NEW INSTRUMENTS AND METHODS OF MEASUREMENT

PHOTOMETRIC PROPERTIES OF SELENIUM PHOTOCELLS WITH A BARRIER LAYER

S. G. Yurov and V. S. Khazanov

1. INTRODUCTION

Physics, chemical production, printing, geophysics, meteorology, lighting engineering, textile production—this is far from a complete list of the fields of science and technology in which measurements of radiation and, in particular, light measurements with the aid of photocells play no small role.

At present, selenium photocells with a barrier layer are most widely used for these purposes. This is explained by three main reasons:

1. Selenium photocells*) are simple to operate, since they do not require external voltage sources.

2. These photocells possess a comparatively high sensitivity, reaching up to 500 μa/lm.

3. Their spectral sensitivity resembles the sensitivity curve of the human eye, which is important for light measurements, which are, of course, the most frequently performed.

A tremendous number of works have been devoted to the theory of operation and to various properties of selenium photocells; however, sufficiently complete surveys of the photometric properties of selenium photocells of various types are practically nonexistent. As a result, users of these photocells who do not have sufficient experience and the possibility of acquainting themselves with their properties from the numerous individual works are not sufficiently familiar with the photometric properties of the photocells being used, which often leads to very large errors, appreciable even in the crudest measurements.

*) For brevity, in what follows the words “with a barrier layer,” which are necessary in substance, will be omitted.

Using the experience of the work of the VEI photometric laboratory and data from the literature, the authors attempt, from a unified point of view, to characterize the most important photometric properties of selenium photocells, without in any way intending to expound the theory of their action.

For a correct assessment of the numerical data given below, it should be borne in mind that they are and can be only approximate, since, as practice has shown, selenium photocells are individual in their properties; moreover, differences are noticeable not only between different makes (firms), but also within one and the same make and even series. The information set forth below gives an idea of possible errors, means of eliminating them, and probable causes of their occurrence.

2. CHARACTERISTIC “PHOTOCURRENT—ILLUMINATION”

The “primary” photocurrent arising as a result of the action of a flux of radiation on a photocell is directly proportional to this flux.^1

Primary photocurrent flows through two parallel circuits: the external circuit (including the measuring instrument), whose resistance remains constant, and the internal circuit (selenium, the blocking layer), whose resistance depends on the illumination of the photocell. Only in the case where the internal resistance of the photocell is considerably greater than the external resistance can the current flowing in the external circuit be regarded, with some approximation, as proportional to the illumination, i.e. it may be assumed that the “current—illumination” characteristic is linear (Fig. 1). The internal dark resistance of the blocking layer depends on the applied voltage and has a value from 1 to 250 KΩ for different makes of photocells (Fig. 2). It decreases with increasing illumination; consequently, the “current—illumination” characteristic will deviate from a straight line the less, the lower the illumination and the lower the resistance of the external circuit, i.e. of the measuring instrument. The e.m.f. of a selenium photocell, for the same reasons, increases nonlinearly with increasing illumination and

Fig. 1. Dependence of the photocurrent on illumination for different load resistances.

when the illuminance increases, approaches a certain constant value.

Numerous experiments\(^{2,3}\), aimed at determining the “current—illuminance” characteristic for selenium photocells, confirm these conclusions in a first approximation.

As an example of the influence of the magnitude of the external resistance on the measurement error that may result from the nonlinear dependence of the photocurrent in the external circuit on illuminance, Table I*) is given.

Table I

From the data of the table it may be concluded that, with a load resistance equal to 10 ohms, most photocells possess, in the interval 400–1000 lx, satisfactory linearity (deviation 2–3%), whereas with an external resistance of 200 ohms the error may reach 25% for a low-ohmic photocell. In the illumination interval—

) 1. The table was compiled from the results of tests carried out at the VEI, and from literature data.
2. The table includes test results only for certain photocells most characteristic of the type.
3. Usually, the smaller the area of the sensitive surface of the photocell, the better linearity it exhibits.
4. Usually photocells with high internal resistance (more linear) have a smooth dark-gray surface, whereas in low-ohmic photocells the sensitive surface is usually uneven, light, with a metallic luster.
*) Selenium photocells SV No. 25 and No. 92 were kindly provided by N. S. Khlebnikov and M. I. Belyaev.

S. G. YUROV AND V. S. KHAZANOV

... illuminances up to 10–30 lx, in some photoelements deviations of the photocurrent from linearity toward more rapid growth are encountered. This probably occurs when the resistances of the selenium layer and of the barrier layer are of the same order of magnitude, and the photoconductivity of the selenium begins to play a noticeable role.

