PROCEEDINGS
of the Tomsk Order of the Red Banner of Labor Polytechnic Institute
named after S. M. Kirov

Vol. 138　　　　　　　　　　　　　　　　　　　　　　　　　　　　1965

ISOTOPE FLAW DETECTION OF MATERIALS AND PRODUCTS

V. I. Gorbunov, V. B. Kuznetsov, V. K. Kuleshov,
Yu. B. Yankelevich, Ya. B. Yankelevich

(Submitted by the scientific seminar of the Research Institute of Electronic Introscopy)

The possibilities of isotope flaw detection are discussed. Results are presented from a study of the counting-spectrometric method in automatic scintillation flaw detection. The industrial automatic scintillation flaw detector VID-3M, which implements this method in practice, is described.

In industrial isotope flaw detection of materials and products, a comprehensive accounting of the characteristics of various radiation detectors as applied to each specific inspection task is very important.

The method, now widely used, of measuring radiation intensity with scintillation counters has opened up new possibilities in flaw detection.

1. Ampoule of the radioactive isotope
2. Container with collimator
3. Object under investigation, moving at a specified speed relative to the collimator window
4. Radiation shielding of the detector
5. Scintillation crystal
6. Light guide
7. Photomultiplier
8. Cathode follower

If the object under investigation (a part, product, material, etc.) is scanned successively from point to point with a narrowly collimated beam of quanta from a radioactive source, while registering the intensity of the radiation that has passed through the object under investigation, one can judge the integrity of the material or product being inspected (Fig. 1).

Such a method for quality control of products without destruction—we shall call it “automatic scintillation flaw detection”—has a number of indisputable advantages in comparison with radiography. While inferior to radiography in the visual clarity of the information obtained about the quality of the object being inspected, automatic scintillation flaw detection surpasses the photographic method in productivity and is more economical. At the same time, its sensitivity can significantly exceed the sensitivity of recording on photographic film. The advantages of the scintillation method become especially evident in flaw detection of rather large thicknesses, already beyond the capabilities of radiography.

In article [1] the author gives a diagram (Fig. 2) showing the ranges of steel thicknesses accessible for radiographic transmission methods using industrial radioactive isotopes.

Steel [mm]

Application ranges of isotopes

□ Regions of uncertain control
■ Regions of reliable control

Fig. 2.

In evaluating the possibilities of isotope flaw detection, the economic effect that is possible provided the existing limits of application of radioactive isotopes in nondestructive testing are expanded becomes obvious.

Isotope radiography of large thicknesses encounters a number of insurmountable difficulties, which also limit its possibilities. The high degree of attenuation of the flux of $\gamma$-quanta in flaw detection of large thicknesses and the low efficiency of recording $\gamma$-radiation by means of photographic film cause a considerable increase in exposure time. Thus, for example, when radiographing a steel product 100 mm thick with $\gamma$-radiation from the standard radioactive isotope $\mathrm{Co}^{60}$—50 g-eq Ra—the exposure time reaches several hours [2]. Reducing the exposure time by further increasing the power of the $\gamma$-source is not promising. The presence of fog on the photographic film substantially worsens image quality. This is explained by the increase, as the thickness of the inspected object increases, in the fraction of radiation scattered in the investigated product within the total flux of $\gamma$-quanta reaching the detector (photographic film) [3—5].

It is obvious that, if one has a detector that makes it possible to select, from the total flux of radiation arriving at the detector, the $\gamma$-quanta carrying information about the integrity of the inspected object, the possibilities of isotope flaw detection can be expanded.

All the advantages of the scintillation detector and its possibilities become evident if the nature of the energy [[unclear: continuation cut off on page]] is studied more deeply.

of the $\gamma$-radiation spectrum behind the inspected object, as well as the processes occurring in the scintillation detector and the photomultiplier when $\gamma$-quanta are registered.

