to determine vn experimentally involves

using temperature sensitivity measurements and relation (16).

The most difficult part in this experiment involves keeping

the field constant as the tip evaporates. Using blunt tips

helps because as the evaporation occurs the average radius

of curvature will not be reduced as drastically as if sharp

tips were used. Perhaps an inert gas with a known "thresh-

old ionization field" could be used to provide a means to

monitor the field strength at the tip. 40)

Experimental determination of Q is achieved also by

measuring the evaporation field and the ionic charge. The

evaporation field can be determined from relation (10) pro-

viding the average field at the tip is known. The applied

potential and tip radius are directly related to the average

field by relation (11) and can be experimentally measured.

The tip radius can be established from transmission electron

microscopy41) (TEM) and scanning electron microscopy (SEM)

profiles of the tip and by averaging the local tip radii

measured from field-ion micrographs.(42) Tip radii can also

be measured from average work functions determined from

Fowler-Nordheim plots(22) of log(I/V2) vs 1/V which yield

straight lines whose slopes are proportional to 3/2/F.

Since this field emission technique requires ultra-high

vacuum, it usually has limited application with ordinary

field-ion microscopes. Although the ionic charge was nor-

mally assumed, accurate values of n can be determined cur-

rently with the atom probe field-ion microscope. Perhaps

if these different experimental means are used to accurately

measure the evaporation field and ionic charge, a more

reliable measure of Q and vn can be obtained.

Imaging Gas Effect

Although the previous discussion considers the evapora-

tion to occur in a vacuum, experimental observations are

made usually in the presence of an imaging gas. The evap-

oration field was observed with tungsten to decrease as the

helium gas pressure was increased. This effect was more

obvious when neon was used. Nishikawa and Miller(43)

accounted for this effect by stating that a sufficient

probability existed for the imaging gas atoms to penetrate

the high field region surrounding the tip without ionizing.

When they strike the surface atoms of mass M with an aver-
MV2 1 2
age kinetic energy, my2- 2 kT + aF2, they transfer a

considerable fraction, 4mM/(m+N)2,of their kinetic energy.

This gas impact promotion effect on lowering the evaporation

field was formulated by Brandon(32) as

-AF 2kT* n (24)
F Q- n (24)
o e

where v* is the bombardment rate and T* is the effective

temperature which relates the tip and imaging gas tempera-

tures with the classical accommodation coefficient. Since

the bombardment rate is proportional to the image gas pres-

sure, the observed evaporation fields were unchanged, if

ke/p was constant, for various W, Mo and Pt tips.43) In

addition, this imaging gas impact effect was observed to

alter the evaporated end form during the transition from a

vacuum evaporation and permitted the preparation of

uniformly shaped tips. Also, if hydrogen was introduced to

the tip, it promoted field evaporation and ionization at a

significant reduction in field.(44) Consequently, this

hydrogen promotion effect afforded another promising exper-

imental technique to observe nonrefractory metals with the

field-ion microscope.

Recent atom probe F-IM results were obtained for the

promotion of field evaporation with noble imaging gases and

suggest that the gas impact mechanism is insufficient. For

instance, the adsorption of the imaging gases(45) and the

detection of molecule ions formed from the image gas and

surface metal atoms( 7,45) was attributed to an electronic

rearrangement whereby the binding of the metal atom is re-

duced when the surface molecule is formed by the overlap of


Pulsed Field Evaporation

Pulsed field evaporation is achieved by superimposing

a high voltage pulse on the dc image voltage to produce a

sufficient field strength to evaporate the surface atoms.

This technique is used in conjunction with the atom probe

F-IM which measures the time-of-flight for the field evapo-

rated ionic species from the tip to the ion detector. From

this value the chemical nature of the ionic species can be

identified. The results obtained from the atom probe

analysis were not predicted by the previously described

field evaporation theory. For instance, unexpected multiply

charged metallic ions and their oxide and nitride compounds

were identified.(9'11,13) The observation of these higher

charged ions appeared to be enhanced significantly when the

evaporation rate was increased to 109 atom layers/sec.

Other significant results include the observation of noble

imaging gas adsorption(1545) and the evaporation of molecu-

lar ions formed with the noble image gases and metal atoms.

Pulsed field evaporation was used by Hren et al. (46)

to produce images of tungsten at 780K which contained bright

bands of variable width parallel to the <111> zones. The

influence of the experimental variables was studied,while

the amplitudes of the pulses were adjusted to achieve a con-

stant evaporation rate, for example, when 5 to 10 pulses

were desired to remove one (110) plane. A two-mechanism

model for field evaporation was proposed to explain the

observed difference in the field evaporated end forms. The

high field evaporation rates produced a more hemispherical

tip, while the slow rates developed an anisotropic-shaped tip.

These reported experimental results for pulsed evapora-

tion rates, particularly the atom probe analysis, indicate

that the current theoretical models for field evaporation

still need modification. The importance of the field evapo-

ration rate for both the pulsed and dc evaporation tech-

niques prompted Tsong and MUller(34) to develop a detailed

procedure for measuring the relative field evaporation rates.

When these proposed methods are tediously followed, fewer

discrepancies will exist and the field evaporation theory

can become more quantitative.

Atom Probe F-IM Application

Since the experimental observations of field evapora-

tion seem contradictory, the need for more experimental data

is obvious. However, the complexity of the field evapora-

tion process requires that sophisticated experimental tech-

niques and intricate instrumentation be developed to gather

useful experimental data. The capabilities of the atom

probe field-ion microscope (745,47) merit its potential

application to this particular problem.

The initial accomplishment of the atom probe was the

detection of the highly charged metallic ions and their

oxide and nitride compounds.19,10,11) Also, the amount of

these unexpected ionic species was observed to increase as

the field evaporation rate increased. Another unexpected

result was that the noble imaging gases, helium, neon and

argon, adsorbed to the metal surface and that their required

evaporation field was sufficient to evaporate the metallic

atoms simultaneously. Another important observation with

the atom probe was the detection of field evaporated

molecular compounds(7,14) comprised of the imaging gas and

surface metal atoms. Thus, the atom probe appears most

useful in obtaining data about the influence of the

environment on the field evaporation of surface atoms,

particularly on the influence of polarization, solute

atoms, the geometric location of atoms, and the presence

of imaging gases. The atom probe enables accurate measure-

ment of both the evaporation field and ionic charges of the

evaporated species required to make rate sensitivity deter-

minations of Q and vn. In addition, if the atom probe has

an ultra-high vacuum capability, work function values for

selected regions of the specimen surface can be obtained

and may be correlated to the electronic configuration of

the surface.




Difficulties encountered in identifying the chemical

nature of specific surface atoms in anF-IM image served as

an incentive for Miller and Panitz(8 ) to develop the atom

probe F-IM in 1967. This instrument is the combination of

a standard F-IM and a time-of-flight mass spectrometer.

The objective of this instrument is to select and identify

individual surface atoms which have been field evaporated.

The atomic resolution of the F-IM and the single-atom

analysis by the time-of-flight mass spectrometer has created

one of the most powerful tools currently available to study

the nature of surfaces. For instance, other devices cur-

rently being used for surface studies, such as the electron

microprobe or the LEED-Auger spectroscope, analyze a minimum

of 109 atoms.

