RCA_PhototubesAndPhotocells_PT-60_1963

$~ 50

Suggested

RCAPrice

phototubes
and
phataeells
TECHNICAL MANUAL PT-60
Vacuum and gas photodiodes
Multiplier phototubes
Solid-state photocells

RADIO CORPORATION OF AMERICA

Electronic Components and Devices •Lancaster, Pa,

CONTENTS

THEORY AND MEASUREMENTS 3

Photoelectric Theory, Photoemission, Semiconductor Photocells, Photo-
conductivity, Photovoltaic Effect, Photoelectric Measurements, Light,
Photometric Units, Illumination From Uniformly Diffusing Surfaces,
Luminous and Absolute Rating Systems, Radiant Energy.

VACUUM PHOTOTUBES 22

Construction and Principles of Operation, Spectral Response, Current-
Voltage Characteristics, Linearity, Frequency Response, Noise, Environ-
mental Factors, Application Considerations.

GAS-FILLED PHOTOTUBES 33

Construction and Principles of Operation, Current-Voltage Character-
istics, Variation of Current with Light Flux, Time- Or Frequency-
Response Characteristics, Noise, Environmental Factors, Application
Considerations.

MULTIPLIER PHOTOTUBES 38

Construction and Principles of Operation, Dynode Properties, Dynode
Configurations, Coupling Dynode System to Cathode, Design for Mini-
mum Transit-Time Spread, Gain Characteristics, Spectral Response,
Dark Current, Noise, Transit-Time Effects, Linearity of Output Current,
Temperature Effects, Effect of Magnetic Fields, Fatigue and Life
Characteristics.

PHOTOCELLS 71

Cadmium-Sulfide Photoconductive Cells, Junction Photocells, Ger-
manium P-N Junction Photocells, Silicon Photovoltaic Cells, Data-

Processing Cells, Germanium and Germanium/Silicon Infrared Detectors

GENERAL APPLICATION CONSIDERATIONS 80

Level of Light or Radiation, Spectral Energy Distribution, Frequency

Response, Typical Applications.

INTERPRETATION OF DATA 85

SELECTION CHART 90

TECHNICAL DATA 91

SOCKET AND SHIELD INFORMATION 182

RCA PHOTOCELL CHART 186

INDEX 190

Copyright 1963 by Radio Corporation of America
(All Rights Reserved)

Trade Marks) Registered 10 - 63
Marca(s) Registrada(s) Printed in USA

RCA

phototubes
and photocells

Tx1s book has been prepared to assist equip-
ment designers in their use .of vacuum, gas, and
multiplier phototubes, solid-state photocells, and
associated circuits. It covers theory, character-
istics, and applications of the four classes of
devices, as well as information on selecting the
right type of device for a specific application. The
section on Technical Data contains detailed infor-
mation on phototubes and photocells presently in
the RCA line.

ACKNOWLEDGMENT 1S given to Dr. Ralph W.
Engstrom for supplying the text material on
Theory and Measurements, Vacuum phototubes,
Gas-Filled phototubes, Multiplier phototubes, and
General Application Considerations, as well as
for technical guidance and review of other text
material in this manual.

RAT C©RPORATION OF AMERICA

Electronic Components and Qerices
LANCASTEP., ~A.

For a broad spectrum of applications
for space, research, medicine, and
industry.

RCA
PHOTC)TU DES
AND
PHOTOCELLS

T~1><eory and
l~~ea~ul~•ernent~

THE photosensitive devices de- the latter part of the 19th century,
the observation of a photovoltaic ef-
scribed in this manual are extremely fect in selenium led to the develop-
versatile tools for extending man's ment of selenium and cuprous oxide
sense of sight. The variety of types photovoltaic cells.
developed during the past few dec-
ades make it possible to equal and The emission of electrons result-
surpass many, if not all, of the ing from the action of light on a
human eye's remarkable capabilities photoemissive surface was a later
for detection and observation. These development. Hertz discovered the
devices exceed the sensitivity of the photoemission phenomenon in 1887,
eye to all the colors of the spectrum, and in 1888 Hallwachs measured the
and even penetrate beyond the visible photocurrent from a zinc plate sub-
region into the ultraviolet and infra-
red. They can observe a bullet in jected to ultraviolet radiation. In
flight or track acosmic-ray particle. 1890, Eaater and Geitel produced a
They can accompany a rocket into forerunner of the vacuum phototube
outer space or explore a hole drilled which consisted of an evacuated glass
deep into the crust of the earth. bulb containing an alkali metal and
an auxiliary electrode used to col-
The availability of these devices lect the negative electrical carriers
has led to a wide range of practical
applications. Vacuum-type photo- (photoelectrons) emitted by the ac-
tubes are used primarily for radia- tion of light on the alkali metal.
tion measurement. Gas-type photo-
tubes made possible the addition of The development of the tran-
sound to motion pictures by convert- sistor and related devices has been
ing sound patterns traced on film closely paralleled by the development
of solid-state photosensitive devices,
into electrical signals. Multiplier such as photoconductive cells, p-n
phototubes, which have tremendous photojunction cells, phototransistors,
amplification capability, are used ex- and silicon photovoltaic cells used as
tensively in photoelectric measure- solar-energy converters.
ment and control devices, and in the
large and growing field of scintilla- Photoemission
tion counting. Photocells are most
widely used in the field of industrial The modern concept of photo-
photoelectric control because of their electricity stems from Einstein's
simplicity, low cost, and high sensi- pioneer work for which he received
tivity.
the Nobel Prize. The essence of Ein-
Photoelectric Theory stein's work is the following equation
for determining the maximum kinetic
energy E of an emitted photoelec-
tron:

The earliest observation of a mv' (1)
photoelectric effect was made by E_ 2 =hv—~
Becquerel in 1839. He found that
when one of a pair of electrodes in Eq. (1) shaves that the maximum
an electrolyte was illuminated, a energy of the emitted photoelectron
voltage or current resulted. During
mv= /2 is proportional to the energy
3

4 RCA Phototube Manual

of the light quanta by less the energy n.= METAL VACUUM
¢, (the work function) which must 00
be given to an electron to allow it to y o:

escape the surface of the photocath- ~c~U~

ode. For each metal, the photoelectric wW ¢-WORK FUNCTION
effect is characterized by a value of ZJ
~, which is usually expressed in WW FERMI LEVEL
electron-volts.
FERMI
In the energy diagram for a ENERGY
metal shown in Fig. 1, the work func-
tion represents the energy which Fig. 1. Energy model fora metal showing
the relationship of the work function
must be given to an electron at the and the Fermi level.
top of the energy distribution to raise
it to the level of the potential barrier for a temperature of absolute zero;
at the metal-vacuum interface. all lower energy levels are filled. As
the temperature is increased, some
According to the quantum of the electrons absorb thermal en-
ergy which permits them to occupy
theory, only one electron can occupy scattered states above the maximum
level for absolute zero. The energy
a particular quantum state of an distribution of electrons in a partic-
atom. In a single atom, these states ular metal is shown in Fig. 2 for
several different temperatures. At
are separated in distinct "shells"; absolute zero, all the lower states are
normally only the lower energy occupied up to the Fermi level. At
states are filled. In an agglomeration higher temperatures, there is some
of atoms, these states are modified by excitation to upper levels. The elec-
interaction with neighboring atoms, tron density at a particular tempera-
particularly for the outermost elec- ture is described by the Fermi-Dirac
trons of the atom. As a result, the
outer energy levels tend to overlap
and produce a continuous band of
possible energy levels, as shown in

Fig. 1.
The diagram shown in Fig. 1 is

to 10 300 0 DEGREES KELVIN

K
F9

8 1033
6W 7

J

O2 fi

5
WU

W4
rc

'~ 3
2N
~2

F

W I 300
JW 0 1033
2~
O 3 4

ELECTRON ENERGY— ELECTRON-VOLTS

~--_—FERMI E2.0N28ERGYF-E-R.-M-I..L~E_V-E-L-- WORK FUNCTION___ VACUUM
fir. = ~= 2.24 LEVEL

~-- OUTER WORK FUNCTION —r
Wa = 4.27

Fig. E. Energy distribution of conduction electrons in potassium at temperatures of 0, 808,
and 1038 degrees Kelvin based on elementary Sommerfeld theory. (ref. 17)

Theory and Measurements 5

energy-distribution function, which can be used to predict the shape of
indicates the probability of occupa-
tion ffor aquantum state having the spectral-response curve near the

energy E: threshold; the work function can

_1 then be calculated from these data.
f — 1 -♦- exp. [(E — E:) /kT] (2)
Although many attempts have
When E is equal to E:, the value been made to calculate entire
of f is i/z. It is customary to refer to
the energy of level E:, for which spectral-response characteristics for
there is a 50-per-cent probability of metals, only order-of-magnitude
occupancy, as the Fermi level. At agreement with experiment has been
absolute zero, the Fermi level cor- obtained. It was formerly assumed
that surface electrons existing in an
responds to the top of the filled
energy distribution. Although the image-type force field accounted for
Fermi level is nearly the same at the threshold emission spectrum.
Early work by Tamm and Schubin'
higher temperatures, as at absolute
postulated that a threshold for a
zero, as shown in Fig. 2, it is actually
slightly lower. This reduction of the "volume photoeffect" occurs at about
Fermi level occurs because the num-
ber of possible energy states in- twice the frequency of that for the

creases in a conductor as the square surface threshold. More recent work
root of E, whereas the probability
of occupancy function is symmetrical by Mayer and his associates' indi-
around a value of E = E:. cates that in fact the threshold for
the volume effect and surface effect
If the energy derived from the may be the same. Because electrons
radiant energy is just sufficient to in the volume of the metal have been
eject an electron at the Fermi level, described as "free", there has been
the following relation exists: a dilemma in the explanation of con-
servation of momentum in the ab-
hvo = ~ (3) sorption of photons. This dilemma
has been resolved by taking into
v, the threshold frequency of the ex- account the interaction of the free
citing radiation, is related to the electrons with the periodic field of
the crystal lattice of the metal.
long-wavelength limit ~o and the
The yield of photoelectrons per
velocity of light c as follows: incident photon must be low because
metals contain large numbers of
~o = C/Ya (4) these free electrons. On the one hand,
these electrons result in high optical
The relationship may be rewritten to reflectivity in the visible and near-
relate the long-wavelength limit to ultraviolet regions. On the other
the work function, as follows: hand, the free electrons scatter the
excited electrons within the metal,
12395 reducing the energy available for
ao = ~ angstroms (5)
escape. Consequently, excited elec-
Because some of the electrons trons originating at a depth of more
occupy states slightly higher than than 10 angstroms have only a slight
the Fermi level, as shown in Fig. 2, chance of escaping. These predictions
excitation of these electrons produces have been confirmed by measure-
an extended response at the red ments of the photoelectric yield in
metals, which indicate a quantum
threshold of the spectral-response efficiency of less than 10-' electron
per incident photon.
characteristic. As a result, there is
no abrupt red threshold at normal Measurement of the work func-
temperatures, and the true work tion and spectral response for clean

function cannot he obtained in a metal surfaces has been of consider-

simple manner from the spectral- able importance in the development
response measurement. However, a
universal function devised by Fowler of photoelectric theory. Table I shows
a number of the results obtained.
Fig. 3 shows spectral-response curves

6 RCA Phototube Manual

TABLE I

Photoelectric wovrkalufuenscftoiornpsuarendmleotnagls-.wl7a,v18elength threshold

Work Work

Element Function Threshold Element Function Threshold

Ag (volts) (angstroms) Na (volts) (angstroms)
Al 2610 Ni
Au 4.73 Pb 2.46 5040*
Ba 2.5 - 3.6 3652 Pd
Bi 4.82 2650 Pt 4.86* 2550
C 2.51 - Rb
Ca 4.32* 2.b2 4920-4940* Rh 3.b - 4.1 2980-3550
Cd (4.7) 2870 Se
Ce 2.706 2565-2615 Sn (B) 6.30 1962
Co 4.24* 4580* Sn (Y)
Cr Sn Liquid 6.20* 2000
Cs 2920 Sr
Cu Ta 2..16 - 2.19 6660-5740*
Fe Th
Ge Ti 4.57 2500
Hg U
K W 5.63* - 4.64* 2200-2670
Li Zn
Mg 2.88* 4300 Zn Single 4.5 2740
Mo 2900, 3000 Zr
4.25, 4.12 2840 4.38 2820
4.36* 6320-6630*
2750-3000 4.21 2925
1.87 - 1.96 2680 2.07* 6000

4.1 - 4.6 2580 4.12* 3010
4.63*
(4.3) 2735 3.40* 3650
4.53 5530* 3.93* 3150
2.24
2.28 6430* 3.65* 3400
3430*
3.61 4.59* 2700
4.35* 2850
3.32 3720

Crystals 3.57 3460

3..76* 3300

* Calculated from 12395/ (angstroms) _ ~ (volts)

for the alkali metals a The curves Because a quantum of radiation

indicate a regular progression of the is necessary to release an electron,

wavelength for maximum response the photoelectric current is propor-
with atomic number. The most red-
sensitive of these metals is cesium, tional to the intensity of the radia-
which is widely used in the activa- tion. This first law of photoelectricity
tion of most commercial phototubes. has been verified experimentally over
a wide range of light intensities.
320 Li 600 For most materials, the quantum
efficiency is very low; on the best
280 a 400 w Z sensitized commercial photosurfaces,
w the maximum yield reported is as
0z 240 K 200 z~
Nu 200 000 a = high as one electron for three light
Rb
~ 160 Os w quanta.
BOO ~w~J Research on commercially useful
a 120
600 Fa o photoemitters has been directed pri-
J
400 w-~ 4
~ 80
~U >., ~-4350 ANGSTROMS
40
200 ~ k3650

0 0 a~ ANGSTROMS

3600 4400 5200 6000 6e00 z~

WAVELENGTH-ANGSTROMS W} 3
F~
Fig. 3. spectral-response characteristiq for
the alkali metals showing regular pro- za
gression in the order of the periodic ~~
table. (ref. 17)
aW a~ 2
The energy distribution of emit- I
N'-'
ted photoelectrons has been meas-
20 J
ured for a number of metals and ~>
photosurfaces. Typical results are
shown in Fig. 4 for a potassium film UW
of 20 molecular layers on a base of W {-
silver.' ~'he maximum emission en-
ergy corresponds to that predicted by wZ
the Einstein photoelectric equation.
0 0.4 0.8 12 16

ENERGY OF PHOTOELECTRONS-
ELECTRON-VOLTS

Fig. 4. Energy distribution of photoelectrons

from a potassium film. (ref. 4)

Theory and Measurements 7

marily toward developing devices shows an energy model fora semi-
sensitive to visible radiation. The conductor photoemitter. The band-
first important commercial photo- gap energy corresponding to the for-
surface was silver-oxygen-cesium. bidden gap between the valence and
This surface, which provides a spec- conduction bands is represented by
tral response designated S-1, is sensi- Er and is expressed in electron-volts.
tive throughout the entire visible The potential barrier from the bot-
spectrum and into the infrared. Al- tom of the conduction band to the
vacuum outside the semiconductor
though it has rather low sensitivity surface is represented by E°, the

and high dark emission, the good electron affinity.

