阅读说明：本技术 用于持续确定薄膜的电阻张量的所有分量的方法 (Method for continuously determining all components of the resistance tensor of a thin film ) 是由 T·科祖比 于 2019-03-19 设计创作，主要内容包括：本发明涉及用于持续确定薄膜(例如各种类型的薄膜电阻器和薄膜传感器)的电阻张量的所有分量的方法。本发明提供了用于持续确定薄膜(例如各种类型的薄膜电阻器和薄膜传感器)的电阻张量的所有分量的方法,其中使用最少数量的触点对薄膜的电阻张量的所有分量进行持续确定,无需对触点进行切换。这在任意形状的均匀薄膜部分(T)中具有至少三个彼此相隔一定距离地布置的触点(K<Sub>i</Sub>至K<Sub>3</Sub>)中得以实现。输入电压U<Sub>i</Sub>(t)施加在每个触点(K<Sub>1</Sub>至K<Sub>3</Sub>)上,检测到流经触点(K<Sub>1</Sub>至K<Sub>3</Sub>)的电流I<Sub>i</Sub>(t),并且通过电压和电流值确定薄膜部分(T)的全部电阻张量ρ。(The present invention relates to a method for continuously determining all components of the resistance tensor of thin films, such as thin film resistors and thin film sensors of various types. The present invention provides a method for continuously determining all components of the resistance tensor of a thin film, such as various types of thin film resistors and thin film sensors, in which all components of the resistance tensor of the thin film are continuously determined using a minimum number of contacts without switching the contacts. In the uniform film portion (T) of arbitrary shape, there are at least three contacts (K) arranged at a distance from each other i To K 3 ) Is achieved. Input voltage U i (t) is applied to each contact (K) 1 To K 3 ) Upper, detecting a current flowing through the contact (K) 1 To K 3 ) Current of (I) i (T), and the total resistance tensor ρ of the thin-film portion (T) is determined by the voltage and current values.)

1. A method for the continuous determination of all components of the resistance tensor in thin films, such as thin-film resistors and thin-film sensors of various types, characterized in that an arbitrarily shaped uniform portion (T) of the thin film has at least three contacts (K) arranged at a distance from one another₁To K₃) Wherein a voltage U is input_i(t) is applied to each contact (K)₁To K₃) At least one of (1) and (b); to obtain a current flowing through the contact (K)₁To K₃) Current of (I)_i(t); and the overall resistance tensor ρ of the portion (T) of the thin film is determined by the voltage and current values.

2. Method according to claim 1, characterized in that three contacts (K) are passed through₁To K₃) At an input voltage U_i(t) the voltage u (t) and the current vector i (t) required to determine the resistance tensor ρ are calculated with constant variation.

3. Method according to claim 1, characterized in that when the measured object has an arbitrary dimension, the number of contacts (K) exceeds the dimension of the measured object by at least 1.

4. Method according to any of claims 1 to 3, characterized in that the input voltage U is made_i(t) the source impedance is increased to measure the effect of the lateral resistance.

5. Method according to any of claims 1 to 4, characterized in that the application is made at a contact point (K)₁To K₃) At an input voltage U_i(t) is a phase-shifted sinusoidal alternating voltage.

6. Method according to any of claims 1 to 5, characterized in that the input voltage U applied at the contact (K) is_i(t) relative to the contact point (K)_i) Respective adjacent input voltages U as a function of the number_i(t) has a phase difference.

7. Method according to claim 6, characterized in that the input voltage U is_iThe phase difference of (t) is 120 ° when there are three contacts (K) and decreases accordingly as the number of contacts (K) increases, the phase difference being on average 90 ° when there are four contacts (K) and 72 ° when there are five contacts.

8. Method according to any of claims 1 to 7, characterized in that the current I is varied by changing the direction of rotation of the voltage vector U (t)_iThe intrinsic timing phase offset of the plurality of measurements of (t) is separated from the effect of the parasitic phase offset.

9. Method according to any of claims 1 to 4, characterized in that the input voltage U is_i(t) is a noise signal.

10. Method according to any of claims 1 to 4 or 9, characterized in that the input voltage U_i(t) is subjected to a noise shaping process.

11. The method of any one of claims 1 to 10, wherein the amplitude and phase are determined directly from the complex result using fourier transforms of the voltage and current waveforms or goertzel filtering.

Technical Field

The present invention relates to a method of continuously determining all components of the resistance tensor of thin films, such as thin film resistors and thin film sensors of various types.