The principle of operation of the atom probe is illus-

trated in Figure 5. The specimen is positioned in the atom

probe's specimen holder and imaged normally on a phosphorus-

coated screen with a central probe hole. By manipulating

the specimen holder, the image of an individual surface atom


aW o






or molecule can be aligned with the probe hole. By super-

imposing a short time, high voltage pulse on the steady dc

imaging potential, the surface atoms can be field evaporated.

During the field evaporation process the surface atoms or

molecules will lose one or more of their electrons and the

resulting ionic species will accelerate radially toward the

phosphorus screen. If an entire layer of surface atoms were

stripped off during field evaporation, only those ions whose

trajectories are aligned with the probe hole would be allowed to

enter the drift tube. These ions will travel down the drift

tube and strike the detector which also is aligned with the

screen's probe hole. By determining the time interval

between the initial evaporation pulse and the detector sig-

nal, the time-of-flight of the ions can be measured.

Since the ion travels from the tip to the detector in

a region essentially at ground potential, its kinetic energy

can be expressed as

1 2
mv = neV (25)

where m and v are the mass and velocity of the ion, n notes

the degree of ionization, e is the electronic charge and V

is the sum of the high voltage pulse and dc imaging poten-

tial, i.e., V = Vdc + Vpulse Also, the ion reaches its

final velocity within a few radii away from the specimen's

tip and can be determined from

v = d/t ,


where d is the distance from the tip to the detector and t

is the measured time of flight. Units normally used are:

kilovolts for V, centimeters for d and microseconds for t.

Substitution of (26) into (25) to eliminate v results in the

following relations for the mass to charge ratio of the ion:

(m) = 2eV( )2 (27a)

) = 2e(Vdc + Vulse) (2 (27b)

The preceding relations reflect the assumption made
that the positive high voltage pulse be superimposed on the
dc imaging potential to achieve field evaporation. This
method has been supported by Miller, Panitz and McLane(9,48)
Stolt,(49) Tong(50) and Hren and Stoltz.(51) However,

another technique that achieves a similar field evaporation

effect is reported by Brenner and McKinney.(52) These

researchers used a negative voltage pulse applied to an open-

ring cathode located in front of the field-ion tip and kept
the screen at a potential different from ground. This means

that the kinetic energy of the ion is equivalent to only the

dc imaging potential, i.e., Vdc = V.
Although the principles involved with the atom probe
F-IM appear relatively simple, the complexity of design,

construction and instrumentation, plus the associated ex-
penses have prevented widespread utilization of this instru-
ment for surface studies. Prior to the publication of the

current review articles by Brenner and McKinney(52) and

Miiller and Tsong(11) which describe various technical design

and construction problems related to the atom probe, the

development and availability of the instrument depended

upon the ingenuity of the individual researchers. Conse-

quently, at this time there are only three research facili-

ties in the world which utilize the atom probe.

Several material problems that exist in current F-IM

studies appear amenable to study with the atom probe. For

instance, work has been initiated for the identification of

bright spots in F-IM images attributed to adsorbed gases,

deposited metals, self-interstitial atoms, bulk impurities

and corrosion products, investigation of short range order

and visibility problems in alloys, interpretations of lat-

tice defects and determination of solute atom distribution.

From a theoretical standpoint, important data can be ob-

tained to improve current field evaporation theories. In

particular, the charge and composition of field evaporated

and desorbed species can be determined for single particles.

Technical Design and Material Selection

One of the most important capabilities desired in the

atom probe is a bakeable, ultra-high vacuum system. When
pressures of 1x10 Torr and less are obtained, the atom

probe can be utilized in both the field ionization and field

emission modes. Another vacuum consideration includes the

need for a titanium sublimation pump to remove gaseous im-

purities and to maintain the "clean" surfaces produced by

field evaporation. To prevent scattering of the field

evaporated ions it is necessary to differentially pump the

drift tube below the imaging gas pressure.

The image formation and alignment are critical factors

in the atom probe. To provide possibilities for imaging

with a wider range of materials and imaging gases, a micro-

channel-plate image intensifier device is needed. In addi-

tion, an optically flat viewing screen, imaging gas ports

and observation of the image are required. To improve the

image resolution,provisions must be made for cryogenically

cooling the specimen and surrounding imaging gases. Utmost

consideration must be taken to align the specimen simul-

taneously with the probe holes in the viewing screen and

mirror, orifice of the drift tube and the time-of-flight

detector. In addition, to allow selection of individual

image spots it is necessary to locate the specimen tip at

the center of a goniometer-type assembly that can be tilted

approximately 450 and rotated 360.

The electrical connection for the high voltage dc

potential must be well insulated and connected to the high

voltage pulse transmission line. Critical impedance match-

ing of the high voltage pulse to the tip is required to

provide a well-defined pulse shape and amplitude at the tip.

Since single particle events must be monitored, a very

sensitive detector is needed for the time-of-flight mass

spectrometer. In addition, appropriate electronic cir-

cuitry and a storage facility are required to measure

accurate time-of-flights for the field evaporated ions.

In order to obtain a high rate of data collection

despite frequent specimen failures, it is highly desirable

to provide convenient specimen exchange and rapid pump down

to a vacuum below 3x109 Torr. Also, extra ports must be

available for pressure gauges, viewing ports, means to

irradiate the tip and means to evaporate material onto the

"clean" tip.

The actual construction of the atom probe represented

a compromise between the desired design features, material

selection and fabrication limitations. In general, any one

or a combination of the following materials were used for

each component in the completed unit: Type 304 stainless

steel, OFHC copper, ultra-pure aluminum, sapphire, glass

and gold. Techniques used to fabricate the precision

machined components included: gas tungsten-arc welding in

protective argon atmosphere, resistance spot welding, high

temperature brazing in either air or protective hydrogen

atmosphere and electroplating. Following their fabrication,

the components were separately cleaned in the following

consecutive solutions: toluene, boiling deionized water,

acetone, methanol and ethanol alcohol (85 and 99 percent

solutions). Ultra-sonic cleaning was utilized when the size

of the component permitted. After drying the components

with a heat gun, they were assembled in a clean area and

all handling was done with protective gloves. During

assembly, any open ports were closed to prevent possible

dust contamination.

Description of Atom Probe

From the previous discussion concerning the require-

ments for the atom probe, it becomes evident that each com-

ponent represents a particular problem in design and fabri-

cation. Consequently, the important components of the con-

structed atom probe, shown in Figure 6, will be discussed

individually in the following order: 1) ultra-high vacuum

system, 2) specimen holder, 3) internal microchannel-plate

image intensifier, 4) time-of-flight detector, 5) high

voltage feedthrough, 6) high voltage supplies, and 6) the

electronic circuitry used to measure ionic time-of-flights.

Vacuum System

A metal system comprised of Type 304 stainless steel,

with the necessary glass viewports and ceramic high voltage

feedthroughs, was designed and constructed to provide a

bakeable, ultra-high vacuum system. Figure 7 represents a

block diagram of the completed system. Two Varian vacsorb

pumps were installed to. rough down the system from atmospheric

Figure 6. Atom probe field-ion microscope.







