_The Problem_
You are a physicist investigating the properties of semiconductors,
which are materials used to make transistors. When you apply a voltage
to one specimen of silicon (a semiconductor), it doesn’t behave quite
like the others that you’ve studied. The electrical properties of this
odd specimen are unusual and interesting and could lead to a new type of
transistor. What makes this specimen different from the others? Very
small amounts of impurities can cause large changes in the electrical
properties of semiconductors. You would like to obtain a chemical
analysis of the material, but your colleagues in chemistry tell you they
would have to dissolve a good size part of your sample to analyze it and
you are reluctant to give it up. How do you do it?
_The Solution_
You decide to try neutron activation analysis. You realize you won’t be
able to detect all the elements, but many of those that might affect
semiconductor performance could be detected quite easily.
What will you need? A source of neutrons to activate the material and a
gamma-ray spectrometer to measure the radiation from the material
afterwards. This spectrometer detects and measures gamma rays and sorts
them according to their energy. You find that your friend down the hall,
who is a nuclear physicist, has a gamma-ray spectrometer that
incorporates a lithium-drifted germanium crystal as a detector and a
pulse height analyzer. The germanium detector is a device that senses
the gamma rays that enter it and gives electrical signals related to the
energy of the gamma rays. It was invented only a few years ago and has a
very fine resolution. That is, it can easily “pick out” gamma rays that
are only slightly different in energy. For example, for gamma rays with
energies of approximately 1 MeV (million electron volts), it is not
unusual to distinguish between gamma rays that differ by only 2 or 3
tenths of a percent. The pulse height analyzer is an electronic device
that sorts the electrical pulses from the detector according to their
energy.
[Illustration: Gamma-ray detector]
A lithium-drifted germanium-crystal gamma-ray detector. The large
container is a reservoir of liquid nitrogen that keeps the detector
cooled to a temperature of -196° Centigrade (321° below zero,
Fahrenheit). The lead brick shield keeps out most of the gamma rays
that come from naturally radioactive materials in the room. The
plastic slots hold cards upon which the samples are mounted for
counting. Sometimes the detector is arranged vertically and samples
are placed on shelves above it.
[Illustration: Gamma-ray detector]
What about the neutrons for the irradiation? Although there isn’t a
suitable nuclear reactor[9] in your city, there is one at a university
only an hour away by jet. Since it may take a few hours to get the
sample to the counter after irradiation, you won’t be able to look for
short-lived activation products, i.e., those with half-lives of up to an
hour. However, this will exclude only a few elements from detection.
[Illustration: Pulse-height analyzer]
A pulse-height analyzer used for gamma-ray spectrometry. A gamma-ray
spectrum is displayed on the television screen. Data is printed out
automatically on the electric typewriter and also may be plotted as a
graph on the paper to the left. In other systems, data may be coded
onto punched paper tape as well. Such tape may be “read” by a computer
that can be programmed to use the data to calculate what radioactive
isotopes are present and their quantities.
Now you are ready to begin the analysis. This will be a qualitative
analysis since you are merely looking for a significantly different
element in that silicon crystal. How much of it is present is only of
secondary interest. Therefore, if you find anything different, you will
rely on an approximate calculation to tell you “how much”.
[Illustration: Nuclear reactor]
This is called a “swimming pool” reactor because the nuclear fuel,
built into metal rods, is held in a framework at the bottom of a deep
pool of water. The water serves as a shield to protect workers from
the radiation and also helps the reactor “go” by slowing down neutrons
to make them more likely to interact with the target atoms. “Swimming
pool” reactors are frequently used for neutron activation analysis and
typically provide neutron fluxes of over 10¹³ (10 million million)
neutrons per square centimeter per second.
[Illustration: Quartz capsules]
These sealed quartz capsules contain samples to be irradiated in a
nuclear reactor. They are about to be placed in the aluminum can,
which will be sealed and positioned at the end of an aluminum pole,
close to the core of a “swimming pool” reactor. Often samples are
placed in plastic tubes and are carried in and out of a reactor by air
pressure in a pneumatic tube system.