In this case an increase in the sensitivity of the photoelement with illuminance is observed^4. For measuring illuminances below 0.1 lx, selenium photoelements are considered inapplicable because of their large initial instability (see § 4). At illuminances up to 30–40 lx, the current and even the e.m.f. for photoelements with sufficiently large internal resistance depend practically linearly on the illuminance^5.

Fig. 2. Resistance of a selenium photoelement as a function of illuminance and applied voltage.

With a further increase in illuminance, the e.m.f. grows nonlinearly, approaching a limit that depends on the spectral composition of the incident radiation and on the specimen of the photoelement^6,7. Beginning with illuminances of approximately \(10^3\) lx, even with a small resistance in the external circuit the photocurrent increases nonlinearly with increasing illuminance^7,8. The form of the characteristic “short-circuit photocurrent—illuminance,” within experimental error, does not depend on the spectral composition of the radiation^9, which is yet another argument in favor of choosing measuring circuits with a small load resistance.

The data set forth above can be regarded only as approximate, since selenium photoelements, even within one and the same brand, possess noticeable individuality. Depending on the requirements for the accuracy of the measurements, the question may arise of checking the linearity. The latter is most often carried out on a photometric bench using the inverse-square law. However, this method, despite its comparative complexity, is insufficiently reliable and accurate. A simpler and most accurate method consists in using an apparatus made up of several lamps burning simultaneously, each of which can be covered ...

opaque shield. Using such an arrangement, Campbell^10 established on several photocells (of different makes) that the short-circuit regime is not the best one for a linear “photocurrent—illuminance” dependence. According to Campbell’s data, for each photocell the most advantageous resistance of the external circuit must be selected, at which deviations from linearity are minimal.

In all probability, the results of these experiments can be explained by changes in the resistance of the selenium layer (not of the blocking layer). In any case, it may be said with sufficient certainty that the usual equivalent circuit^1 of a selenium photocell is inadequate for a complete description of its photometric properties.

3. SPECTRAL CHARACTERISTICS

Figure 3 gives curves of spectral sensitivity for photocells manufactured by various firms, and, for comparison, the visibility curve. These curves are plotted in relative units of sensitivity on the basis of literature data and measurements carried out at VÉI.

Attention is drawn to the diversity in spectral sensitivity for individual specimens of photocells. Investigation of the causes entailing differences in spectral sensitivities^12,11 showed that it is determined by the presence of impurities in the selenium. Thus, for example, impurities of sulfur and naphthylamine (a) shift the spectral characteristic into the short-wavelength part of the spectrum, while an impurity of tellurium shifts it into the long-wavelength part. On the basis of the above-noted similarity between the curve of spectral sensitivity of the photocell and the visibility curve, in practical photometry selenium photocells are often used for comparison of sources of different spectral composition. This almost always leads to very appreciable errors, even in production and field measurements. These errors are the greater, the greater the difference in the spectral composition of the radiations being compared and the greater the difference between the photocell’s spectral-sensitivity curve and the visibility curve. As an example, Fig. 4 gives the magnitude of the error obtained when using different photocells to compare radiations with color temperatures from \(1900^\circ\text{K}\) to \(2320^\circ\text{K}\).^2 Considering Fig. 4, one may conclude that, when comparing radiations \(T_{\text{c}} = 1900^\circ\text{K}\) and \(T_{\text{c}} = 2320^\circ\text{K}\), an error may be obtained both greater than \(20\%\) and less than \(3\%\), depending on the course of the photocell’s spectral-sensitivity curve. On the other hand, measurements carried out at VÉI showed the possibility of photometric comparison by means of a GOI photocell (see Fig. 3) of the radiation of an incandescent lamp \((T_{\text{c}} = 2360^\circ\text{K})\) with the radiation of a luminescent lamp \((T_{\text{c}} = 6000^\circ\text{K})\) with an error not exceeding \(13\text{--}15\%\). This is explained by the fact that the spectral sensitivity of these

Fig. 3. Spectral sensitivity of photocells of various brands: I—ФАИ, II—ГОИ, III—ЛЭТИ, IV—Weston and Photronic, V—Tungstam, VI—Eye.

PHOTOMETRIC PROPERTIES OF SELENIUM PHOTOCELLS

...of photocells comes closer to the visibility curve of the average eye than that of photocells produced by other firms. Their comparatively low sensitivity in the near infrared and ultraviolet regions is especially favorable.

Fig. 4. Error produced by various photocells in comparing radiations of different spectral composition:
I—Weston, II—NPL, III—Electrocell, IV—NPL, V—SAF and Tungstam.

