rus
Here we consider peculiarities and aspects of negative luminescence, radiant refrigeration and apparent temperature, response time, quantum yield and efficiency terms with respect to their use in the papers that refer to mid-IR (λ=3–7 μm ) optoelectronics particularly to LEDs and photodiodes.
Теги: mid-ir leds and photodiodes negative luminescence quantum efficiency quantum yield radiant refrigeration response time ик-свето- и фотодиоды квантовая эффективность квантовый выход и быстродействие отрицательная люминесценция радиационное охлаждение
I
n the scientific literature we can meet publications the authors of which argue that, under certain assumptions, the quantum efficiency of medium wave infrared (IR) photodiodes (PD) is greater than one. Of course, it goes beyond the accepted physical concepts and requires explanations that usually accompany such statements [1,2]. The value of coefficient of efficiency of the LEDs exceeding one also looks unusual [see, for example, 3]. But the result being very strange at first glance is becoming clear after a careful reading of the literature on the infrared LEDs. The authors of some works, considering the effects of radiative cooling (thermoelectric excitation) and photoelectric phenomena in semiconductor p-n structures use the terminology adopted in the infrared optoelectronics too broadly, or not quite correctly. What is the matter?
Negative luminescence
and radiative cooling
The term "negative luminescence" (NL) has long ago ceased to amaze physicists. It is firmly adopted in the vocabulary of the researchers involved in optoelectronic devices for the wavelength range of 3–7 microns, and the Encyclopedia Britannica. But sometimes "negative luminescence" is replaced with the term "radiative cooling" [6]; the degree of cooling (∆Т) often has a dimension of temperature [K]. The methods for determining such radiative cooling, followed by the emergence of a negative flow of photons are explaned in detail in the respective surveys [7]. Due to the terminological "democracy" the unsophisticated reader can be confused, because the "radiative cooling" can be interpreted as two different physical phenomena. Firstly, the actual negative luminescence occurring during the reverse bias of p-n junction and representing the prevalence of absorption over radiation in the spectral range close to the edge of own absorption of the active area [6–8]. Secondly, the effect of "heat pump" or thermoelectric excitation during the bias of the p-n junction in forward direction [3–6]. In the second case, due to the interaction phonon and photon fields, a photon, leaving a semiconductor, receives the additional energy. In our opinion, in order to avoid future misunderstandings, the term "radiative cooling" should be attributed to the effects that arise only in the case of the forward bias of the p-n junction.
The quantum efficiency of photoconversion.
In contrast to the radiative cooling, the cases mentioned above and related to the photodiodes can be analyzed without the involvement of knowledge of mathematical physics, they can be considered within the ordinary concepts of geometrical optics and the law of Bouguer-Lambert. Indeed, for the proper determination of the effectiveness of the photodiode work, the key parameter is the exact value of the number of quanta (photons) entering the electrically active area of the photodetector. That is why the measurements typically use the model of ablosute black body (ABB) at a predetermined (fixed) temperature (T); at the same time, the radiation density is determined by the "Nobel" Planck’s formula and, of course, by the distance between the ABB and the PD. However, despite the known number of photons emitted by the ABB in the selected spectral range, it is not possible in some cases to specify the exact number of photons entering the active zone of the PD. In other words, the number of photodiodes has such a structure that it is impossible to determine the area of radiation collection, or it is possible only to approximately estimate it. Moreover, we can almost always confidently say that the area of the radiation collection in the PD not containing the external radiation hubs is always greater than the area of the p-n junction (i. e., the area of the electrically active parts) [1, 2, 9–11]. This statement is due to the fact that part of the radiation that has entered into the PD chip, can be redirected to the p-n junction by the multiple reflection from the faces, curved surfaces or other components of the heterostructure, for example, from the ohmic contacts. This last remark is especially important for the PD with a flip-chip structure with a wide ohmic contact [1, 2, 11].
