DE102021208363B4 - Method and device for determining a parameter for electronic components - Google Patents

Abstract

The invention relates to a method and a device for determining a parameter for electronic components, namely for determining the influence of radiation on the electronic components or for determining a symmetry factor (z) of CMOS inverters (16), using a ring oscillator (2). which has a number of timers (4) and each timer (4) is constructed from two serially arranged CMOS units (6, 8, 10) and the CMOS units (6, 8, 10) of a respective timer (4) have different switching delays. The parameter is determined using a measured variable of the ring oscillator (2), specifically a parameter index (AI) is determined from the ratio between the duration of the high level (H period) and the duration of the low level (L period). and this index (AI) is used to determine, for example, an asymmetry factor (y) and/or a degeneration factor (DEG) for the PMOS and/or NMOS switches used due to the ionizing radiation (TID) as a parameter.

Description

The invention relates to a method and a device for determining a parameter for electronic components, in particular semiconductor-based components, namely for determining the influence of ionizing radiation on the electronic components or for determining a symmetry factor of CMOS switches.

In the case of electronic components, in particular semiconductor components, it is generally known that ionizing radiation, for example cosmic radiation, X-rays, gamma radiation, etc., have an influence on the properties, especially switching properties, of these semiconductor components and thus lead to (premature) aging of the semiconductor components . The reason for this can be seen in the fact that the ionizing radiation has an influence on the charge carriers in the semiconductor component. In particular, the charge carrier zone between source and drain is influenced, which leads to a change in the switching behavior. In particular, the switching time is shifted, at which there is a switch between two switching levels, namely the high level (H level) and the low level (L level), and the duration of the high period (H) and the low Period (L) is measured.

In the case of CMOS components, this leads to an asymmetric aging effect of these two switches of the CMOS component due to their structure consisting of a PMOS switch (field effect transistor with p-type channel) and an NMOS switch (field effect transistor with n-type channel). In particular, investigations have shown that the PMOS switch is significantly more affected by ionizing radiation than the NMOS switch.

The undesired influence of the ionizing radiation increases with increasing miniaturization and with the increasingly lower supply and control voltages for the electronic components and is therefore gaining in importance.

The so-called "total ionizing dose" (TID) is used as a parameter for the influence of ionizing radiation. The overall influence depends on this TID.

In particular, the switching time, also referred to below as the switching delay, of the respective switch is influenced by the ionizing radiation.

In many electronic circuits, the period of a clock signal is crucial, i.e. the duration between two H states, since the period specifies, for example, process and processing cycles within which certain tasks are processed, such as the transmission or processing of data packets . Typically, the rising edge of the H level, for example, is used as a trigger for the individual processing cycles.

In some applications, the so-called DDR (Double Data Rate) technology, for example, is used to speed up the processing, in which a processing cycle is typically carried out both on the rising edge and on the falling edge of the clock signal. In this case, in addition to the period, the duration of the H level (H period) is also particularly important.

A change, specifically a shortening of the duration of the H level or the L level as a result of ionizing radiation can, under certain circumstances, lead to the intended tasks not being processed or not being processed completely (duty cycle distortion, distortion of the duty cycle). This can lead to processing errors and, in complex systems, to errors in the system. This should be avoided, especially in safety-critical applications.

There is therefore an increasing need to be able to identify radiation-related influences and aging effects on electronic components, particularly in real time, for example as part of online monitoring.

As explained above, a CMOS switch, also referred to below as a CMOS inverter, consists of a PMOS switch and an NMOS switch. As a rule, the CMOS switch is constructed symmetrically so that the switching times of these two switches are matched to one another. In the case of an unbalanced CMOS switch, the PMOS switch and the NMOS switch have different switching times. These can also affect the process cycles of a circuit. Therefore, there is often an interest in knowing the relationship between the switching times of the PMOS switch to NMOS switch. This ratio is referred to here as the symmetry factor. The symmetry factor 1 then designates a symmetrical CMOS switch with the same switching times (switching delays) for the PMOS and NMOS switches. Furthermore, the deterioration or also the acceleration (eg in the case of NMOS switches in specific dose intervals) of the switching delay of the CMOS switches is of interest.

From the US 10 705 552 B1 a self-optimizing circuit can be found in which the negative effects of ionizing radiation, temperature drifts and aging effects are reduced.

The effect of ionizing radiation on CMOS devices, for example, is described in the scientific publication "Modeling lonizing Radiation Effects in Solid State Materials and CMOS Devices, Hugh J. Barnaby et al., IEEE Transactions on Circuits and Systems - I, Regular Papers, 2009- 08, Vol. 56 (8)”, pp. 1870-1883.

From the DE 10 2005 054 745 A1 a method for influencing the operating frequency of a quartz oscillator can be found in which the quartz oscillator is exposed to electromagnetic radiation.

the U.S. 2020/0 350 893 A1 describes an electronic circuit that is designed to compensate for voltage-temperature fluctuations in semiconductor components.