Background

Understanding or determining the parameters of the film is of great importance for practical applications. For this reason, determination of the resistance tensor of the sheet resistance is indispensable, which is of great significance in studying the characterization transport phenomenon, and in reading out various types of thin film sensors, resistance determination in the industry is indispensable.

The general resistance tensor of an object consists of 9 components (a 3x3 matrix), where in the case of thin films one spatial direction (layer normal) is meaningless, so that a thin film can be understood as a two-dimensional body. Thus, the electrical properties of this two-dimensional volume can be described by a reduced tensor. The reduced sheet resistance tensor consists of four components, which form a 2x2 matrix.

However, in this 2x2 matrix, only 3 values are physically independent. In the case of isotropic conductive films, there are even only two physically independent variables, namely the machine direction resistance and the transverse direction resistance. In this case, the results of the four components of the resistance tensor are as follows:

an example of an arrangement for measuring the resistance according to the so-called four-point method is disclosed in DE 3123427C 2. In particular, a measuring device for a cathode sputtering apparatus with a planar movable substrate holder is described. In this case, the four measuring contacts are guided from below via the substrate holder to a reference substrate situated thereon, which has four low-resistance contacts. In addition, the measurement can also be carried out by means of direct current or alternating current according to a two-point measurement method. In the case of direct resistance measurement, a constant current is applied to the resistance to be measured, the voltage drop across the resistance being the measured variable.

If the longitudinal resistance of the film is to be determined accurately, a so-called four-point measurement is usually used, in which two outer contacts conduct the applied current through the layer strip (a layer strip) and two inner contacts are used to measure the voltage drop.

The four-point sheet resistance (four-point sheet resistance) is calculated using the measured values of the voltage value and the current value, thereby determining the sheet resistance. However, analog geometric corrections are still required for this purpose.

If the lateral resistance is to be measured, a cross structure should be used, in which the voltage measurement points are as diametrically opposed to each other as possible with respect to the current flow direction.

In both methods, there is the problem that the measurement results are corrupted by inevitable geometric errors in terms of yield values, which result from the mixing of longitudinal and transverse resistances at the measurement contacts.

One possible solution to this problem is van der pol measurements. In this case four contacts are attached to the edge of an arbitrarily shaped uniform test portion of the membrane, each at a distance from the other. By alternately switching the contacts to current or voltage contacts and then offsetting the respective measurements, the longitudinal and transverse components of the resistance tensor for the portion can be inferred.

The van der pol measurement has the disadvantage that switching the current direction can be problematic due to stray inductance between the test part (membrane) and the measurement electronics, since voltage peaks can occur during switching, which can damage the membrane or the measurement electronics. Furthermore, the decay time of the voltage peak, within which the measurement may be affected, has to be expected.

In the Review of Scientific Instruments (Review of Scientific Instruments), 1999, 70 th, 2177 th page 2178, KIM, G.T. [ et al ]: in the non-switching van der Waals technique (non switching van der Pauw technical using two different modulation frequencies), a method of measuring the resistance of a sensitive sample using two independent AC power supplies and two lock-in amplifiers with two different frequency modulations is described.

In Review of Scientific Instruments, 2008, 79, 73901-173901-3, RISS, O. [ etc.: a similar approach is described in Offset reduction in Hall effect measurements using an on switching van der Pauw technology. It is a switchless van der pol technique that uses two electrically isolated ac power supplies operating at different frequencies and two lock-in amplifiers, and is proposed for hall effect measurements. The typical parasitic offset voltage of this measurement method is reduced by simultaneously averaging two simultaneously acquired data sets.

In "metrology", 2007, 44, pages 15-23, SCHURR, J. [ et al ]: the AC quantum Hall effect as a primary standard for impedance (The AC quantum Hall effect as a primary impedance of impedance) states that quantum Hall resistances measured at frequencies in The kilohertz range exhibit frequency-dependent and current-dependent deviations from quantified dc resistance values. This is due to the capacitive effect, which is reflected in the ac longitudinal resistance. The quantum Hall resistor irrelevant to frequency and current can be obtained by zeroing the alternating-current longitudinal resistor.

Another example of an application of the four-point method for dynamically compensating the offset voltage of a hall device is disclosed in DE 69736944T 2, in which a plate-shaped structure has four contacts which are arranged on the outer circumference of a conductive area in the form of pairs of opposing contacts.