In order to eliminate or, more precisely, reduce errors in the photometry of heterospectral radiations with a selenium photocell, several methods have been proposed. Among the most widespread are the following:

1. The galvanometer measuring the photocurrent is provided with a specially graduated shunt, which changes the fraction of the photocurrent passing through the galvanometer by the required number of times when measuring radiation sources of known spectral composition. In practice, such a shunt can be made only for a comparatively small number of radiation sources, such as mercury, sodium, neon, etc., lamps.

The positive aspect of the method is its simplicity, but nevertheless it has not become widely used.

1. The photocell is supplied with a table of correction coefficients giving the ratios of the sensitivity of the photocell for the radiations of the sources being measured to its sensitivity for the radiation of an incandescent lamp with a definite color temperature. The method possesses all the advantages and disadvantages of the method described above.

2. The most convenient and, in practice, the most widespread method of reducing the error is to approximate the sensitivity curve of the photocell to the visibility curve by means of a compensating light filter.

Thus, it may be concluded that photocells with compensating filters can be used quite effectively where great accuracy is not required, for example, in constructing “objective” luxmeters. However, their use for precise photometric measurements, without introducing the corresponding corrections, cannot be recommended.

In the event that a photocell with a compensating filter is used to measure light falling upon it at different angles, for example, in measuring the illumination produced by the sky and in other similar cases, it should be borne in mind that compen-

S. G. Yurov and V. S. Khazanov

The compensation filter greatly worsens the cosine characteristic of the photocell.

The use of a compensation filter appears very attractive, owing to the simplicity and universality of the method. Over the last 10–15 years several types of photocells with compensation filters have been developed and brought to market. The most widespread have been the photocells of GOI[^13], Weston[^14], Dressler[^15], and Rix[^16].

At VEI, a comparison was made of the photometric properties of these photocells[^17], which showed that the best are the GOI and Weston photocells. However, even with comparatively small differences in spectral composition, such as, for example, an incandescent lamp \((T_{\mathrm{col}} = 2860^\circ\mathrm{K})\) and a fluorescent lamp \((T_{\mathrm{col}} = 6500^\circ\mathrm{K})\), the errors for a Weston photocell may reach 5%, and for a GOI photocell \(-(1 \div 3)\%\)*.

Among the properties of selenium photocells that are unfavorable for precision measurements is the dependence of the curve of spectral sensitivity on the resistance of the external circuit and on the load current[^4].

Fig. 5. Change in the ratio of the sensitivity of a photocell in the blue part of the spectrum to its sensitivity in the red part of the spectrum as a function of external resistance and load current.

In the graph: vertical axis, \(S_{\text{blue}}/S_{\text{red}}\); horizontal axis, short-circuit current in \(\mu\mathrm{A}\). Curves are marked \(R=\infty\), \(R=10000\,\Omega\), \(R=1000\,\Omega\), \(R=100\,\Omega\), \(R=0\).

For example, Fig. 5 gives curves of the dependence of the spectral sensitivity of an NPL photocell on external resistance and load. This dependence is yet another confirmation of the desirability of using selenium photocells in the short-circuit regime or close to it.

When working with photocells one should bear in mind the nonuniformity of sensitivity over the surface, reaching as much as 100% from the center to the edges. Within the accuracy of measurement, the spectral sensitivity over the surface of a photocell does not change[^18].

4. INITIAL INSTABILITY

By initial instability is meant the change in photocurrent in the initial period after switching on the illumination of the photocell, all other conditions remaining unchanged. In speaking of initial instability,

* The “minus” sign means that the luminous intensity of the fluorescent lamp, as measured by the photocell, is less than its true value.

usually limited to the concept of fatigue, by which is meant a gradual decrease of the photocurrent with time under constant illumination of the photocell. However, other terms are also encountered for characterizing various manifestations of initial instability: the Kriech effect (Kriecheffekt), or the effect of photocurrent growth, as well as terms characterizing instability in general, irrespective of the direction of the change in current, such as, for example, the drift effect (Drifteffect).

Fig. 6. Initial instability: I—GOI No. 1896, 1947; II—GOI No. 1900, 1947.

The initial instability of selenium photocells is of interest primarily from the following points of view:

1) Its influence on measurements of radiation from incandescent lamps, especially with a color temperature corresponding to the most common standard lamps (approximately \(2200^\circ\mathrm{K}—2600^\circ\mathrm{K}\)).

2) Its influence on the accuracy of spectral measurements, when the selenium photocell is placed as the receiver at the entrance slit of a monochromator.

The determination of initial instability was carried out under illumination of the photocells by the light of an incandescent lamp with \(T_{\mathrm{c}} = 2360^\circ\mathrm{K}\), and also under irradiation by monochromatic fluxes \((\lambda = 460\,m\mu,\ 550\,m\mu,\ \text{and}\ 680\,m\mu)\). The instability of the photocurrent for most photocells is noticeable for 10–15 minutes. In the case of illumination of the photocell by the light of an incandescent lamp, the initial instability

may consist of 1) an initial rise and then a fall of the current, 2) a gradual rise, or 3) a gradual fall of the current.