20-400 L / SEC.



Figure 7. Ultra-high vacuum system for atom probe.

to less than 10 microns pressure. Besides physically adsorb-

ing the gaseous constituents (except inert gases) on the

liquid N2 cooled xeolite, these pumps afforded the advantage

of eliminating the backstreaming of a mechanical forepump's

oil. Coupled with the vacsorb pumps are four 20 A/sec Ultek

DI-pumps. Three of these noble gas ion pumps are located

on ports in the atom probe body and the remaining one is

located on the drift tube. The three pumps provide maximum

pumping speed for the large atom probe body which houses the

internal microchannel-plate image intensifier and specimen

holder. It must be noted that the 35p diameter microchannel-

plate and the specimen holder's flexible copper cooling

cable represent approximately 10 square meters of surface

area. Because of this enormous area the three pumps were

necessary to insure sufficient pumping capacity to reduce

the atom probe body to UHV pressures within a reasonable

time. From Figure 7, it is noted that two of the Ultek DI-

pumps can be isolated from the system while remaining at

UHV pressure. This feature reduced significantly the time

needed to pump down the system from atmospheric to UHV


Once the vacuum system proved capable of reaching

pressures in the 10-0 Torr range, then it became adaptable

for both the field ionization and field emission modes.

Since both modes required final specimen tip cleaning by

field evaporation and field ionization, necessary design

features for imaging were considered. First, it was desir-

able to keep the drift tube pressure about two orders of

magnitude less than the image gas pressure in the atom probe

body to prevent scattering of the field evaporated ionic

species. To accomplish this differential pressure between

the two sections, a 2.8 mm diameter orifice of length 5 cm

was machined for the drift tube. In addition, the fourth

20 /sec noble gas ion pump was attached to the drift tube

section as shown in Figure 7. The second consideration for

imaging the tip involved the means of introducing the imag-

ing gases to the atom probe body. Three Varian bakeable

leak valves were used. Two of these were located purposely

so that the imaging gases must pass over freshly deposited

titanium before entering the atom probe body. This feature

allowed the gaseous contaminants within the imaging gases

to be initially gettered by the titanium to reduce their

subsequent coverage of 'the tip. Imaging gas mixtures

were possible since the vacuum gauges in the atom probe body

allowed monitoring of the partial pressure of each constitu-

ent. In addition to introducing imaging gases, the third

leak valve can be used to return the system to atmospheric

pressure with ultra-pure dry nitrogen. This feature was

important for minimizing the time needed to reach UHV con-

ditions after the specimen exchange.

The third important factor of imaging that was included

in the vacuum system was the titanium sublimation pump. As

reported by several researchers, (710 152) the titanium

sublimation pump provides means to eliminate gaseous con-

taminants and to maintain the desired imaging gas pressure.

From Figure 7 it is observed that the Varian titanium sub-

limation pump was enclosed in a water-cooled chamber and

its 6-inch diameter flange was directly coupled to the atom

probe body. Due to this large conductance path a conserva-

tive 400 A/sec pumping speed was realized. Once the field

evaporated tip has developed a "clean" surface, the

titanium pump also provided means to retain its cleanliness

when the field was removed and during the time required to

reach the necessary UHV pressure for the field emission


Other vacuum considerations featured in the atom probe

include available view ports for observing the image and

specimen holder, specimen exchange port, monitoring of atom

probe body with milli-torr and nude ion gauges, flexible

coupling of drift tube and atom probe sections to alleviate

alignment of time-of-flight detector with the precision

aligned specimen tip, image intensifier probe holes and

drift tube orifice, high voltage feedthrough ports, cryo-

genic cooling port, extra ports for specimen irradiation

and evaporation sources, plus electroplating gold on the

interior surface of atom probe body to maintain vacuum

cleanliness. In addition, with a special procedure developed

for the removal and insertion of the atom probe specimens,

routine times of approximately four hours are required to
return the atom probe to pressures of 1x10 Torr and less.

This short time actually doubles the number of specimens

normally observed, as reported by workers with similar in-

struments,(7,9,10,16) and emphasizes the efficiency of the

vacuum system for the atom probe.

Specimen Holder

Since it is desirable to maneuver any section of the

tip's image to the probe hole, the specimen holder must be

capable of tilting and rotating the tip while it images.

This requirement meant that the specimen holder must provide

simultaneously cryogenic cooling, electrical insulation,

UHV vacuum conditions and the necessary mechanical manipu-

lation of the specimen tip. This difficult design problem

was accomplished initially with the specimen holder shown

attached to the support yoke and upper flange in Figure 8(a).

Figure 8(b) shows the specimen holder linked to the mechan-

ical manipulator. The physical arrangement of the specimen

holder and manipulator is shown schematically in Figure 9.

With the initial specimen holder the liquid hydrogen

was poured in the top of the flange into the stainless steel

cold finger and passed down through the flexible stainless

steel bellows to the specimen holder. In the specimen holder

the liquid hydrogen cooled the entire holder and particularly

the tip which was mounted in an OFHC copper holder that was

intimately secured between a pair of sapphire disks. The

(a) Specimen holder attached to support yoke and upper

(b) Specimen holder linked to mechanical manipulator.

Figure 8. Initial atom probe specimen holder.






P .

nrn r=

tn to



*14 r

p4 ,-




electrical lead was incorporated into the manipulator

assembly with a coaxial cable extending solidly from the

power supply to the ball joint. The tip was loaded by remov-

ing the entire specimen holder on the large 8-inch flange.

To do this the high voltage lead housed in the manipulator

rod and ball joint assembly was disconnected from the speci-

men holder. Once the specimen holder was outside the

vacuum chamber, the specimen tip was installed and aligned

by optical means to center the tip within 0.3 mm.

Unfortunately, after limited service the stainless

steel bellows used for cryogenically cooling the specimen

holder developed vacuum leaks. Also, a sufficient amount

of stress was developed in the manipulator assembly to

fracture the ceramic insulators that shielded the 30 kV dc

potential. In addition, the time required to exchange

specimens prevented more than one specimen to be examined


Consequently, several design modifications were inves-

tigated and tested prior to selecting another workable model.

This specimen holder and mechanical manipulating assembly

are shown in Figure 10(a) and (b). In Figure 10(a) the

stainless steel bellows used to cryogenically cool the

specimen holder was replaced with a flexible bundle of gold-

plated OFHC copper wires. Attached to both ends of the

flexible copper wires were two machined OFHC copper pieces

that were brazed with a.BAG-19 (silver-copper alloy) in

(a) Specimen holder.

(b) Specimen holder linked to mechanical manipulator.

Figure 10. Final specimen holder and mechanical
manipulating assembly.

protective hydrogen atmosphere. These end pieces were

fastened with respect to the specimen yoke (OFHC copper)

and the specimen holder. Because of the intimate contact

of these end pieces, excellent thermal conductance from

the stainless steel cold finger to the specimen holder was

assured. In addition, if extended manipulation of the

specimen holder caused the flexible copper wires to work

harden and fracture, another flexible copper bundle with

similar machined end pieces can be installed without remov-

ing the specimen holder from its precision aligned position

in the atom probe body.