As a source of $\gamma$-quanta, let us consider the Co$^{60}$ isotope, widely used in flaw detection of large thicknesses. As is known, this isotope is the most widespread in practice, is inexpensive, and has a fairly high $\gamma$-radiation energy—on average 1.25 MeV. The gamma radiation of Co$^{60}$ has two monochromatic lines with $E = 1.17$ MeV and $E = 1.33$ MeV (Fig. 3). However, this simple spectrum is transformed into a more complex one after the $\gamma$-quanta pass through the thickness of the object being radiographed (Fig. 4).

Fig. 3.

As a result of the interaction of $\gamma$-radiation with the material of the product, not all quanta reach the scintillation detector. Some of them are removed from the beam due to photoabsorption; another part, after undergoing one or several acts of Compton scattering, is photoabsorbed or simply does not enter the detector collimator (the effect of pair production for energies of 1.17 and 1.33 MeV may be neglected).

Fig. 4.

Some of the quanta scattered in the product and entering the detector collimator “smear out” the peaks in the energy-distribution curve and transform the discrete $\gamma$-radiation spectrum of the isotope into a continuous “white” spectrum extending from the $\gamma$-radiation energy of the radioactive isotope down to zero.

Figure 4 qualitatively shows the instrumental energy spectrum of Co$^{60}$ (1.17 and 1.33 MeV), recorded with a scintillation spectrometer NaI (Tl) $70 \times 50$ and an FEU-13 photomultiplier, behind steel ($\mu d = 6$) for various collimator diameters.

Figure 5 qualitatively presents the true spectrum of the $\gamma$-radiation of $\mathrm{Co}^{60}$ behind an absorber made of steel ($\mu d = 6$). For processing the instrumental spectra, a numerical matrix of amplitude component distributions for a NaI (Tl) 70×50 crystal, taken from [4], was used.

Fig. 5.

In Fig. 5 the peaks corresponding to energies of 1.17 and 1.33 MeV are less clearly visible (because of single scattering in the absorber); the continuous spectrum extends down to zero, with a rise in the region of 150–200 keV. This rise is due to the entry into the detector collimator of quanta multiply scattered in the absorber. In [7, 8, 11], a theoretical and experimental estimate was made of the contribution of scattered quanta to the total radiation flux incident on the detector. The buildup factor $B$ is estimated (the ratio of the intensity of the radiation incident on the detector behind the object under study to the intensity of only the primary, unscattered quanta that have passed through the object being inspected and entered the detector); for different experimental geometries, in each specific case it is possible to substantially reduce the magnitude of $B$.

The authors carried out a series of experiments with a scintillation spectrometer to determine the dependence of the buildup factor on the area of the collimator window. As can be seen from Fig. 4, with an increase in the area of the detector-collimator window, the contribution of radiation scattered in the absorber material to the total flux of quanta increases. This increase is especially pronounced in the energy region of 150–200 keV, i.e., in the range of energies of the multiple-scattering peak for steel [4]. In the present case, the buildup factor $B$ was determined as the ratio of the area under the energy-distribution curve to the area under the photopeak. On the basis of the spectral distributions obtained, the dependence of the buildup factor $B$ for a steel absorber with $\mu d = 6$ on the area of the collimation aperture was constructed (Fig. 6).

As can be seen from Fig. 6, $B$ increases with increasing collimator-window size and in the limit tends to the value 6.25—the buildup factor for steel ($\mu d = 6$) at an energy of 1.25 MeV under broad-beam conditions [9]. The values of $B$ obtained for collimator areas of $\varnothing 25$ and 40 mm are somewhat overestimated, since in correcting the instrumental spectra no correction was made for edge effects in the NaI (Tl) 70×50 scintillation crystal, although for such comparable dimensions

of the crystal area and the collimator-window area they are rather large. This assumption was made deliberately in order to obtain data more convenient for practical calculations of the sensitivity of the method of automatic scintillation flaw detection.

Fig. 6.