Proceeding from this, the invention is based on the object of specifying a method with the aid of which a parameter for electronic components, namely the influence of ionizing radiation on electronic components and/or the symmetry factor in CMOS switches, can be determined.

The object is achieved according to the invention by a method and a device for determining a parameter for electronic (semiconductor) components, namely for determining the influence of, in particular, ionizing radiation on the electronic components or for determining a symmetry factor of CMOS switches. A special ring oscillator is used for this purpose, which has a number of timing elements, with each timing element being constructed from two serially connected CMOS units. The individual CMOS units are connected in series and form a chain consisting of 2n such CMOS units or n timers. A respective CMOS unit forms an inverter in each case, so that the units can also be referred to as inverter units.

Furthermore, it is provided that the two CMOS units of a respective timing element have different switching delays. Finally, in the method, the desired parameter is determined using a measured variable of the ring oscillator.

The individual CMOS units of a timer are in particular arranged directly one after the other.

If reference is made here to a serial arrangement, this means that a respective output of a CMOS unit (drain connection) is connected to the gate connection of the subsequent CMOS unit. This means that the output level of a respective CMOS switching unit forms the input level for the gate connection of the subsequent CMOS unit.

Ring oscillators and their structure are known in principle. They regularly consist of a series of individual inverter units. Due to the ring-shaped, i.e. closed, wiring and arrangement, this leads to an oscillating circuit whose natural frequency and thus period duration is determined using a measuring device such as an oscilloscope or an internal measurement of the H and L period (edge detection, counter, system frequency). To measure the H and L period, a so-called PWM (pulse width modulation) analyzer is used, for example, which is arranged, for example, together with the CMOS units on a common chip. Based on this (total) period duration of the ring oscillator, if the number of installed inverter units is known, conclusions can then be drawn about the properties of the individual inverter units, such as their frequency and thus switching delay.

In the case of the ring oscillator, typically several 100 or even several 1000 such inverter units are arranged in series.

In the case of the ring oscillator, in the present case a chain of (logical) timers, with each timer consisting of (precisely) two CMOS units and thus two inverter units, the level at the input of the chain is the same as at the output of the chain after running through the total of n timers on. If the level at the input changes, the level at the output of the chain also changes after an overall switching delay for the entire chain.

In order to form the ring oscillator, there is also an inversion between the input and output of the chain, so that after running through the entire chain, an inverted level is present at the input of the chain. If the entire ring oscillator is considered, then after two runs through the entire chain, a total period results, which is made up of the time for a high level and a low level.

In the case of the special ring oscillator described here for determining the parameter, a special feature can be seen in the fact that the two CMOS units of a respective timing element have different switching delays. This is generally understood to mean that the switching time and thus the switching delay of the PMOS unit of the first CMOS unit (i.e. of the first type of CMOS units) differs from the switching delay of the PMOS unit of the second CMOS unit (i.e. of the second type of CMOS -units) differs by a factor x. The same applies to the NMOS units of both types, i.e. their switching delays also differ by a factor of x. In this case, the factor x is preferably identical for both the PMOS and the NMOS units.

This configuration is based on the consideration that in a conventional ring oscillator with normal, identically constructed CMOS units, the influence of the ionizing radiation and the symmetry factor cannot be detected or cannot be reliably detected due to the special design of the CMOS units and their inverting character. As will be explained in more detail below, only this special configuration with the CMOS units configured differently within a respective timing element allows a reliable determination of the influence of the ionizing radiation. In other words: The desired characteristic values can only be reliably determined by combining two types of CMOS units with different switching delays in pairs.

The influence of the ionizing radiation on the CMOS components used in the ring oscillator is determined by the special ring oscillator used—in the variant for determining the influence of ionizing radiation. In this way, a parameter of the influence of ionizing radiation on electronic components is generally determined, which are also exposed to the same ionizing radiation as the ring oscillator with its CMOS components. In this case, the ring oscillator can be arranged, for example, on a common chip with further semiconductor switches, specifically CMOS switches. Overall, these are circuit arrangements such as ICs, FPGA, ASICs, SOCs, etc. In a preferred embodiment, these are designed as a structural unit together with the ring oscillator described here.

With regard to the parameter of the symmetry factor, this is determined for the CMOS inverter units used in the ring oscillator. From this determined symmetry factor, conclusions can then be drawn, for example, about the symmetry factor of further CMOS inverters which, for example, belong to the same production batch.

In a preferred embodiment, a parameter index is determined, which is correlated to the ratio between an H level and an L level of the ring oscillator described here. Specifically, this characteristic index is defined by the ratio between the H level and the L level of the ring oscillator.

As previously explained, the ring oscillator forms an oscillating circuit with the individual timing elements, so that depending on the physical characteristics of the respective CMOS timing elements, a frequency and thus a period and also defined periods of time for the H level and for the L level are set . These variables, especially the duration of the H level and the L level, are measured using the aforementioned measuring device (oscilloscope) or an internal measurement using edge detection. In a preferred embodiment, the characteristic index is therefore determined by measuring the time of the H level and the L level.

