LETTERS TO THE EDITOR

REGARDING I. D. KONOZENKO’S ARTICLE “SEMICONDUCTOR BOLOMETERS”

(UFN, Vol. LVI, No. 2, p. 283)

Recently, semiconductor materials have been used for the manufacture of thermal receivers of infrared radiation—bolometers. Interest in such applications is quite understandable, since semiconductor materials usually have large temperature coefficients of resistance and thereby promise a significant increase in the sensitivity of thermal receivers.

In I. D. Konozenko’s article an attempt is made to systematize data on a number of semiconductor bolometers developed up to the present time abroad, and to characterize the position of such bolometers among other thermal receivers of infrared radiation (mainly metallic, superconducting, and dielectric bolometers).

In our opinion, however, the article suffers from a number of serious shortcomings, as a result of which the state of affairs with semiconductor bolometers may disorient the reader. We therefore consider it our duty to draw attention to them.

1. The main defect of the review is the author’s ignoring of the measurement time in estimating the threshold sensitivity of bolometers. It is well known that the limiting sensitivity of any radiation receiver is bounded by its own noise, whose root-mean-square value is inversely proportional to the square root of the time constant of the measuring device at whose output this noise is registered1. At present, the threshold sensitivity of a receiver is defined as the power of the radiation flux producing an electrical signal at the output of the measuring system equal in magnitude to the root-mean-square value of the receiver noise measured at the output of this same measuring system2. It is quite obvious that a threshold determined in this way will depend on the properties of the measuring system (for example, on the time constant, etc.), and therefore they must either always be chosen to be constant or be very clearly specified. Let us see how I. D. Konozenko approaches this question.

On p. 294 I. D. Konozenko has placed Table 1, where “theoretical data on the sensitivity threshold obtained in the works of a number of authors” are “presented.” From a comparison of these data he concludes that the theoretical sensitivity threshold of bolometers with a receiving-surface area of 0.1 cm² is a power of \(5 \cdot 10^{-11}\) watts. However, by indicating neither in the table nor in the explanations to it to what measurement time the data cited refer, I. K. Konozenko has thereby deprived this table of meaning. It must be noted here also that the data refer to different measurement times. Thus, for example, Becker and Moore experimentally (and not theoretically, as Konozenko states) determined the sensitivity threshold of their bolometer for a measurement time of 15 seconds3. At the same time, K. Jones and R. Gevens give the value of the threshold sensitivity for a time constant of the measuring instrument equal to the time constant of the receiver, and the figures cited by Konozenko correspond to its value equal to one second456.

On p. 284 I. D. Konozenko gives a number of figures characterizing the sensitivity threshold of metallic and semiconductor bolometers and concludes that “in individual cases it proves possible to create samples of semiconductor bolometers that have” a better sensitivity threshold than metallic bolometers. This conclusion is also made on the basis of comparing quantities obtained for different measurement times and is therefore incorrect.

On p. 304 I. D. Konozenko gives Table IV, and states that “the data presented there are borrowed from various literary sources without the corresponding recalculations to uniform conditions.” By means of this table the author invites the reader “to judge the place already occupied by semiconductor radiation bolometers among other types of bolometers.” It is clear, however, that without the corresponding recalculations to uniform conditions it is impossible to compare bolometers even approximately. I. D. Konozenko tries to justify presenting the table in such an unprocessed form by saying that “in bolometric technology, standard criteria for measuring these or

...quantities for assessing the quality of bolometers. In reality, however, the situation is far from hopeless: even if standard measurement criteria do not yet exist, in any case solid rules for this kind of assessment have become established in the literature. As early as 1940–1946 K. S. Vul'fson carried out a comparison of various thermal radiation receivers, used mainly in the prewar period and designed for operation with galvanometers[^7]. The principal parameter here was the signal-to-noise ratio of the receiver, obtained under specified comparative measurement conditions. The most convenient and appropriate measurement conditions for comparing various modern radiation receivers were proposed in 1947–1949 by K. Jones. On the basis of data published in the literature, as well as on the basis of his own information, he compared 50 thermal infrared-radiation receivers with one another by their signal-to-noise ratio, reduced to the proposed conditions of comparison5. Subsequently these conditions were also used by other authors. In one of K. Jones’s papers4, where his comparison conditions are presented, there is also a reference to I. D. Konozenko; however, the author did not use K. Jones’s method, since in that case his Table IV would have looked quite different. I shall permit myself to cite excerpts from Tables I and II of K. Jones’s paper5 so that one may judge what place the semiconductor bolometer occupied in 1949 among other modern low-inertia radiation receivers.

Here $\tau$ is the time constant of the receiver, $F$ is the area of its receiving surface, and $H_m$ is the minimum perceptible radiation flux under K. Jones’s comparative conditions (the basis of which is equality of the time constants of the measuring instrument and of the receiver itself). The receivers listed are among the best (as applied to the most frequently encountered problems) among samples of the given type. The data refer to 1949, but, judging from the literature, no special changes have since occurred in the situation with thermal infrared-radiation receivers and, in particular, with semiconductors abroad.

Thus, despite the variety in the values of $\tau$ and $F$, K. Jones’s method makes it possible to characterize various receivers by mutually comparable values of the quantity $H_m$.

From the table given it is evident that modern semiconductor bolometers are still far from perfection and are considerably inferior to other infrared receivers.