The mechanical manipulator assembly was modified to

eliminate the weak ceramic electrical insulators. In fact,

the new internal slotted tube and outside manipulator rod

were machined from stainless steel and were copper brazed

in dry hydrogen to the precision machined stainless steel

ball joint. Since another ball joint connection was located

on the back portion of the specimen holder, the necessary

linkage between the ball joint in the outer flange and the

specimen holder was provided with a push rod sliding in the

slotted tube. Due to the location of the two ball joints,

the desired motion of the specimen holder was achieved with

a minimum motion of the outside manipulator rod. Vacuum

integrity was maintained with a flexible stainless steel


The high voltage lead was introduced to the specimen

holder from a Ceramseal 40 KV vacuum feedthrough. Electrical


52 -

insulation of this flexible OFHC copper lead was provided

with specially designed glass beads. The flexibility of

this insulated lead enabled the specimen holder to be tilted

a minimum of 450 and rotated 3600 by the external manipulator

rod while imaging the specimen tip. Precise translational

movement of individual image spots to the probe hole was

realized with the two-axial micrometer assembly. This

aluminum structure was mounted to the large 8-inch end port

on roller bearings for easy rotation. Sequentially, the speci-

men holder is manuevered to a central position and the

micrometer stage is clamped. Next, the two micrometers are

adjusted to align the desired image spot with the screen's

probe hole. Continuity of the coaxial shielding was main-

tained by keeping the specimen holder and chassis at common

ground, while isolating the tip holder and flexible lead.

The tip holder was isolated electrically from the specimen

holder's shield with 2.5 mm thick sapphire disks. These

disks also provide thermal conductance to the tip, since

they are in intimate contact with the cryogenically cooled

specimen holder.

Another important aspect of this particular specimen

holder is illustrated in Figure 11. In this figure one

notes that the specimen holder and manipulator rod assembly

are biased into a position that faces the small port adja-

cent to the large view port. This means that the specimen

tips can be inserted and removed through this port rather

Figure 11. Specimen holder and manipulating rod
assembly in biased position.

than removing the entire specimen holder as required pre-

viously. The biggest gain in this modification was the

reduction in time needed to exchange specimens.

Special care was taken to assure precise tip alignment

with the probe holes in the microchannel-plate, screen and

mirror and the drift tube orifice. An alignment stand wi-th

a flexible stainless steel bellows was mounted on the large

upper flange of the atom probe body to support and provide

precise alignment of the pivoting specimen holder's yoke.

With an optical source located outside the drift tube ori-

fice the specimen tip was positioned within 0.1 mm of the

previously aligned probe holes in the microchannel-plate

image intensifier and reflecting mirror. Further electronic

alignment was undertaken during field evaporation and is

described later.

Internal Microchannel-plate Image Intensifier

A microchannel-plate image intensifier was constructed

and mounted in the mirror assembly unit located inside the

atom probe body. This device improves the observation and

photography of the helium-, neon- and argon-ion images plus

it affords the possibility of imaging a greater number of

materials. Similar devices have been reported by Turner

and Southon53) Von Oostrom and Manley,(3) Brenner and

McKinney(52) and Hren et al.54) Briefly, the principle

of its operation is illustrated in Figure 12. The ions



1 1 1 1II I III I11 1I I 1 I I I I I I I I 1 1 1 1 11
+17Kv t% e- (103)
+3Kv ____lllllllllilll/!_

Figure 12.



Principle of microchannel-plate image

repelled from the specimen tip during field ionization

strike the incident surface of the microchannel-plate and

generate secondary electrons. These secondary electrons

cascade through the individual channels of the plate and

are proximity focused onto the ground phosphor screen.

Typically, measured gains of 102 to 103 are reported(55)

for these microchannel-plate irrespective of the imaging

gas used.

Figure 13(a) and (b) depicts,respectively, the device

schematic and the assembled unit prior to its installation

in the atom probe body. The 37.5 micron diameter micro-

channel-plate and phosphor-coated screen were mounted in a

gold-plated stainless steel holder. Four stainless steel

bolts fastened the device's components. Electrical insula-

tion was provided for these bolts with precision ground

glass cylinders. Because the emerging electrons are-

proximity focused, the glass insulators were carefully

ground to maintain a 1.8 mm separation between the micro-

channel-plate and phosphorus-coated screen. Hat shaped

glass insulators were used to shield the four stainless

steel springs needed for supporting the phosphor screen.

The high voltage leads extending from the Ceramseal MHV

vacuum feedthroughs to the mounted device were insulated

with glass to prevent possible arcing when the 1 to 1.2 KV

and 3 to 5 KV potentials were applied to the bottom of the

microchannel-plate and screen, respectively.


(1-X "' I



Figure 13. Microchannel-plate image intensifier for
atom probe.


HR$M Mm P____isss aia

To assure precise alignment of the centered 1.5 mm

diameter holes in the microchannel-plate, phosphorus-coated

screen, aluminum-coated reflecting mirror and drift tube

orifice, a particular procedure was devised during the

device's assembly and mounting in the mirror support assem-

bly. Initially, the 1.5 mm diameter holes in the micro-

channel-plate and phosphorus-coated screen were aligned with

a light source prior to fixing their positions in the holder.

Once the holder was tightened, the device was mounted in the

mirror support cylinder. Two threaded supports with oppo-

sitely positioned spring supports in this polished aluminum

cylinder were used to position the probe holes in both this

device and the reflecting mirror. Thus, when the light

source was located at the drift tube orifice, the holes in

the reflecting mirror and microchannel-plate image intensi-

fier could be aligned with respect to each other and the

drift tube orifice simultaneously. Following this align-

ment, the complete mirror assembly was installed in the

atom probe body and the movable specimen holder was properly


Bendix Spiraltron Detector

The detection of either selected field-ion image points

or field emission regions is an important requirement for

the atom probe. The Bendix Spiraltron electron multiplier

was chosen for this function. Bendix(56) developed this

detector to be windowless and sensitive to electrons, ions,

short wavelength ultraviolet radiation and soft x-rays with

the spectral range of 2 to 1,500 A. Laboratory and space

experimental results(57) indicated that these devices have

a minimal gain of 5x107 with a 3 KV dc applied potential

difference, low background noise, narrow pulse height dis-

tributions and low power consumption. In addition, the

special semiconducting interior coating of the Spiraltron

can be exposed to air when not operating without degradation

in performance.

The large conical input aperture (10 mm diameter) of

the Bendix Spiraltron CEM-4028 presented a slight accuracy

problem in defining the tip-to-detector distance. This

uncertainty was reduced to less than 1 mm by tilting the

detector 450 with respect to the incident radiation. This

tilting, however, reduced the effective input area by 30

percent of its maximum value (perpendicular to incident

radiation). The Bendix Spiraltron was housed at the end

of the drift tube in a standard Varian tee. The detector

was supported by connecting its two nickel leads to two

vacuum sealed MHV high voltage feedthroughs. Alligator

clips were used to allow optimum positioning of the input

aperture with respect to the drift tube orifice, reflecting

mirror and microchannel-plate image intensifier probe holes and

the specimen tip. A third MHV high voltage feedthrough

was connected to an OFHC copper anode that was

carefully positioned within 1 mm of the output orifice of

the Spiraltron CEM-4028.