response in the red and near-infrared ~T VACUUM
Ea+ EG
regions provides a spectral response AFfICNITYN Ea
CONDUCTION BAND
which is a good match to the emis-
sion of a tungsten lamp. One of the FORBIDDEN E
most important early uses of this GAP i9'
photoemitter was in the sound head
for motion-picture projectors. VALENCE /
BANDj/
Although other emitters using
rubidium and potassium as the prin- Fig. 5. Energy model 04 s semiconductor
cipal element were developed, none photcemitter.
was of great commercial value until
the development of the cesium-anti- When radiation of sufficient
mony photoemitter,° which provides energy by excites an electron in the
an S-4 or 5-11 spectral response. The valence band causing it to move to
spectral response of the cesium-anti- the conduction band, photoconductiv-
ity may result; however, photoemis-
mony photocathode includes most of sion does not take place until the
the visible region; maximum re- electron escapes the barrier to the
vacuum. Therefore, the minimum
sponse is in the blue and the ultra- photon energy necessary to produce
threshold emission is Eg plus E°.
violet regions. A recently developed
photoemitter,° described as multi- For the cesium-antimony photo-
alkali (Sb-K-Na-Cs), provides an cathode, the value of E° plus E. is
5-20 spectral response and is re- 2.05 electron-volts; the value of E.
markable for its high quantum is 0.45 electron-volt. Corresponding
efficiency and its extended red re- values for the multialkali photocath-
sponse. Spectral-response curves for ode are 1.55 and 0.55 electron-volts?
Although both of these emitters are
these designations are shown in the believed to have impurity (p-type)
levels in the band-gap region which
TECHNICAL DATA section. account for minor effects, the photo-
Because emitters such as the emission is primarily intrinsic (that
is, originating from the valence
cesium-antimony and the multialkali band).
cathode are semiconductors, the
theory developed for metals does not An important advantage of us-
apply in all respects. Ina semicon- ing semiconductors as photoemitters
ductor, the highest filled energy band is that their quantum efficiency in the
for electrons is called the valence visible spectrum is much higher than
band. Immediately above the valence that of metals (up to 30 per cent
band, there is an energy gap for
which no electron-energy states exist.
This gap is referred to as the for-
bidden gap. In a metal, this gap does
not exist and the continuum of energy
states directly above the filled energy
levels permits conduction. Above the
forbidden gap in a semiconductor,
there is a band of permitted energy
states referred to as the conduction
band, which at ordinary temperatures
contains very few electrons. Fig. 5

8 RCA Phototube Manual

instead of up to 0.1 per cent). In a CONDUCTION VACUUM VACUUM
semiconductor, aphoto-excited elec-
tron is less likely to lose its energy eANo CONDUCTION
~ BAND
by electron-electron collision than in r
a metal. The principal mechanism of FORBIDDEN 1
energy loss is through the process of FORGAPDEN__-LEVEL
impact ionization and pair produc- GAP--- FERMI ~y
tion, i.e., the creation of a "hole" and
electron across the band gap. Usually LEVEL ~~~~~~

the threshold energy for such pair ~~~~~~~ VALENCE
production is several times larger BAND
than the band-gap energy. In semi-
VALENCE
conductors in which the electron af- BAND
finity is small compared with the
threshold energy required for pair A
production, the depth from which an
electron can escape may be expected CONDUCTION VACUUM' VACUUM

to be large; consequently, the quan- BAND CONDUCTION
tum yield should be large, as is •FERMI BAND
actually the case with materials
~sooo~LEVEL
which are practical photoemitters.
DONOR
Semiconductor IMPURITY
Photocells
LEVEL
A photocell is defined as a
photosensitive device in which the ACCEPTORy
movement of electrons is in a solid IMPURITYogo00_
(in contrast to a phototube in which FERMI
the movement takes place in a LEVEL ~ LEVEL
vacuum or gas). For photocells,
VALENCE ~~~~
therefore, the electron affinity and BAND VALENCE
emission outside the solid material BAND

need not be considered. c0
The behavior of semiconductor
Fig. 6. Energy models for intrinsic and
photocells is most easily explained impurity-type semiconductors and for an
by reference to energy-state models. insulator. (A) an insulator, which has
Energy models for several types of a large forbidden gap. (B) an intrinsic
semiconductors and for an insulator semiconductor, which has a smaller for-
are shown in Fig. 6. The distinction bidden gap. (C) an n-type impurity
between semiconductor and insulator semiconductor. (D) a p-type impurity
is arbitrary; materials with band
gaps smaller than two electron-volts semiconductor.
are usually called semiconductors;
those with larger band gaps are electrons from the filled-band level
called insulators. In an insulator (A),
essentially no thermal excitation of leaves vacant levels in the energy
electrons from the valence band to structure, which are referred to as
the conduction band takes place at holes. Electrons of the filled band
normal temperatures. In an intrinsic may then move under the influence of
semiconductor (B), the forbidden gap the applied field to fill these holes.
is sufficiently small to permit some As a result, the holes may also con-
thermal excitation; the amount of tribute to the conductivity as they
excitation increases exponentially travel across the semiconductor in
with temperature. The thermally ex- the opposite direction from the elec-
cited electrons contribute to a con- trons.
duction current. The removal of
Conduction is also possible in
substances which are insulators in
the pure state because of the pres-
ence of certain impurities or imper-
fections. Two types of impurity
semiconductors are possible; n-type
(donor, C) and p-type (acceptor, D).
For an n-type semiconductor, the
highest filled quantum level of the
impurity lies just below the conduc-

tion band of the original or parent

lattice (Fig. 6). Conduction becomes

Theory and Measurements 9

possible when electrons from the CONDUCTION BAND
donor impurity are thermally or
otherwise excited to the conduction 567
band. STIMULATE

In the p-type semiconductor, the z hvI a J by
unfilled level of the impurity lies HEAT IR
just above the filled band of the by E
parent lattice. Excitation of an elec- G3 OUENCH BAND GAP
tron from this filled band to the EG
acceptor impurity is then possible
and conduction takes place by means hvlR B 9HEAT
of holes in the valence band.
VALENCE BAND
The Fermi levels of these vari-
ous types of materials are also Fib. 7. Basic electronic processes in an
indicated in Fig. 6. For the insulators n -type photoconductor such an CdB.
and intrinsic semiconductors, the
Fermi level lies approximately half- photoconductor such as CdS. Arrow
way between the valence and conduc- 1 illustrates the case in which the
tion bands; for an impurity-type
semiconductor, the Fermi level usu- host crystal is excited by an absorp-
ally lies in the forbidden band near
the impurity level—close to the con- tion of light having energy equal to
duction band in the case of a donor-
type impurity, and close to the or greater than the band gap. In this
valence band in the case of an ac- case, one electron-hole pair is formed
ceptor-type impurity. for each photon absorbed. Arrow 2
represents the excitation of a bound
Photoconductivity electron at an imperfection level.
These imperfections may be either
In photoconductors, currents flow impurities or crystal defects such
when light excites electrons from the as vacancies. Arrow 3 is the capture
filled band to the conduction band. of aphoto-excited hole by an im-
Conduction may also be produced by
the excitation of electrons from an perfection center. Arrow 4 is the
impurity level to the conduction band
in an n-type semiconductor or by the capture of aphoto-excited electron
excitation of electrons from the filled
band to an acceptor level in a p-type by a center which has previously
semiconductor (in the latter case
conduction is by means of holes). captured a photo-excited hole,

Although excited electrons or thereby resulting in a recombina-
holes may travel freely through the tion of the carriers. Arrow 5 is the
photoconductor to the electrodes, capture of aphoto-excited electron
they may also be captured by vari- by an electron-trapping center. Ar-
ous types of imperfection centers row 6 is the thermal freeing of a
in the photoconductors. If such a cap- trapped electron. The optical free-
ture site provides only temporary ing of a trapped electron and a cap-
retention of the carrier and there tured hole is shown by arrows 7 and
is good probability of thermal re- 8, respectively. Arrow 9 is the
excitation, it is called a trapping thermal freeing of a trapped elec-
center. However, if the capture re- tron.
sults in good probability of re-
combination with a carrier of the Transitions 1 and 2 determine
opposite polarity, the center is re- spectral response; 3 and 4 determine
ferred to as a recombination center. free-electron lifetime and, hence,
photoconductivity; 5 and 6 fre-
Fig. 7 shows the various modes quently determine speed of re-
of electron transition in a n-type sponse; 7 causes stimulation of
conductivity; 8 and 9 correspond to
optical and thermal quenching of
photoconductivity when the centers
involved are those having small
cross-section for free electrons.
Transitions 3 and 4 may be either
radiative, that is, give rise to lumi-
nescence emission, or nonradiative.

18 RCA Phototube Manual

Because of the relatively low Photovoltaic Effect

mobility of carriers in semicon- The same type of energy-state
model that applies to photoconduc-
ductors, space charge usually limits tors also applies to photovoltaic
the currents to low values. The effect cells. In a photovoltaic cell, however,
a junction of two different materials
of the charge is greatly increased by produces a contact potential. The two
materials may be a semiconductor
the presence of trapping centers. In and a metal, or they may be an n-
some cases, the space-charge fields of and a p-type semiconductor. The
silicon solar cell is this latter type of
trapped holes result in the ejection of photovoltaic cell. An n-type silicon
wafer is formed with a p-type layer
electrons from the negative terminal on one surface. The operation of such
a cell is illustrated in Fig. 9. The p-
(or those of trapped electrons in the type material is shown at the left
and the n-type material at the right.
ejection of holes from the positive

terminal) giving rise to a secondary

current. As a result of such secon-

dary processes, it is possible to have

a quantum efficiency greater than

one; that is, one photon results in the
movement through the semiconductor

of more than one electron. Time de-

lays are observed in the photocur-

rent ~~hich are associated with the

time ospr ernetcobmybincuatriroenntcecnaterrriseBrs in
traps
BARKER
Fig. 8 shows some of these p-TYPE REGION REGION

processes for an n-type photoconduc- EXCESS n-TYPE

ELECTRONS~,,,, REGION CONDUCTION

tor. At condition (1), the absorption ^/ BAND

of a photon forms a free electron- ENERGY, I~.-FERMI LEVEL.
by ENERY GAP (I.I ev)
hole pair; (2) under the applied elec- ~~~~~~~~~ VALENCE BAND

tric field, the photo-excited electron WAFER THICKNESS

moves toward the anode; (3) simi- Fig. 9. Energy model of a silicon solar cell.

larly, the photon-excited hole moves

toward the cathode; (4) the hole is

captured at an imperfection center;

(5) after the initial electron has left For the p-type, the Fermi level lies
near the bottom of the forbidden
the photoconductor at the anode, the gap; for the n-type, it is near the
top. When the two types of silicon
residual positive space charge due are in intimate contact, a potential

to the excess captured hole leads to adjustment takes place across the

the entrance of an electron into the boundary. Electrons flow from the
n-type material to the lower vacant
photoconductor from the cathode. levels of the p-type material, and
holes flow across the boundary in
Photocurrent continues until a free the opposite direction. When the

electron recombines with the cap- Fermi levels are at the same height,

tured hole. the current ceases to flow, as shown

OHMIC CONTACT in the potential-energy model. A con-
— ,rte cONOG tact potential equal to the original
cr/Q. difference in Fermi levels exists; the
p-type material is negative. When
LOCALIZED the area in the neighborhood of the
IMPERFECTIONS junction is illuminated, hole-electron

ANODE' pairs are created. The minority car-

CATHODE \~ riers created (holes in the n-type

// \ QNo silicon and electrons in the p-type
silicon) then flow across the junction
Fig. 8. An n-type photoconductor in opera- and constitute the current developed
tion. Two ohmic metallic contacts to the by the photovoltaic cell.
photoconductor are assumed.

Theory and Measurements 11

Photoelectric Measurements 3

Light 2

Historically, photoelectric meas- i
urements and specifications have
been related to the measurement of ~o
illumination. It is appropriate, there-
fore, to consider the characteristics -I
of the eye and related photometric
units. Although characteristics of the N -2
human eye vary from person to per-
son, standard luminosity coefficients w -3

for the eye were defined by the ~ -4
Commission Internationale de Eclair-
age (International Commission on Il- a -5
lumination) in 1931. These standard
C.I.E. luminosity coefficients are oJW: —s
listed in Table II. They represent the
relative luminous equivalents of an o -7
equal-energy spectrum for each
wavelength in the visible range, as- "' -s
suming foveal vision. An absolute
"sensitivity" figure established for -9
the standard eye relates photometric
units and radiant power units. At -10
5550 angstroms, the wavelength of 3000 5000 7000 9000 1100(1,
maximum sensitivity of the eye, one WAVELENGTH —ANGSTROMS
watt of radiant power corresponds to
680 lumens. Fig. 10. Relative spectral sensitivity of the
dark-adapted fovea and peripheral
The sensitivity of the eye outside retina. (ref. 10)
the wavelength limits shown in Table
II is very low, but not actually zero. have shown that the eye is sensitive
Studies with intense infrared sources to radiation of wavelength at least as
long as 10500 angstroms. Fig. 10
shows a composite curve given by
Griffin, Hubbard, and Wald10 for the
sensitivity of the eye for both foveal
and peripheral vision from 3600 to
10500 angstroms. According to
Goodevel' the ultraviolet sensitivity
of the eye extends to between 3125
and 3023 angstroms. Below this level,
the absorption of radiation by the

TABLE II

Standard (C.I.E.) spectral luminous-effir,iency values (relative to
unity at 555 millimicrons wavelength). These values represent the
relative capacity oprfodraudcieanvtiseunael rsgeynsoaf tivoanrsio.lu2 s wavelengths to

Wavelength C.I.E. Wavelength C.I.E. Wavelength C.I.E.
Value (angstroms) Value Value
(angstroms) (angstroms)

3800 0.00004 6100 0.503 e400 0.176

3900 0.00012 6200 0.710 6600 0.107
4000 0.0004 6.^,00 0.862 6600
4100 0.0012 6700 0.061
6400 0.954 6800 0.032
4200 0.0040 6500 0.995 6900 0.017
0.0116 6600 7000 0.0082
4300 0.023 0.995 7100 0.0041
4400 0.038 6700 0.952 0.0021
4500 0.060 6800 7200 0.00105
4600 0..091 b900 0.870 7300 0.00052
4700 0.139 6000 0.757 7400 0.00025
4800 0.208 6100 0.631 7500 0.00012
4900 0..323 6200 0.503 7600 0.00006
6000 6300 0.381
0.265

12 RCA Phototube Manual

proteins of the eye lens apparently A suitable substandard for prac-
limits further extension of vision tical photoelectric measurements is
into the ultraviolet. Light having a the developmental-type calibrated
wavelength of 3023 angstroms is lamp, RCA Dev. No. C70048, which
detected by its fluorescent effect in
the front part of the eye. operates at a current of about 4.5

Photometric Units amperes and a voltage of 7 to 10
volts. A typical lamp calibrated at a
Photometry deals with the meas-
urement of light in reference to the color temperature of 2870 degrees
effect produced on the theoretical Kelvin provides a luminous intensity
standard C.I.E. observer. Measure- of 55 candelas. The luminous inten-
ments are made by visual comparison sity of a tungsten lamp measured in
or by some equivalent photoelectric candelas is usually numerically some-
method. Units, standards, and sys- what greater than the power de-
tems of measurement have been livered to the lamp in watts.
developed to correspond to the ef-
fect as observed by the eye. Luminous flux is the time rate of
flow of light energy—that character-
Luminous intensity (or candle- istic of radiant energy which pro-
power) is a measure of a light source duces visual sensation. The unit of
which describes its luminous flux per luminous flux is the lumen, which
unit solid angle in a particular direc- is the flux emitted in unit solid angle
tion. For many years the standard by a uniform point source of one
measure of luminous intensity was candela. Such a source produces a
the international candle established total luminous flux of 4 ~ lumens.
by a group of carbon-filament lamps
at the Bureau of Standards. In 1948 A radiant source may be evalu-
the International Commission on Il-
lumination agreed on the introduc- ated in terms of luminous flux if the
tion of a new standard of luminous radiant-energy distribution of the
intensity and recommended the adop- source is known. If W (~) is the to-
tion of the name candela to dis- tal radiant power in watts per unit
tinguish it from the international wavelength, total radiant power over-
candle. The term candela is now
widely used abroad and is coming all wavelengths is 00
into general use in the United States; W (7~) d7~,
the older term candle is sometimes
still used, but refers to the new 0
candle or candela.
and the total luminous flux L in lu-
The candela is defined by the mens can be expressed as follows:
radiation from a black body at the
temperature of solidification of plat- L —~ ~ [680 W (a)7 [Y (~)] d~
inum. Acandela is one-sixtieth of
the luminous intensity of one square 0
centimeter of such a radiator. The (6)
major advantage of the new standard
is that it may be reproduced in any where y (A) represents the luminos-
laboratory. The effective change in
the value of the candle as a result ity coefficient (Table II) as a func-
of the 1948 agreement is of the order tion of wavelength.
of tenths of one per cent and, there-
fore, is negligible in practical meas- The lumen is the most widely
urements. used unit in the rating of photo-
emissive devices. For diode photo-

tubes, typical test levels of luminous
flux range from 0.01 to 0.1 lumen;
for multiplier phototubes, the range
is from 10-4 to 10~` lumen (0.1 to 10
microlumens).