The contact pairs are here arranged offset by 90 °, wherein each contact pair supplies a periodic alternating current, so that the phase shift of the supply current corresponds to the spatial phase shift of the contact pair (i.e. 90 °). The superposition of the supply currents produces a continuous rotating current vector in the hall device. By measuring the voltage between the respective terminals simultaneously, the signal consisting of the hall voltage and the periodic offset voltage can be isolated. The offset voltage may be eliminated by averaging the signal over at least one period.

Another disadvantage of the four-point measurement described above is the number of contacts required and the consequent cost of circuitry and wiring. This is also why two-point measurements are preferred in integrated circuits.

Disclosure of Invention

It is a general object of the present invention to provide a method for continuously determining all components of the resistance tensor of a thin film, such as thin film resistors and various types of thin film sensors, in which all components of the resistance tensor can be continuously determined without switching contacts using a minimum number of contacts.

This is achieved by providing at least three contacts on an arbitrarily shaped uniform portion of the film to be connected to an input voltage, obtaining a current flowing through the contacts, and determining the overall resistance tensor of the portion of the film from the voltage and current values, the contacts being arranged at a distance from each other, wherein the input voltage is applied at each contact.

In a first design of the invention, the voltage and current vectors required to determine the resistance tensor are calculated by the continuous variation of the input voltage at the three contacts. In a development of the invention, the number of contact points for a measured object having any dimension exceeds the dimension of the measured object by at least 1.

Since the measurement of the lateral resistance is usually performed at a high resistance, it is advantageous to increase the impedance of the input voltage to measure the lateral resistance effect.

The input voltage applied to the contacts is preferably a sinusoidal alternating voltage with each phase shift, wherein the input voltage applied to the contacts has a phase difference with respect to respective adjacent input voltages, which are a function of the number of contacts.

For three contacts (three-phase current principle), the phase difference of the input voltage is particularly preferably 120 °, and as the number of contacts increases, the phase difference decreases accordingly; the phase difference is on average 90 ° for four contacts and 72 ° for five contacts.

Furthermore, the inherent timing phase offset of multiple measurements can be compensated by changing the current rotation direction.

Alternatively, the input voltage may be a noise signal and/or the input voltage is subjected to a noise shaping process.

In addition, the voltage and current waveforms may be fourier transformed or lattice filtered to directly determine the amplitude and phase from the complex results.

The present invention provides the option of minimizing the number of contacts required in precision measurements and allows the components of the resistance tensor to be resolved without disturbing the current and voltage switching. On the one hand, the resulting advantages are that the circuit complexity of the commercial product can be reduced, which can save resources and increase durability. On the other hand, the advantage of precision measurements in science is that the negative effects of stray inductances, which may lead to serious limitations, especially on long connection lines between layers and electrons, for example in major research facilities, are mitigated.

The present invention is explained in more detail below based on exemplary embodiments.

Drawings

The figure shows a uniform portion T of arbitrary shape of the film, having at least three contact points K₁To K₃The three contacts are arranged at a distance from each other.

Detailed Description

Time-varying input voltage U_i(t) applied to contact K₁To K₃And a time-varying current I_i(t) is determined. The 3 voltage and 3 current vectors are added to obtain time-varying voltage vector u (t) and time-varying current vector i (t) results. These vectors are connected to each other by the resistance tensor ρ according to u (t) ═ ρ i (t).

A system of linear equations may be formed from a matrix equation having a known vector U and a known vector I. If there are exactly two value pairs (U (t), I (t))₁And (U (t), I (t))₂Then the system of equations is uniquely solvable for the film. Linearly independent means that the two current vectors from the two value pairs are not parallel, and the two voltage vectors from the two value pairs are also not parallel. If more than two value pairs (U (t), I (t))_jThe resistance tensor ρ therefore cannot generally be uniquely determined.

However, a regression method may be specified that fits all the pairs of values as good as possible according to the least squares deviation method. Therefore, if the measurement is affected by noise (e.g., measurement noise), the accuracy of the determination of the resistance tensor ρ can be improved by measuring two or more values.

After the determination of the resistance tensor ρ by the method, further variables of importance can be calculated. The conductance tensor σ is thus obtained by matrix inversion of the resistance tensor ρ.

Another advantageous operation is to triangulate the resistance tensor ρ by Schur decomposition (Schur decomplexion). This characterization in resistance tensor ρ_SchurThe independent components of the resistance tensor are explicitly generated. For example, in the case of a two-dimensional film, the resistance tensor ρ is a 2 × 2 matrix. At rho derived from the schuler's theorem_SchurThe highest and lowest longitudinal resistances are generated on the main diagonal, and the lateral resistances are generated on the upper adjacent diagonal elements.