In a preferred embodiment, the characteristic number for the influence of the ionizing radiation is determined from the characteristic index according to the following formula: $$\text{Al}=\left(1+\text{xyz}\right)/\left(\text{x}+\text{Y Z}\right)$$

Where Al is the parameter index, y is the parameter for the influence of ionizing radiation, z is the symmetry factor and x is a factor with which the switching delays of the CMOS units of the timers differ. The symmetry factor z is determined, for example, by a previous calibration measurement or is known per se. The factor x, which indicates the ratio of the switching delay, is determined by the physical-technical configuration of the individual CMOS units and is preferably known. The key figure y remains as the only variable. This is also referred to as a degradation index, since it characterizes the degradation / aging as a result of radiation.

In particular, it should also be pointed out that for very large x values, the limit of equation F1 approaches yz, ie. Limes AI(x->infinity) = yz.

A percentage degradation index y′ as a result of the ionizing radiation is preferably determined from the index Al of parameters by replacing the index y in the above formula as follows: $$\text{y}=\text{y}\text{'}+1\left[\%\right]$$

This results in the characteristic index as follows $$\text{Al}=\left(1+\text{xz}+\text{xy}\text{'}\text{e.g}\right)/\left(\text{x}+\text{e.g}+\text{y}\text{'}\text{e.g}\right)$$

As before, the values for x, z are known or determined, so that the percentage degradation index y′ can be determined directly.

In a preferred embodiment, the symmetry factor z itself—if it is not already known—is determined from the index of parameters according to the following formula, preferably without the components being exposed to ionizing radiation (y=1 or y′=0): $$\text{Al}\left(\text{y}=1\right)=\left(1+\text{xz}\right)/\left(\text{x}+\text{e.g}\right)$$

Where Al is the characteristic index and x is the factor by which the switching delays of the CMOS units of the timers differ.

In particular, it should also be pointed out that for very large x values, the limit of equation F4 approaches z, ie. Limes AI(y=1, x->infinity) = z.

According to a preferred embodiment, the index for the influence of the ionizing radiation and/or the symmetry factor is determined using the determined index of parameters and the above equations.

Alternatively or additionally, the measured parameter index can also be assigned to the characteristic number for the ionizing radiation or to the symmetry factor by comparison with a stored (look-up) table. This was created in advance, for example, either by reference measurements or by simulation calculations

The individual CMOS inverters are usually, and therefore also preferably, constructed symmetrically, ie have a symmetry factor z=1. The switching delay of the PMOS switch and the NMOS switch of a respective CMOS inverter are therefore equal, at least in an initial state before being exposed to ionizing radiation.

It is preferred--particularly in the event that the symmetry factor is not known--that this is determined as part of a type of "calibration measurement", specifically in particular before the ring oscillator is exposed to ionizing radiation. In this case, the key figure is y=1, so that the symmetry factor z can be derived from the above formulas: $$\text{e.g}=\text{f}\left(\text{x},\text{Al}\left(\text{y}=1\right)\right)=\left(1-\text{Al}*\text{x}\right)/\left(\text{Al}-\text{x}\right)$$

In principle, the symmetry factor z can therefore be determined independently of the determination of the influence of ionizing radiation, as described.

Knowing z and x, the currently measured AI parameters can then be used to deduce the parameter y for the influence of the ionizing radiation: $$\text{y}=\text{f}\left(\text{x},\text{e.g},\text{Al}\right)=\left(1-\text{Al}*\text{x}\right)/\left(\text{e.g}*\left(\text{Al}-\text{x}\right)\right)$$

Alternatively or additionally, in a preferred embodiment, a (reference) total period To is determined in a calibration measurement before the ring oscillator is irradiated. This results from the sum of the durations of the high level and the low level of the ring oscillator T ₀ =H0+L0. This calibration measurement serves as a basis for comparison to determine the influence of the radiation on the total period T.

According to a preferred variant, a degeneration factor is determined for the respective PMOS and/or NMOS transistors used in the CMOS units. The degeneration factor gives a measure of the influence of the radiation on the switching delay of the PMOS or NMOS transistors used and therefore also represents a parameter for the aging of the respective transistors.

$$\text{DEG}\left(\text{PMOS}\right)=\text{yzt}/\left({\text{zt}}_0\right)=\frac{y\left(1+z\right)T}{\left(1+yz\right)T_0}$$

In this case, the determination is made in particular in accordance with the following formulas. The determination can in turn be arithmetic or with the help of a lock-up table: $$\text{DEG}\left(\text{NMOS}\right)=\text{t}/{\text{t}}_0=\frac{\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0}$$

$$\text{DEG}\left(\text{PMOS}\right)=\text{yzt}/\left({\text{currently}}_0\right)=\frac{y\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0}$$

With the method described here, the radiation-related change in the switching delay of the switches used for the CMOS units can therefore also be determined in a simple manner. For the NMOS transistors, the degeneration factor DEG(NMOS) indicates the ratio of the switching delay to of the unaged transistor (i.e. in particular in the initial state without irradiation) to the switching delay t of the aged transistor (i.e. in particular after irradiation). In the case of PMOS transistors, the degeneration factor DEG(PMOS) indicates the ratio of the two switching delays t, to weighted with the characteristic value y.