The theoretical limit of sensitivity of the method of scintillation isotope flaw detection may be taken to be the magnitude of the change in the intensity of the radiation that has passed through the product, exceeding by \(Z\) times the value

\[ \delta = \frac{100}{\sqrt{Nt}}, \tag{1} \]

which determines, in percent, the statistical fluctuations of the radioactive decay of the radiation source. The quantity \(Z\) is the conditional ratio of the signal from the minimum defect to the fluctuation “noise.” The smaller \(Z\) is, the less accurately the defect dimensions can be estimated. In formula (1), \(N\) is the counting rate of pulses arriving at the detector, and \(t\) is the time for measuring the intensity at the given point.

The number of pulses arriving per unit time from the detector may be expressed as follows:

\[ N = 3.7 \cdot 10^{10} \cdot A \cdot \gamma G \cdot \varepsilon_d \cdot e^{-\mu d} B, \tag{2} \]

where

\(A\) — activity of the source (curies);

\(\gamma\) — number of quanta per one decay;

\(G\) — geometric attenuation factor;

\(e^{-\mu d}\) — attenuation of the beam of \(\gamma\)-quanta by the material of the test specimen of thickness \(d\);

\(\varepsilon_d\) — efficiency of registration of \(\gamma\)-quanta by the scintillation counter;

\(B\) — accumulation factor,

\[ G = G_1 + G_2 + G_3 + G_4. \ [10] \tag{3} \]

The geometric factor of attenuation of radiation in a cone bounded by the edge of the collimator is

\[ G_1 = 0.5 \left[ 1 - \frac{1}{\left(1 + \frac{4r^2}{R^2}\right)^{1/2}} \right], \tag{4} \]

where \(r\) is the radius of the collimator aperture;
\(R\) is the distance from the source to the detector crystal.

The terms \(G_2\), \(G_3\), \(G_4\) in formula (3) take into account the radiation scattered in the collimator, partially absorbed in the collimator, and passed through the shielding of the detector. According to [9], the sum \(G_2 + G_3 + G_4\) may be neglected, since for the geometry shown in Fig. 1 it amounts to less than one percent of the value of \(G_1\).

\(\varepsilon_d\) in formula (2) may be determined as

\[ \varepsilon_d = \left(1 - e^{-\mu_1 d_1}\right)\cdot P \cdot e^{-\mu_2 d_2}, \tag{5} \]

where \(\mu_1\) is the coefficient of absorption of \(\gamma\)-radiation by the scintillator material;
\(d_1\) is the length of the crystal;
\(P\) is a coefficient taking into account the shape of the energy spectrum of the pulses arriving from the photomultiplier at the recording electronic apparatus;
\(\mu_2\) is the coefficient of absorption of \(\gamma\)-radiation by the material of the crystal package;
\(d_2\) is the thickness of the crystal package.

Fig. 7.

The role of the coefficient \(P\) in formula (5) becomes clear from a more detailed evaluation of the amplitude spectra of the pulses arriving from the photomultiplier of the scintillation counter. In [12], instrumental spectra are given for NaI(Tl) crystals of various sizes (Fig. 7). As can be seen, a small NaI(Tl) crystal, at an incident quantum energy of \(1.25\) MeV, has a continuous distribution which, in area, exceeds by several times the area under the full-energy peak. The larger the dimensions of the crystal, the greater the fraction of the quanta incident on the crystal that is recorded in the photopeak.

It should be noted that Compton-scattered quanta in the crystal, which determine the continuous distribution in the spectrum of pulses from the photomultiplier tube, carry the same amount of information about the integrity of the article being radiographed as do the quanta registered in the photopeak. As indicated above, in the range of the continuous Compton distribution (0–1 MeV), $\gamma$-quanta scattered in the absorber material and entering the detector collimator window are recorded (hence the pronounced peak of multiple scattering). In the range 0–30 keV on the NaI (Tl) scale lie the noise pulses of the photomultiplier tube, which make a substantial contribution to the continuous distribution of the spectrum.