For the determination of the resistance tensor in the procedure, the number of contacts K_iThe dimension of the object to be measured must exceed at least 1. Thus for a film which is generally considered to be a two-dimensional object, three contacts K are required₁、K₂、K₃To fully determine the resistance tensor. The number of contacts is reduced by one from the number of contacts required by previous methods.

However, this procedure is also conceivable in other dimensions N, in which the measurement result of the two electrical conductors according to the invention is a special case when N is 1. The 9-component resistance tensor of the three-dimensional object (N ═ 3) can be fully obtained by four contact points. Theoretical extensions of abstract high-dimensional objects are also conceivable.

The input voltage U is described below_i(t) a specific case that simultaneously generates a plurality of items of information on the test object and is easy to implement simultaneously. For a two-dimensional test portion and several contacts Ki of 3, three voltage values can be selected at each time point. This allows three boundary conditions to be met simultaneously. An advantageous boundary condition is that the sum of the voltage values is equal to zero, thereby ensuring that the test portions are always at the same average potential. The direction and length of the voltage vector u (t) can be randomly determined by the remaining two boundary conditions.

Thus, not only the resistance tensor ρ is determined, but also its dependence on the excitation frequency (by spectral analysis of the noise) and its dependence on the amperage are described in the measured values. Thus, the frequency and current dependence of the components of all the resistance tensors can be determined simultaneously, so that a separate measurement may be superfluous.

By further limiting the possible values of direction and length of the voltage vector u (t), the excitation noise can be shaped such that it can describe particularly effectively a particular dimension, for example, the frequency dependence of the resistance tensor ρ between just 125 and 1000 hertz, or between 50 and 1000 hertz, and any other value in the ampere-frequency space of the resistance tensor ρ.

The following section describes a particular case in relation to a measured two-dimensional object. In this case, the tensor resolution of the resistances can also be achieved without solving the system of equations.

This particular case relates to phase shifting the input voltage U_i(t) at contact K₁To K₃A sinusoidal voltage curve of (a). In this case, a difference evaluation may be performed in which the system of linear equations does not have to be solved to determine the resistance tensor ρ. Instead, the time-dependent absolute values of voltage | U | (t), current | I | (t), and phase angles of the voltage and current vectors are calculated from the time-dependent voltage and current vectors U (t) and I (t)And

the longitudinal and transverse resistances can then be calculated from these dimensions. This method always separates the longitudinal and transverse resistances, but only yields the average longitudinal resistance in the case of longitudinal resistance anisotropy.

For this reason, a time-varying variable is observed during the rotation of the voltage and current vectors.

The longitudinal resistance is given by:

the lateral resistance is then given by:

in a method for the continuous determination of all components of the resistance tensor ρ of a thin film, for example of thin-film resistors and thin-film sensors of various types or in the case of thin conductive layers, for example metal layers, it is assumed that there are sections T which have in each case at least three contacts K₁To K₃To be connected to different input voltages U_i(t) wherein the contact K₁To K₃The resistance tensor ρ of the film is to be determined around an arbitrarily large surface portion T of the film. The maximum number of contacts for implementing the method according to the invention is not limited in theory, but the wiring and calculation overhead for carrying out the method increases with the number of contacts and is therefore subject to economic limitations.

Unlike the van der Pauw method, in this method, at least four contacts K_iApplied at the edge of a uniform portion of arbitrary shape, each at a distance from the other, and alternately switching contacts K₁To K₄As current or voltage contacts to determine the resistance tensor ρ of the section by shifting the respective measurement results, in the method according to the invention the voltage vector u (T) of the section T of the film enclosed by the contacts and thus the current vector i (T) are varied by applying different input voltages at the contacts.

The resulting input voltage vector u (t) may be defined by a vector addition of the input voltages, and the resulting current vector i (t) may be defined by a current flowing through contact K_iIs defined by a vector addition of the input currents. The resistance tensor ρ can be reconstructed from the ratio of the magnitude and phase of the input voltage and current vectors u (t), i (t).