The alignment of the Spiraltron's input aperture with

the previously aligned probe holes and specimen tip is very

critical and influences the detector's efficiency. This

alignment was performed initially by optical means. A

blank copper gasket with a central 1 mm diameter hole was

attached to the input port of the vacuum connection which

supported the detector and a light source was positioned

behind it. Once the detector's incident aperture was cen-

trally positioned, the Spiraltron's supporting stainless

steel flange was rotated 450 and fixed permanently. Next,

the tee was attached to the drift tube and a light source

was aimed at the 2 mm diameter drift tube orifice from the

atom probe body. Since the drift tube was connected to the

atom probe body and vacuum roughing lines with flexible

metal bellows, the end of the drift tube could be manuevered

until the input aperture of the Spiraltron was optimally

located. This position of the detector was fixed rigidly

with the adjustable support tripods. In the following

chapter, the final laser and electronic alignment of the

detector and its calibration will be discussed.

Pulse and dc High Voltage Feedthrough

Two primary concerns in designing and constructing

this atmospheric high voltage feedthrough were to minimize

the uncertainty of the mass resolution of the field evapo-

rated species and to reduce the severity of tip "flashing."

In order to reduce the reported uncertainties in the mass

resolution,(48) it is desirable to have a square pulse with

a very fast rise time and short duration. In general, the

time of flight for the field evaporated species is 5 micro-

seconds, so pulse durations of 50 nanoseconds or less are

necessary. Also, the rise time of the square pulse signal

should be less than 1 nanosecond to preserve its sha p lead-

ing edge.

Since the high voltage pulser, Microwave Associates

Nanosecond Pulser, is a transmission line type, it was

necessary to match its impedance with the high voltage feed-

through circuitry. Consideration was also given to the

fact that the line terminated with an open end, the speci-

men tip, which created reflections up and down the line and

further complicated the mass resolution of the field evapo-

rated process. Sharp bends and discontinuities of the

transmission line and shielding were avoided to reduce

shunt capacities and lead inductances.

The circuitry for this high voltage feedthrough is

shown in Figure 14(a). Resistor, R1, provides the proper

impedance termination (50) of the transmission line of

the Microwave Associates Nanosecond Pulser. Capacitor, C1,

is a coupling capacitor for the evaporating pulse between

the high voltage pulser and the specimen tip. Resistor R2






Figure 14. Atmospheric high voltage feedthrough for
atom probe.

is included in this circuit to reduce the amount of dis-

charging of capacitor C2 through the specimen tip upon its

"flashing." For instance, if the tip "flashes" the time

constant is considerably greater with R2 since it is deter-

mined by multiplying C2 times the sum of R2 and the tip

resistance rather than just C2 times the tip resistance.-

When the tip flashes, the potential decreases and slowly

builds up to its initial discharge potential. The time re-

quired for this increase is given by T = C2R3 = 10 seconds.

Normal operation indicated that this estimated time is con-

servative. Experimental measurements made with the electro-

static voltmeter indicated a recharging time of 40 seconds.

The primary purpose of the 40 KV, 1,000 pf capacitor

C2 is to isolate the high voltage pulser from the 0-30 KV

dc high voltage supply. Resistor R3 can withstand a 40 KV

potential and maintain a 1010 ohm resistance. This resistor

limits effectively the current available to the tip and

diminishes catastrophic "flashing" of the tip. This limited

current is equivalent to the applied tip potential divided

by R3 (1010 ohms).

Other steps taken to maintain the field evaporation

pulse's sharp leading edge and short duration are shown in

Figure 14(a). The cross-hatched lines were minimized in

length to reduce possible shunt capacitances and lead in-

ductances. Also, the transmission line contained a minimum

number of interfaces and sharp bends. The continuous

shielding that housed this circuitry was made from OFHC

copper and was terminated securely to the atom probe body.

Since this feedthrough must be capable of withstanding

potentials of 30 KV in atomsphere, it was necessary to line

the interior surface of the copper tee with acrylic tubing

plus construct special high voltage connectors insulated

with Teflon to prevent discharging. Other features of this

component include: quick connect and disconnect coupling

with the UHV Ceramseal 40 KV feedthrough, unit can be easily

removed while baking out atom probe body, and provisions for

exchanging the electrical components to achieve maximum

restoration of the incident high voltage pulse at the tip.

The completed unit is shown attached to the atom probe body

in Figure 14(b).

High Voltage Power Supplies

Generation of high voltage square pulses with durations

less than 50 nanoseconds can be accomplished by at least two

methods.(52) After evaluation, it was decided to purchase

the commercially available Microwave Associates Nanosecond

Pulser. This particular model 961E produces square high

voltage pulses by discharging a transmission line into the

output line with a mercury switch. The electrical pulses

are produced with an impedance level of 51 ohms either

manually or at a repetition rate of 60 cps. Primary posi-

tive features of this supply(52) include: a positive square

pulse shape with a half nanosecond rise and fall time and

an adjustable width of 2 to 20 nanoseconds; pulse amplitude

range of 0 to 3,000 volts with a jitter of less than 4

percent; and a simultaneously produced trigger pulse with

an amplitude of about 2 percent of the high voltage pulse;

and either a positive or negative leading.edge with a half

nanosecond rise time. Although the mercury-wetted switch

is asynchronous,(52) the trigger pulse has sufficient energy

to initiate the horizontal sweep of the Tektronix 549

Storage Oscilloscope used in the electronic circuitry to

identify the field evaporated species.

The dc high voltage for imaging the specimen tip was

provided with a Kilovolt Corporation's Model KVR30-1S. This

particular power supply produces a highly regulated positive

or negative dc potential that ranges from 0 to 30 KV.

Following an initial 30-minute warmup, the stability is

within 0.01 percent per day. A high voltage electrostatic

voltmeter was used to determine the deviation of the indi-

cated potential and the actual output value. No deviation

was measured up to the 10 KV limit of the voltmeter.

Electronic Circuitry

To reduce the uncertainty in measuring the m/n (i.e.,

mass to charge) ratios of the field evaporated ionic species,

it was necessary to obtain the best possible resolution for

the individual displays of the Tektronix Type 549 Storage

Oscilloscope. Noticeable improvement was obtained when a

preamplifier was located adjacent to the collector of the

Bendix Spiraltron detector. This unit was designed and

constructed to minimize shunt capacitances and lead induc-

tances associated with the collector and coaxial leads to

the storage oscilloscope. A schematic diagram of this pre-

amplifier is shown in Figure 15. Physically, this inte-

grated circuit board was mounted in a solid copper box.

All the MHV feedthroughs were soldered to the copper box to

assure proper termination.

After completion the preamplifier was tested to check

its performance. The output signal had a characteristic

50 ohm impedance which was consistent with the remaining

circuitry. Generated square wave pulses were sent through

the preamplifier and the resulting oscilloscope traces indi-

cated that the output pulses retained the leading sharp

edge and duration of the incident signals without evidence

of ringing. The amplitudes of the output pulses, however,

were reduced by approximately 10 percent.