Stars of various magnitudes are
frequently measured photoelectri-

cally. The flux in lumens L from a
star of magnitude m which is re-
ceived by a telescope having a diam-

eter of d inches can be expressed

as follows:

Theory and Measurements 13

2.5 login (L) — 7.57 — 30 l04
-{- 5 loges (d) — m

(7)

An increase of one magnitude indi- 10 3

cates adecrease of 100 in illumi- W io2

nation. 0
a
Illumination is the density of ~ 10
0
luminous flux incident on a surface. a0.
zI 1
A common unit of illumination is the 0
footcandle, which is the illumination aIO~
z
produced by one lumen uniformly
J 102
distributed over an area of one J

square foot. It follows that a source 103

of one candela produces an illumina-

tion of one footcandle at a distance

of one foot.

Table III lists some common l04

values of illumination encountered in

photoelectric applications. Further 5
10_4 -3 -2 -I
information concerning natural radi- 0 23

ation is shown in Fig. 11, which HOURS BEFORE AND AFTER SUNSET
indicates the change in natural il-
Fig- 11. Natural illumination on the earth
i lumination at ground level during, for the hours immediately before and
before, and after sunset fora con- after sunset with a clear sky and no
dition of clear sky and no moon" moon.

Photometric luminance (or levels of illumination. The term
brightness) is a measure of the luminance describes the light emis-
sion from a surface, whether the
luminous flux per unit solid angle surface is self-luminous or receives
leaving a surface at a given point its light from some external luminous
body.
i in a given direction, per unit of pro-
jected area. The term photometric For a surface which is uniformly
diffusing, luminance is the same re-
luminance is used to distinguish a gardless of the angle from which
physically measured luminance from the surface is viewed. This condition
a subjective luminance. The latter results from the fact that a uni-
formly diffusing surface obeys Lam-
varies with illumination because of

the shift in spectral response of the

eye toward the blue region at lower

TABLE III

Typical values of illumination for various conditions.

Condition Illumination

(Footeandlea)

Average solar illumination, 42° N latitude noon, 8800
June 21, measured in a plane perpendicular to the
sun's rays.

Recommended for reading. 30 to 70

Moonlight. 0.02

Natural night illumination, clear, no moon.• 9 x 10;

Natural night illumination, heavily cloudy, 2 x 10~
no moon.

' Although published data are in some disagreement, it appears that starlight itself

is a substantial but not a major component of the quoted figure. Note also that zodiacal
radiation in the near-infrared is many orders of magnitude higher than in the visible
region.

14 RCA Phototube Manual

bert's law, or the cosine law of L in lumens from the area a may be
emission. Thus, both the emission obtained by integration over the
per unit solid angle and the pro- hemisphere as follows:
jected area are proportional to the
cosine of the angle between the di- ~~r/2 1
rection of observation and the sur- L = — (A cos e)
face normal.
0 '~
A logical unit of luminance
based on the definition given above (2a sin e) dA = A (g)
is a candela per unit area. When the
unit of area is the square meter, this In other words, the total flua from
unit is called a nit; when the unit a uniform diffuser having a lumi-
of area is a square centimeter, the nance of one footlambert is one
unit is a stilb. It is also possible to lumen per square foot.
refer to a candela per square foot.
However, none of these units is as Illumination From
commonly used in photoelectric Uniformly Diffusing
measurement as the footlambert,
which is a unit of photometric lumi- Surfaces
nance equal to 1/,r candela per square
foot. The advantage of using the An advantage of the above rela-
footlambert for a uniform diffuser is tionship is that the illumination at
that is it equivalent to a total emis- a surface in front of and parallel
sion of one lumen per square foot to an extended and uniformly diffus-
from one side of the surface. This ing surface having a luminance of
relationship can be demonstrated by one footlambert is equal to one
the following conditions as shown in lumen per square foot or one foot-
Fig. 12: an elementary portion of a
candle. As a result, an instrument
Fig. 12. Diagram illustrating Lambert's law reading illumination in footcandles
and the calculation of total luminous indicates photometric luminance or
flua Yrom a diffuse radiator. brightness in footlamberts if the in-
strument is illuminated essentially
diffusing surface, having an area of from the entire hemisphere. (This
A square feet has a luminance of one statement neglects the possible per-
footlambert, or 1/,r candela per turbation caused by the measuring
square foot. Consider the light flux instrument.)
which is emitted into an elementary
solid angle, 2a sin A d0. At an angle In a typical application, a uni-
of A, the projection of the elementary formly diffusing radiating surface
area is equal to A cos A. Because the may be of such small size that it
luminous flux in a particular direc-
tion is equal to the product of the can be considered practically a point
source strength in candelas and the
solid angle, the total luminous flux source. However, if the radiator is

assumed to be a flat surface radiat-

ing according to Lambert's law, the
distribution of flux about the point

is not the same as for an ordinary
point source. In this case, if the sur-
face luminance is one footlambert
and the area is A square feet, the
flux per steradian in a direction
normal to the surface would be
A(1/,r) lumens, or at an angle A
with respect to the normal line the
flux would be A(1/~) (cos e) lumens
per steradian.

Table IV provides a reference of
luminance values for a number of
common sources.

All photometric data in this
manual are presented in units of

candelas, lumens, footcandles, and

Theory and Measurements 15

TABLE IV

1is Typical values of lumvainrioaunsces(opuhrcoetosm.l9etric brightness) for

Source Luminance

(Footlamberts)

Sun, as observed from Earth's surface at meridian 4.7 x 108

Moon, bright spot, as observed from Earth's surface 730

Clear blue sky. 2300

Lightning flash. 2 x 1010

Atomic fission bomb, 0.1 millisecond after firing, 6 x 1011
90-feet diameter ball.

Tungsten filament lamp, gas-filled, 16 lumen/watt. 2.6 x 108

Plain carbon are, positive crater. 4.7 a 108

Fluorescent lamp, T-12 bulb, cool white, 430 ma, 2000
medium loading.

footlamberts; Table V permits con- the case of an infrared detector,
version to other units if such con- luminous sensitivity is practically
version is required. meaningless although luminous sen-
sitivity is sometimes arbitrarily de-
Luminous and Absolute
Rating Systems fined as the response of the device

Although the practice of using to the whole radiation of the test
luminous ratings for photosensitive source (usually a tungsten lamp)
devices is almost universal in the and the radiant flux from the lamp
photoelectric industry, there are is defined by its luminous flux. Some-
situations in which an absolute rat- times, in order to convey this par-
ing system, or system based on ticular meaning, the term hololumen
radiant power units instead of photo- is used in place of lumen. At other
metric units, is more appropriate. times, the device may be rated di-
rectly in terms of radiant power—
All photodetectors do not have for example, by use of the power in
the spectral-sensitivity range of the watts projected on the device from
eye; most do not even closely ap-
proximate its spectral response. In a black-body source of prescribed

temperature.

TABLE V

Conversion table for various photometric units.

1 footcandle (lumen/fts) = 10.764 ]ux (meter candle)
1 footlambert (1/~ candle/fte) (lumen/meters)

= 0.001076 phot (lumen/em2)

= 0.0010764 lambert (1/~ candle/cros)

= 1.0764 millilamberts

= 3.426 nits (candle/meters)
= 0.0003426 stilbs (candle/ems)
= 0.3183 candle/fts
= 10.764 apostilbs (1/.,,. candle/meters)

16 RCA Phototube Manual

In general, the specification of in the visible spectrum. If the radiat-
the luminous sensitivity of a device ing body is one which may be de-
is not sufficiently definitive; the dis- scribed technically as "black", its
tribution of radiant energy from the behavior may be accurately described
source as well as the spectral re- by the laws of radiation. Because the
sponse of the device and the spectral black-body radiation is used as a
sensitivity of the eye must be known.
However, because many photoelectric standard for the infrared region and
devices are fabricated to approxi-
mate the visual range, the rating of because other sources may be de-
these devices in luminous terms is
convenient, although not entirely scribed in terms of the black body,
unambiguous. a brief review of black-body radia-
tion laws and standards is given be-
A practical advantage of using low.
luminous standards for testing
A black body is one which ab-
photoelectric devices is that the test sorbs all incident radiation; none is
standard usually is a tungsten lamp transmitted and none is reflected.
calibrated for color temperature and Because, in thermal equilibrium,
luminous intensity. Secondary stand- thermal radiation balances absorp-
ards can readily be prepared by tion, it follows that a black body is
photometric comparison. the most efficient thermal radiator
possible. A black body radiates more
Because the spectral emission of total power and more power at a
the tungsten lamp is known, the particular wavelength than any other
radiant sensitivity of a photodevice thermally radia+ing source at the
can be calculated by rather simple same temperature. There are two
procedures if the relative spectral- general theoretical laws which de-
response characteristic of the device
is known:" Therefore, the luminous scribe the radiation from a black
rating is not just an isolated value.
body.
For devices sensitive primarily The Stefan-Boltzmann law de-
in the ultraviolet region, the tung-
sten lamp is a poor standard because scribes the total radiant flux W per
of the very small amount of ultra-
violet radiation it provides. Such unit area from a black body as a
devices are sometimes rated in terms function of temperature, as follows:
of watts of monochromatic power.
Similarly, in the infrared region W — o T' (9)
luminous reference is of doubtful
meaning. For this reason, and be- where c is equal to 5.6819 x 10-"
cause infrared-sensitive devices are watt per square centimeter per de-
frequently used as detectors of ther- gree', and T is the temperature of
the radiator in degrees Kelvin.
mal radiation, such devices are
often rated in terms of the radiation Planck's radiation law describes
of a black body at a temperature the spectral distribution of black-
that can provide a spectrum rich in
the energy of the spectral region body radiation as follows:
characteristic of the device.
C~/~T
Radiant Energy
J~ — C~ ~-` (e —1)-' (10)
Black-body radiation:* As a body
is raised in temperature, it first where Ja is the power in watts in the
emits radiation primarily in the in- complete solid angle of 2a steradians
visible infrared region; then, as the on one side of the black-body plane
(Lambert's cosine law applying) p¢r
temperature is increased, the radia- unit area for an increment of wave-
length of one centimeter; C~ is equal
tion shifts to the shorter wavelength to 3.7413 x 10-" square centimeter-
watts (the first radiation constant);
~ is the wavelength in centimeters;
and C~ is equal to 1.4380 centimeter-

degree (the second radiation con-

stant). Fig. 13 illustrates the dis-

Theory and Measurements 17

125 I ~~~`4000°ABSOLUTE

ZOa 100 3000°
0aa: 2500°
0 75 2000°
>ocw~-:
W 50
w

F
~w 25

0 4000 8000 12000 16000 20000

WAVELENGTH—ANGSTROMS

Fig. 13. Distribution of black-body radiant energy as a function of wavelength at various
temperatures.

tribution of black-body radiant en- ards for the visible range are
ergy as a function of wavelength for frequently given in terms of color
a number of different temperatures temperature, instead of true tem-
as calculated from Eq. (10).
perature. The color temperature of
Although no materal is ideally a selective radiator is determined by
black, the equivalent of a theoretical comparison with the true tempera-
black body can be achieved in the ture of a black body; when the out-
laboratory by providing a hollow
radiator with a small exit hole. The put of the selective radiator is the
closest possible approximation of a
radiation from the hole approaches perfect color match in the range of

that from a theoretical black radia- the visual sensitivity, the color tem-

tor if the area of the cavity is large perature is numerically the same as
the black-body temperature. For a
compared with the area of the exit tungsten source, the relative distri-
hole. The characteristic of 100-per- bution of radiant energy in the vis-

cent absorption is achieved because ible spectral range is very close to
any radiation entering the hole is that of a black body although the
reflected many times inside the absolute temperatures differ. How-
cavity. For many years experimen- ever, the match of energy distribu-
tal physicists had to build their own tion becomes progressively worse in
black-body radiators; however, well-
designed commercial radiators are the ultraviolet and infrared spectral

now available. regions.

For a radiation source which is Radiation sources and character-

not black, the radiation may be cal- istics. Relative spectral-emission
characteristics for a number of im-
culated from black-body radiation
portant radiation sources are shown
laws provided the emissivity as a
function of wavelength is known. in Fig. 14. Each of these sources is
Spectral emissivity e\ is defined as described briefly belo~v.
the ratio of the output of a radiator
at a specific wavelength to that of Tungsten lamps are probably
a black body at the same tempera- the most important type of radiation
ture. Tungsten sources, for which source for photoelectric applications
tables of emissivity data are avail- because of their availability, reli-
able,lfi are widely used as practical ability, and constancy of operating
standards, particularly for the visi- characteristics. Commercial photo-
ble range. Tungsten radiation stand- tube design has been considerably

18 RCA Phototube Manual

MERCURY LAMP
40

FLUORESCENT LAMP —DAYLIGHT X30
60
w
z
w
w20

>~ 40 Q
w 10
as
wz 20 V ~.A `,.
i 5000 6000 "~~
rc W 0 0 4000 5000 6000 7000
70 WAVELENGTH— ANGSTROMS
3800 7000
ZIRCONIUM CONCENTRATED
WAVELENGTH —ANGSTROMS ARC LAMP-100 WATT

y9 CARBON-ARC LAMP so

y7 _\ r
w
~- 6 4000 5000 6000 7000 ~ 50
z WAVELENGTH—ANGSTROMS w
W 40
Fz~ 5 W
Q ~ 30
p4
Q a
~3 w 20
a rc
F2
V 10
al
N 0 J~

3000 4000 5000 6000 7000

WAVELENGTH —ANGSTROMS

SOLAR ENERGY DISTRI@UTION TUNGSTEN LAMP 2870° K SSTAN DARD TEST

Y 1.6 I
~i
2
g z F0.8
a~
ow-o 1.2 Sm

¢~ ¢ 0.6
W LL I-
300.8 WN
oaNu, ~ ~ 0.4
a~
a31Z 0.4 J~
~ z 0.2
a ~~/- w
I0
2000 6000 10000 14000 (6000
WAVELENGTH —ANGSTROMS O 800 1600 2400 3000

WAVELENGTH —ANGSTROMS

Fig. 14. Relative spectral-emission characteristics for several representative radiation source8.

influenced by the characteristics of cal curve for a fluorescent lamp of
the tungsten lamp. the "daylight" type.