Vector addition is described below, with voltage vector u (t) and current vector i (t) being derived by vector addition. Applied to M contacts K_iAt a voltage U_i(t) is a scalar voltage value indicative of a potential difference relative to a common reference potential (e.g., ground). Contact current I_i(t) is also a scalar current value representing the value of the current passing through each contact K_iThe dimension of the current entering the portion T of the film. To form a three-dimensional voltage vector from these scalar voltage values, M arbitrary n-dimensional spatial directions n are selected which are not parallel to each other_i. All spatial directions n_iAre all unit vectors, which means | n_i1. The voltage vector u (t) and current vector i (t) may be determined by:

selection of spatial directions n_iExcept for the condition of being non-parallel to each other. In particular n_iNeed not be in contact with the contact K_iCorresponds to the actual spatial direction of (a). Selected n_iAdvantageously along the M contacts K_iAnd the portion T is advantageously a regular M-polygon, wherein the contacts are each located on one of the M axes of symmetry of T and are M times rotationally symmetrical to each other.

If the sinusoidal input voltages U have a predetermined phase difference from each other according to the three-phase current principle_i(t) is used as the input voltage, rotation of the input voltage vector u (t) is most easily induced, wherein more than three phases can also be considered.

Or, input voltage signal U_i(t) may also be an independent noise signal, resulting in a voltage vector u (t) that varies continuously in absolute value and direction. These noise signals may be formed by spectrally-unshaped (white) noise or may be subjected to a spectral noise shaping process.

The following exemplary embodiments relate to a method according to the invention for determining all components of a resistance tensor ρ in an arbitrarily shaped homogeneous portion T with three electrical contacts K₁To K₃In which the contact point Ki is formedThe corners of the triangle.

At these contacts K_1-3Having a sinusoidal input voltage U applied thereto_i(t) in each case with respect to the respective adjacent contact K₁；K₂；K₃The phase difference of the voltages is 120 °, i.e. the three-phase current principle is used to determine the resistance tensor ρ. If more than three contacts are used to determine the resistance tensor ρ, the phase difference between the input voltages becomes small, i.e., the phase difference averages 90 ° in the case of four contacts K, 72 ° in the case of five contacts K, and so on.

Input voltage U_i(t) summing their known phase difference vectors to determine a voltage vector u (t) that rotates due to the phase difference of the sinusoidal voltage applied at contact K.

In the process, the input current I flowing through the contact K is adjusted_i(t) measurements are taken and added according to the phase difference vector to determine the instantaneous current vector i (t), which also rotates.

The measured values are provided for further digital processing by a multi-way a/D converter or by a simultaneous a/D converter.

For contact K₁To K₃Enclosing the ideal case of a perfectly symmetrical triangle without transverse resistance, the current vector i (t) is always in phase with the voltage vector u (t), so that their ratio describes the longitudinal resistance.

The resistance tensor ρ may be determined by the ratio of the magnitude and phase of the voltage and current vector U, I.

However, in practice it must be assumed to be an arbitrary triangle, where the current vector i (t) is usually out of phase with the voltage vector u (t). If the phase difference of the two vectors is integrated over a full rotation of 360 deg., the result is zero. This means that if a complete rotation of the voltage vector is observed, the deviation from the ideal triangle is negligible.

For example, during a rotation around 360 °, the current vector may first have a positive phase difference with respect to the voltage vector. If it now rotates further, this phase difference decreases, crosses zero, and becomes negative again. Thus, during rotation, the phase difference itself swings up and down. During a complete revolution the phase difference will be as negative as positive and zero at 360 deg.. This situation is generally valid also for non-ideal triangles. An ideal triangle is a special case where the phase difference is always 0.

If the section T has a lateral resistance (e.g. hall effect) in addition to the longitudinal resistance, a lateral current will also flow during the application of said voltage, so that the current vector i (T) is deflected throughout the rotation. The result is that the integral of the phase difference between current and voltage is no longer zero, but results in a value that depends on the lateral resistance.

Therefore, in order to determine the resistance tensor of the portion T, it is necessary to measure the effect of the lateral resistance, which typically occurs in the high resistance case. However, since the ideal voltage source is low-resistive, it is reasonable to increase the impedance of the voltage source during the measurement of the lateral resistance. This can be easily achieved by using an additional resistor, so that the low lateral current can be measured better.

In the case of contacts K or connecting lines of the layer geometry which contain inductances or capacitances, these reactive influences can thus influence the measurement of the component of the lateral resistance.

Furthermore, a time offset of 6 measurements on the multi-channel a/D converter results in a significant phase offset. By changing the direction of rotation of the current, these effects can be separated from the effects of the transverse resistance inherent to the thin layer. The inherent phase shift also creates a swap.

Method for continuously determining all components of the resistance tensor of a thin film

List of reference marks

K_iContact point

U_i(t) input voltage

I_i(t) input current

I (t) Current vector

U (ton) voltage vector

T surface portion

Rho resistance tensor

Number of M contacts

n_iDirection of space