A scintillation detector makes it possible to select from the recorded flux of $\gamma$-quanta only that part of the quanta which carries information about the integrity of the object being radiographed, since it permits amplitude discrimination of pulses. The correct choice of the parameters of the scintillation detector, the geometry of the experiment (radiography) [7], and the amplitude-discrimination threshold makes it possible to bring the sensitivity of the method as close as possible to its theoretical limit. Figure 8 schematically shows the amplitude distribution

Fig. 8.

of pulses arriving from the photomultiplier tube. The shaded regions conventionally show the area under the full-energy peak and under the continuous Compton distribution, i.e., the regions characterizing pulses that carry useful information. The unshaded part characterizes the contribution of pulses to the overall energy distribution due to photomultiplier-tube noise and scattering in the absorber material. The choice of a scintillation crystal of sufficient size makes it possible to increase the number of pulses registered in the photopeak by sharply reducing the Compton distribution. The choice of the optimal radiographic geometry [7] makes it possible to reduce the contribution of radiation scattered in the material of the articles to the energy distribution of pulses from the photomultiplier tube.

Such consideration of the nature of the amplitude spectrum of signals from the photomultiplier tube and of the factors affecting this distribution makes it possible to make the proper choice of the energy range of the pulse-discrimination threshold in order to increase the “signal-to-noise” ratio (in Fig. 8, the range $A—B$). Proceeding from the above, we can give more concrete-

but the expression for the coefficient \(P\) (in formula (5))—the ratio of the number of quanta recorded by the counting device to the number of quanta recorded by the scintillation detector (in formula (2), the intrinsic sensitivity threshold of the photomultiplier is not taken into account). In [12] the optimal choice of the energy discrimination threshold is discussed, and the advisability of amplitude discrimination of pulses is indicated, even in the case where the ratio of the number of pulses carrying information and recorded by the detector to the total number of pulses arriving from the photomultiplier is small. When testing large thicknesses, where the accumulation factor becomes a significant quantity, the coefficient \(P\) may reach 0.2–0.1 and lower. Such a useful reduction in the number of pulses arriving at the input of the counting apparatus, by selecting only those pulses that carry useful information, reduces the requirements imposed on the counting apparatus.

It is clear that, in order to carry out accurate discrimination of pulses by amplitude, it is necessary to ensure the highest possible stability of the threshold device. Under this condition, the pulse-counting method for measuring radiation intensity will make it possible to realize all its advantages over other methods of measuring intensity [12].

Let us determine the maximum total relative error of the method [13]. Usually, the value taken as the maximum error is equal to three times the root-mean-square error. Thus, for the maximum total relative error of the measurement method, the following approximate formula is valid:

\[ \varepsilon \simeq 3 \cdot \sqrt{\frac{1}{9}\varepsilon_{\mathrm{por}}^{2}+\rho^{2}}, \]

where

\[ \varepsilon_{\mathrm{por}} \]

is the maximum relative error of the threshold device, which is a function of the parameters of its electronic circuit and of the level of the continuous distribution in the discrimination region.

On the basis of the counting-spectrometric method of intensity measurement, the authors developed a flaw detector based on an industrial gamma installation GUP—Co-60. The inspection scheme is shown in Fig. 1. The block diagram of the installation is given in Fig. 9.

Fig. 9.