In the above formulas, DEG(NMOS) denotes the degeneration factor for the NMOS transistor, DEG(PMOS) the degeneration factor for the PMOS transistor, y the index for the influence of ionizing radiation, z the symmetry factor, x the factor , with which the switching delays of the CMOS units (6, 8, 10) of the timing elements (4) differ, T the (aged) total period and To the (non-aged) reference total period of the ring oscillator. The characteristics x, y, z and T, To are previously determined in a suitable manner—as described here—or are known.

In a preferred embodiment, the switching delay of the two CMOS units of a logic timer differ by a factor x, which is preferably greater than or equal to 1, in particular greater than 5 and which is also specifically in a range between 2 and 20 and in particular between 4 and 10 or can also be much higher (theoretically infinite, see Limes F1 or F4). According to the formula (F1) or (F3), it can be seen that the effect of the parameter (influence of the ionizing radiation or the symmetry factor) becomes more apparent the higher this factor is selected. This is also illustrated in Tables 1 and 2 below. However, as the factor increases, so does the structural complexity. The selected ranges enable a reliable determination of the parameter with moderate effort.

To set the different switching delays, the different CMOS units have different physical and technical structures.

For example, according to a first variant, there is the possibility that a respective CMOS inverter is designed differently physically and technically, for example with regard to geometric and/or Electrical parameters (conductance W/L, width/length) for example of the charge carrier zone between source and drain connection or also due to different doping etc.

In a preferred embodiment, it is provided that, in order to set the different delay times of the two CMOS units, they have a different number of individual, preferably identical, parallel-connected CMOS inverters. Specifically, it is provided that one CMOS unit is formed by a single CMOS inverter and the other CMOS unit by a plurality of CMOS inverters. The ratio of the number of CMOS inverters of the two CMOS units correlates with the factor x for the different switching delays, specifically the ratio of this number corresponds to the factor x. If, for example, a single CMOS inverter is used for the first CMOS unit and five individual CMOS inverters for the second CMOS unit, the result is at least a first approximation of a factor of 5. Connecting the individual CMOS inverters in parallel simplifies shown - a switching resistance is reduced, resulting in faster switching.

In the present application, the actual CMOS component consisting of precisely one PMOS switch and one NMOS switch is generally understood to be the CMOS inverter or switch. The term CMOS unit designates a number of CMOS inverters, in the simplest case just one inverter.

1. 1. Es wird - speziell bei einem Einfluss ionisierender Strahlung - auf einen unzulässigen Alterungszustand geschlossen, falls die Kenngröße für den Einfluss der ionisierenden Strahlung einen Grenzwert überschreitet. In diesem Fall wird eine geeignete Maßnahme ergriffen, beispielsweise eine Fehlerwarnung erzeugt oder ein Hinweis, dass eine ordnungsgemäße Funktion des Bauteils nicht mehr zwingend sichergestellt werden kann und daher ein Austausch vorgenommen werden sollte.
2. 2. Es wird basierend auf der Kennzahl für die ionisierende Strahlung eine Bestrahlungsdosis ermittelt, die der Ringoszillator ausgesetzt war. Hierzu wird beispielsweise eine - in vorab durchgeführten Referenzmessungen - ermittelte Korrelationen zwischen der Kennzahl und der tatsächlichen Bestrahlungsdosis (z.B. TID) hinterlegt.
3. 3. Auf Basis der gemessenen Kenngröße wird ein von einer elektronischen Schaltung ausgeübter Arbeitsprozess modifiziert und angepasst. Dies beruht auf der Überlegung, dass - speziell bei Anwendungen, die die DDR (Double Date Rate) nutzen - eine Veränderung der Zeitdauer des H-Pegels zu einer Beeinflussung der Prozesse führen kann. Da mit den Ringoszillator unmittelbar eine Aussage über den H-Pegel ermittelt wird, wird dies herangezogen, um die durchgeführten Arbeitsprozesse an den gegebenenfalls geänderten H-Pegel anzupassen.

In a preferred embodiment, one or more of the following measures is carried out on the basis of the determined parameter:

1. 1. An inadmissible state of aging is concluded - especially in the case of an influence of ionizing radiation - if the parameter for the influence of the ionizing radiation exceeds a limit value. In this case, a suitable measure is taken, for example an error warning is generated or an indication that proper functioning of the component can no longer be guaranteed and that an exchange should therefore be carried out.
2. 2. An exposure dose to which the ring oscillator was exposed is determined based on the ionizing radiation index. For this purpose, for example, a correlation—determined in reference measurements carried out in advance—between the index and the actual radiation dose (eg TID) is stored.
3. 3. On the basis of the measured parameter, a work process carried out by an electronic circuit is modified and adjusted. This is based on the consideration that - especially in applications that use the DDR (Double Date Rate) - a change in the duration of the H level can affect the processes. Since a statement about the H level is determined directly with the ring oscillator, this is used to adapt the work processes carried out to the possibly changed H level.