As an example, we can point to the dependence of current sensitivity at a wavelength of 4.2 microns SI having the dimension [A/W] from the p-n junction area and mesa depth square ratio indicated in [11]. It shows that with increasing height of the mesa sloped walls (i. e., its depth) the radiation collecting area is increased and thus the PD photocurrent is also increased. Using the mesa sloped walls to increase the radiation collecting area, i. e., establishment of internal radiation gubds is widely known and has long been used in semiconductor optoelectronics, for example, when creating a large area diode panels with the least possible value of the reverse current. Such panels are used for the protection of photodetector arrays from the external thermal noise [7, 8]. While making such panels the scientists generally use a set of identical PDs activated in the reverse direction, i. e. emitting negative luminescence.
On the other hand, for some PD designs it is possible to rather accurately specify an upper limit of the number of quanta entering the p-n junction area. One of these designs is the "immersion" PD, i. e. the PD conjugated with the lens using immersion, more often, by an optical adhesive [12,13] (Figure 1). The immersion layer ("adhesive") is not shown in the diagram, in this case it means that the immersion is ensured by absence of the air gap between the lens and the PD surface. It is understood that the distance between the lens surface and PD is much smaller than the radiation wavelength. This conjugation requires the high-precision equipment in order to achieve "optical" quality of the surfaces of FD and lenses. When using the optical adhesive the quality requirements of the conjugated surfaces in the immersion PD, of course, are less strict.
For the PD with an optical immersion the optical area is substantially increased, while the number of photons entering the p-n junction area can not exceed the number of photons entering the lens. This amount can be rather accurately estimated for a lens with a known diameter. Therefore, the parameters for the immersion PDs can serve as some "standard" to characterize the properties of the p-n junction and the active area in the original PD chip[2]. Indeed, as it is shown by measurements, the quantum efficiency is reduced in the case of attachment of the immersion lens to the "bare" PD chip (Figure 2) [13]. In practice, it means a reduction in the current sensitivity of the PD. Among the possible reasons for reducing this important parameter we can indicate the optical losses caused both by parasitic absorption and incomplete radiation focusing on the active PD area. However, despite the decline in the current photosensitivity, for the customer the use of the immersion PD is almost always preferred and useful, because it is possible to achieve a substantial increase in the signal/noise ratio in optical measurements of weak signals. For the most common option – the detectability D* – such an increase depending on the vision angle is from n 2 to n 4, where n – the index of refraction of the less dense optical material used in the PD design [12]. For medium wave PD made of heterostructures on the basis of indium arsenide (n≈3,4) with the diameter of active area of ~ 280 microns, bonded to a silicon lens (n = 3,5) using the chalcogenide glass (n = 2,4) the increase in D*λ in average is about 10 for the lens with a working diameter of 3.3 mm (see. Figure 2б).
Increasing the photon collection area in the PD with a thin annular contact on the front side of the narrow-band part of the structure can be achieved also by the beams reflected from the back side of the base coat GaSb or InAs. The authors of [9, 10] have found that the semiconductor receiver with the advanced back surface of the base coat, along with the extended band of the spectral sensitivity has also increased efficiency in the medium IR region of the spectrum (1–5) microns due to the additional absorption in the active region of the heterostructure of the photons repeatedly re-reflected from the curved surfaces of depressions in the semiconductor base coat (Figure 3). In fact, while creading a relief to the original streams A and D, absorbed in the PD surface layer with a flat back surface of the base coat, are combined with the flows generated by the flows B and C and their subsequent "echoes". The increase in photosensitivity in the wavelength region of 1–2.5 micrometers (GaSb base coat) and 1.5–4.mu.m (InAs base coat), according to the above works, is small (~ 30%). Therefore it is clear that the proposed approach is capable in the future to provide a significant increase in sensitivity during the use of other, more transparent base coats, for example, GaAs for the wavelength range λ = 1–2,5 m, and GaSb for λ = 2–4 microns.
In some PD types the current sensitivity (quantum efficiency) can be increased by increasing the coefficient of reflection from the back surface of the structure. Such a device may include the FDs having a flip chip structure and a reflective contacts such as the anode (R = 0,6 [11]). In such diodes the part of photons not absorbed during the first pass can be redirected towards the active region as a result of reflection from the contact that will provide the increase in the photocurrent. Naturally, such a mechanism of increasing the quantum efficiency is relevant for radiation with a low absorption coefficient, i. e., close to the long-wavelength absorption of the active region. In the short-wave part of the spectral curve it is not necessary to expect the increase in the quantum efficiency due to the complete absorption of radiation by the active area – the effects due to reflection from the contact for this spectrum area can be ignored. It is also clear that while shading all PD parts except for the active region the quantum efficiency can not exceed one.