Characteristic curves of the initial instability for these cases are shown in Figs. 6 and 7. Photocells are also encountered which, under these illumination conditions, do not possess an initial instability.

When photocells are illuminated with blue and yellow-green light, either constancy or an increase of current is observed, whereas irra-

Fig. 7. Initial instability of NBS photocells.

diation with red light always causes a comparatively rapid fall of current. When photocells are illuminated with deep-red light with \(\lambda = 700\,m\mu\), the fall of current occurs still more rapidly.

The results of the experiment are summarized in Table II.

The initial instability was evaluated by the greatest difference of photocurrents over 10 minutes of testing and was expressed as a percentage of the initial value of the photocurrent. The illumination of the photocells for different spectral compositions of the incident radiation was varied in such a way that the generated photocurrent remained approximately constant. Thus the data of Table II for \(\lambda = 460\), 550, and \(680\,m\mu\) should not be regarded as contradicting the results obtained under irradiation with light of spectral composition corresponding to \(T_{\mathrm{цв}} = 2360^\circ\mathrm{K}\), since the energy illuminances on the photocell in these cases (at the same photocurrent) are very different.

The investigation of the initial current instability when photocells are illuminated by incandescent-lamp light was carried out by Berbrou and Preston\({}^{19}\). Berbrou gives the dependence of the photocurrent on time,

Table II

Initial instability as a percentage of the initial current value

analogous to the majority of the curves obtained at VEI. In order to obtain reproducible measurement results, he recommends using the graph shown in Fig. 8, which represents the result of processing a series of dependences analogous to those shown in Fig. 7. Along the ordinate is plotted the time at which the photocurrent maximum occurs; along the abscissa, the illumination of the photoelement. The work carried out by Preston with photoelements from various firms gave results in relation to which Berbrow’s data are a special case. Preston also observed, for individual photoelements, a rise and fall of the current.

Fig. 8. Time at which the photocurrent maximum occurs as a function of illumination \((T_{\mathrm{col}} = 2360^\circ\mathrm{K})\).

Elvegard^20^ and Knoll^9^ investigated the dependence of the initial instability on the wavelength of the incident radiation. Elvegard worked with photoelements manufactured by the firm Tsirold.

and TsAF and obtained results close to those obtained at VEI. A discrepancy exists only for the case of illuminating the photoelements with monochromatic blue light, in which Elvegård observed a decrease in photocurrent. Knoll plotted the dependence of fatigue and current increase on wavelength over the spectrum and obtained the results shown in Fig. 9. The general course of the fatigue curves agrees with the VEI data and with Elvegård’s results.

Fig. 9. Change in initial instability over the spectrum: I — current increase (Weston); II — current decrease (Tungstram); III — current decrease (TsAF).

In those cases where a definite spectral sensitivity is not required from the photoelement, in order to reduce the initial instability one should decrease the sensitivity in the red part of the spectrum by introducing a corresponding light filter. In spectral measurements with photoelements (especially in the red part of the spectrum), to reduce the error due to initial instability, the reading on the galvanometer scale should be taken after a fixed interval of time from the beginning of irradiation. Practice shows that such a method appreciably increases the reproducibility of the measurement results.

5. STABILITY OF PHOTOELEMENTS

The constancy of sensitivity of a selenium photoelement depends on the method of its manufacture. This is the cause of contradictory data concerning the stability of selenium photoelements from different firms. One of the main causes of changes in sensitivity is considered to be a change in the resistance of the barrier layer ^{5, 2, 21}. However, while for GOI photoelements Freiberg indicates a decrease in sensitivity with time caused by a decrease in the resistance of the barrier layer, Bergmann, as well as Pelz and Lange, report an increase in current and an increase in the resistance of the barrier layer as a result of aging. The least sensitive photoelements are the most constant ^3, provided the load resistance is small. Thus, of two low-sensitivity TsAF photoelements illuminated for \(\sim 150\) hours at \(6000\) lx, the short-circuited one showed excellent constancy, whereas an identical photoelement, connected to the galvanometer only during measurements, showed by the end of the test period a decrease in sensitivity of 5%.

High-sensitivity SAF photoelements under the same conditions give a current drop of 20–40%. At lower illuminances (100 lx) these photoelements possess a high constancy of sensitivity. The Weston firm, taking into account the use of selenium photoelements for meteorological measurements, manufactured, according to Bergmann’s prescription, a photoelement distinguished by especially high constancy[^22].