The method described here is used specifically in safety-critical applications for monitoring the degradation of electronic components, for example in the field of particle accelerators (CERN), aviation or space technology, nuclear technology or medical technology (radiation medicine (therapy / diagnosis)).

• 1 eine schematisierte Darstellung eines Ringoszillators mit einem Zeitglied bestehend aus zwei CMOS-Einheiten mit unterschiedlichen Schaltverzögerungen,
• 2 eine Schaltbilddarstellung einer möglichen Umsetzung des in der 1 dargestellten Zeitglieds mit den unterschiedlichen Schaltverzögerungen.

An embodiment of the invention is explained in more detail below with reference to the figures. These show in partly greatly simplified representations:

• 1 a schematic representation of a ring oscillator with a timer consisting of two CMOS units with different switching delays,
• 2 a circuit diagram representation of a possible implementation of in the 1 shown timing element with the different switching delays.

Based on 1 the preferred structure of a ring oscillator 2 and some formula-based relationships between a parameter of the ring oscillator and the influence of the ionizing radiation or the symmetry factor are first explained.

The ring oscillator 2 generally has a number n of timers 4 . A respective timing element 4 has exactly two CMOS units 6 which are arranged in series with one another. The two units 6 are units of a first type 8 and a second type 10. The two types differ in terms of their switching delays.

Switching delay is generally understood to mean the length of time that elapses between the point in time of a change in the signal (level) present at the input (gate connection) and the change in the output level (drain connection).

Since a respective timing element 4 includes an inverter pair, the level between the input and output of such a timing element 4 does not change. Depending on the desired resolution or accuracy, the number n of timers 4 is typically in the range of more than 100 and for example also more than 1000 and is for example in the range between 100 and 1000 or even 10,000 individual units 6. The individual units 6 form a chain (delay chain or delay line) due to their serial arrangement. A serial arrangement means that the drain connection of a respective unit 6 is connected to the gate connection of a following unit 6 .

If an H level is present at the input of this chain, this is also present at the output of this chain, delayed by a total switching delay. The output of the chain is connected to an input-side inverter which, in the exemplary embodiment, is in the form of an inverting OR gate (NOR gate 12). whose in the 1 The second input shown is designed as an initialization input, for example as a reset. The input-side NOR gate 12 therefore inverts the output level of the chain and provides an inverted level at the input of the chain. There is therefore a switching between the level states. Overall, therefore, the level at the output of the chain changes periodically between an H level and an L level. These levels are measured with the aid of an oscilloscope 14, or also by internal measurements as described above, so that the duration of the respective level state is measured.

The total period T of the resonant circuit formed by the ring oscillator results from the sum of the durations of the two levels H and L.

In the 1 the switching times or switching delays of the individual switches (transistors) are also shown. First of all, each unit 6--as is usual with CMOS switches--has a PMOS unit and an NMOS switch unit (cf. also 2 with the PMOS switches 18 and the NMOS switches 20). These can differ in their switching delays. If the switching delay is the same, one speaks of a symmetrical CMOS switch or of a symmetrical CMOS unit 6. If the switching delay between the PMOS unit and the NMOS unit differs, then the structure is asymmetrical. The ratio of the switching delays between the PMOS unit and the NMOS unit is represented by a symmetry factor z.

The switching delay, that is to say the switching time required for a switching operation, specifically by the NMOS switch unit, is characterized here as t. Taking the symmetry factor into account, this means that the switching delay for the NMOS unit is t and the switching delay for the PMOS unit is t*z.

In the case of irradiation with ionizing radiation, an asymmetrical degradation occurs, so that the switching times of the PMOS unit and the NMOS unit also change relative to one another as a function of the radiation dose. In the present case, this is represented by the factor y, which is a key figure for the influence of the radiation. Thus, given a switching delay t in the NMOS unit, there is a switching delay y*z*t in the PMOS unit overall.

As already explained, each timer 4 contains different types 8, 10 of CMOS units, which differ in terms of their switching delay. This is shown in the diagram 1 represented by the factor x. In the example shown, the CMOS unit 6,10 of the second type is therefore designed with a switching delay that is lower by a factor of x.