Operation speed of the PD
The term "fast PD" also requires clarification, because different authors put different meaning in this concept. As an example, we will consider the works in which the term "high-speed" or "ultra high-speed" PD is included in the name of the article [14–16]. In the work [15] the PD operation speed from InAs means the time constant of the RC-chain calculated for R = 50 Ohm and is 30–80 ns for the PD with the diameter D = 0,2–0,3 mm. Sometimes we are limited to the usage of the FD capacitance values, for example, C = 2–5 pF in the case of the reverse bias (U = – (0,2–0,4 V)) for the PD with the diameter Ø = 50 mm [16]. The work [17] provides the time constant that is less than half of the value indicated in [15], (τ = 15 ns) for the PD on the basis of p-InAsSbP/n-InAs with the comparable dimensions, obtained by the measurements using semiconductor laser based on GaAs. Even lower value for the time constant can be found in [18] (the calculated value of τ = 1–6 ns for diode with an area of 1 mm 2), as well as the specifications of the commercially available photodiode J12-LD2-R250U made of indium arsenide with the diameter Ø = 250 mm, for which the term "high-speed" (τ <3 ns) is applied only to the description of its connection layout, rather than to the device as a whole [19]. In this connection it seems reasonable to use in the title of the article the value relating to the operation speed, for example, values of the product of the multiplication factor and the bandpass of the avalanche photodiode (see, for example, [14]) or the bandpass (see, for example, [16], but without broad interpretation and dissemination of the same data on all types of the studied PDs.
Radiation power
The very common adjective used to describe the medium wave LEDs is the adjective "powerful" [20–22] or "high powerful" [23]. It is understood that the concept of "powerful" is meant by the authors as the "optically powerful" LED. It often turns out that the "high powerful" (in the original, the "high power", 5 mW, I = 1,4 A [23]) is second in the optical power to the simply "powerful" (5.5 mW, I = 9 A [20]) at comparable pump currents. A simple calculation also shows that the conversion efficiency defined, for example, for the LEDs of nominally identical structures with the active region, ade of indium arsenide [20–23], and having the dimension of mW/A gradually decreases with increasing volume serial number of the file. It should be also noted that the reduction in the conversion factor in the later works of these authors [see, for example, 20] is received, despite the substantial improvements in the LED chip design, i. e. druing the transition from design with the point contact to the p-region [21] and the remoteness of the p-n junction from the heat sink [21, 22] to the chip designwith a point contact to n-region and the closest approximation of the p-n junction to the heat sink, i. e. while mounting the epitaxial layer down on the body [20]. According to all the prerequisites and reasons given in [24, 28] the LEDs [20] would have to be more efficient and powerful than the LEDs in [21, 22]. The issue of non-compliance of expectations with the actual data has already been considered in [24] with respect to the measurements using the infrared microscopes.
The internal quantum efficiency
The internal quantum efficiency wuite often provided for by the authors and obtained from the experimental values of the optical power has a value that does not match the data aggregate relating to the object of research, especially the temperature dependence of its optical power. We have already drawn attention to such a discrepancy earlier in the review [24] and noted that the power increase by 1–2 orders standard for the LEDs of indium arsenide during cooling from 300 to 77 K means that at the room temperature, the internal quantum efficiency can not exceed 1–10%. Unfortunately, some authors [for example, 20] neglect the simple estimates of the quantum efficiency, based on the assumption of 100% quantum efficiency in InAs at the temperatures below 77 K, and the transmittion immutability of the layers in the heterostructures with changes in temperature. In our opinion, the overestimated value of power and, consequently, the internal quantum efficiency, is due most likely to the methodological errors in the calculation of power. This assumption is appropriate, given the large difference in the values of power in the pulsed and continuous modes [20–23]. This difference, of course, should be available, but only at the high pump currents when the LED chip is heated by Joule heat (see, for example, data in [24, 25]).