1. Secondly, an equally substantial defect of the review is its divergence from the literature data. Unfortunately, a list of all such defects would be too long, and therefore we shall dwell only on the most important ones, in our opinion:

a) I. D. Konozenko writes (p. 293): “K. S. Vul'fson gave the theoretical value of the sensitivity threshold for any receiver as $7\cdot10^{-9}$ W, but this is incorrect, since we know receivers with a threshold of $10^{-9}$ W.” But in Vul'fson’s work it is clearly stated that he defines the threshold as a power 100 times greater than the noise; moreover, the figure $7\cdot10^{-9}$ applies only to a thermocouple with a surface of $1$ cm$^2$;

b) further, in the same place (p. 293) formula (21), taken from Vul'fson’s paper[^7], is given, and from it the minimum power sensed by a bolometer is calculated; however, completely incorrect quantities are substituted into this formula: Vul'fson has

$$ \alpha=\frac{d\rho}{dT}, $$

whereas I. D. Konozenko substitutes the value corresponding to

$$ \alpha=\frac{1}{\rho}\frac{d\rho}{dT}, $$

and as a result makes an error of approximately 1000 times (if at the same time the exponent of $T_0$ is corrected from $1/2$ to $2$, since apparently this is simply either a typographical error or a slip of the pen);

c) it is asserted that Wulfson’s formula (21) was derived without taking into account the effect of the change in bolometer heating due to the current flowing through it when it is illuminated, whereas this formula was obtained under the condition that such heating does occur, but that it exactly compensates the loss of heat due to radiation, since in this case the optimum sensitivity of the bolometer will be realized;

d) also incorrect is the statement that formula (21) was derived under the assumption of bridge equilibrium, since in fact Wulfson’s derivation considers a nonequilibrium bridge as the more advantageous one.

Another example. I. D. Konozenko considers the best semiconductor bolometer to be the bolometer developed by Becker and Moore³. He refers to the corresponding note by the authors, where they give the principal parameters of their bolometer. However, the data for this bolometer given by Konozenko (p. 299) do not agree at all with the authors’ data. Their bolometer resistance is \(2 \cdot 10^6\) ohms; Konozenko’s is \(6 \cdot 10^6\) ohms; their receiving-surface area is \(0.6\ \mathrm{mm}^2\), Konozenko’s is from 0.5 to \(5\ \mathrm{mm}^2\); their minimum detectable power, which was recorded with an apparatus having a time constant of 15 seconds, is \(10^{-10}\) watt, while Konozenko gives the figure \(5 \cdot 10^{-10}\) watt. And, finally, in Table IV Konozenko gives the supply voltage for this bolometer as 100 volts, whereas the authors do not give this figure at all, and the only figure in their paper that is close to this (110 V a. c.) denotes the supply voltage of the amplifying device with which the authors made their measurements. Further, the data for the bolometer of Brattain and Becker⁸, placed by I. D. Konozenko on p. 299, to a considerable extent do not correspond to the data given in Table IV, although in both cases the reader is referred to one and the same paper.

1. In addition to the defects noted above, I. D. Konozenko’s article contains a number of statements that are incorrect from the standpoint of modern measuring technique. For example, on p. 285 I. D. Konozenko writes: “since semiconductor bolometers have high resistance, it is possible to use amplifying tubes. This eliminates the necessity of using exclusively galvanometers as signal recorders, and one may also use other instruments: a cathode voltmeter, a cathode oscillograph, etc.” This statement can only be understood to mean that, in the case of receivers having low resistance, the use of amplifier tubes is impossible. This is entirely incorrect. Electronic amplifiers are successfully used with receivers having a resistance of \(0.2\)—\(0.3\) ohm (in superconducting bolometers), and amplification of signals from receivers with a resistance of 5—20 ohms is carried out in numerous instruments intended for broad use in research laboratories and in industry⁹, ¹⁰, ¹¹, ¹².

As for the substance of the question of semiconductor bolometers, the following may be said on this subject: in the 13 years that have elapsed since the appearance of semiconductor bolometers, the main efforts have been directed toward finding the material most suitable for them, which should possess a high temperature coefficient of resistance and at the same time permit the fabrication from it of thin (\(\sim 0.1\ \mu\)) and stable layers that do not give additional electrical noise. However, to this day such a material has not been found. Foreign-made semiconductor bolometers made of various materials, in terms of threshold sensitivity (the principal parameter essential for bolometers), still fall considerably behind (by as much as 20-fold or more) the best specimens of other types of thermal infrared receivers operating at room temperature, although their high volt sensitivity constitutes a certain advantage when used in comparatively rough instruments, where the fluctuation limit is not reached.

M. Markov

Literature Cited

1. V. L. Granovskii, Electrical Fluctuations, ONTI, 1936.
2. R. Clark Jones, J. Opt. Soc. Am. 37, 879 (1947).
3. Becker, Moore, J. Opt. Soc. Am. 37, 354 A (1948).
4. R. Clark Jones, J. Opt. Soc. Am. 39, 327 (1949).
5. R. Clark Jones, J. Opt. Soc. Am. 39, 344 (1949).
6. R. Havens, J. Opt. Soc. Am. 36, 355 (1946).
7. K. S. Vul’fson, Proceedings of the VEI, issue 41 (1940).
8. Brattain a. Becker, J. Opt. Soc. Am. 36, 354 (1946).
9. N. Fuson, J. Opt. Soc. Am. 38, 845 (1948).
10. White, M. Liston, J. Opt. Soc. Am. 40, No. 1 (1950).
11. Baird, O’Bryan, J. Opt. Soc. Am. 37, 754 (1947).
12. Wright, Herscher, J. Opt. Soc. Am. 37, 216 (1947).