This photoelement can be exposed to the open sky for a month without a change in sensitivity. The course of recovery of the sensitivity of a selenium photoelement after overexposure, when the latter is kept in the dark, is shown as an example in Fig. 10. Complete recovery of sensitivity was observed only after 10 hours[^9].

Fig. 10. Establishment of the sensitivity of a photoelement after overexposure.

Summarizing, it may be said that, by choosing a photoelement with not very high sensitivity and placing it in a measuring circuit that provides a small load resistance, one may expect considerable constancy of the sensitivity of a selenium photoelement.

Fig. 11. Oscillogram of the photocurrent after the illumination is switched on.

6. INERTIA

In the first seconds after illumination is switched on to a photoelement, the current increases gradually. Thus, the inertia of a selenium photoelement is readily noticeable. The increase of the photoelement current after switching on illumination was studied with the aid of an oscillograph by Bergmann and Pelz[^3]. For individual photoelements they observed a difference in inertia exceeding 100%. The characteristic course of the change in current is presented in Fig. 11.

Fig. 12. Inertia of a photoelement under irradiation with monochromatic radiation.

S. G. YUROV AND V. S. KHAZANOV

The inertia of a photoelement depends on the spectral composition of the radiation incident on it (Fig. 12) and, as a rule, is the greater the lower the illuminance. At very low illuminances (of the order of \(0.01\) lx) the inertia increases so much\({}^{23}\) that the photoelement becomes unsuitable for measurement purposes.

7. APPLICABILITY OF TALBOT’S LAW

If a selenium photoelement is illuminated with intermittent light, then the readings of a galvanometer connected to the photoelement are obtained, in a first approximation, the same as under constant illumination with the same effective illuminance as the intermittent illumination. However, on more detailed consideration one may pose three questions:

1. With what accuracy do individual photoelements obey Talbot’s law?

2. Do the galvanometer readings depend on the frequency of the flashes and on the ratio between the durations of the dark and light parts of the period?

3. Does the accuracy of obedience to Talbot’s law change with a change in the spectral composition of the flashes?

There are data\({}^{2,24}\) indicating that selenium photoelements\({}^{*)}\) obey Talbot’s law with an accuracy of approximately \(\pm 1\%\) up to frequencies of \(200\)—\(300\) cps. However, even among those photoelements that were tested, several were found which showed fairly large deviations (up to \(3.5\%\)) from Talbot’s law. These deviations were especially noticeable when the apertures of the sector disk were small. At the same time, a quite definite dependence of the readings of the galvanometer connected to the photoelement on the frequency was also observed. Thus, for example, one of the photoelements, when the frequency was changed from \(20\) cps to \(300\) cps, showed an increase in photocurrent of \(20\%\).

It was found\({}^{25}\) that the deviations from Talbot’s law are the smaller, the smaller the external resistance of the selenium photoelement circuit. This fact makes it possible to explain the deviations from Talbot’s law.

When a sector disk is used, the actual illuminance on the photoelement during the “light” part of the period is greater than the equivalent mean illuminance. Thus, during the “light” part of the period the photoelement operates at another point of the “current—illuminance” characteristic than in the case of continuous illumination. The deviation of this characteristic from linearity may lead to deviations from Talbot’s law, especially with sector disks of small aperture.

\({}^{*)}\) In Preston’s work\({}^{2}\), 5 photoelements manufactured at NPL were tested, as well as Weston, Tungstram, SAF, and Elektrocell photoelements, 15 in all.

When the frequency increases, the capacitance of the photoelement, amounting to tenths of a microfarad, begins to play a role.

The check carried out at VEI on the compliance of selenium photoelements with Talbot’s law concerned both the detection of the dependence of the photocurrent on the flicker frequency and the correspondence of the photocurrent to the mean value of intermittent illumination.

In the range of flicker frequencies from 4 to 100 cps, and within the limits of “transmission” by a rotating sector from 10 to 50% of the radiation incident upon it, for all photoelements there was observed independence of the current from frequency when the photoelements were illuminated by white and colored light sources. To check the correspondence of the photocurrent to the mean value of intermittent illumination, photoelectric measurements were made of the “transmission” of a sector disk with an aperture equal to 30%. The results are summarized in Table III, showing the deviation from the true value in percent.

Table III

It is interesting to note that the accuracy with which photoelements obey Talbot’s law may differ even for one and the same series of photoelements.

8. EFFECT OF TEMPERATURE

The temperature coefficient of selenium photoelements differs for different specimens both in magnitude and in sign. Usually\(^{8,26}\) the photocurrent decreases with increasing temperature, and the direction and rate of change depend on the illumination and the external resistance.

The results of measurements of the temperature coefficient of selenium photoelements, reported by individual authors, differ greatly from one another, which does not permit these data to be averaged.