The common fluorescent lamp, A very useful point source1e is
which is a very efficient light source,
consists of an argon-mercury glow the zirconium concentrated-arc lamp.
discharge in a bulb internally coated
with a phosphor that converts ultra- Concentrated-arc lamps are available
having ratings from 2 to 300 watts,
violet radiation from the discharge and in "point" diameters fx•om 0.003
into useful light output. There are
numerous types of fluorescent lamps to 0.116 inch. Operation of these
having different output spectral dis-
lamps requires one special circuit to
tribution depending upon the phos- provide a high starting voltage and
another well-filtered and ballasted
phor and gas-filling. The spectral circuit for operation.
response shown in Fig. 14 is a typi-
Although many types of elec-
trical discharge have been used as

Theory and Measurements 19

radiation sources, probably the most 0

important are the mercury arc and If the total power of the source
is known (as it usually is) and the
the carbon are. The character of the relative distribution of the radiant
light emitted from the mercury arc flux is known from data such as Fig.
♦aries with pressure and operating
conditions. At increasing pressures, 14, an absolute value of W(~) may
the spectral-energy distribution from
the arc changes from the typical be obtained by performing the inte-
mercury-line spectra characteristic gration indicated in Eq. (11) with
to an almost continuous spectrum the known relative values W,.,(~)
of high intensity in the near-infra- and solving for the calibration con-
red, visible, and ultraviolet regions.
Fig. 14 includes the spectral-energy stant Vv(~)
distribution from a water-cooled
mercury arc at a pressure of 130 wre,(~)
atmospheres. The carbon arc is a
source of great intensity and high If the receiver has an acceptance
color temperature. A typical energy- area A normal to the line from the
distribution spectrum of a do high- source to the receiver, and the dis-
intensity arc is shown in Fig. 14. tance from the source to the receiver
is d, the total power intercepted by
Matching sources and receiver. the device (assuming a uniform
In applications involving a photo- spatial distribution of radiant flux)
detector and a radiation source, it can be expressed as follows:
is frequently necessary to estimate
WTotal A A co
or calculate the expected response
from the combination. Fora tung- (12)
sten-lamp source operating at normal
brightness, the response is the Let R(~) represent the spectral-re-
sponse characteristic function for the
product of the luminous flux from receiver in amperes per watt. Be-
the lamp and the sensitivity of the cause the response of the device at
photodevice specified in terms of any wavelength is the product of the
response per lumen for a tungsten power intercepted at that wavelength
source (a standard color tempera- and the sensitivity at that wave-
ture of 2870 degrees Kelvin is length, the total response I in am-
usually specified). For other com-
peres of the device to the source for
binations of source (or source tem-
all wavelengths is given by
perature) and receiver, the same
A ~ W(a) R(~) da
simplification does not follow, unless I= 0
the spectral distribution of the
~4 d' (13)
source in the range of the receiver
A linear relationship between re-
corresponds closely to that of the
tungsten lamp with which the re- sponse and radiation is assumed.
ceiver has been calibrated.
Given only the relative spectral
In general, the calculation of the
expected response of a particular response of the photosensitive device
r(a), normalized to unity at the
photodetector and a radiation source wavelength of maximum response,
involves a consideration of the point- and the luminous sensitivity of the
by-point match of the spectral emis- device s in amperes per lumen from
sion and the spectral response. a tungsten lamp at a specified color
Specifically, the response may be cal- temperature, the absolute sensitivity
culated as follows: of the device, R(~), may be calcu-
lated18 as follows:
If W(~) is the radiant power
from the source in watts per unit If o is the absolute sensitivity of
wavelength, the total power from the device at the wavelength of peak
the source is given by

20 RCA Phototube Manual

sensitivity in amperes per watt, the through their representation in the
sensitivity R (~) is given by Electronic Industries Association
(EIA) have set up a series of regis-
The term w(~) is used to rep- tered S-numbers which indicate the
resent the energy distribution of the relative spectral response of the de-
radiation from the tungsten lamp vice, i.e., a combination of the trans-
falling on the active area of the mission characteristic of the en-
photosensitive device. velope or window as well as the
basic photosensitivity of the detector
The light flux in lumens on the element. These numbers are often
device is then given by misused to indicate the photosensi-
tive material or the basic photosen-
L — 680 ~ y(>`) w@) d~ sitivity of the detector element
alone; it is important to note the
0 distinction. Table VI lists the cor-
(15) rect meaning of particular S-num-
bers.
(See Eq. 6.) As in Eq. (13), the total
current I developed is given by References

w(~) Q r(~) dA 3 Tamm and Schubin, "Theory of the Photo-
effect of Metals", Z. Physik, 68 (1931)
(16)
$Mayer, H. and Thomas, H., "External
Because the sensitivity s is spe- Photoeffect of Alkali Metals", Z. Phyaik,
cified in amperes per lumen, the 147 (1957)
equation for s can be written as fol-
loWs: ° Seiler, E. F., "Color Sensitiveness of Photo-
sensitiveness of Photoelectric Cells", Aatro-
8=I/L= w(~) o r(1.) dl~ phys. J., 52 (1920)
~~ 0
E Brady, J. J., "Energy Distribution of
680~~ y(A) w(~) d7~ Photoelectrons as a Function of the Thick-
0 ness of a Potassium Film", Phys. Rev., 46,
(17) (1934)

Finally, the absolute sensitivity of s Goerlich, P., Z. Phys, 101, (1936)
the device in amperes per watt at the
wavelength of peak sensitivity can Sommer, A. H., "New Photoemissive Cath-
be obtained by solving for o. odes of High Sensitivity", Rev. Sci. Inatr.,
26, No. 7, (1955)
680s~~ y(J~) w(1.) dl~
0 a Spicer, W. E., "Photoemissive, Photocon-
ductive, and Optical Absorption Studies of
O= Alkali-Antimony Compounds", Phys. Rev.,
112, No. 1, (1955)
w(1~) r(1,) dA
0 s Bube, R. H., Photoconductivity of Solids,
John Wiley &Sons, New York, (1960)
(1$)
a Rose, Albert, "An Outline of Some Photo-
Because w(~) appears in both the conductive Processes", RCA Review, 12,
numerator and the denominator, it (1951)
is not necessary that this value be
any more than a relative function to Griffin, D. R., Hubbard, R. and wald, G.,
showing the distribution of energy "The Sensitivity of the Human Eye to
from the tungsten lamp. Infrared Radiation", JOSA, 37, No. 7,
(1947)
Spectral-response designation. To
facilitate the designation of the spec- n Goodeve, C. F., "Vision in the Ultraviolet",
tral response of photodetectors, the Nature, (1934)
manufacturers of such devices
v I.E.S. Lighting Handbook, Illuminating
Engineering Society, New York, N.Y.,
(1959)

Theory and Measurements 21

TABLE VI

Typical combinations of photosensitive surfaces and window
materials which can provide the basic spectral-response
designations standardized by E.I.A.

Spectral Photosensitive Material Envelope
Response Type of Ag-O-Cs Lime-glass
Number Photodetector

S-1 Photocathode Ag-O-Rb Lime-glass
S-2• Cs-Sb Lime-glass
S-3 Photocathode Cs-Sb UV-transmitting glass
S-4 Photocathode Na Unspecified
S-5 Photocathode Cs-Rb-O-Ag Pyrex
5-6 Photocathode Cs-Bi Lime-glass
S-7 Photocathode Cs-Sb (semitransparent) Lime-glass
S-S Photocathode Ag-Bi-O-Cs Lime-glass
S-9 Photocathode
5-10 Photocathode (semitransparent) Lime-glass
Lime-glass
5-11 Photocathode Cs-Sb (semitransparent)
5-12 Photoconductor Fused silica
CdS (crystal with plastic Lime-glass
5-13 Photocathode coating)
5-14
Photojunetion Cs-Sb (semitransparent)
5-16 (Photocell )
Ge
5-16 Photoconductor
CdS (sintered) Lime-glass
S-17 (Photocell )
Photoconductor CdSe Lime-glass

(Photocell ) Cs-Sb Lime-glass
(reflecting substrate) Lime
Photocathode
SbzSa
S-18 Photoconductor
(Vidieon) Cs-Sb Fused silica
5-19 Sb-K-Na-Cs Lime-glass
S-20 Photocathode UV-transmitting glass
(semitransparent)
Photocathode
Ca-Sb (semitransparent)
5-21 Photocathode
S-22 Presently Rb-Te Fused silica
Na,KSb Lime-glass
5-23 unspecified
5-24 Photocathode
Photocathode

~ Now obsolete. Formerly a variation similar to S-1, discarded by EIA action to reduce
confusion.

References cont'd iT Zworykin, V. K. and Ramberg, E. G.,
Photoelectricity and its Application John
a Engstrom, R W., "Absolute Spectral- Wiley and Sons, Inc., New York, (1949)
Response Characteristics of Photosensitive
Devices", RCA Review, 21, (1960) is Hughes, A. L. and DuBridge, Photoelectric
Phenomena, McGraw-Hill Book Co., Inc.,
u Forsythe, W. E., Measurement of Radiant New York (1932)
Energy, McGraw-Hill Book Co., N.Y. (1937)
1B Walsh, J. W., Photometry, Constable and
~ DeVos, J. C., Physics, 20, (1954) Co. Ltd., London, (1953)

to Buckingham, W. D. and Deibert, C. R,
"Characteristics and Applications of Con-
centrated-Arc Lamps", J. Soc. Motion
Picture Eng., 47, (1946)

Vacuum

Pl~ototubes

Construction and Principles positive potential, approaching that
of Operation of the anode. If the applied voltage
is too low to provide an effective

I N a vacuum phototube, one of the secondary-emission ratio* greater
than unity, the glass becomes
simplest of photodetectors, the es- charged negatively to approximately
sential elements are a photocathode,
an anode, an envelope, and a suit- cathode potential.
Fig. 15 shows a perfectly cylin-
able termination or base. The shape
drical electric field in which the elec-
of the photocathode is determined by tron-emission energy is E electron-
the particular optical requirements volts and the emission velocity is
of the application and the general tangent to the surface of the cathode
electron-optical requirement that and in a plane normal to the axis of
electrons emitted from the cathode the tube. When the electron path is

must be collected by the anode. The

most common cathode shape is semi-

cylindrical; one side is open to admit

light and an anode rod is located ap-

proximately at the center line of the

cylinder. An ideal geometrical lay-

out for omnidirectional response

without interference from the anode

is provided by evaporating cathode

material on the inside of the en-

velope window to forma semitrans-

parent cathode and locating the

anode in the center of the bulb.

Cup-shaped cathodes have been used

with ring-shaped anodes.

The anode of a vacuum photo-

tube is usually made small to mini-

mize obstruction of the light falling

on the photocathode. As a result, the Fig. 15. Paths of electrons i a tube having
anode may not collect all the elec- cylindrical geometry. nDirections of
emission velocities are (a) tangential
trons emitted from the cathode. This and (b) radial.

situation arises because of the tan-

gential component of the electron-

emission velocity which causes some just tangent to the anode, or when
the electron path impinges on the
of the electrons to miss the anode anode, a consideration of conserva-
and strike the glass window. When tion of angular momentum and en-
the energy of the electrons striking ergy leads to the following relation:
the glass wall is sufficient, secondary

electrons are emitted from the glass

and may be collected by the anode. • Effective secondary-emission ratio is de-
If the secondary-emission ratio is fined as the ratio of collected secondary-
substantially greater than unity, the electron current (total secondary-emission
current less return current) to primary
glass window becomes charged to a
current.

22

Vacuum Phototubes 23

b~ _ V+E (19) increase in the collected current. The
a' ~ E
top of the loop represents the condi-
where b is the cathode radius, a is
the anode radius, and V is the poten- tion in which the glass is positively
tial difference between anode and charged. Although the case of Fig.
cathode. For an electron energy of 16 is an exaggerated one because of
one volt and an applied potential of the small anode size, some instability
100 volts, the anode radius must be of the output current is occasionally
greater than one-tenth the radius of observed in conventional tubes. An
anode voltage of about 250 volts
the cathode in order to collect the eliminates the condition entirely.
emitted electrons. In a typical con-
struction, the radius of the cathode The construction of the photo-
tube stem and envelope follows the
cylinder is approximately 8 milli- standard practices of the electron-
meters and the anode radius is O.b tube industry. A typical construction
millimeter. In this case, the number is shown in Fig. 17. Choice of metals
and glasses for specific sealing con-
of electrons which strike the glass
ditions is governed principally by
depends upon the wavelength of the their rates of thermal expansion.

exciting radiation. Table VII lists expansion coefficients
Fig. 16 shows the sort of erratic for the more common glasses and
metals used in the manufacture of
behavior that occurs when the anode phototubes:
radius is too small (in this case the
ratio of the cathode radius to the
anode radius is about 30:1). Data

3.0- `UNSTABLE
ww UNSTABLE--~~

rc w ARROWS INDICATE THE CATHODE
~a ORDER IN WHICH DATA
owaa~ 2.0-~ ACTIVATING
PELLET ~
UV WERE TAKEN
Fig. I7. 1~pical construction of a vacuum
J i 1.0 phototube.

0U An essential process in the
manufacture of most phototubes is
o zo ao r so ~~ r Isio Iso zoo the introduction of cesium. Although
in some special phototubes cesium
so roo Izo lao is introduced from aside tube, the
most common method of introduc-
APPLIED VOLTS ing this activating material is by
means of an activating pellet, as
Fig. 16. Erratic current-voltage character- shown in Fig. 17. A mixture of
istic in a vacuum phototube caused by cesium chromate and silicon is
bulb charges which result when the
anode is too small to collect all emitted
photoelectrons.

shown are for a nonstandard 935

vacuum phototube having an anode-

wire diameter of 0.020 inch (one-
half the normal dimension). The out-
side of the envelope was wrapped in
s metal foil except for the window;
the foil was at cathode potential.
The phenomenon was also observed
when the foil was at anode poten-
tial and when no foil was used.