Radiation temperature
For most applications of the IR LEDs the paramount importance is provided by their luminosity, or brightness, but not by the integral radiation power. This is due to the fact that for efficient use of radiation and the electric power supplied to the emitter, it is important that the size of the LED active area is significantly smaller than the size of the focusing optical elements, such as immersion lenses or mirrors used in the measuring device. For example, when using the LEDs with a lateral dimension of the active region of 430 microns, microimmersion lens with the size of ~ 1 mm and spherical mirrors with the diameter of 68 mm it was able to create a measurement channel for the wavelength of 3.3 microns with the optical length of up to 100 m [26]. At the same time it is becoming clear that the key parameter in the LED is radiated power per unit of its active area (mW/cm 2). The characteristics of such a "specific" power very often uses the term "radiation temperature" or "apparent temperature" (Ta) [24, 27]. The analytical expressions for calculating the radiation temperature can be found, for example, in [27]; Fig. 4 shows the data for the brightest modern medium wave IR LEDs for which the effective radiation temperature at a wavelength of 3.3 microns is 1250 K [24, 28]. The effective radiation temperature (Ta) is indicated for the maximum pulse current value (4.5 A) for the LEDs with the deep etching mesa and textured surface, obtained by chemical etching of the light deferential surface of the n + -InAs. In addition to the high brightness the LEDs proposed in [28] are suitable for the optical fiber connections that creates prerequisites for their use in the fiber-optic sensors, for example, the sensors of chemical composition of the fluid working according the disappearing wave method.
Instead of conclusion.
We believe that understanding the above mentioned nuances of using the terminology of infrared optoelectronics is important for proper assessment of the applicability of a particular PH or LED design to solution of a particular technical problem. Such problems have recently become more and more, the balance between high speed, sensitivity, power, brightness and the possibility of PH/LED using in the fiber-optical systems, for example, to measure the objects heated to low temperatures [29], have greater importance.
The author expresses his gratitude to the personnel of the MIRDOG diode optocouplers group of the infrared optoelectronics laboratory, Ioffe Physical and Technical Institute of the Russian Academy of Sciences for assistance in work.
The appearance of this article is due to the implementation of the project "Development of technology for semiconductor photosensitive materials for matrix infrared photodetectors and thermal imagers". The contract code is 14.576.21.0057.
--------------------------------------------------------------------------------
[1] В английской литературе исходный чип часто называют "голым" чипом (bare chip).
[2] In the English literature the original chip is often called "bare chip".
n the scientific literature we can meet publications the authors of which argue that, under certain assumptions, the quantum efficiency of medium wave infrared (IR) photodiodes (PD) is greater than one. Of course, it goes beyond the accepted physical concepts and requires explanations that usually accompany such statements [1,2]. The value of coefficient of efficiency of the LEDs exceeding one also looks unusual [see, for example, 3]. But the result being very strange at first glance is becoming clear after a careful reading of the literature on the infrared LEDs. The authors of some works, considering the effects of radiative cooling (thermoelectric excitation) and photoelectric phenomena in semiconductor p-n structures use the terminology adopted in the infrared optoelectronics too broadly, or not quite correctly. What is the matter?
Negative luminescence
and radiative cooling
The term "negative luminescence" (NL) has long ago ceased to amaze physicists. It is firmly adopted in the vocabulary of the researchers involved in optoelectronic devices for the wavelength range of 3–7 microns, and the Encyclopedia Britannica. But sometimes "negative luminescence" is replaced with the term "radiative cooling" [6]; the degree of cooling (∆Т) often has a dimension of temperature [K]. The methods for determining such radiative cooling, followed by the emergence of a negative flow of photons are explaned in detail in the respective surveys [7]. Due to the terminological "democracy" the unsophisticated reader can be confused, because the "radiative cooling" can be interpreted as two different physical phenomena. Firstly, the actual negative luminescence occurring during the reverse bias of p-n junction and representing the prevalence of absorption over radiation in the spectral range close to the edge of own absorption of the active area [6–8]. Secondly, the effect of "heat pump" or thermoelectric excitation during the bias of the p-n junction in forward direction [3–6]. In the second case, due to the interaction phonon and photon fields, a photon, leaving a semiconductor, receives the additional energy. In our opinion, in order to avoid future misunderstandings, the term "radiative cooling" should be attributed to the effects that arise only in the case of the forward bias of the p-n junction.