Since the chain consisting of the timing elements 4 is completely run through twice in a total period T, the following relationship results for the total period T in Equation 1: $$T=H+L=n\left(y\mathrm{e.g}t+\frac{t}{x}+t+\frac{y\mathrm{e.g}t}{x}\right)=nt\left(1+x\right)\left(1+y\mathrm{e.g}\right)/x$$

Furthermore, a characteristic index is also referred to below as the aging index AI, which is formed by the ratio of the durations of the two levels H, L. The following relationship results for this characteristic index AI according to equation 2 $$AI=\frac{H}{L}=\frac{y\mathrm{e.g}t+\frac{t}{x}}{t+\frac{y\mathrm{e.g}t}{x}}=\frac{1+xy\mathrm{e.g}}{x+y\mathrm{e.g}}$$

ergibt sich dann ein Zusammenhang zwischen dem Kenngrößen Index Al und einem prozentualen Kennwert y' für den Einfluss der ionisierenden Strahlung wie folgt gemäß Gleichung 2 $$\text{AI}=\left(1+\text{xz}+\text{xy}\text{'}\text{z}\right)/\left(\text{x}+\text{z}+\text{y}\text{'}\text{z}\right)$$

With $y=y\text{'}+1\left[\%\right]$

there is then a relationship between the parameter index Al and a percentage parameter y' for the influence of the ionizing radiation as follows according to equation 2 $$\text{Al}=\left(1+\text{xz}+\text{xy}\text{'}\text{e.g}\right)/\left(\text{x}+\text{e.g}+\text{y}\text{'}\text{e.g}\right)$$

Equation 2 shows on the one hand that the parameter index AI, which is nothing more than the duty cycle between the duration of the H level and the L level (also referred to as duty cycle D, D = Al/ (1+ Al)), is independent of the index y, in the event that the two units 6 of a respective timing element 4 would have an identical switching delay, ie if x=1. In this case, the denominator and numerator are each determined by the expression 1+yz, resulting in a constant value of 1 for the parameter index Al, i.e. the times for H and L are identical, which corresponds to a duty cycle (duty cycle D) of 50% corresponds.

This therefore shows that the influence of ionizing radiation can only be detected and determined by the configuration according to the invention, according to which the CMOS units 6 differ in terms of their switching delay times and different types 8, 10 of CMOS units 6 are used.

The symmetry factor z is preferably known and is selected to be 1, in particular by suitably designing the individual CMOS units 6 as symmetrical units. At least the symmetry factor z is known, for example through an initial calibration measurement, i.e. in the event that no irradiation has yet taken place and in this case y=1 (cf. also explanations of formulas F4 and F5 above). Starting from Equation 2, this then leads to the following Equation 4: $$AI\left(y=1\right)=\frac{H}{L}=\frac{1+x\mathrm{e.g}}{x+\mathrm{e.g}}$$

This relationship is used here to determine the symmetry factor z. Here, too, it can be seen that the determination of the symmetry factor z using the measurement setup described here with the ring oscillator 2 and the measurement of the time durations for H level and L level is only possible if different delay times are set for the different types 8,10 .

The relationship according to Equation 4 is either used directly to determine the symmetry factor z in order to then carry out appropriate further measures such as adapting a work process, etc. Alternatively, the determination of the symmetry factor z serves as a calibration measurement as part of the determination of the influence of the ionizing radiation. This means that especially in the case when the symmetry factor is not known or not known reliably, the determination of the index for the influence of the ionizing radiation is preferably carried out in two stages and the symmetry factor z is first determined in a first stage as part of a calibration measurement. This takes place before the individual components are exposed to ionizing radiation.

The table below shows the relationship between the parameter index AI and the symmetry factor z with different factors for x according to equation 4. Table 1 Al(y=1) z=r/f x 1 2 3 4 5 6 7 8th 9 10 e.g 0.5 1,000 0.800 0.714 0.667 0.636 0.615 0.600 0.588 0.579 0.571 0.6 1,000 0.846 0.778 0.739 0.714 0.697 0.684 0.674 0.667 0.660 0.7 1,000 0.889 0.838 0.809 0.789 0.776 0.766 0.759 0.753 0.748 0.8 1,000 0.929 0.895 0.875 0.862 0.853 0.846 0.841 0.837 0.833 0.9 1,000 0.966 0.949 0.939 0.932 0.928 0.924 0.921 0.919 0.917 1 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1.1 1,000 1,032 1,049 1,059 1,066 1,070 1,074 1,077 1,079 1,081 1.2 1,000 1,063 1,095 1.115 1.129 1,139 1.146 1.152 1.157 1.161 1.3 1,000 1,091 1,140 1,170 1,190 1.205 1.217 1,226 1,233 1,239 1.4 1,000 1.118 1,182 1,222 1,250 1,270 1,286 1,298 1,308 1,316 1.5 1,000 1.143 1,222 1.273 1,308 1,333 1,353 1,368 1,381 1,391

This Table 1 forms a lookup table so that if the factor x is known and the measured parameter index AI (fields with a gray background) the respective symmetry factor z can be read or calculated directly from the equations using external software algorithms (see formulas F4 and F5).