On the lower branch of the

"hysteresis" loop, as the voltage is

increased, the glass becomes nega-
tively charged. At the unstable point
on the right, the secondary-emission
ratio reaches unity; the result is an

24 RCA Phototube Manual

TABLE VII

Expansion Coefficients for Common Glasses and Materials Used
in the Manufacture of Phototubes

Soft Glasses Coefficient of Expansion
X 10- 7 (per °C)
0080
0120 92
7285• 89
*Radioactivity less than 10 epm/kilogram 96

Hard Glasses

7040 47
7052 46
7720 (Nanex)
7740 (Pyrex) 36
32
7750
9741 41
39

Quartz Temperature - °C 8
400
7912 (Vyeor) b0
500 sss
Lithium Fluoride 1000
833
Sapphire 903

Parallel to C-axis

Perpendicular to C-axis 50 50
500 77
Metals 1000 831

Tungsten Temperature Range - °C 46
Molybdenum 56
Nickel-Iron 0 — 500 63
25 — 500
(42 alloy) 42 per cent Ni 20 — 400 95
Nickel-Iron
30 — 310 60
(52 Alloy) 60 per cent Ni 91
Kovar 25 — 300 108
Platinum 25-100 92 radial
Chrome Iron 26 — b00 65 axial
Dumet

formed into a pill and held between Properties of
Vacuum Phototubes
the two sides of a metal container.
During processing the activator con- Spectral Response

tainer is heated to a dull-red color The spectral response of a
by the use of radio-frequency heat- vacuum phototube depends upon the
ing. An exothermic reaction takes photocathode material and the spec-
place in which silicon chromate is tral transmission of the bulb win-
formed and metallic cesium is re- dow. Although pure metals have
photoemissive properties, their re-
leased. sponse is usually rather poor and
predominantly in the ultraviolet re-
Although microphonic problems gion; for most practical purposes,
therefore, pure metallic photo-
are not common in phototubes, it is cathodes are of little value. Use-
able sensitivity has been obtained
sometimes important to prevent the only with surfaces involving alkali
introduction of a modulated current
caused by relative movement of the
elements of the phototube. Special
beads and "snubbers" are sometimes
introduced for this purpose and
metal parts are usually spot-welded

together.

Vacuum Phatotubes 25

metals. One of the first photoelectric are prepared by oxidizing a porous
surfaces to be used commercially was silver base in an oxygen glow dis-
potassium sensitized with hydrogen. charge to a degree determined by
This type of surface is produced by
the evaporation of a thin layer of the appearance of interference colors
potassium in vacuum. The surface is
sensitized by exposing it to a glow at the surface. Cesium is then intro-

discharge in a hydrogen atmosphere duced and the surface is baked at

to form a surface layer of potassium about 250 degrees centigrade until
hydride, which is presumed to be
covered with a potassium film. The a characteristic straw color results.
resultant surface may be symboli- Maximum sensitivity at the 8000-
cally represented as (K) — KH angstrom peak corresponds to only
— K' This sensitization process about 0.8-per-cent quantum effi-
greatly increases the photoelectric ciency, but the spectral range is
yield, but produces little change in wide. In addition, this type of photo-
either the threshold or the wave- cathode can tolerate a somewhat
length of maximum response. The higher ambient temperature (100 de-
sensitivity at the wavelength of grees centigrade maximum) than
maximum response, 4400 angstroms, most photocathodes.

has been reported as high as 0.023 A closely related photocathode
in which rubidium replaces cesium is
ampere per watt (about 7-per-cent used in tubes having aspectral-re-
quantum efficiency). However, as sponse characteristic identified as
shown in Fig. 18, the spectral re- S-3. Although the sensitivity of this
sponse is so limited (epecially surface is not high, it has proved use-
for applications involving tungsten ful in color-matching applications.
lamps) that the surface is not gen-
erally used for commercial applica- The most important photocath-
tions. ode currently used in vacuum photo-
tubes is the cesium-antimony "alloy"
~wo- surface developed by Goerlich a In the
a S-5 formation of this photocathode, an
evaporated layer of antimony is
W treated with cesium vapor at 170
f~aa lo---_ degrees centigrade. The resulting
photocathode, which is believed to be
J a semiconductor Cs3Sb,* is character-
J_ ized by high sensitivity in the visible
f
spectrum. The spectral response for
> I_ the cesium-antimony surface depos-
r ited on a solid backing mounted in
a lime-glass bulb is shown in Fig. 18;
E- (K)-KH-K s-I it is identified as S-4. Quantum ef-
z ficiency is occasionally as high as 31
per cent at 4000 angstroms, the
h 0.1 wavelength of peak response.
1000 ~~3000 5000 7000 9000 11000
The envelopes of most photo-
WAVELENGTH —ANGSTROMS
tubes are made of Corning 0080 lime
Fig. 18. Spectral response characteristics of
glass, which cuts off transmitted
vacuum phototubes: (K) - KH - K; S-1 ;
S-8 ; S-4 ; S-6. radiation in the ultraviolet region at

The silver-oxygen-cesium photo- about 3000 angstroms. Envelopes
cathode (used in tubes having S-1 have also been made of ultraviolet-
response) is more important com- transmitting Corning 9741 glass. A
mercially because it is more sensitive cesium-antimony cathode having the
to long wavelengths than the potas-
sium-hydride photocathode. Com- latter type of window provides the
mercial photocathodes of this type spectral response identified as S-b in

Fig. 18. Some special phototubes
have fused-silica windows, which
further extend the spectral response

in the ultraviolet region.

zs RCA Phototube Manual

to A s / GG F DC
~ H
~Z 6I
J
Vw
w ~,~~ CURVE MATERIAL THICKNESS
a G~
z A I A LITHIUM FWORIDE Imm
0 s G
Nvi 4i ~ 8 SUPRASIL 10 '
f c i
z C SAPPHIRE 1
aA e D FUSED QUARTZ 10 HC
f 21 F 4 S 6 78 9
E E VYCOR 1
D l00000
F TYPE 9741 GLASS I
aoo G TYPE 9823 GLASS 1

H LIME-GLASS 1

H 1 1G -
_ _ _ __ 3
l0000

WAVELENGTH —ANGSTROMS

Fig. 19. Transmission characteristics of various glasses used in phototube manufacture.

Transmission curves of several the output circuit. In the normal op-
glasses used in phototube manufac- erating range (region C), the in-
ture are shown in Fig. 19. These crease in current (approximately 5
curves are for a typical thickness of
1 millimeter, except as noted; ultra- per cent) is caused by a number of
violet cutoff is critically dependent factors. Some of the increase is
on the thickness of the glass. The the result of improved collection
transmittance T of glass at a partic- efficiency at higher voltage. Photo-
emission is also slightly increased
ular wavelength is described by the
following relationship: as a result of the applied electric

field at the cathode, which aids

T = K10-Rt (20) 5 C

where K is a factor (approximately a ENLARGED
w SCALE
0.9) dependent upon the surface re- ~3
flectivity, p is the coefficient of a A SECTION
absorption, and t is the thickness. 0 FI.0
Near the cutoff wavelength it is im- ¢2
portant to use as thin a glass as
practicable. Some experimental I
phototubes have windows made of a 0
thin inverted bubble of glass.
0
loo 200 300
VOLTS

Current-Voltage Fig. 20. Current-voltage characteristic for
Characteristics a typical vacuum phototube showing the

A typical current -voltage char- various regions of interest.

acteristic for a vacuum phototube is in the emission by reducing the
shown in Fig. 20. At the foot of the voltage barrier at the surface as-
curve (region A), the energy of the sociated with the photoelectric work
photoelectrons is sufficient to permit function. The increase in emission
some collection by the anode, even from this field effect is primarily
against an opposing field. As the observed in the neighborhood of the
voltage is increased in the positive
direction (region B), more of the long-wavelength cutoff of the spec-
emitted electrons are collected. How-
ever, because of the finite size of the tral characteristic. Therefore, a
anode, some of the electrons which
escape and strike the bulb are lost to phototube operated at a higher volt-
age is slightly more red-sensitive. In
some phototubes, the vacuum may

Vacuum Phototubes 27

not be sufficiently low to prevent all charge. At high current levels
ionization; this condition usually (recommended absolute-maximum
results in an increased slope of the current ratings are listed in the
current-voltage characteristic in the Phototube Data section) the tube
C range. may suffer both temporary and per-
manent fatigue, resulting from a
linearity change in surface composition. Be-
cause such fatigue is usually a func-
Vacuum phototubes are charac- tion of both current and time,
terized by a photocurrent response phototubes can often be used safely
which is linear with incident light at very high light levels if the
level over a wide range—so much so exposure is brief. The use of pulsed
that these tubes are frequently used light makes it possible to develop
as standards in light-comparison large photocurrents (up to the point
measurements. Fig. 21 shows the
linear current-light relationship where the space charge limits the
characteristic of a vacuum Photo-
tube. If a tube is to be relied upon as output current) without excessive
a standard because of its linearity fatigue effects.
characteristic, the voltage used
should be sufficient to prevent in- The problem of currents limited
stability of the type shown in Fig. 16. by space charge between coaxial
cylinders has been worked out by
120 Langmuir and Blodgett.` In practical
units, the expression for the current
loo I in amperes per unit length 1 of
axis is given by
W eo
I _ 14.66 X 10-" Va/'
w 1 r(i' ( 21)

~ so where 1 is the length of the cylinder,
0 r is the radius of the anode cylinder,
v 40 V is the applied voltage between
cathode and anode, and S is anon-
20 dimensional function of the ratio of
the cathode and anode radii. Fig. 22
O I 2345 6 shows an experimental current-volt-
age curve fora 1P39 vacuum Photo-
LIGHT FLUX-LUMENS tube; the current is plotted on a
2/3-power scale to show the limita-
Fig. 21. The linearity of current as a funo- tion resulting from space charge. The
lion of light for a vacuum Phototube. theoretical line is for an assumed
cathode radius of 0.8 centimeter, an
Because the photosensitivity of the anode radius of 0.051 centimeter, and
photocathode may vary across the a cathode length of 2.22 centimeters.
surface, the same area on the photo- Because the cylinder is only halt
cathode should be used throughout closed, there are large end effects.
the measurement. Caution should Although Eq. (21) does not take into
also be taken to avoid the effects of account the initial velocity of the
shadowing by the anode. The anode electrons, it is adequate for gen-
should either always occupy the same eral order-of-magnitude evaluations.
relative area in the light beam, or the Space charge is usually not a limita-
shadow effects should be avoided by tion in vacuum phototubes because
placing the light beam to one side.
steady currents of the magnitude
Two effects may limit the linear
operation of the Phototube at high necessary to produce space-charge
light levels, fatigue and space limitation would usually first produce
severe fatigue limitation.

28 RCA Phototube Manual

EXPERIMENTAL Although the calculation indi-
CURVE cates that vacuum phototubes may
theoretically be used at light-modu-
w i lation frequencies approaching 10'
w cycles per second (one gigacycle)
i practical difficulties preclude the
~w realization of such performance. For
0 i example, the 1P39 has an interelec-
i trode capacitance of 2.6 picofarads
i and its associated circuit further
TWO-THIRDS adds to this value; the total capaci-
POWER LAW tance is approximately a minimum of
(ASSUMING ZERO 10 picofarads. In order that the cir-
cuit time constant not limit the
W ~ INITIAL VELOCITY) response, the equivalent resistance-
i capacitance time constant of the cir-
VO cuit should be less than 0.6 nanosec-
ond. For a capacitance of 10 pico-
a it farads, the load impedance would
have to be less than 60 ohms. Unless
i the light level were high, such op-
eration would not be possible. Never-
to Is 20 theless, for light pulses of high
magnitude and short duration, the
VOLTS vacuum phototube is capable of very
short response time when coupled
Fig. EE. Experimental current-voltage curve with a minimum load resistance.
fora 1P39 taken at high values of cur-
rent to show the space-charge limitation. Noise
Note that the current scale is drawn on
a two-thirds power scale so that the As the light level and the photo-
space-charge law becomes a straight line. current become less, it becomes dif-
ficult to distinguish the photocurrent
Frequency Response from the dark current (current re-
sulting from sources other than
The inherent response time of a radiant flux on the photocathode).
vacuum phototube is exceedingly One limit of detection is in the fluctu-
short. No time delay between the ation of the dark current, or dark
incidence of light and the emission noise.

of electrons has been measured° For Dark currents in vacuum photo-
tubes arise from various sources.
a vacuum phototube such as the When both the anode and cathode are
1P39 (anode radius of 0.051 centi-
meter, cathode radius of 0.8 centi- terminated in a base attached to one
meter) with an applied potential of
100 volts, the transit time for an end of the tube, the leakage across
electron having an initial velocity of the base may be a major source of
dark current. Such leakage is par-
zero has been calculated to be 3.9 ticularly troublesome in an atmos-
phere having high humidity, espe-
X 10-' seconds (3.9 nanoseconds). cially if the base of the tube is dirty.
The transit time itself does not In a vacuum phototube, internal elec-
limit frequency response; rather it is trical leakage usually results from
limited by the spread in transit time excess photocathode activating ma-
resulting from differences in initial terial. Tubes having anode and cath-
electron velocities. For an electron ode terminations at opposite ends of
energy of 1 volt and emission veloc- the tube, especially when the separa-
ity directed toward the anode, the tion is also maintained inside the
transit time for an applied potential
of 100 volts is 3.3 nanoseconds. Thus,
a total spread in transit time of ap-
proximately 0.6 nanosecond may be
expected.

Vacuum Phototubes 29

envelope, have a minimum of elec- (Johnson noise) associated with the
trical leakage. resistor is given as follows:

In such phototubes, the main 1" i/a
source of dark current may be dark Ve~f = [ 4 kTR pf ,
emission of electrons from the photo- (23)
cathode. Dark noise in the vacuum
phototube is shot noise caused by where k is Boltzmann's constant
such random dark emission. The rms
fluctuation current in such cases is (1.38 X 10-~ Jeule per degree), T is
given by the following shot-noise
the absolute temperature, and pf is
relationship: the bandpass.

I' pf i/a The Johnson noise may be com-
1/~ — pared with the voltage across the
2e idpf ] (22) lead resistance resulting from the
fluctuating shot-noise current: R(2e
where id is the dark emission current, idpf)~/'. The signal voltage and the
e is the electron charge (1.6 X 10-1' shot-noise voltage across R both in-
coulombs), and pf is the bandpass crease directly as R, whereas the
of the measuring circuit. When the
Johnson noise voltage increases only
signal current is just equal to this as the square root of R.

rms fluctuation current, the condition Consequently, if the load re-
is that of minimum signal detection. sistance is made sufficiently high, the
signal-to-noise ratio from the circuit
This minimum signal detection
is rarely realized in vacuum photo- improves until the limitation result-
tubes. The noise associated with the ing from shot noise alone is reached
dark emission is usually very small
compared with the circuit noise. For (see Fig. 23). The resistance R for
example, for a load resistor of value which the two noise sources are equal
R, the rms-thermal-noise voltage may be determined as follows:

R(2e iapf)1~' — (4kTRpf)1"

10 s
SIGNAL VOLTAGE ACROSS LOAD RESISTANCE R CAUSED BY
PHOTOCURRENT PROVIDING 101 SIGNAL-TO-NOISE RATIO

104 SHOT PLUS
JOHNSON NOISE
N
~ Ip 5 ~i~

JOHNSON-NOISE VOLTAGE
ACROSS LOAD RESISTANCE

106 NOISE VOLTAGE ACROSS R
RESULTING FROM SHOT NOISE
,OF THE DARK-EMISSION CURRENT

10—~ ~t 106 109 1010 ID'I 1012
106 10? LOAD RESISTANCE—OHMS

Fig. 23. Variation of signal voltage, Johnson-noise voltage, and shot-noise voltage developed
across the load resistor by the dark emission—all as a function of the load resistor. Data
are for a vacuum phototube having S-1 response, dark emission of 10 picoamperes, and
temperature of 300° K, and for a bandwidth of 1 cycle.