Thus, Lange\(^{5}\), for the temperature range from −25° to +35° C, gives for the short-circuit photocurrent the dependence \(i = i_0(1+\alpha t)\), where \(\alpha = \pm 0.0003\). Bergmann and Pelz\(^{3}\), for the same temperature interval, give \(\alpha = +0.001\), while Putseiko\(^{26}\), between +15° and +30° C, for GOI photoelements gives \(\alpha\) varying from \(\alpha = -0.006\) to \(\alpha = -0.03\), and reports that at GOI it proved possible to obtain photoelements with an almost zero temperature coefficient of the short-circuit current.

For the photo-e.m.f. the temperature coefficient is usually 2–3 times greater than for the photocurrent.

*) Photoelements marked “K” have a compensating filter.

A “hysteresis” character of the temperature effect and the dependence of the temperature coefficient on the resistance of the external circuit are clearly visible in Fig. 13.

Bernard’s investigations⁴ showed that the temperature coefficient also depends on the spectral composition of the light. According to his data, the sensitivity to different colors has the following temperature coefficients (Table IV).

Fig. 13. Influence of external resistance on the dependence of current on temperature:
I — illumination 400 lx, \(R_{\mathrm{ext}}=1000\ \Omega\);
II — illumination 400 lx, \(R_{\mathrm{ext}}=0\).

Table IV

In laboratory conditions, when measurements of the standard and the tested source usually follow immediately one after the other and the temperature remains at approximately a constant level, the influence of temperature is insignificant. But when precise measurements are carried out, temperature changes of \(1\text{–}2^\circ\mathrm{C}\) may already have a substantial effect, increasing the error by \(1\text{–}1.5\%\). Considerably larger errors may occur in those cases when a photometer with a selenium photoelement, calibrated in the laboratory, is used for production or field measurements, with temperature fluctuations that may reach \(20\text{–}30^\circ\mathrm{C}\).

9. MEASUREMENTS OF POLARIZED LIGHT

When measuring polarized light with selenium photoelements, it should be borne in mind that their readings depend strongly on the angle of incidence of the light and on the orientation of the plane of polarization relative to the plane of incidence of the radiation⁷.

10. MEASUREMENT OF ILLUMINANCE PRODUCED BY OBLIQUE LIGHT BEAMS

Luxmeters in which the light receiver is a selenium photocell are usually calibrated with the aid of a standard lamp normally illuminating the surface of the photocell. These luxmeters, however, are often used to measure the illuminance produced by oblique beams or by diffuse light*). Yet the current delivered by the photocell for oblique incidence of light does not decrease in proportion to the cosine of the angle of incidence, which sometimes leads to very considerable errors. Deviations from the cosine law arise for the following reasons:

1. The housing of the photocell screens the beam of light incident upon it.
2. Fresnel reflections from transparent materials covering the photosensitive layer (protective glasses and lacquer coatings, compensation filters) increase as the angle of incidence of the light increases.

Fig. 14. Deviations from the cosine law for various photocells at small angles of incidence of the radiation.

1. In oblique incidence, the light falling on the photosensitive layer proves to be partially polarized.

At small angles of incidence both positive\(^{29}\) and negative\(^{2}\) deviations of the photocurrent from the cosine curve were observed (see Fig. 14). For large angles (Fig. 15) a negative deviation is observed, the greater the larger the angle of incidence.

To reduce the deviation of the photocurrent, under oblique incidence of radiation, from the cosine law, a plate of diffusely transmitting material is often placed in front of the photocell; the material usually used is ground opal glass\(^{2,30}\). However, as measurements carried out at the VEI show (see Fig. 15a), the use of such plates (tests were made with glass ground on one and on two sides, and also with ground opal glass) not only does not reduce the deviations from

*) For example, in measurements of street illuminance or of illuminance produced by the sky.

of the cosine dependence, but, on the contrary, increases them in comparison with the deviations observed for a photocell without additional glasses, or even with a transparent protective glass. From Fig. 15a it is evident that, for example, for an angle of incidence of light equal to \(60^\circ\), the deviation from the cosine dependence is 7% for a photocell without protective glass, 15% with a transparent protective glass, 20% with a deeply frosted glass, and 35% with a glass frosted on both sides.

To approximate the cosine dependence, various kinds of attachments are also used\(^{28,32}\), which, when light falls at large angles, artificially increase the luminous flux incident on the photo-

Fig. 15a. Dependence of the photocell current on the illumination angle \(\alpha\): \(I\)—\(\cos \alpha\); \(II\)—photocell without protective glass; \(III\)—photocell with transparent protective glass; \(IV\)—photocell with deeply frosted glass; \(V\)—photocell with glass frosted on both sides; \(VI\)—photocell with frosted glass, with the frosting toward the illuminator.

cell. These devices greatly reduce sensitivity and have not taken root in practical photometry.