The individual electronic components of the atom probe

were arranged as shown schematically in Figure 16 to form

the control circuitry for identifying the evaporated species.

With this arrangement the field evaporation pulse and

trigger output pulse are initiated simultaneously from the

Microwave Associates Nanosecond Pulser. The high voltage

pulses ranging from 100 to 3,000 volts with two or twenty


100 OHMS
100 OHMS
100 K-OHMS
220 K- OHMS

C, : 0.1 L f 100 VDC
Cz: 0.01 pf 35 VDC
C3: 0.1 / f 100 VDC
C4: 0.001 jf 6 Kv
T : 2N930
DI,D2,D D3 D4: 5A6-D

Figure 15. Circuit diagram of preamplifier.

VI :

Z a
z -

"J" 3










nanosecond durations are transmitted from the Microwave

Associates Pulser through the pulse circuit section of the

high voltage "tee" feedthrough to the tip. Meanwhile, the

dc high voltage signals travel through the 1010 ohm resis-

tor in the high voltage "tee" feedthrough and are coupled

to the high voltage pulses before reaching the specimen tip.

The trigger output pulse in the meantime is sent by a

short coaxial transmission line to the oscilloscope where

it initiates the horizontal sweep. The following procedure

was used to determine the necessary oscilloscope settings

for identifying the ionic species. The delay time multi-

plier of the oscilloscope can be adjusted to produce a full

spectrum display of the expected evaporated species. This

is accomplished by first setting the desired width of the

spectrum on the A sweep time/cm control. Next, the expected

time of flight for the fastest speciesis divided by the

setting on the B sweep time/cm control. By putting this

value on the Delayed Time Multiplier the beginning of the

spectrum is located at the left edge of the screen. For

example, suppose a normal time spectrum extends from 5 to

6 microseconds. The width of the screen represents 1 micro-

second, so the A sweep time/cm dial is set at 0.1 micro-

second/cm, nonmagnified. To begin the time flight display

at 5 microseconds the B sweep time/cm is set at 2 micro-

seconds/cm. By dividing this value into the expected time

of flight for the fastest particle, one obtains

2 usec/cm = 0.40
5 psec/cm

which is set on the Delay Time Multiplier.

To distinguish the mass resolution of an expected

individual ionic particle from others in the spectrum, one

subtracts half the length of the A sweep time/cm from the

expected time of flight and divides this value by the set-

ting on the B sweep time/cm dial. This value is then set

on the Delay Time Multiplier. For instance, if the evapo-

rated specie's expected time of flight, calculated from

relation (3), is 5.2 psec, the A sweep time/cm switch is

set at 0.1 psec/cm, magnified. This gives a screen width

of 0.2 psec. Half of this width is 0.1 psec. Subtracting

0.1 psec from the expected time of flight, 5.2 psec,

results in 5.1 psec. Dividing 5.1 psec by the 2 psec/cm

B sweep time/cm setting gives 2.55 which is set on the

Delay Time Multiplier. This arranges the spectrum so the

expected 5.2 psec event will be recorded at the center of

the screen.

From the control circuit, Figure 16, one observes that

the field evaporated surface atoms aligned with the 1.5 mm

probe holes in the microchannel-plate intensifier and

screen pass through the drift tube orifice, travel the

length of the drift tube and strike the Bendix Spiraltron

detector. To avoid possible divergence of this trajectory,

the applied potentials on the microchannel-plate image

intensifier are removed prior to the pulsing. When each

of the ionic species strike the Spiraltron, the produced

secondary electrons are accelerated with a +3 KV potential

to the end of the detector. A positive 225 potential is

used between the detector's output and the collector to

prevent significant path divergence of the emerging elec-

trons. From the collector the signal passes through a short

lead to the preamplifier and along a coaxial line to the

input of the Type 549 Tektronix Storage Oscilloscope. After

each pulse, the trace position is relocated on the storage

oscilloscope until the accumulated individual signals com-

pletely fill the storage screen. These are photographed

and the individual signals identified.




After the atom probe F-IM was assembled and its vacuum

capabilities determined, it was necessary to calibrate the

instrument and evaluate the performance of the individual

components. Since so many components and different func-

tions were evaluated, the procedures and results will be

discussed separately. For instance, the specimen holder,

high voltage feedthrough,microchannel-plate image intensi-

fier and Spiraltron detector will be discussed in the

component section, while the observed defocusing effect,

detector efficiency and mass resolution will be described

in the operational performance and calibration section.

Performance of Individual Components

Specimen Holder

Since the specimen tip must be precisely positioned

within 0.3 mm of the optical axis, it was necessary to

initially align the center of the specimen holder's support

yoke with the probe holes in the microchannel-plate,

phosphor-coated screen and drift tube orifice. The arrange-

ment of these components was shown in Figure 16 in the pre-

vious chapter. This was accomplished by placing a high

intensity light source at the end of the drift tube and

maneuvering the yoke with the adjustable support assembly

until the centrally positioned crosshairs on the yoke coin-

cided with the center of the light beam passing through the

1.5 mm diameter probe holes. After the support assembly was

rigidly fixed, the specimen holder was installed by care-

fully adjusting the support bolts to assure the centers of

the specimen holder and yoke coincided. After the specimen

holder was attached, the glass-insulated, flexible copper

high voltage lead was connected to the Ceramseal high volt-

age vacuum feedthrough. Finally, the ball joint was fixed

to the specimen holder and connected to the push rod.

After the push rod assembly was positioned properly to

allow the specimen holder to rotate to its biased position,

it was guided into the slotted manipulator rod which in

turn was fixed to the atom probe end flange.

The atom probe body was evacuated to the 10-3 Torr

range and high voltage was applied to determine the effec-

tiveness of the specimen holder's electrical insulation.

This pressure was selected because it represents the highest

possible image gas pressure used during operation. As the

dc potential was carefully applied to the specimen holder,

observations were made to detect possible voltage breakdown.

Up to 15 kV no visible breakdown occurred. Between 15 and

24 kV, small discharges were observed, but after repeated

discharging at a particular position and potential they would

disappear. Apparently, sharp surface asperities of dust and

surface conductance caused these discharges and with repeated

discharging "healed" themselves. Testing-was discontinued

at 24 kV, since this would be the maximum dc potential

normally applied to the specimen tips. Also, the opera-

tional imaging gas pressure for the atom probe does not

exceed 1x105 Torr and this two-order improvement in vacuum

will allow a conservative 30 percent increase in the effec-

tiveness of the electrical insulation.

Cryogenic cooling of the specimen holder was achieved

by connecting the OFHC copper cooling block to the specimen

holder with a firmly attached bundle of flexible 0.75 mm

diameter copper wires. Due to the amount of copper wire in

the bundle and the minimal heat conductance losses, the tip

temperature is less than 20K above the cryogenic coolant's

temperature. If the specimen tip temperature needs to be

monitored during field evaporation studies, platinum or

carbon thermocouples can be attached to the electrostatic

shield of the specimen holder. By prior correlation of this

thermocouple's temperature with the temperature of a similar

thermocouple mounted at the tip position, a reasonably

accurate description of the specimen temperature is achieved.