30 RCA phototube Manual

or signed for environments of extreme
R_ 2kT shock or vibration. Ina nonrugged-
ized tube, the cathode would probably
ise be distorted in its position relative to
(24) the anode by such environments,
even to the extent of causing a short.
For example, for a vacuum Even if no permanent damage is
done to the phototube, difficulty may
phototube having S-1 r1e0sp_uonasme paenrde arise from modulation of the photo-
total dark emission of current resulting from vibration of
the photocathode in the beam of light
(10 picoamperes) at room tempera- because different areas of the photo-
cathodes may have different sensitiv-
ture, the value of R from Eq. (24) is ities. Atypical variation in cathode
sensitivity (sometimes resulting
5000 megohms. For tubes having from the heat used to seal the bulb
to the stem) is from low sensitivity
lower dark emission, such as those near the stem to high sensitivity
at the upper end of the cathode.
with S-4 response, the value of load
Under normal operating condi-
resistance which must be exceeded to tions, the vacuum phototube is one
of the most stable photosensitive
override Johnson noise becomes or- devices available. When stored in the
dark at normal temperatures and
ders of magnitude above practicabil- operated at low photocurrents, vac-
uum phototubes can be used as a
ity. Even with a 5000-megohm load reasonably good laboratory standard.
They usually show a loss in sensi-
resistance, the time constant of the tivity during continuous operation,
depending upon the magnitude of the
circuit limits the response to only a current. Fig. 24 shows typical life
characteristics of vacuum phototubes
few cycles per second. For applica- having S-4 and S-1 spectral re-
sponses for continuous operation un-
tions requiring detection of very low der the test condition.

light levels with high-frequency re- Most vacuum phototubes are
rated to several hundred volts; nor-
sponse, it is advisable to use a mally there is no need for higher
voltages. Spacings of the leads in the
multiplier phototube. stem and base are not designed for
very high voltages which would
Environmental Factors cause arc-overs, usually across the
base. For best operation, high tem-
As a rule, the sensitivity of a peratures, high humidity, high volt-
vacuum phototube is only slightly age, dirty or greasy environments,
and excessive vibration or shock
affected by the ambient temperature. should be avoided.

However, a small reversible effect Application Considerations
may be observed in the spectral
range near the long-wavelength cut- Vacuum phototubes are used to
best advantage in applications which
off where increasing the temperature exploit their stability, good fre-
tends to increase the wavelength for quency response, flat current-voltage
cutoff. Permanent changes may characteristic, and linearity of photo-
result from redistribution of the current with radiant flux.
sensitizing metals at elevated tem-
peratures.

Most phototubes are rated to a
maximum temperature in the range

of 75 to 100 degrees centrigrade.
Above the maximum rated tempera-

ture, the phototube usually suffers
permanent loss of sensitivity depend-
ing upon the length of exposure time.

As the temperature approaches the
temperature to which the tube is sub-
jected during processing sensitiza-
tion (in the range from 150 to 250
degrees centigrade), serious loss of
sensitivity occurs in less than an
hour. As the temperature is in-
creased, the dark emission also in-
creases exponentially; a typical
increase is 2:1 for every 10 degrees

centigrade.
Most vacuum tubes are not de-

Vacuum phototubes 31

p.100 922 (S-p
Wz IP39 (S-4)
U
CW
' 9D —
F
2
W

u 60 —
w0
a0z
w 40 —

F TEST CONDITIONS:
aJw
IP39-Ebb=250 V, Rb=1 MEGOHM, Ib=b MICROAMPERES AVERAGE INITIALLY
922-Ebb=500 Y, Rb=I MEGOHM, Ib=4 MICROAMPERES AVERAGE INITIALLY

~ 20 —

1 11 i i f1 Ii
50
100 150 200 250 300 350 400 450 500

TIME—HOURS

Fig. 24. Typical life characteristics for two types of phototubes having S-4 and S-1 spectral-
response characteristics.

In applications requiring the ob- phototube is not critical because of
the flat current-voltage characteris-
servation of light pulses of short tic. Aminimum of approximately 20
volts is usually recommended for
duration, or light modulated at rela- most phototubes to provide an ade-
tively high frequencies, the vacuum quate collection field, although more
than 100 volts may be desirable to
phototube performs better than gas-
prevent bulb charging caused by
filled phototubes or most solid-state
photocells. Because vacuum photo- initial electron velocities. Voltages
tubes are relatively stable over long higher than the maximum rating
(usually around 250 volts) should not
periods, they may be used as stand- be used because they may result in

ards of reference or in applications electrical breakdown between ex-
requiring long periods of operation ternal elements' of the tube.
without recalibration. The vacuum
phototube, because of its linear char- Usually, the life of a vacuum
acteristic, may be used in many ap- phototube (for a given decrease in
plications as an instrument for sensitivity) is related approximately
measuring light flux. inversely to the current drawn
through the tube. More stable and
The maximum ratings provided reliable performance results if small
in the data for vacuum phototubes areas of concentrated illumination on
should be carefully adhered to, espe-
cially in commercial applications in the cathode surface are avoided.
which many tubes are used in iden-
tical circuits. In laboratory appli- phototubes should not be stored
cations, however, it is possible to in light when not in use. Blue and
exceed published ratings if the ex- ultraviolet light especially can cause
perimenter takes into consideration photochemical changes in the cathode
which result in changes in sensitiv-
the behavior of the device and the ity. It is especially important to
reasons for the stated limitations. avoid exposure to intense illumina-
The minimum do load resistance tion such as sunlight even when no
value shown in the data for each type voltage is applied to the tube. Perma-
is recommended to prevent damage nent damage may result if the tube
to associated circuit components in is exposed to light so intense that it
the event of a short circuit in the causes excessive heating of the
phototube, which normally serves as
a high series resistance. cathode. Tubes should not be stored

The voltage supply for a vacuum

32 RCA Phototube Manual

for long periods at temperatures silicones, or other non-hygroscopic i
insulators. For example, in the case
near the maximum rating of the of a tube having atop-cap connec-
tube; high temperatures almost al- tion, acontinuous band of wax ap-
ways result in loss of sensitivity of proximately ahalf-inch wide around
the tube. the top cap or around the bulb is
sufficient to interrupt all external
A vacuum phototube may be op- leakage paths.
erated with either do or ac applied
voltage. Usually a do supply is Some phototubes have special
preferred, especially for measure- nonhygroscopic bases which provide
a substantial advantage in critical
ments involving very small currents. applications; special sockets can also
However, in some applications an ac be obtained which minimize leakage
supply can be used to advantage be- in humid conditions. Teflon is one of
cause of the time relationship pro- the best materials for such sockets.
vided; the ac supply may also be less
expensive in some applications. Be- In many applications, it is ad-
cause the vacuum phototube acts as vantageous to modulate the light flux
which is to be detected by the photo-
a rectifier, the use of steady illumi- tube. Modulation can be achieved in
nation and an ac applied voltage a variety of ways: the most common
results in approximately square method is the use of a rotating disk
waves of unidirectional current flow. or "chopper" which has a number of
In this case, a do current meter holes that modulate the light beam;
other methods use a vibrating reed or
would indicate an average current diaphragm, rotating mirrors, or a
approximately half that indicated Kerr cell.
when a do power supply is used.
Undesirable modulation may oc-
Whenever a small ac signal cur if vibration of the phototube
from the phototube is to be observed causes a shift in the position of the
and the amplifier gain is high or the light spot oh the photocathode be-
load resistance is large, it is recom- cause the photosurface may not be
mended that shielding be provided uniformly sensitive or because the
for the phototube and the signal out- anode may interrupt more or less of
put loads. It is advisable to make the the light. In general, it is desirable
signal lead as short as possible to to use a large spot of light on the
avoid pickup and stray capacitance. photocathode to minimize micro-
phonic effects and cathode current
This precaution is important if fre- density.
quency response is a consideration.
References
Because a phototube is a high-re-
1 Kohl, W. H., Material and Techniques for
sistance device, it is important that Electron Tubes, Reinhold Publishing Cory.,
New York (1960)
insulation of associated circuit parts
' Lukirsky, P. I. and Rijanoft, S., "Depend-
and wiring be adequate. In very ence of Photcemission of Potassium on
critical applications it may be desir- Arrangement of Atomic Hydrogen and
able to use a phototube in which the Potassium Layers on Its Surface", A.
signal lead (either anode or cathode Physik, 75, (1932)
depending upon polarity of signal
desired) is terminated through an • Gcerlich, P., "On Composite Transparent
insulated bulb-top cap. The power Photocathodes", Z. Physik, 101, (1936)
supply should be connected between
~ Sommer, A., "Photoelectric Alloys of Alkali
ground and the phototube element Metals", Proc. Phys. Soc. (London), 55,
not used for signal output to avoid (1943)

unnecessary pickup of extraneous ' Langmuir, I., and Blodgett, K. B., "Cur-
rents Limited by Snace CharSe between
signals. Coaxial Cylinders", Phys. Rev., 22, (1923)
For maximum sensitivity of
• Lawrence, E. O. and Beams, J. W., "In-
phototube circuits, leakage resistance stantaneity of the Phot-oElectric Effect",
of circuit parts and wiring insulation (abstract), Phys. Rev., 29, (1927)
should be high. Leakage across mois-
ture films on the surface of the glass
can be prevented by coating the
glass with pure white ceresine wax,

Gas-Filled
Photo tubes

Construction and Principles results in an ionization. As the
of Operation voltage on the phototube anode is
increased above the ionization po-

MOST gas-filled phototubes have tential, the amplification of the

the same general construction as photocurrent increases. At 90 volts,
vacuum phototubes except that an an electron averages 3 ionizing col-
inert gas such as argon at a pressure lisions while traversing the gap
of approximately 0.1 millimeter of between cathode and anode. This
mercury is introduced before the degree of ionization results in an
final sealing of the tube. Ionization of eight-fold amplification (28) of the

the molecules of the inert gas results photocurrent.
However, in practice, the actual
in amplification of the primary
amplification is greater than that
photoemission. This amplification which results from ionization because
secondary effects become more im-
provides an important advantage portant as the voltage on the tube
over the vacuum phototube for appli- is increased. The most important of
cations in which the primary photo- these effects is the release of second-
currents are small and it is necessary ary electrons when positive ions
to minimize external amplification. strike the photocathode. Other ef-
With gas-filled tubes, amplification fects of minor importance are the
factors of from 5 to 10 become quite release of secondary electrons by
practical; even higher amplification metastable atoms produced by elec-
can be used under carefully con- tron excitation in the gas, ionization

trolled conditions. by positive ions, and electron emis-
sion from photons created in the gas.
When electrons are emitted from
the cathode by photoelectric action, The combination of these effects
they are accelerated through the gas
by the applied voltage. If the energy produces an amplified current i
of the electrons exceeds the ioniza- described by the following equation:

tion potential of the gas (15.7 volts i=i, e ad
in the case of argon), collision of an 1 — y(e ad _1) (25)

electron and a gas molecule can

result in ionization, that is, the cre- where is the initiating photoelec-io

ation of a positive ion and a second tric current, a is the number of ions
electron. The probability of an ioniz- formed per electron per unit length

ing collision in a gas depends upon across the tube from cathode to an-
the energy of the electron and the ode, and y is a lumped constant

density of the gas. The mean free (nominally the number of secondary
path of an electron also depends upon electrons emitted from the cathode
per impacting positive ion, but actu-
the electron energy. In argon the ally including the other minor and
secondary sources of regenerative
mean free path is of the order of 1.4 current in the tube).
to 3 millimeters at a gas pressure of
0.1 millimeter of mercury. (For a Eq. (2b) shows that, as the volt-
general discussion of electrons in a age is increased on the tube and both
gas and related phenomena, see ref- a and y increase, a point is ultimately
erences 1 and 2.) Not every collision

33

34 RCA phototube Manual

reached at which the denominator ap- la 1
proaches zero as a result of the 0(OPERATING PO1NT I1
combined effect of the primary and FOR RL•I ME60NM)
secondary mechanisms. At this point, ~o~f -
a state of uncontrolled current is 120 I
reached, and the currentinereases to
the limit of the circuit or until a GR•=P
glow discharge sets in, which may
result in permanent damage to the WKK)
photocathode.
d W~ KDOWN VOLTAGE
Properties of
Gas-Filled Phototubes a 80 ~% JJ a

Current-Voltage ~iV< se ~'Fcoy~ ~a"
Characteristics 0
i as 2 r~
Fig. 25 shows the increase in ~
anode current of a gas-filled photo-
tube as the voltage is increased. Most zo om
commercial gas-filled phototubes are z' P
designed to operate with a 90-volt
supply. The intersection of the load 20 40 60 80 KM 120 140
line and the anode-current character- ANODE VOLTS
istic defines the operating point. The
ratio of this current to the current at Fig. 25. Current-voltage characteristic for a
25 volts (with the same load) for a
specified light flux (usually 0.1 gas-filled phototube illustrating gas-
lumen) and a specified load (usually ratio (GR), load lines, operating point,
and breakdown voltage.

1 megohm) is referred to as the gas-
ratio or GK. The breakdown voltage
is that voltage at which, with no light
on the tube, an uncontrolled dis-
charge occurs. This voltage is well
above the 90-volt maximum operat-
ing voltage to provide for stable

performance.

40

LIGHT SOURCE IS A TUNGSTEN-FlLAMENT
LAMP OPERATED AT COLOR TEh1PERATURE
OF 2870° K

30

N , CUM
W eNs,a.i
K
aW

02

J} O~
O'

W

Z V~~ ~6
a v

0

A
O~

-----_' 40 60 0Q' 100
0 2U ANODE VOLTS 80

Fig. 26. Current-voltage characteristics for various light levels for a type 918 gas-filled
phototube. Some nonlinearity with light is observed at maximum currents.

has-Filled Fhototubes 35

Variation of Current which linear response is required
With Light Flux over a wide range of light levels,
it is best to use a vacuum photo-
A series of current-voltage tube.

curves for various values of light Time- Or
flux is shown in Fig. 26 for type Frequency-Response
918. As the level of light increases,
the current increases more than Characteristics
linearly. This relationship is illus-
trated more specifically in Fig. 27, The tsme of response of a gas-
which shows the current developed filled phototube, unlike that of a
as a function of light flux for agas- vacuum tube, is limited by the sec-
filled phototube. The nonlinear be- ondary effects associated with gas
havior is caused by positive-ion amplification. Fig. 28 shows the fre-
space charge. The field strength quency-response characteristics of
gas-filled phototubes having cesium-
near the cathode is low because of antimony photocathodes and silver-
the cylindrical construction. The mo- oxygen-cesium photocathodes. Be-
bility of positive ions is much less cause response becomes increasingly
than the mobility of the electrons; poor above 10,000 cycles per second,
at a current of approximately 20 applications are limited to the audio
microamperes, the accumulation of range. Gas-filled phototubes are
positive-ion space charge is suffi- widely used in pickups for sound
cient to distort the cylindrical field reproduction, both in theaters and
and increase the electrical gradient in 16-millimeter sound systems.
near the photocathode. This in- The frequency-response character-
creased gradient provides a more istic shown in Fig. 26 was obtained
efficient field distribution for the by passing light through a toothed
production of multiple ionization wheel driven by a variable-speed
than when the bulk of the voltage motor and then through a fixed
aperture and onto the photocathode.
drop is concentrated near the anode, The teeth were so shaped that in

as in the case of the undistorted
cylindrical field. For applicatons in

40

3G

N
W

c
aw

a
0
~ 20

a00wz
io

0 0.02 0.04 0.06 0.08 0.1 012

LIuHT FLUX—LUA4ENS

Flg• 'L7. Anode current (at 90 volts, zero series resistance) in a gas-filled phototube as a
function of light flux showing the increasing nonlinearity at high levels of light flux.