For photocells with a compensating filter, the dependence of the current on the angle of incidence of light is determined to a considerable extent by the spectral composition of the incident radiation. The influence of the spectral composition of the radiation is explained by the fact that, with oblique incidence of light, the path of the ray through the light filter increases, which is equivalent to an increa-

with an increase in the optical density of the latter. The greater the density of the filter for some wavelength of the radiation, the more noticeable is the decrease in spectral transmittance for this wavelength, caused by oblique incidence of light. Hence it follows that the deviation from the \(\cos \alpha\) curve for a photoelement with a compensating filter will be greater for the blue and red regions of the spectrum, where the optical density of the compensating filter is greater, and smaller for the green part of the spectrum.

The experiment completely confirms this assumption. From Fig. 15b it is seen that for a photoelement with a compensating filter

Fig. 15b. Dependence of the photoelement current on the angle of incidence of light for radiations of different spectral composition: \(I\)—\(\cos \alpha\); \(II\)—photoelement with a compensating filter under illumination with green light; \(III\)—photoelement with a compensating filter for radiation with \(T_{\mathrm{цв}} = 2850^\circ\ \mathrm{K}\); \(IV\)—photoelement with a compensating filter under illumination with red light; \(V\)—photoelement with a compensating filter under illumination with blue light.

the deviation from \(\cos \alpha\) for \(\alpha = 60^\circ\) is minimal for green light and amounts to \(20\%\); for white light the deviation is \(30\%\), for red light it increases to \(45\%\), and for blue light this deviation increases to \(50\%\).

For measurements of illuminance under field conditions, an instrument of the following design has recently been proposed\({}^{29}\): six spectrally corrected photoelements, in order to obtain greater sensitivity, are connected electrically and placed in a common box, covered

plate of matte plexiglass. This device, called a “multi,” gives a practically sufficient approximation to the cosine.

For the purpose of approximately determining the error obtained when measuring scattered light with different photoelements, a calculation was made for three photoelements, the curves of the dependence of whose photocurrent on the angle of incidence are given in Figs. 15a and 15b. The deviations of the photocurrent of these receivers, in percent of the current given under the same conditions by a receiver obeying the cosine law, were determined for the case in which the photoelement is located at the center of a uniformly bright hemisphere. The results of the calculation are summarized in Table V.

Table V

The data of Table V may be used as corrections to the results of measurement by means of a photoelement of the illumination produced by scattered light.

11. CONNECTION CIRCUITS FOR SELENIUM PHOTOELEMENTS

As noted above, when the resistance of the external circuit of a selenium photoelement is decreased, the following occurs:

Fig. 16. Campbell–Fries circuit: $\Phi$ — photoelement, $G$ — galvanometer, $A$ — microammeter, $R$ — voltage divider, $B$ — battery.

1. In most cases the linearity of the “current—illumination” characteristic is improved, in any case if an accuracy higher than 2–3% is not required.

2. The change in spectral sensitivity as a function of the load current is reduced.

3. The accuracy of measurements when using a Talbot disk is increased.

4. The influence of temperature on sensitivity is reduced.

5. The stability of sensitivity is improved during long-term use of the photoelement.

All this indicates the desirability of using such measuring circuits as would reduce the load resistance of the photoelement to a minimum value.

In most such circuits, the voltage drop across the active load resistance is compensated by introducing into the external circuit

... of the photoelement by some additional e.m.f. These include the Campbell and Freeth circuit, shown in Fig. 16.

The circuit consists of a microammeter connected in series with the photoelement and a voltage divider, and a galvanometer \(G\), connected directly to the terminals of the photoelement.

With the input potential difference varied during operation until zero deflection of the galvanometer \(G\) is obtained. In this mode of operation the microammeter measures the short-circuit current of the photoelement. In order to simplify the described circuit, which requires two measuring instruments, Budd\(^{23}\) proposed an analogous circuit (Fig. 17) with one galvanometer.

Fig. 17. Budd circuit: \(\Phi\)—photoelement, \(G\)—galvanometer, \(R\)—voltage divider, \(M\)—sliding contact, \(K\)—key, \(B\)—battery.

Fig. 18. Frommer circuit: \(\Phi\)—photoelement, \(G\)—galvanometer, \(A\)—milliammeter, \(W\)—rheostat, \(R, r\)—fixed resistances, \(B\)—battery.

By moving the slider \(M\), one obtains zero potential difference at the terminals of the photoelement. This position can easily be found, since closing the key \(K\) then causes no change in the current through the galvanometer.

The linearity of the “photocurrent—illumination” characteristic and the stability of the photocurrent when this circuit is used prove to be quite satisfactory.