The quantum efficiency of photoconversion.
In contrast to the radiative cooling, the cases mentioned above and related to the photodiodes can be analyzed without the involvement of knowledge of mathematical physics, they can be considered within the ordinary concepts of geometrical optics and the law of Bouguer-Lambert. Indeed, for the proper determination of the effectiveness of the photodiode work, the key parameter is the exact value of the number of quanta (photons) entering the electrically active area of the photodetector. That is why the measurements typically use the model of ablosute black body (ABB) at a predetermined (fixed) temperature (T); at the same time, the radiation density is determined by the "Nobel" Planck’s formula and, of course, by the distance between the ABB and the PD. However, despite the known number of photons emitted by the ABB in the selected spectral range, it is not possible in some cases to specify the exact number of photons entering the active zone of the PD. In other words, the number of photodiodes has such a structure that it is impossible to determine the area of radiation collection, or it is possible only to approximately estimate it. Moreover, we can almost always confidently say that the area of the radiation collection in the PD not containing the external radiation hubs is always greater than the area of the p-n junction (i. e., the area of the electrically active parts) [1, 2, 9–11]. This statement is due to the fact that part of the radiation that has entered into the PD chip, can be redirected to the p-n junction by the multiple reflection from the faces, curved surfaces or other components of the heterostructure, for example, from the ohmic contacts. This last remark is especially important for the PD with a flip-chip structure with a wide ohmic contact [1, 2, 11].
As an example, we can point to the dependence of current sensitivity at a wavelength of 4.2 microns SI having the dimension [A/W] from the p-n junction area and mesa depth square ratio indicated in [11]. It shows that with increasing height of the mesa sloped walls (i. e., its depth) the radiation collecting area is increased and thus the PD photocurrent is also increased. Using the mesa sloped walls to increase the radiation collecting area, i. e., establishment of internal radiation gubds is widely known and has long been used in semiconductor optoelectronics, for example, when creating a large area diode panels with the least possible value of the reverse current. Such panels are used for the protection of photodetector arrays from the external thermal noise [7, 8]. While making such panels the scientists generally use a set of identical PDs activated in the reverse direction, i. e. emitting negative luminescence.
On the other hand, for some PD designs it is possible to rather accurately specify an upper limit of the number of quanta entering the p-n junction area. One of these designs is the "immersion" PD, i. e. the PD conjugated with the lens using immersion, more often, by an optical adhesive [12,13] (Figure 1). The immersion layer ("adhesive") is not shown in the diagram, in this case it means that the immersion is ensured by absence of the air gap between the lens and the PD surface. It is understood that the distance between the lens surface and PD is much smaller than the radiation wavelength. This conjugation requires the high-precision equipment in order to achieve "optical" quality of the surfaces of FD and lenses. When using the optical adhesive the quality requirements of the conjugated surfaces in the immersion PD, of course, are less strict.
For the PD with an optical immersion the optical area is substantially increased, while the number of photons entering the p-n junction area can not exceed the number of photons entering the lens. This amount can be rather accurately estimated for a lens with a known diameter. Therefore, the parameters for the immersion PDs can serve as some "standard" to characterize the properties of the p-n junction and the active area in the original PD chip[2]. Indeed, as it is shown by measurements, the quantum efficiency is reduced in the case of attachment of the immersion lens to the "bare" PD chip (Figure 2) [13]. In practice, it means a reduction in the current sensitivity of the PD. Among the possible reasons for reducing this important parameter we can indicate the optical losses caused both by parasitic absorption and incomplete radiation focusing on the active PD area. However, despite the decline in the current photosensitivity, for the customer the use of the immersion PD is almost always preferred and useful, because it is possible to achieve a substantial increase in the signal/noise ratio in optical measurements of weak signals. For the most common option – the detectability D* – such an increase depending on the vision angle is from n 2 to n 4, where n – the index of refraction of the less dense optical material used in the PD design [12]. For medium wave PD made of heterostructures on the basis of indium arsenide (n≈3,4) with the diameter of active area of ~ 280 microns, bonded to a silicon lens (n = 3,5) using the chalcogenide glass (n = 2,4) the increase in D*λ in average is about 10 for the lens with a working diameter of 3.3 mm (see. Figure 2б).