36 RCA Phototube Manual

+I ANODE-SUPPLY VOLTS=90 B
VOLTAGE DROP IN LOAD VERY SMALL, A
,„ O
J CAPACITANCE EFFECTS MADE NEGLIGIBLE
m CURVE A: PHOTOTUBE HAVING S-I OR
U
S-3 RESPONSE

CURVE B: PHOTOTUBE HAVING S—-4~R'-ESPONSE

I
f

f
N
2
N2
W

aH

J
W
C-

-4 100 200 400 1000 2000 4000 10000 20000
IQ 20 40

FREQUENCY —CYCLES PER SECOND

Fig. 28. Frequency response of gas-filled phototubes:
(a) Response of a tube having S-1 spectral response (Ag-O-Ca photocathode)
(b) Response of a tube having S-4 spectral response (Ca-Sb).

combination with the aperture they Fig. 29 shows this component
produced a sinusoidal variation of
the light flux. of the delayed current for a special

The loss in high-frequency re- gas-filled Phototube designed for use
sponse is chiefly the result of the
transit time of the positive ions in- in studying the mechanism of delay
volved in the gas-amplification in gas amplification.a The very slow
process. The loss of frequency re- component of the gas-amplified
sponse becomes more severe as the photocurrent results from second-
gas amplification is increased. On ary-electron emission by metastable
the other hand, when an argon-filled atoms. The transit time of metastable
atoms (of the order of 10~ second)
Phototube is operated at a voltage is governed by diffusion time and
below the ionization point for argon, is not affected by the electric field.
it behaves very much like a vacuum
Phototube with no gas amplification z loo ~-
and little loss in frequency response. ~Z so
2 3 45 6
At normal operating voltage ~U 60
for agas-filled Phototube, the tran- TIME —MILLISECONDS
sit time of the positive ions is less >W ao
than 10 microseconds. Cumulative
effects of the regenerative process a ~ 20
cause slightly longer delay times
for part of the current. However, a J
small component of the current is
o
delayed by too great a factor to be
the result of positive-ion transit- Fig. 29. Time response to a square wave
time effects. Ordinarily, only a small flight for a special gas-fiilled Phototube
designed to emphasize the lag resulting
percentage of the total current
from secondary effects of metastable
shows this effect. The slight falling argon atoms (Ref. 3).

off of the frequency-response curves Noise

(Fig. 28) near 1000 cycles per sec- The gas-amplification process is
ond and less which results from this not entirely noise-free because the
effect increases as the gas amplifi- gas ratio for an individual photo-
electron is a statistically variable
cation is increased. quantity. However, additional noise

Gas-Fi!!ed Phototubes 37

resulting from the gas-amplification portant use of these tubes is in mo-s,.
statistics is only a fraction of that tion-picture sound-on-film sensor
which results from the randon emis- systems for theater and home pro-
sion of electrons from the photo- jection equipment.
cathode. The Equivalent Noise Input
for agas-filled phototube is nearly It is especially important not to
the same as for a vacuum photo- exceed the absolute maximum volt-
tube having equal photocathode age and current ratings of gas-filled
sensitivity provided the vacuum type phototubes; excessive voltages can
is followed by a noiseless amplifier cause damage from ionization ef-
having a gain equal to the gas ratio fects, and excessive currents can re-
of the gas tube and a bandpass
limited by the frequency-response sult in loss of sensitivity.
characteristic of the gas-filled photo-
tube. The principal advantage of the Because the gas-filled photo-
gas amplification in realizing low tube does not have the flat current-
equivalent-noise input is to reduce voltage characteristic of the vacuum
the value of the load resistance for phototube, it is usually not feasible
which the resistor noise is equal to to use large load resistances without
the thermionic shot noise of the great loss in linearity of response,
as shown by the current-voltage
tube. characteristic and load lines in Fig.
25. However, in special applica-
Environmental Factors tions it is possible to use a large
load resistance provided the light
The previous discussion of the
level is low so that the drop in volt-
effects of temperature on vacuum
phototubes applies also to gas-filled age across the load is negligible.
phototubes. Because the number of
gas molecules does not change with In fact, the maximum recommended
temperature (even though the pres- operating point can be exceeded to
sure does), the gas amplification
does not vary appreciably with tem- advantage in such cases; the large
perature.
load resistance protects the tube and
In other respects (shock, vibra- circuit elements from damage in
tion, humidity) the behavior of the case the glow potential should be
gas tube is similar to that of the exceeded. Under very carefully con-
vacuum phototube. However, be- trolled conditions, gas-filled photo-
cause the positive ions bombard the tubes can be operated at very high
photocathode during operation, the gas ratios (of the order of 100);
life and the stability of a gas-filled however, because of the inherent in-
phototube are not as good as the stability of the tube under these
life and stability of vacuum photo- conditions, such operation is nor-
tubes operated at the same current. mally not recommended.

Application Considerations In some gas-filled phototubes,
there is a slight tinting of the glass
Gas-filled phototubes are used envelope opposite the photocathode.
to best advantage in applications This tinting, which does not materi-
which exploit the simplicity of the ally affect tube operation, is caused
circuit associated with the tube and by sputtering of cathode material
the low cost with which the addi- as a result of ion bombardment dur-
tional sensitivity is achieved. Be- ing the processing and aging of the
cause linearity and frequency re- tube. Further tinting may occur dur-
sponse are reasonably good, gas- ing long and especially severe
filled tubes may be used for a wide
variety of practical applications, operation of the phototube.
particularly when the more precise
characteristics of the vacuum photo- References
tube are not needed. The most im-
1 Cobine, J. D., Gaseous Conductors, McGraw-
Hil] Book Co., New York (1941)

■ Lceb, L. B., Basic Process of Gaeeone Elec-
tronics, Univ. of Calif. Press (1965)

s Engstrom, R. W. and Huxford, W. "Time
Lag Analysis of the Townsend Discharge in
Argon with Activated Cesium Electrodes",
Phys. Rev., 58 (1940)

11'~ultiplier
P~otot~x~es

Constrncti®n and Principles trons near the surface of the emit-
of ®peration ter; some reach the surface and
overcome the work function (in the

A LTHOUGH photoelectric emis- case of metals) or the electron af-
finity (for semiconductors) and es-
sion is a relatively efficient process cape into the vacuum. In general,
on aper-quantum basis, the primary the number of secondaries created

photocurrent for low light levels is increases as the primary electron
energy is increased. However, the
so small that special amplification depth from which the electrons must
escape also increase as the primary
techniques are required for most ap- energy increases because of the
greater primary penetration; this
plications. The multiplier phototube, factor tends to reduce the number
which uses secondary electron emis- of secondaries at higher dynode
sion to provide current amplifica- voltages.
tion in excess of 108, is a very use-
ful detector for low light levels. Fig. 30' shows the number of
emitted secondary electrons S per
In a multiplier phototube, the primary electron (secondary emis-
photoelectrons emitted by the photo- sion coefficient) as a function of the
cathode are, in general, electrostat- energy of the primary electrons for
ically directed to a secondary emit-
ting surface called a dynode. When a number of practical dynode ma-
terials. The same general pattern
normal operating voltages are ap- is observed for metals, but the yield
is insufficient for use in multiplier
plied to the dynode, 3 to 6 secondary phototubes.

electrons are emitted per primary An important property of the
secondary electrons is the energy
electron. These secondaries are fo- distribution of the emitted second-
aries. A typical distribution curve
cused to a second dynode, where the is shown in Fig. 31. The peak at the
process is repeated. In addition to extreme right corresponds to the
6 to 14 dynodes, a multiplier photo- energy of the primary electrons,
tube may contain other electrodes
for focusing the electron stream, re- and probably represents elastically
ducing space-charge effects, or ac-
scattered primary electrons. The
celerating the electrons to reduce
true secondaries are represented by
transit-time effects. The last dynode
is followed by an anode which col-
lects the electrons and serves as the

signal-output electrode in most ap-

plications.

the peak at the left. Although the

Dynode Properties spread of secondary-electron veloci-
ties of good secondary emitters is
Secondary emissionl-e in many generally much less than that shown
respects is similar to photoelectric in Fig. 31, it is nevertheless large
emission. The impact of primary in comparison with that of the
electrons rather than incident pho- photoelectron velocities. This ve-
tons causes the emission of elec- locity spread dictates to some ex-
trons. One primary electron excites tent the type of electron optics
several low-energy secondary elec- needed for efficient utilization of
the secondary electrons.

38

Multiplier Phototubes 39

ZW B ~~- i ~~

U ' ~ / ~~
~ '/
W — ~Aq~-Ca20, AgCa
~~ ~- Ca3 Bb
0B ' /~
/~ -- Aq Mq0-Ca
0 — — Cu Ba0-Ca

N

f4
r

0
z
0 2

w

0 100 200 300 400 500 600 700 BOO 900 1000
ACCELERATING VOLTAGE OF PRIMARY ELECTRONS

Fig. 30. Secondary emission coefficient for a number of dynode materials.

The materials silver-oxygen- emission and instability of the
cesium (Ag-O-Cs) and cesium-anti- Ag-O-Cs surface, it is no longer used
mony (Cs,Sb) used in the photo- commercially.
cathodes of phototubes having spec-
tral responses of S-1 and S-4 are The CsaSb emitter has the high-
also useful as secondary emitters.
They are practical from a manu- est secondary emission of the mate-
facturing standpoint because the ac- rials in the practical working range
near 100 volts. This material, how-
tivation process is nearly identical ever, has certain limitations: it can-
to that used in the production of the not tolerate exposure to air, it is
corresponding photocathode. How- damaged by temperatures in excess
of 75 degrees centigrade, and it does
ever, it is very difficult to produce
both the cesium-antimony and the not have stable characteristics when
silver-oxygen-cesium emitters in the
same envelope. Furthermore, be- subjected to current densities in ex-
cause of the generally high dark
cess of approximately 100 micro-
amperes per square centimeter.
During manufacture, the cesium-

RELATIVE NUMBER OF EMITTED SECONDARY ELECTRONS 1

20 40 60 80 100 120 140 160
ELECTRON ENERGY—VOLTS

Fip. 31. 1S~pical secondary-electron energy distribution* for a silver tarSet: primary electron
energy is approximately 150 volts.

40 RCA Phototube Manual

antimony dynode requires a slightly Secondary emission and stability are
similar to those of the Ag-Mg-O
different technique to achieve opti- dynode, although copper-beryllium
mum secondary emission and stabil- has some advantages in ease of
ity than does the cesium-antimony handling and dynode manufacture.
photocathode. Although this differ-
ence is not of major consequence, Dynode Configurations
the result is that on the average the
photocathode sensitivity in tubes One of the primary problems of
having cesium-antimony dynodes is design in a multiplier phototube is
slightly less than that achieved with the shaping and positioning of the
certain other combinations. dynodes (usually in a recurrent geo-
metrical pattern) so that all stages
A very practical secondary are properly utilized and no elec-
trons are lost to support structures
emitter can be made from an oxi- in the tube or deflected in other
dized silver-magnesium alloy con- ways. Although it is not necessary
taining approximately 2 per cent of that the electrons come to a sharp
focus on each succeeding stage, the
magnesium. Oxidation by means of shape of the fields should be such
low-pressure water vapor or carbon that electrons tend to return to a
center location on the next dynode,
dioxide produces a concentration of even though the emission point is
MgO on the surface which does not not at the optimum location of the
occur when the alloy is heated in preceding dynode. If this require-
oxygen directly (probably because ment is not met, the electrons in-
creasingly diverge from the center
the large H2O or COZ molecules do of the dynode in each successive dy-
not diffuse as far into the surface, node stage. This effect in turn leads
and therefore Mg migrates to the to skipping of stages and loss of
surface before oxidation). When gain. Magnetic fields may be com-
bined with electrostatic fields to
cesium vapor is present during the provide the required electron op-
tics, although today most multiplier
processing of multiplier photo- phototubes are electrostatically fo-
cathodes, it has the further benefit cused.

of increasing the secondary-emission A number of different dynode
ratio s-~ Although Ag-Mg-O dynodes configurations (See Figs. 32 and
do not have as high a secondary- 33) are used in multiplier photo-
emission ratio as CsaSb dynodes, the tubes. The circular arrangement of
material is easily processed and is dynodes of the 931A and similar
more stable at relatively high cur- types permits a compact layout, but
rents. In addition, it can tolerate
higher temperatures, and can be out- allows little flexibility in adding dy-
gassed at higher temperatures dur-
nodes beyond the circle; however,
ing exhaust. This surface has a low fewer than the full circle of nine
thermionic background emission, dynodes can be used. The collection
which is important in applications between stages and the transit-time
requiring detection of low-level light. dispersion is remarkably good for
Without the cesium activation, the the circular cage which was one of
oxygen-activated silver-magnesium the earliest systems developed.°

layer is used in demountable sys- The Rajchman linear-dynode
structure° (as in type 6810A) pro-
tems for detecting ions and other vides agood recurrent-field system,
particles. although the dynode shapes are
rather complex. This design is fur-
A material with characteristics ther complicated by a curvature not
similar to those of Ag-Mg-O can be
formed from an oxidized layer of
copper-beryllium alloy$ in which the
beryllium component is about 2 per
cent of the alloy. Oxidation of the
beryllium is accomplished in a man-
ner similar to that used to oxidize
the magnesium in the Ag-Mg emit-
ter; secondary emission is enhanced
by the bake-out in cesium vapor.

Multiplier Phototubes 41

cap

GRILL

INCIDENT
LIGHT

SHIELD

0=PHOTOCATHODE
10=ANODE
I-9=DYNODES

(b)

SEMI— —++
TRANSPARENT ~

PHOTOCATHODE 1

FACEPLATE ~ INCIDENT
LIGHT

FOCUSING I

ACCELERATING I
ELECTRODE ELECTRODE /

SHIEL I
INTERNAL /
CONDUCTIVE ~
COATING I

I-14=DYNODES L/NEAR TYPE 6B/OA
15=ANODE

tc~

SEMI— \~

TRANSPARENT

II ~~ PHOTOCATHODE—~►~
9
a 5 3•" J FOCUS RING 1

PHOTOCATHODEjLLY I INCIDENT
uGHT
/ (CTO

I FACEPLATE

FOCUSING
ELECTRODE

INTERNAL
Y CONDUCTIVE

COATING /

I-10=DYNODES L/NEAR TYPE 7746
II•ANODE

Fig. 32. Various dynode configurations in general use: (A) circular-cage ty~i (B) and
(C) linear types.