Table 2 below shows the relationship between the parameter index AI and the percentage parameter y' for the influence of ionizing radiation, using the example of symmetrically designed units 6, ie the symmetry factor z=1 according to equation 3 above. Table 2 Al(z=1) x 1 2 3 4 5 6 7 8th 9 10 y' 0 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 1,000 0.1 1,000 1,032 1,049 1,059 1,066 1,070 1,074 1,077 1,079 1,081 0.2 1,000 1,063 1,095 1.115 1.129 1,139 1.146 1.152 1.157 1.161 0.3 1,000 1,091 1,140 1,170 1,190 1.205 1.217 1,226 1,233 1,239 0.4 1,000 1.118 1,182 1,222 1,250 1,270 1,286 1,298 1,308 1,316 0.5 1,000 1.143 1,222 1.273 1,308 1,333 1,353 1,368 1,381 1,391 0.6 1,000 1.167 1,261 1,321 1,364 1,395 1,419 1,438 1,453 1,466 0.7 1,000 1,189 1,298 1,368 1,418 1,455 1,483 1,505 1,523 1,538 0.8 1,000 1.211 1,333 1,414 1,471 1,513 1,545 1,571 1,593 1,610 0.9 1,000 1,231 1,367 1,458 1,522 1,570 1,607 1,636 1,661 1,681 1 1,000 1,250 1,400 1,500 1,571 1,625 1,667 1,700 1,727 1,750

This table 2 is also a lookup table, by means of which the percentage characteristic number y′ can be read off directly using the characteristic variable index A1 recorded by measurement. In the present case, Table 2 was determined as an example by means of a simulation calculation. In principle, such a table can also be created using reference measurements. External software algorithms can be used to calculate the value for y' directly from the measured value AI=H/L and Equation 3.

A measured AI value of 1.308 for x=5 (marked dark gray in Table 2) indicates that the ionizing radiation has resulted in an asymmetry of 50%.

Overall, on the basis of the previously described relationships between the index Al and the symmetry factor z or the index y or y′ for the degradation due to ionizing radiation, desired parameters for the influence of the ionizing radiation and/or the symmetry factor z can be reliably and safely determined determine. In this case, the implementation factor x must be selected to be greater than 1, with the accuracy or resolution for determining the asymmetry increasing by selecting a larger x-factor.

Coming back to equation (1), the switching delay t of the NMOS transistor is generally obtained from 1 : $$\text{t}=\text{f}\left(\text{T},\text{x},\text{y},\text{e.g},\text{n}\right)=\frac{xT}{n\left(1+x\right)\left(1+y\mathrm{e.g}\right)}$$

With this equation, one can now determine the original switching delay to without irradiation, ie with y=1, even if n, x, z are known and T ₀ =H ₀ +L ₀ . According to the calculation, the degradation of the NMOS transistor is: $$\text{DEG}\left(\text{NMOS}\right)=\text{t}/{\text{t}}_0=\frac{\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0}$$

Likewise, for the degradation of the PMOS transistor ( 1 ): $$\text{DEG}\left(\text{PMOS}\right)=\text{yzt}/\left({\text{currently}}_0\right)=\frac{y\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0}$$

With the measured H or Ho and L or Lo values before and after irradiation, the periods T or To can be determined and with the known factors x and z, as well as with the y determination according to Equation 2 (or Formula F6) not only determine the asymmetry between PMOS and NMOS, but also their respective absolute degradation (DEG).

To determine the parameter, especially to determine the influence of ionizing radiation, the becomes the 1 described ring oscillator 2 used. By means of the oscilloscope 14, for example, or an internal measurement as described above, the durations of the levels H, L are measured and set in relation to one another to determine the parameter index AI.

In order to set different delay times between the different types 8 , 10 of CMOS units 6 , a preferred embodiment provides for several individual CMOS inverters 16 to be connected in parallel with one another for the second type 10 . This means the output level of a preceding CMOS unit 6 is simultaneously at the inputs of several CMOS inverters 16, whose outputs (drain connections) are connected to one another and form a common output of the CMOS unit 6.10 of the second type. This is in the 2 shown as an example. In this case, the CMOS unit 6 of the second type 10 has a total of 5 inverters 14 connected in parallel, whereas the CMOS unit 6 of the first type 8 is merely formed by a single CMOS inverter 16 . A factor x=5 is therefore set in this embodiment variant. the 2 therefore shows a total - now by detailed circuit symbols for the individual CMOS inverter 16 - in the 1 shown timer 4.

A respective CMOS inverter 16 has a PMOS switch 18 and an NMOS switch 20 switch. The PMOS switches 18 are—generally in the entire ring oscillator 2—connected to a supply voltage, specifically a respective source connection of the PMOS switch 18 is connected to the supply voltage. In contrast, the source connection of the NMOS switch 20 is connected to a reference potential (ground potential, for example ground). The output, ie the respective drain connection, of both the PMOS switch 18 and the NMOS switch 20 form the output of the CMOS inverter 16. The input is defined by the respective gate connection.

The invention is not limited to the embodiment described above. Rather, other variants of the invention can be derived from this by the person skilled in the art without the to leave the subject of the invention. In particular, all of the individual features described in connection with the exemplary embodiment can also be combined with one another in other ways without departing from the subject matter of the invention.

