GB2257520A - Method and device for measuring the thickness of thin layers - Google Patents

Abstract

The device comprises two coils 18, 36 concentric with a common core 13, the coils being arranged so that they see the thin layer and a layer supporting it in different ways. The coil outputs are fed to a device which computes the thickness. As shown the inner coil 18 is wound on core 13 and the outer coil 36 on sleeve 28 and a shielding sleeve 43 surrounds both coils. When operated at low frequencies (20Hz-1kHz) an excitation winding is required, at eddy current frequencies (100kHz-10MHz) single windings are required for the coils and at super high frequencies (10MHz-1GHz) the coils may be replaced by cavity resonators open towards the object. <IMAGE>

Description

Method and device for measuring thin layers The invention relates to a method in accordance with Claim 1 and a device in accordance with Claim 9. The preamble of Claim 1 proceeds from a prior art such as is disclosed, for example, in German Offenlegungsschriften 3,902,095; 3,437,253; 2,556,340, etc.

Non-destructive methods for layer thickness measurement, which use the widespread magneto-inductive method or the eddy-current method, are based on the variation in a low-frequency or high-frequency electromagnetic field as a function of a layer applied to the measured object. The field used for measurement has a spatial extent, and thus there is not only a desired dependence on the layer thickness, but also on the shape of the measured object.

In the case of the magneto-inductive, lowfrequency method, which is used to measure non-magnetic or electrically non-conductive layers on a magnetic base material, the permeability of the measured object also features in the measurement as a further disturbance variable. In the eddy-current method, which is principally used to measure electrically non-conductive or weakly conductive layers on nonferrous metals, the influence of the geometrical shape of the measured object is substantially more strongly pronounced. Instead of the permeability, which with nonferrous metals can be set virtually equal to 1, the electrical conductivity of the base material also features as a further disturbance variable. It has recently become possible to use circuit engineering in order to exclude this undesired influence over a wide range for the latter disturbance variable.

In both methods, the geometrical shape of the measured object remains a non-negligible influencing variable. It is therefore necessary in measurement practice to perform a so-called calibration on the measured object. This is carried out by simulating the measured value 0 firstly on the non-coated measured object and then on a measured object with a known layer, which is applied either permanently or in the form of a foil on the non-coated measured object. The indicated value corresponding to the measured value is set to the known layer thickness of the measured object. This calibration can be carried out using a plurality of layers in order to match the characteristic better to the relevant measurement task.Since the measured object is not generally available in a non-coated form, and often has very complex shapes, it is mostly very difficult to measure layer thicknesses on concave or convex surfaces.

One of the possibilities of reducing the geometrical influencing variables resides in the design of the measuring probe. The smaller the measuring probe, the smaller is the spatial extent of the measuring field and thus the dependence on the geometrical shape of the measured object. However, such measures are limited, since for the layer thicknesses in the range 0 - 300cm, which are principally of interest in practice, the probes would have to be kept so small in size that an implementation of such designs is no longer technically possible.

It is the object of the invention to eliminate, over a wide range, the undesired dependence of the measured value on the geometrical shape of the measured object by means of an arrangement. At the same time, the design of the measuring probe is still to be easy to master in terms of production engineering, and is to have a high sensitivity in the measuring range of approximately 0 - 500cm. According to the invention, this object is achieved, with respect to the method by the features of Claim 1, and, with respect to the device by the features of Claim 9.

It should be pointed out that in this technique the layer to be measured is not measured against air, but that the layer to be measured is always seated on a carrier material.

Thin layers in this sense are in the range from 0 to a few hundred micrometers, but can also be, for example, up to 15 mm thick, if bitumen or ceramic layers are measured, for example.

If magneto-inductive, that is to say in the language of this technique "low-frequency", measurements are made, two windings, specifically an excitation winding and an induction winding are required, such as the windings 33, 36 from German Offenlegungsschrift 3,437,253, for example. In this regard, the frequency band is approximately between 20 Hz and 1 kHz. If measurements are made using the eddy-current method, only a single coil is required, because the latter is situated in the resonant circuit of an oscillator and varies its frequency depending on the damping. Here, the frequency band used is between 100 kHz and 10 MHz.

If measurements are made at super high frequency, it is possible to use cavity resonators open towards the measured object, which in this frequency band generate a magnetic field analogous to a coil. The frequency band can then be roughly 10 MHz to 10 GHz.

A "coil device" in the meaning of the claims can therefore be a single coil, can be constructed from two or more coils, can be a super high frequency component which generates a magnetic field, or can be produced using multilayer technology or a similar technology.

The invention will now be described with reference to two exemplary embodiments. In the drawing: Figure 1 shows a cross-section through the lower part of a probe, which is not otherwise further represented, substantially enlarged but to scale, Figure 2 shows the exploded representation of an entirely similar exemplary embodiment, Figure 3 shows a graph to explain the invention, Figure 4 shows a computing device for the inven tion, Figure 5 shows a first table representing the performance of the invention, and Figure 6 shows a second table of the same sense.