42 RCA Phototube Manual

(al
SEMITRANSPARENT I
PHOTOCATHODE

NCIDENT
LIGHT

ACCELERATING
GRIDS

i

I-9=DYNODES
10=ANODE

(bl

FOCUSING FACEPLATEpl y
'~'~ELECTRODE

~ti ti ~ ~~'J '//~I \ , i RATION
/ ~ ~ ~ ~^' V ~

\~~/~~/,~ t SEMI`

I ~ /' TRANSPARENT
PHOTOCATHODE

_ I `1 I `I ~I I
10 8 6 4 2
INTERNAL
CONDUCTIVE I/

COATING

i-10=DYNODES
11=ANODE

Fig. 33. Various dynode configurations in general use: (A) bos type. (B) venetian-blind
type.

only as shown in the drawing, but ode, as in the 931A, the cathode area
also at right angles to the plane is too small for many applications.
of the drawing. This curvature pro-
vides a focusing field which main- Box-type dynodes provide very
tains the electron stream near the efficient collection of electrons be-
tween boxes, except for losses to the
middle of the dynode structure and grid wire. However, because of the
prevents bombardment of the sup- lack of specific focusing properties
porting spacers at the edges of the and the wide variation in with-
dynodes. For this reason, a larger drawal fields, the dynodes do not
number of dynodes can be used suc- provide a good transit-time disper-
cessfully without problems caused
by lateral spreading of the electron sion characteristic.
stream. In general, linear-style dy- The venetian-blind type of dy-
node systems have good transit-time
characteristics because of the focus- node (as in type 8053) can be
ing properties and the good with- coupled simply and in a relatively
drawal fields at the dynode surface. small space. More dynodes can easily

Although focused-dynode arrays be added to the chain if the system
is opaque to feedback either by light
have a minimum of stray electrons or by ions. A disadvantage of this
type of focusing is the rather low
between stages, the acceptance area value of electric field at the emitting
of the first stage is generally small. surfaces, which results in relatively
If the first stage is used as a cath-
large transit-time dispersion. Some

Multiplier Phototubes 43

electrons are lost from one dynode a multiplier phototube for scintilla-
to the next because of the low with- tion counting because it can be
drawal field, and some electrons are mounted parallel to the photocathode
lost to the wire grid used to prevent and has a relatively large accept-
field interaction between dynodes. ance area. This arrangement per-
mits the design of tubes having
At the present time, the trans- larger photocathodes (8054) and
mission type of dynode is only used good collection efficiency at the first
experimentally in multiplier photo- dynode.
tubes. The dynode is a thin mem-
brane on which primary electrons Design For Minimum
impinge from one side and secondary Transit-Time Spread
electrons are emitted from the other.
Dynodes are mounted in closely When a tube is required for
spaced parallel planes. For this rea- scintillation-counting applications in
son, the transit time and transit- which a minimum transit-time
time spread can be made very small. spread is desired, the venetian blind
Transmission secondary-emission dy- dynode is not suitable. Mathesonu
nodes are not practical for ordinary has designed a special focused cage
applications because of the difficulty structure designed to solve both the
and expense of construction and be- problem of high speed and the prob-
cause present dynodes have very lem of good collection. In this de-
poor life at ordinary current levels. sign (type 7746), the front end of
the cage deviates from a strictly
Coupling Dynode System linear construction to present a
To Cathode large effective area for the collec-
tion of photoelectrons on the first
One of the design problems of dynode.
a dynode system is the coupling of
the recurrent dynode chain to the The ideal arrangement should
photocathode. In the circular-cage also provide for equal transit time
type, the first stage of the circle for all photoelectrons to the first
serves as cathode (type 931A); dynode. Fig. 34 also shows the
however, for many applications this Matheson" solution to this prob-
limited-area cathode is too small. lem: acurved cathode and annular
In scintillation counting, it is de- rings just above the first dynode to
sirable to have relatively large flat correct and shape the potential field
photocathodes at the end of the tube between the cathode and first dy-
for efficient coupling of the tube node.
to the scintillation crystal. When
the first stage of the circular cage A multiplier phototube devised
is used as a dynode, as it is in the by G. A. Morton, R. M. Matheson,
6342A, the effective size of the first and M. H. Greenblattll provides
dynode is critically small for collect- minimum interdynode transit-time
ing all the electrons from the photo- spread; accelerator electrodes placed
cathode. between dynodes, as shown in Fig.
35, are connected to a highly posi-
In the linear-cage type (type tive potential. The proximity of the
6810A), the problem is increased by high voltage provides a large with-
the requirement of housing the drawal field for the electrons, and
whole assembly in an axial con- although they are slowed down after
figuration. In this case the first dy- passing the accelerator electrode, the
node is at an angle which presents transit time and transit-time spread
almost a minimum of projected area are very short.
to the photocathode.
The output sections of some
The venetian-blind type of dy- multiplier phototubes have special
node is well suited to the design of terminals for very high-speed pulse
counting and analysis. The construc-
tion may be specially designed to

44 RCA Phototube Manual

PHOTOCATHODE
0 VOLTS

. \
10 ~~~ \\

i ~
~ 25 V i~ ~~ ~

FOCUS RING i , 50 Vj \ \\ \
(AT CATHODE, ~ / ~

POTENTIAL) / 100 ~ \ ~~

FOCUSING 1 / 150 V \
ELECTRODE'
(SHIELD) /i / ~ ~1
% I2oo v I ~~

2n. /
DYNODE !I' •

_ /I~~ DYNODE
DYNODE 1

•
3RD

DYNODE

5TN
DYNODE

$rig. t4. Matheson-type front-end configuration showing equipotential lines and electron
trai ectories feeding into a modified linear-type dynode cage which exposes more of the
first dynode area to the photoelectron stream.

provide a maximum pulse current In practice, some of the elec-
before space-charge limits the re- trons may skip stages, or become
lost to the amplification process by
sponse of the tube. impinging upon nonproductive sec-
ondary-emission areas.
Properties of
Multiplier Phototubes It is customary to describe the
gain of the multiplier phototube as
Gain Characteristics a function of the applied voltage.
Fig. 36 shows two such curves on a
yVVhen several secondary-emis- semilog scale. These curves illus-
sion stages are coupled together, so
that the secondary electrons from trate the wide range of amplification
one become the primary electrons
of the next, the total gain µof the in a multiplier phototube. They also
multiplier phototubes is given by indicate the necessity of providing
a well regulated voltage supply for
µ = S° (26)
the dynode stages.
where S is the secondary emission It is possible to operate a mul-
per stage (assumed to be equal for
each stage) and n is the number of tiplier phototube so that each stage
stages. It is also assumed in this ex- is at the voltage required for maxi-
pression that all the secondary elec- mum secondary emission, as shown
trons are collected at the next stage.
in Fig. 31. In such cases, the gain
could be made practically inde-
pendent of voltage over a small
range. However, such a condition
would require approximately 500
volts per stage; thus the total volt-

Multiplier Phototubes 45

C 5800

C 5800
5000 3404800 ~- v=400

v=200 5800

G 2600
~ 4200

4200 10
2600 1200 — v=0

C 5800 l
Y=400

10 003400 5000 5800
v=200
C1800

v=600

Fig. 35. Interdynode accelerator-electrode system designed by G. A. Morton to provide mini-
mum transit-time spread.

w' I I I I I age required would be very high for
the amount of gain achieved.
I()6 IP2I
In the design or operation of a
10 5 6342A multiplier phototube having a fixed
supply voltage, the number of stages
i POINT OF MAXIMUM POINT OF MAXIMUM can be chosen so that the gain of the
0 104 GAIN PER VOLT GAIN PER VOLT tube is maximum. For this purpose,
the optimum voltage per stage is
a that value at which a line through
the origin (unity gain on the log-
LL gain scale) is tangent to the curve,
aJ 103 as shov~n in Fig. 36. This point is
a identified on the graph as the point
of maximum gain per volt. (Note
l02 that this argument neglects the volt-
age used between the last dynode
to and the anode and any discrepancy
resulting from nonuniform distribu-
io II tion of voltage per stage.) In most
applications of multiplier photo-
20 40 60 60 100 120
tubes, the tubes are operated above
VOLTS PER STAGE
the point of maximum gain per volt.
Fig. 36. Log of gain as a function of volts It is customary to present tube data
per stage for a tube (1P21) with Cs-Sb with both the gain and the voltage
dynodes and for a tube (6342A) with on a logarithmic scale; over the nor-
Cu-Be dynodes. mal range of operation the resultant

46 RCA Phototube Manual

curve is then closely approximated The first transmission-type
by a straight line. cesium antimony cathodes were used
in tubes having a spectral response
Spectral Response
designated as S-9. Later,~vhen tubes
Photocathodes developed for began to be used in scintillation-
diode phototubes are also used in counting applications, the processing
multiplier phototubes. Photoelec- was modified to increase the blue
trons emitted from the photocathode sensitivity of the cathode because
are directed to the first dynode of scintillators typically have a blue
the tube instead of to the anode as emission. This modified response was
in a photodiode. However, several designated 5-11; both response
special photocathodes which are
rarely used in photodiodes have been characteristics are shown in Fig. 37.
used in multiplier phototubes.
loo 1 J~ \~1
Cathodes of the transmission I~
type are often used in multiplier so
phototubes, in contrast with the h ly ,
}
opaque type used in most photo- I
diodes. A transmission-type photo- 60
cathode is one in which a semi- W /~
transparent layer is applied to the
inside surface of the envelope.win- W ~ 1I
dow. Light impinges on the outer
(glass) side of the photocathode, ~ 40 I
and electrons are emitted on the in-
ner or vacuum side. The vacuum- it I
evaporation and processing of a
transmission-type cathode require zo 1~
careful control to achieve uniformity II
and high sensitivity. The most com- 0
mon transmission-type photocathode 1
is made of antimony and cesium, the I
same elements used for the opaque
photocathode in tubes having an r
S-4 spectral response. During the
processing of the tube, antimony is I
vacuum-evaporated onto the inner
surface of the window. A metallic I1
substrate is often put on the window
first to improve the conductivity of I~
the photocathode or to facilitate the I~
activation process. The antimony is
usually evaporated from a heated Jt ~
filament pre-beaded with antimony; 3000 5000 7000
the beads are positioned to provide
a uniform layer of antimony. The WAVELENGTN —ANGSTROMS
thickness of the evaporated layer is
usually monitored photoelectrically Fig. 37. Comparison of the S-9 and S-11
by measurement of the transmission spectral response characteristics. Both
of light through the layer during the curves are for transmission-type cesium-
evaporaton procedure. After the antimony photocathodes. The 5-11 re-
antimony is evaporated, cesium sponse was evolved to provide maximum
vapor is allowed to react with the blue response for scintillation counting.
antimony. The resultant photo-
cathode has a chemical composi- The principal difference in
tion of approximately CseSb. processing which results in the 5-11
response is the use of a thinner layer
of antimony. A photocathode layer
of Cs,Sb tends to absorb blue light
and transmit red, as shown in Fig.
38. Two effects combine to explain
the dependence of the spectral char-
acteristic on the thickness of the
photocathode. (1) As the thickness

Multiplier Phototubes 47

loo

r-

zw 80

U
LL'

aW
60

O
Fa
0
am 40
a

U

~a 20
O

0 400 500 600 700 B00 900

WAVELENGTH a—MILLIMICRONS

Fig. 38. Optical absorption of Cs35b layer having a typical 5-11 spectral response.

of the cathode is increased, more The multialkali photocathode" is
very difficult to process to uniformly
light is absorbed; this increased ab- high sensitivity. The process is
sorption tends to increase the sensi- complicated and involves alternate
tivity in direct proportion. (2) As treatment with evaporated antimony
the thickness is increased, the photo- and alkali vapors. The resultant
electrons must emerge from a photocathode is the most sensitive
greater depth to escape into the known for the region from the ultra-
violet to the red end of the spectrum.
vacuum; this effect tends to reduce However, compared with the cesium-
antimony photocathode, the multi-
photoemission because of the ab-
sorption of photoelectrons. The spec- alkali photocathode has only a
tral-response characteristic S-9 was
evolved for use with a typical tung- slight advantage in the blue region.
sten light source. The resulting
loos r/ \S '~o
rather thick surface provided better s~0
I- s
response at the red end of the spec- f-
trum. A compromise to a thinner
cathode improved the blue response 4
(5-11) with the loss of some red re- 3
sponse.
z
Two transmission-type photo-
cathodes of importance, particularly w
in the red region of the spectrum,
are the bismuth-silver-oxygen-cesium a IsB

(Bi-Ag-O-Cs) cathode used in tubes ~

having 5-10 spectral response, and J4
the multialkali ([Cs]Na2KSb) cath-
ode, used in tubes having 5-20 spec- z

tral response. These spectral-re- I
sponse characteristics are shown in
Fig. 39. r
s
The semitransparent Bi-Ag-O- 6
Cs cathode is prepared by first
evaporating a thin layer of bismuth f- a
and then a thin layer of silver. The
silver is then oxidized, and cesium z
vapor is allowed to react with the
layer. vw, 2
o.l

2000 4000 6000 8000
WAVELENGTH —ANGSTROMS

Fig. 39. Spectral response characteristics for
5-10 (Bi-Ag-O-Cs photocathode) and
S-20 [(Cs) Na$ KSb photocathode].

48 RCA Phototube Manual

The bialkali photocathode low dark current. Its tentative spee- i
(NazKSb) also deserves mention
although at the present time its tral response is also given in Fig 40.
use is only experimental. The NazKSb
cathode has a spectral response At the present time, it seems that
(tentatively identified as 5-24) simi- the most promising application is in
lar to the 5-11 (See Fig. 40), but scintillation counting, perhaps for
has the advantage of a lower low energy work, although the dark
thermionic emission at room tem- current and temperature character-
perature of the order of 10-1° amperes istics may not be quite as promising
per square centimeter. In fact, some as the NazKSb cathode.
data have indicated values as low as
10-19 amperes per square centimeter. As in vacuum photodiodes, the
Another advantage of the NazKSb spectral response of multiplier
cathode is its ability to withstand phototubes is generally limited in
somewhat higher temperatures than the ultraviolet region by the type
other cathodes; it can be used to of window used in the device, as
about 100 degrees centigrade. The shown in Fig. 19. Fig. 41 compares
bialkali cathode promises to be use-
ful as a cathode for low-energy the spectral responses of multiplier
scintillation counting because of its
good blue sensitivity and very iow phototubes having several photo-
dark current.
cathode-window combinations.
70 In recent years, considerable

F developmental effort has been ex-
pended on cathodes made of the
a 60 following materials: CszTe, RbzTe,
3 and CsI and CuI. All of these cath-
ode materials are useful in the
N ultraviolet region, particularly be-
cause of their lack of sensitivity in
it 50 the longer wavelength region. They
are generally combined with a win-
a dow made of a material such as LiF
or sapphire.
a
40 The only photocathode useful in

vi 30 the infrared region is the silver-
z oxygen-cesium (Ag-O-Cs) cathode
a used in tubes having S-1 response, as
shown in Fig. 18. For wavelengths
~ zo longer than 8000 angstroms to its
limit of response, about 11000 ang-
J stroms, it is the most sensitive
photocathode.
a
a Some multiplier phototubes have
photosensitive dynodes—Cs-Sb, for
v 10 example. When the transmission-

W type cathode is quite thin, as in the

41 case of Cs-Sb photocathodes used in
tubes having 5-11 response, trans-
0 mitted light may strike the first
dynode and cause the emission of
3000 4000 5000 6000 7000 photoelectrons. This effect is of sec-
WAVELENGTH —ANGSTROMS ond order because photoelectrons
emitted from the first dynode do not
PYQ. 40. Comparison of the absolute spectral have the benefit of multiplication
sensitivities of three antimony alkali by the secondary emission from the
photocathodes: Cs,Sb, Na„KSb, and first dynode; nevertheless, the ef-
(K,Ca)°Sb. The 5-11 curve shown ie fect is observable. The increase in
representative of the cathode on type sensitivity occurs primarily at the
5053: the 5-24 and (K, Cs),Sb curves red end of the spectrum, where the
are tentative. transmission of the photocathode ;s

Another two-alkali antimony
cathode has recently been announced
by A. H. Sommer, (K,Cs)°Sb. Al-
though its properties have not been
thoroughly explored, it is apparent
that it has high blue sensitivity and