Frommer proposed yet another circuit (Fig. 18).

The resistance \(R\) is taken approximately 1000 times larger than \(r\). By varying the rheostat \(W\), the current in the auxiliary circuit is adjusted so as to obtain zero deflection of the galvanometer \(G\). In this case \(iR=(I-i)r\), whence

\[ i=I\frac{r}{R+r}, \]

or, since \(R \gg r\),

\[ i=I\frac{r}{R}. \]

Fig. 19. \(\Phi\)—photocell, \(G\)—galvanometer.

With a ratio \(\frac{R}{r}\) reaching a thousand, an instrument of low sensitivity may be used to measure the current. At the All-Union Electrotechnical Institute, in laboratory practice, when a large reserve of sensitivity is available, the circuit shown in Fig. 19 is used successfully. The greater the illumination determined by the photocell, the smaller the resistance \(r\) that may be used. The resistance \(R\) serves for smooth variation of the current in the galvanometer circuit \(G\), and also so that the small resistance \(r\) does not damp the galvanometer.

12. CONCLUSION

The foregoing may give rise to the opinion that selenium photocells are of little suitability for more or less precise measurements. This opinion can hardly be regarded as correct, since it is known\(^{34}\) that with the aid of selenium photocells one can obtain an accuracy of photometric measurements up to \(\pm 0.1\%\). However, in order to obtain high accuracies it is necessary to observe a number of conditions and precautions, which are determined both by the content of the given photometric problem and by the individual properties of the photocell used for its solution. The authors hope that the present article will facilitate the identification of those features of a photocell which may reduce the accuracy of measurement, and will help in particular cases to choose the most suitable conditions for carrying out various photometric measurements.

CITED LITERATURE

1. S. Yu. Lukyanov, Photocells, Acad. Sci. Press, Moscow–Leningrad (1948); P. Gerlich, Photocells, Gostekhizdat, Moscow–Leningrad (1948).
2. J. Preston, JIEE 79, 424 (1936).
3. L. Bergmann, R. Pelz, Zeits. techn. Phys. 18, 177 (1937); L. Bergmann, Phys. Zeits. 35, 450 (1934).

1. G. Barnard, Proc. Phys. Soc. 51, 222 (1939).
2. B. Lange, Zeits. Instrumentenkunde 53, 344 (1933).
3. R. Houstoun, A. Howatson, Phyl. Mag. 36, 279 (1945).
4. R. Sewig, ATM 67, 62 (1937).
5. DIN No. 5038.
6. O. Knoll, Licht 5, 16 (1935).
7. J. Atkinson, N. Campbell, E. Palmer and G. Windr, Proc. Phys. Soc. 50, 934 (1938).
8. E. Putseiko, ZhTF 8, 645 (1938).
9. G. Barnard, Proc. Phys. Soc. 47, 477 (1935).
10. E. Putseiko, ZhTF 11, 485 (1941).
11. M. Fogle, Trans. Ill. Eng. Soc. 31, 773 (1936).
12. A. Dresler, Licht 3, 41 (1933).
13. J. Rieck, Zeits. techn. Phys. 21, 184 (1940).
14. M. Kapnik, S. Yurov, ZhTF 18, 573 (1948).
15. H. Krefft, Licht 2, 203 (1932).
16. L. Barbrow, Journ. Res. NBS 25, 703 (1940).
17. E. Elvegard, S. Lindroth, E. Larsson, JOSA 28, 33 (1938).
18. C. Freivert, ZhTF 6, 1855 (1936).
19. H. Poole, W. Atkins, Phil. Trans. Roy. Soc. A235, 1 (1935).
20. H. Hamaker, W. Beezhold, Physica 1, 119 (1933).
21. P. Görlich, Zeits. techn. Phys. 14, 144 (1933); W. Leo, C. Müller, Phys. Zeits. 36, 113 (1935); B. Schönwald, Ann. d. Phys. 15, 395 (1932).
22. P. Gleason, Phys. Rev. 45, 745 (1934).
23. E. Putseiko, ZhTF 7, 10 (1937).
24. L. Bergmann, Phys. Zeits. 33, 17 (1932).
25. R. Sewig, W. Vailliant, Licht 4, 57 (1934).
26. C. Dows, Ill. Eng. 37, 103 (1942).
27. Bloch, L. u. L. 20, 179 (1931).
28. K. Larché, Licht 12, 90 (1942).
29. N. Campbell, M. Freeth, Journ. Sc. Instr. 11, 125 (1934).
30. L. Wood, Rev. Sc. Instr. 7, 157 (1934).
31. H. König, Rev. Sc. Instr. HPA, 8, 211, 505 (1935).