Increasing the photon collection area in the PD with a thin annular contact on the front side of the narrow-band part of the structure can be achieved also by the beams reflected from the back side of the base coat GaSb or InAs. The authors of [9, 10] have found that the semiconductor receiver with the advanced back surface of the base coat, along with the extended band of the spectral sensitivity has also increased efficiency in the medium IR region of the spectrum (1–5) microns due to the additional absorption in the active region of the heterostructure of the photons repeatedly re-reflected from the curved surfaces of depressions in the semiconductor base coat (Figure 3). In fact, while creading a relief to the original streams A and D, absorbed in the PD surface layer with a flat back surface of the base coat, are combined with the flows generated by the flows B and C and their subsequent "echoes". The increase in photosensitivity in the wavelength region of 1–2.5 micrometers (GaSb base coat) and 1.5–4.mu.m (InAs base coat), according to the above works, is small (~ 30%). Therefore it is clear that the proposed approach is capable in the future to provide a significant increase in sensitivity during the use of other, more transparent base coats, for example, GaAs for the wavelength range λ = 1–2,5 m, and GaSb for λ = 2–4 microns.
In some PD types the current sensitivity (quantum efficiency) can be increased by increasing the coefficient of reflection from the back surface of the structure. Such a device may include the FDs having a flip chip structure and a reflective contacts such as the anode (R = 0,6 [11]). In such diodes the part of photons not absorbed during the first pass can be redirected towards the active region as a result of reflection from the contact that will provide the increase in the photocurrent. Naturally, such a mechanism of increasing the quantum efficiency is relevant for radiation with a low absorption coefficient, i. e., close to the long-wavelength absorption of the active region. In the short-wave part of the spectral curve it is not necessary to expect the increase in the quantum efficiency due to the complete absorption of radiation by the active area – the effects due to reflection from the contact for this spectrum area can be ignored. It is also clear that while shading all PD parts except for the active region the quantum efficiency can not exceed one.
Operation speed of the PD
The term "fast PD" also requires clarification, because different authors put different meaning in this concept. As an example, we will consider the works in which the term "high-speed" or "ultra high-speed" PD is included in the name of the article [14–16]. In the work [15] the PD operation speed from InAs means the time constant of the RC-chain calculated for R = 50 Ohm and is 30–80 ns for the PD with the diameter D = 0,2–0,3 mm. Sometimes we are limited to the usage of the FD capacitance values, for example, C = 2–5 pF in the case of the reverse bias (U = – (0,2–0,4 V)) for the PD with the diameter Ø = 50 mm [16]. The work [17] provides the time constant that is less than half of the value indicated in [15], (τ = 15 ns) for the PD on the basis of p-InAsSbP/n-InAs with the comparable dimensions, obtained by the measurements using semiconductor laser based on GaAs. Even lower value for the time constant can be found in [18] (the calculated value of τ = 1–6 ns for diode with an area of 1 mm 2), as well as the specifications of the commercially available photodiode J12-LD2-R250U made of indium arsenide with the diameter Ø = 250 mm, for which the term "high-speed" (τ <3 ns) is applied only to the description of its connection layout, rather than to the device as a whole [19]. In this connection it seems reasonable to use in the title of the article the value relating to the operation speed, for example, values of the product of the multiplication factor and the bandpass of the avalanche photodiode (see, for example, [14]) or the bandpass (see, for example, [16], but without broad interpretation and dissemination of the same data on all types of the studied PDs.
Radiation power
The very common adjective used to describe the medium wave LEDs is the adjective "powerful" [20–22] or "high powerful" [23]. It is understood that the concept of "powerful" is meant by the authors as the "optically powerful" LED. It often turns out that the "high powerful" (in the original, the "high power", 5 mW, I = 1,4 A [23]) is second in the optical power to the simply "powerful" (5.5 mW, I = 9 A [20]) at comparable pump currents. A simple calculation also shows that the conversion efficiency defined, for example, for the LEDs of nominally identical structures with the active region, ade of indium arsenide [20–23], and having the dimension of mW/A gradually decreases with increasing volume serial number of the file. It should be also noted that the reduction in the conversion factor in the later works of these authors [see, for example, 20] is received, despite the substantial improvements in the LED chip design, i. e. druing the transition from design with the point contact to the p-region [21] and the remoteness of the p-n junction from the heat sink [21, 22] to the chip designwith a point contact to n-region and the closest approximation of the p-n junction to the heat sink, i. e. while mounting the epitaxial layer down on the body [20]. According to all the prerequisites and reasons given in [24, 28] the LEDs [20] would have to be more efficient and powerful than the LEDs in [21, 22]. The issue of non-compliance of expectations with the actual data has already been considered in [24] with respect to the measurements using the infrared microscopes.