Reference List

2

ring oscillator

4

timer

6

CMOS units

8th

first type

10

second type

12

NOR gate

14

oscilloscope

16

CMOS inverters

18

PMOS switch

20

NMOS switch

T

total period

H

High level or duration of the high period

L

Low level or duration of the low period

t

switching delay

y

Indicator for the influence of radiation (asymmetry factor)

y'

percentage indicator for the influence of radiation

e.g

symmetry factor

x

Factor for switching delay

Al

Parameters - index (degradation index)

w/l

Conductance of the transistor (Width/Length)

DEG(NMOS)

Degradation of NMOS transistors

DEG(PMOS)

Degradation of PMOS transistors

Claims (12)

Method for determining a parameter for electronic components, in particular semiconductor-based components, namely for determining the influence of ionizing radiation on the electronic components or for determining a symmetry factor (z) of CMOS inverters (16), using a ring oscillator (2), which has a number n of timers (4) and a respective timer (4) is constructed from two serially arranged CMOS units (6, 8, 10) and the CMOS units (6, 8, 10) of a respective timer ( 4) have different switching delays and the parameter is determined using a measured variable of the ring oscillator (2). Method according to the preceding claim, wherein a characteristic index (AI) is determined which is correlated to the ratio between the duration of the high level (H period) and the duration of the low level (L period) of an individual CMOS -Timer (4). Method according to the preceding claim, in which the characteristic index (AI) is measured using a measuring device, in particular using an oscilloscope (14) or using an internal measurement of the H and L periods. Procedure according to one of claims 2 until 3 , where a number (y) for the influence of the ionizing radiation is determined from the parameter index (AI) according to the following formula: $$\text{Al}=\left(1+\text{xyz}\right)/\left(\text{x}+\text{Y Z}\right),$$

where Al is the characteristic index, y is the characteristic number, z is the previously determined or known symmetry factor and x is a factor by which the switching delays of the CMOS units (6, 8, 10) of the timers (4) differ. Procedure according to one of claims 2 until 4 , where the symmetry factor (z) is determined from the characteristic index (AI) according to the following formula: $$\text{Al}=\left(1+\text{xz}\right)/\left(\text{x}+\text{e.g}\right),$$

where Al is the characteristic index and x is a factor by which the switching delays of the CMOS units (6, 8, 10) of the timers (4) differ. Procedure according to one of claims 2 until 5 , whereby a key figure for the influence of the ionizing radiation or the symmetry factor is determined using the determined characteristic index (AI) and a lookup table. Method according to one of the preceding claims, wherein a symmetry factor (z) and/or a total reference period (T ₀ =H0+L0) of the ring oscillator (2) is determined in a calibration measurement before the ring oscillator (2) is irradiated with ionizing radiation. Method according to one of the preceding claims, wherein at least one degeneration factor DEG(PMOS, NMOS) for the PMOS and/or NMOS transistors used in the CMOS units is determined, in particular according to one of the following formulas $$\text{DEG}\left(\text{NMOS}\right)=\text{t}/{\text{t}}_0=\frac{\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0}$$

and or $$\text{DEG}\left(\text{PMOS}\right)=\text{yzt}/\left({\text{currently}}_0\right)=\frac{y\left(1+\mathrm{e.g}\right)T}{\left(1+y\mathrm{e.g}\right)T_0},$$

where DEG(NMOS) is the degeneration factor for the NMOS transistor, DEG(PMOS) is the degeneration factor for the PMOS transistor, y is an index for the influence of ionizing radiation, z is a symmetry factor, x is a factor, with which the switching delays of the CMOS units (6, 8, 10) of the timers (4) differ, t is the switching delay of the respective (aged) transistor, to is the original switching delay of the (non-aged) transistor, T is the (aged) Total period and T ₀ is a (non-aged) reference total period of the ring oscillator (2). Method according to one of the preceding claims, wherein the switching delays differ by a factor (x), the factor (x) being greater than or equal to 1, preferably greater than 5 and in particular in the range between 2 and 20, especially between 4 and 10. Method according to one of the preceding claims, in which, in order to set the different delay times of the two CMOS units (8,10), these units (8,10) have a different number of individual CMOS inverters (16) connected in parallel. Method according to one of the preceding claims, wherein one or more of the following measures are carried out on the basis of the determined parameter a) it is - especially in the case of an influence of ionizing radiation - concluded on an impermissible state of aging if the parameter for the influence of ionizing radiation exceeds a limit value b) based on the index for the ionizing radiation, a radiation dose to which the ring oscillator was exposed is determined, c) based on the measured parameter, a work process performed by an electronic circuit is modified and adjusted. Device for determining a parameter for electronic components, namely for determining the influence of radiation on the electronic components or for determining a symmetry factor (z) of CMOS inverters (16), a ring oscillator (2) being provided which has a number n of Has timers (4) and a respective timer (4) is constructed from two serially arranged CMOS units (6, 8, 10) and the CMOS units (6, 8, 10) of a respective timer (4) have different switching delays .

Images