You carefully scrape off a small amount of material, weigh it on a
sensitive balance, and put it into a short piece of pure quartz tubing.
You do the same with an ordinary piece of silicon for comparison and
then seal both tubes with an oxygen-gas torch. Although the tubes are
both ¼ inch in diameter and about 1 inch long, the first tube is just
slightly longer so you will be able to determine which is which after
the irradiation.
Off it goes to the reactor in a carefully wrapped package along with
instructions to irradiate the tubes for 12 hours in a neutron flux of
about 10¹³ neutrons per square centimeter per second and to return them
as quickly as possible after they are removed from the reactor.
The following week, the samples are delivered about 4 hours after they
were removed from the reactor. Working quickly but carefully, you note
that they are radioactive but easily handled by ordinary laboratory
techniques. You break the quartz tubes one at a time and attach each of
the two pieces of silicon to a card with self-sticking tape. Then you
place each card, in turn, on a holder close to the gamma-ray detector
for a period of 10 minutes. A spectrum, which is a graph of the quantity
of radiation recorded in each increment of energy over the range
observed for each of the samples, is plotted automatically at the end of
the counting period and you may now compare the compositions of the two
samples. (See the figure on the next two pages.)
The two spectra are virtually identical except that the suspect sample
has one obviously different peak in channel 157 and a somewhat smaller
peak in channel 183. Referring to an energy calibration curve for the
pulse height analyzer, you find that these channels correspond to 0.559
and 0.657 MeV respectively. A search of a table of nuclides, arranged by
gamma-ray energy, reveals that this combination is emitted by
arsenic-76, which would be the activation product for arsenic. Other
data also indicate that for arsenic there should be a number of smaller
peaks, including some corresponding to energies of 1.216, 1.228, 0.624,
and 1.441 MeV. A closer look at the spectrum of the suspect sample
reveals that these are also present.
Finally, noting that the half-life of arsenic-76 is approximately 27
hours, you wait a day and count the sample again in the same position as
the previous count. A decrease in the heights of the 0.559 and 0.657 MeV
peaks, by a little less than half in 24 hours, confirms that arsenic is
the unusual element in this sample. It may not be the only impurity
causing the peculiar behavior of this semiconductor, but it does seem a
likely candidate.
[Illustration: Graph: “Counts in 20 minutes per 3.8 KeV channel” _vs_
“Channel Number”]
Element Counts Channel Number
Gold 1300 125
Sodium 900 140
Antimony 650 150
Antimony 550 160
Scandium 220 240
Iron 150 300
Scandium 60 310
Cobalt 110 320
Iron 45 350
Cobalt 50 360
Sodium 900 370
Sodium 140 425
(values estimated from the graph)
The gamma-ray spectrum obtained after activation of a sample of “pure”
silicon having “ordinary” properties of this type of semiconductor.
Only very small quantities of various trace impurities are indicated.
[Illustration: Graph: “Counts in 20 minutes per 3.8 KeV channel” _vs_
“Channel number”]
Element Counts Channel Number
Gold
Sodium
0.559 Arsenic-75 1500 150
Antimony
0.624 Arsenic-75 190 165
0.657 Arsenic-75 135 175
Scandium
Iron
Scandium
Cobalt
1.216 Arsenic-75 100 330
1.228 Arsenic-75 60 335
Iron
Cobalt
Sodium
1.44 Arsenic-75 25 380
Sodium
(distinct values estimated from graph)
The gamma-ray spectrum obtained after activation of a sample of
silicon having “unusual” electrical properties. While most of the
spectrum is identical with that obtained from the ordinary material,
there is an interesting difference.
Using the equation given on page 12, the approximate known values for
half-life, sample weight, neutron flux, and periods of irradiation and
decay after irradiation, and an estimated value for the number of
arsenic-76 atoms measured by the gamma-ray spectrometer, you calculate
that the arsenic content of the sample is approximately 44 parts per
million (ppm). (See appendix.)
With this information as a starting point, you are now ready to proceed
with further research on the properties of your semiconductor, e.g., if
you double the concentration of arsenic, how will that affect its
properties?