The internal quantum efficiency
The internal quantum efficiency wuite often provided for by the authors and obtained from the experimental values of the optical power has a value that does not match the data aggregate relating to the object of research, especially the temperature dependence of its optical power. We have already drawn attention to such a discrepancy earlier in the review [24] and noted that the power increase by 1–2 orders standard for the LEDs of indium arsenide during cooling from 300 to 77 K means that at the room temperature, the internal quantum efficiency can not exceed 1–10%. Unfortunately, some authors [for example, 20] neglect the simple estimates of the quantum efficiency, based on the assumption of 100% quantum efficiency in InAs at the temperatures below 77 K, and the transmittion immutability of the layers in the heterostructures with changes in temperature. In our opinion, the overestimated value of power and, consequently, the internal quantum efficiency, is due most likely to the methodological errors in the calculation of power. This assumption is appropriate, given the large difference in the values of power in the pulsed and continuous modes [20–23]. This difference, of course, should be available, but only at the high pump currents when the LED chip is heated by Joule heat (see, for example, data in [24, 25]).
Radiation temperature
For most applications of the IR LEDs the paramount importance is provided by their luminosity, or brightness, but not by the integral radiation power. This is due to the fact that for efficient use of radiation and the electric power supplied to the emitter, it is important that the size of the LED active area is significantly smaller than the size of the focusing optical elements, such as immersion lenses or mirrors used in the measuring device. For example, when using the LEDs with a lateral dimension of the active region of 430 microns, microimmersion lens with the size of ~ 1 mm and spherical mirrors with the diameter of 68 mm it was able to create a measurement channel for the wavelength of 3.3 microns with the optical length of up to 100 m [26]. At the same time it is becoming clear that the key parameter in the LED is radiated power per unit of its active area (mW/cm 2). The characteristics of such a "specific" power very often uses the term "radiation temperature" or "apparent temperature" (Ta) [24, 27]. The analytical expressions for calculating the radiation temperature can be found, for example, in [27]; Fig. 4 shows the data for the brightest modern medium wave IR LEDs for which the effective radiation temperature at a wavelength of 3.3 microns is 1250 K [24, 28]. The effective radiation temperature (Ta) is indicated for the maximum pulse current value (4.5 A) for the LEDs with the deep etching mesa and textured surface, obtained by chemical etching of the light deferential surface of the n + -InAs. In addition to the high brightness the LEDs proposed in [28] are suitable for the optical fiber connections that creates prerequisites for their use in the fiber-optic sensors, for example, the sensors of chemical composition of the fluid working according the disappearing wave method.
Instead of conclusion.
We believe that understanding the above mentioned nuances of using the terminology of infrared optoelectronics is important for proper assessment of the applicability of a particular PH or LED design to solution of a particular technical problem. Such problems have recently become more and more, the balance between high speed, sensitivity, power, brightness and the possibility of PH/LED using in the fiber-optical systems, for example, to measure the objects heated to low temperatures [29], have greater importance.
The author expresses his gratitude to the personnel of the MIRDOG diode optocouplers group of the infrared optoelectronics laboratory, Ioffe Physical and Technical Institute of the Russian Academy of Sciences for assistance in work.
The appearance of this article is due to the implementation of the project "Development of technology for semiconductor photosensitive materials for matrix infrared photodetectors and thermal imagers". The contract code is 14.576.21.0057.
--------------------------------------------------------------------------------
[1] В английской литературе исходный чип часто называют "голым" чипом (bare chip).
[2] In the English literature the original chip is often called "bare chip".
Readers feedback