After the initial cooling of the specimen holder, the cold

finger will maintain liquid nitrogen temperature for at

least 30 minutes before being refilled.

In order to replace'the specimen tip the:following'pro-

cedure was found to be the most efficient. The heating

tapes were turned on prior to introducing ultra-pure nitro-

gen to the atom probe. After the atom probe reached a

minimum of 1600C and its pressure reached 1 atmosphere, the

view port was removed and the specimen holder was maneuvered

to its biased position, as shown in Figure 11 in the previous

chapter. The copper piece that contains the specimen tip

is removed from the specimen holder and the new specimen is

carefully installed. After the specimen holder is returned

to a position facing the microchannel-plate, the view port

is replaced and the bakeout temperature increased to 2800C.

After 10 to 15 minutes, the system is evacuated by standard

vacuum methods. Repeated specimen exchanges indicate that

four to five hours are required to return the atom probe

to less than 1xl09 Torr.

It is essential that the rotating specimen tip remain

axially aligned with the probe holes and Spiraltron detector,

as the individual atomic sites are translated to the probe

hole. To accomplish this the optical axis position in the

specimen holder was determined. In particular, the dis-

tance from the end of the tip to the bottom of the copper

mounting sleeve must be positioned within 0.001 cm. This

distance must be ascertained prior to crimping the tip

securely into the copper sleeve. In order to assure the

tip will be aligned within 0.3 mm of this axis, the mounted

specimen is inserted in an exact duplicate of the specimen

holder's hat. This unit is then rotated slowly and the tip

position adjusted to its center of rotation. Following

these mechanical alignments the tip is observed optically

to check for possible bending or breaking.

The rotational range of the specimen holder is best

illustrated by Figure 17. In this figure the probe hole is

positioned on the (110), (111) and (112) planes of a tung-

sten tip imaged in a helium-10 percent neon mixture with an

applied potential of 8.5 kV. If the average tip radii had

been larger the (100) plane could also be rotated to the

probe hole center. The angular motion of the specimen holder

is easily controlled within a few tenths of a degree by care-

ful translation of the manipulator rod with the two microm-

eters mounted in the goniometer stage. During atom probe

analysis the desired region of the specimen surface is located

first and if the analysis of individual atomic sites are

desired, these image spots are translated to the center of

the probe hole.

High Voltage Feedthrough

Another important factor in the atom probe performance

is an accurate measure of the applied dc potential at the

tip. To establish the uncertainty of the dc potential at

the tip the high voltage feedthrough was connected to the

(a) Probe hole positioned on (110) plane.

(b) Probe hole positioned on (111) plane.

Figure 17. Specimen manipulation provides observation
of nearly entire surface of tip.

(c) Probe hole centered on (112) plane.

Figure 17.


atom probe and a high voltage electrometer was connected

to the tip position. Calibration was conducted by comparing

the applied dc potential to the potential measured at the

tip position. From zero to 10 kV the high voltage dc supply

and the measured tip potential appeared identical. Cali-

bration of higher potentials was not possible due to the

electrometer limitation; however, deviation from the extrap-

olated results would be less than 0.1 percent, based upon

the stability of the high voltage supply.

Another concern for the high voltage feedthrough was

its capability to withstand potentials above 10 kV at one

atmosphere. The applied potential was increased slowly to

24 kV to test for possible high voltage discharging. Al-

though small discharges occurred above 20 kV, these were

quickly "healed" and the dc potential remained steady. Also,

at the 24 kV dc potential the pulsing portion of the feed-

through circuitry was adequately protected by the high volt-

age blocking capacitor, C2, as shown in Figure 14(a).

Microchannel-plate Image Intensifier

Initially, the microchannel-plate image intensifier was

tested for possible electrical breakdown. The gas pressure

of the atom probe was again maintained at 10-3 Torr and the

applied positive potential was slowly increased to 1.2 and

5 kV for the microchannel-plate and phosphor-coated screen

respectively. No electrical discharging was observed with

repeated application, so the glass-insulated assembly

appeared satisfactory. Next, the atom probe was evacuated
to less than 2x10 Torr and a tungsten specimen was imaged

in the field emission mode. With the appropriate potentials

applied to the intensifier a bright image resulted. Also,

it was observed that the pressure increased to 7x109 Torr.

Apparently, there remained gaseous species in the optical

fibers of the microchannel-plate after the atom probe bake-

out and upon electron excitation these were outgassed.

After two hours, the system pressure was in the 10 Torr

range and the field emission image appeared much sharper

in contrast.

Image formation parameters were established for the

microchannel-plate image intensifier for helium, neon and

helium-neon mixtures. In particular, the applied positive

potentials to the microchannel-plate and phosphor-coated

screen were varied to establish the best image for the

different imaging gases. For example, no image intensifi-

cation was evident unless 600 volts were applied to the

microchannel plate. If the potential exceeded either 950
volts at lx106 Torr image gas pressure or 900 volts at

1x10 Torr gas pressure, the resolution of the image dimin-

ished. Since the image intensifier depends on proximity

focusing of the secondary electrons emitted from the micro-

channel-plate, it was necessary to establish the best focus-

ing potential for the phosphor-coated screen. Visual obser-

vation indicated that the optimum image resolution was

obtained at 3.5 kV. Noticeable improvement in the observed

images was obtained with the helium-10 percent neon mixture.

This particular composition afforded improved atomic resolu-

tion over both the elemental gases and confirmed recent

results reported by Chen and Seidman. (8)

Spiraltron Detector

In order to predict the sensitive area of the specimen

tip that can be analyzed by the Spiraltron CEM-4028 detector,

it was necessary to determine the following geometric factors

for the atom probe. First, it was necessary to measure the

listed distances: tip-to-microchannel-plate, 86.9 mm;

tip-to-screen, 88.9 mm; tip-to-mirror, 122 mm; and the tip-

to-detector, 100 cm. Since the tip-to-detector distance

was determined within 1 mm the uncertainty for this param-

eter for the time-of-flight measurements is 0.1 percent.