1. Source
2. Product
3. Motor
4. Scintillator
5. Photomultiplier
6. Shielding
7. Cathode follower
8. Pulse discriminator
9. Spectrometric amplifier
10. Pulse discriminator with stable triggering threshold
11. Pulse shaper
12. Integrator
13. Self-recording instrument

The flux of \(\gamma\)-quanta from the radioactive source \(\mathrm{Co}^{60}\) (1) passes through the product under investigation (2), which is fixed on the table of a specially developed manipulator (3), providing reversible rotation and axial displacement of the table at prescribed speeds. Beyond the pro—

Next is a scintillation counter (4), (5) (NaI (Tl) 70×50 crystal and FEU-13 photomultiplier tube) in lead shielding. The crystal is optically coupled to a light guide made of organic glass and then to the photocathode window of the photomultiplier tube. In the remote unit a cathode follower (6) is placed for matching with the cable connecting the remote unit to the operator’s console. Pulses from the photomultiplier tube are fed to a diode discriminator with an adjustable threshold (7); the discrimination threshold is set on the basis of a study of the pulse spectrum from the photomultiplier tube for the particular product being radiographed. Pulse amplification takes place in a spectrometric amplifier (8), assembled using secondary-emission tubes. The amplifier makes it possible to operate reliably at high counting rates (up to 0.5–1 Mcps). The amplified signals are fed to a highly stable threshold device (9), assembled using BLPP-type tubes with infinite slope. Pulses with an amplitude above the established discrimination threshold are shaped in amplitude and duration in the shaping stage (10). The normalized pulses are integrated and fed to a tube voltmeter, in the bridge diagonal of which a self-recording potentiometer EPP-09 is connected.

The use of circuits for recording and processing pulses from the photomultiplier tube with high amplitude and time resolution makes it possible to use the flaw detector in a high-count-rate mode (up to 500,000 pulses/sec after the discriminator). This ensures high sensitivity (up to 0.5% of the thickness of the section being radiographed, with a collimator diameter of 1.2 cm) when inspecting large steel thicknesses (up to 15 cm), with high productivity (up to 3 m²/h).

When radiographing welded seams, the flaw detector was operated according to a differential scheme; a second reference channel of the same type was introduced, which tracked changes in the thickness of the welded products being inspected.

Small changes in the design of the detector and of the recording apparatus, associated with the specifics of the technological process, will evidently make it possible to reliably monitor the quality of rolled stock at metallurgical plants, the thickness and integrity of thick-walled pipes, etc., that is, wherever it is necessary to continuously and automatically monitor the quality of products without destroying them.

REFERENCES

1. E. Shiebold, E. Becker. F. Vorsprach, W. Z. No. 3, 209–217, 1960.
2. E. Shiebold, E. Becker. F. Vorsprach, W. Z. No. 1, 283–292, 1961.
3. O. I. Leipunskii et al. Propagation of Gamma Quanta in Matter. Fizmatgiz, 1960.
4. E. P. Stoliarova et al. Investigation of spectral-angular distributions of scattered gamma radiation in protective barriers. Collection of MEPhI, issue 3, 1962.
5. V. I. Kukhtevich et al. Angular distribution of the dose of scattered gamma radiation in water. Atomic Energy, vol. 5, No. 6, 1958.
6. G. E. Butakova et al. The count-spectrometric method in isotope flaw detection of materials. Collection “Nondestructive Testing Methods,” Moscow, 1964.
7. V. I. Gorbunov et al. Investigation of the spectral composition of cobalt-60 gamma radiation behind an absorber. Report at the Conference on Nondestructive Testing Methods, Kiev, 1963.
8. D. P. Osanov. Dependence of buildup factors on the position of a protective barrier between source and detector. Collection of MEPhI, issue 3, 1962.
9. N. G. Gusev et al. Gamma Radiation of Radioactive Isotopes and Fission Products, 1961.
10. M. Geisler und Brumer. Atomkern Energie No. 2, 60–65, 1960.
11. M. Witer, J. Canad. Phys., No. 7, 33–96, 1955.
12. V. I. Gorbunov et al. High-speed isotope flaw detection. Report at the Conference on Nondestructive Testing Methods, 1963.
13. T. K. Tkachenko. Radioactive Isotopes in Instrument Engineering. 1960.
14. V. Sh. Dzhagatsparian. Application of Radioactive Isotopes in Chemical Technology, 1963.