With the above distances and the diameters of the micro-

channel-plate (1.5 mm) and Spiraltron orifice (10 mm), it

was calculated that the specimen tip must be located within

0.3 mm of the optical axis to allow the field evaporated

ionic species to strike the Spiraltron cone. In addition,

it was noted that the sensitive area of the tip to be ana-

lyzed was not limited by the probe hole diameter, but by

the size of the incident aperture of the Spiraltron. Brenner

and McKinney(52) estimated this value from the following



R 2
A = AD 62 (28)

where AT is the sensitive area of the tip, AD is the area

of the detector incident aperture, RT is the tip radius,

RD is the tip-to-detector distance, and B is an image com-

pression factor dependent upon the tip geometry and usually

ranges from 1.5 to 1.8. Substituting appropriate values

for the atom probe, the sensitive tip area ranges from 31.4

to 50.2 A with respect to assumed tip radii of 400 and

800 A. This AT value indicates that the number of atoms

analyzed from a surface will depend upon their crystallo-

graphic orientation. For instance, if the atoms are in the

(110) plane, four could be analyzed, whereas if they were

located in the (111) plane, only two could strike the


The Bendix Spiraltron CEM-4028 can be operated in

either the analog or pulse counting modes. The analog cur-

rent mode, illustrated in Figure 18, was used to measure

both the ion and electron currents depending upon the selec-

ted atom probe's mode of operation, i.e., field ionization

or field emission. Schmidt(56) reported that the CEM-4028

was linear in operation for output currents below approxi-

mately 10 percent of the bias current, i.e., (V/109 x 0.1)

amps, where V is the applied potential. Examination of the

analog circuitry indicates that the measurement of both

types of currents (electron and ion) requires an appropriate



0 o
> U)

+ a

0- 0

S_ 0





Z 0


selection of polarity input and output apertures and main-

tenance of a 225 volt bias between them and the output ori-

fice and the collector. The pulse counting mode, 'refer to

Figure 16 Chapter III, is used primarily for the time-of-

flight measurements. Results from this mode will be dis-

cussed later.

Prior to performing the final electronic alignment of

the Spiraltron with the tip position, it was necessary to

determine the best operational parameters for the Spiral-

tron's two modes of operation. The analog mode was ana-

lyzed with an iridium tip imaged at 10 kV with a neon gas
pressure of 5x10 Torr. The resulting ion current was

detected by the Spiraltron that was differentially pumped

in the drift tube to 5x10-7 Torr and recorded by a Keithley

electrometer. The observed current represented the actual

current times the gain achieved by the Spiraltron, i.e.,

Measured = CIactual' where C is the gain. Electro-Optics

Division of Bendix(56) reported a gain of 108 was obtained

during normal operation and this value was assumed for the

actual current calculations. The resulting data are plotted

in Figure 19. Note the linear increase in neon current as

the applied potential increases from 500 to 2,750 volts.

Above 2,800 volts, the current fluctuations were so extreme

that it was impossible to obtain a steady state value. The

range of the actual current extended from 1017 to 1014

amps. These current values were comparable with reported

0 C


I U 4

_- o

a.. o

'c Z

0 v oo


I I I I .-I
2o b
(, ) N


b bb b b

(Sd VV) IN3d8flO NOI N03N

data by Brenner and McKinney(52) for helium currents ob-

tained with a similar Spiraltron detector.
The neon imaging gas pressure was reduced from 5x104
to 1.4x104 Torr and the resulting ion current indicated

that the linear portion of Figure 19 was extended to 3 kV.

Also, the tip potential was decreased from 10 to 9 kV and
the image gas pressure remained constant at 5x10 Torr.

These changes indicated no alteration in the actual current

plot of Figure 19. Apparently, the 1 kV tip potential drop

did not affect noticeably the ionization efficiency.

From these ion current measurements, it was established

that the initial optical alignment between the tip and de-

tector was sufficient to allow the emitted ion current to

be detected by the Spiraltron. Next, the phosphor-coated

screen was removed and the microchannel-plate image intensi-

fier was assembled and aligned optically with respect to

the drift tube orifice, as described in the previous chapter.

The final alignment of the Spiraltron detector with the

specimen tip and probe holes in the microchannel-plate,

phosphor-coated screen, reflecting mirror and drift tube

orifice was accomplished by two means. The first method

(refer to Figures 9 and 16 in previous chapter) involved

the following steps: remove the Spiraltron detector and

end blank flange from the drift tube, locate a translucent

grid on the specimen holder and carefully aim a He-Ne laser

from the open end of the drift tube through the drift tube

orifice and the probe holes of the reflecting mirror and

microchannel-plate image intensifier. By viewing the grid

on the specimen holder, the specimen holder could be

adjusted until the laser beam coincided with the center of

rotational axis of the specimen holder. Following this

alignment, another translucent grid was superimposed over

the open end of the drift tube and the drift tube was maneu-

vered until the laser beam was centered on the grid. The

aligned drift tube was rigidly fixed with the two support

tripods and the Spiraltron was repositioned and fixed when

its incident aperture was centered on the laser beam.

The second means to assure optimum alignment of the

tip and detector was accomplished electronically by measur-

ing both the ion and electron currents. The ion current
was measured with an iridium tip at 10 kV in a 5x104 Torr

atmosphere of neon. Prior to the measurement, the applied

potentials to the microchannel-plate image intensifier were

removed. Again, the drift tube section was differentially

pumped to less than 2x106 Torr. With the flexible bellows

connecting the drift tube and atom probe it was possible to

adjust the Spiraltron position in two directions. Initially,

the detector was moved laterally with the adjustment screws

on the support tripods from the previously fixed position

determined by the laser beam alignment. The actual current

observed in the initial position decreased in both the

lateral directions and the two height adjustment directions.

Consequently, the Spiraltron detector was locked rigidly

into the optimum position established by the laser beam


The iridium tip was replaced with a tungsten tip and

the atom probe was altered to the field emission mode. The

Spiraltron was coupled to a Monsanto Counter-Timer Model 101B

to determine accurately the detected electron current as a

function of the field emission potential. Initial measure-

ment of the electron current, however, was not possible unless

the potential was removed from microchannel-plate and ground

screen for at least three minutes. The resulting data indi-

cated that the electron current increased from an average

4.4 to 2,009 counts per sec (cps) as the tip potential was

increased from 2 to 2.7 kV. When the tip potential was de-

creased the previous values were reproducible. Besides

further confirmation of the tip and detector alignment, these

data suggest the feasibility of utilizing the atom probe to

obtain work function values from previously field ionized

"clean" metal surfaces. Since the sensitive area of the
tip ranges from 30 to 50 A emission measurements can be

made on selected regions of the tip surface. For instance,

comparative work function differences can be measured in

relation to crystallographic orientation, atomic location

in a particular plane, region in vicinity of lattice defect

emerging at tip surface, degree of surface coverage by

gaseous absorbants, and.effect of solute concentration.

Operational Performance and Calibration

Defocusing of Ionic Species

After the ion and electron current measurements con-

firmed the alignment of the tip and the Spiraltron detector,

the next operation to evaluate was the application of a 2

or 20 nanosecond high voltage pulse to the specimen tip and

detect the field evaporated ionic species. Prior to the

pulsed field evaporation it was necessary to increase the

dc potential above the BIV (best image voltage) until the

surface dynamically field evaporated at a reasonably con-

stant rate. After establishing this field evaporation

strength, the dc potential was decreased approximately 500

volts and the Microwave Nanosecond Pulser was activated.

The amplitude of the 20 nanosecond pulses was increased by

50 to 100 volt increments until visual observation confirmed

evaporation of the selected plane. In order to maintain a

constant field evaporation rate for the selected plane, the

amplitude of the pulse was adjusted until an atomic layer

could be removed repeatedly by a certain number of pulses.

After the parameters were established, the positive poten-

tial applied to the microchannel-plate and phosphor-coated

screen were turned off. Muller(7) and Brenner and McKin-

ney(2) report that these high potentials must be switched

off prior to pulsing the tip or they will divert the tra-

jectory of the field evaporated species and reduce the

detector efficiency and mass resolution.