A probe 11 serves measurement using the eddycurrent principle. It has a geometrical longitudinal axis 12. Provided in relation thereto is a coaxial core 13 of ferrite, which has an outside diameter of 1.4 mm. Other dimensions can be derived from this dimension. In its lower region, the core is reduced by a step 14 to a coil core 16 which has a mounting spherical cap 17 of abrasion-resistant material at the bottom. Seated on the coil core 16 is a coil 18 of copper wire, whose two connecting wires 19, 21 are led upwards and which - in so far as they pass the core 13 - extend in a longitudinal groove 22 of the core 13.The connecting wires 19, 21 are shielded by means of a symbolically represented shield 23 and are fastened at the top to contacts 24, 26 of a connection plate 27, which is rigidly mounted in a way not represented on the probe body (not represented).

The center of the magnetic field of the coil 18 and the center of the coil core 16 coincide with the geometrical longitudinal axis 12, which penetrates the surface of the mounting spherical cap 17 exactly at its center.

Seated coaxially with the geometrical longitudinal axis 12 is an outer sleeve 28 of ferrite, whose lower annular end face 29 is perpendicular to the geometrical longitudinal axis 12 and ends just above the mounting spherical cap 17. A cavity 31, in which the connecting wires 19, 21 run in a mechanically and electromagnetically protected fashion, is formed above the core 13.

Between the circular cylindrical inner wall 32 and the likewise configured outer wall of the core 13 there is a gap (not representable in the figure), which permits a relative displacement along the geometrical longitudinal axis 12, that is, however, negligible as a magnetic resistance. The outer sleeve 28 has an inwardly directed step 33 approximately halfway up. This results in a circular cylindrical coil tube 34 which is coaxial with the geometrical longitudinal axis 12 and has approximately half the wall thickness. Seated thereon is a coil 36 of thin copper wire, whose magnetic field coincides at its center, during operation, with the geometrical longitudinal axis 12. Connecting wires 37, 38 of the coil 39 are led through a longitudinal groove 39, which is open outwards, to contacts 41, 42 of the connection plate 27.

A shielding sleeve 43 of magnetizable, high permeablity material is coaxial with the geometrical longitudinal axis 12 and is an annular cylinder. A gap between the inner wall 44, the shielding sleeve 43 and the outer wall 46 of the outer sleeve 28 is insignificant from the point of view of the magnetic resistance, but permits a guided relative movement of the shielding sleeve 43 on the outer sleeve 28. The lower end face 47 is perpendicular to the geometrical longitudinal axis 12, and permits the end face 29 and thus also the mounting spherical cap 17 to emerge downwards. The shielding sleeve 43 is sealed at the top with the outer sleeve 28.

All the mutually displaceable parts can be connected nondisplaceably to one another, for example by means of a bonding agent, after adjustment.

The design produces, inter alia, two circular cylindrical annular gaps 48, 49, in which the coils 18, 36, which are wound on the whole of the total available heights, are seated.

The second exemplary embodiment according to Figure 2 differs from the construction just outlined only in that here the coils 51, 52 have a rectangular crosssection 53, 54. In this case, the cross-section 54 in accordance with Figure 2 is relatively slimmer than the cross-section 53.

The annular gaps 48, 49 can also be provided per se in one piece in a cup-type core. At present, however, no cup-type cores are known which could be purchased in bulk and would thus be sufficiently cheap and also at the same time have the required dimensional stability and coaxiality with respect to the geometrical longitudinal axis 12.

If the measured object has a deviation from the plane shape with an infinite extent, a so-called geometrical influence is present. The deviation from the said plane shape has a larger influence on the outer coil 36 than on the inner coil 18. A signal is derived from the outer coil 36 which is used to compensate for the geometrical influence on the measured variable.

This is implemented by exciting the coil 18 using another frequency than the outer coil 36. The frequencies have a ratio of 1:2 to 1:10, the inner coil 18 expediently being excited using the higher frequency.

There are thus available two signals of different frequency, which can be decoded using suitable circuits.

The characteristics, which are required for explanatory purposes, of an eddy-current probe, which is designed for measuring nonconductive or weakly electrically conductive layers on nonferrous metal, are plotted in Figure 3. The nominated measured variable is obtained by determining electrical reference values on a reference sample having a plane surface which is to be regarded as infinite with respect to the dimensions of the probe and has a conductivity of 10 - 60 mS/m. These measured values can be present both as voltages and in the form of frequencyencoded signals. When a measuring system is mounted on the plane reference surface, the inner coil 18 obtains the measured value x0l, which is stored in a memory. In the outer compensation coil, a second measured value x02 is stored in a further memory.By lifting the probe from the plane reference surface, a further pair of measured values x,1, x,2 is obtained, which are stored once again in memories as so-called saturation values, x5l referring analogously to the coil 18, and xS2 to the coil 36. The distance from the measured object must be so large that it is no longer possible to perceive any influence due to it. As a rule, this distance is approximately 4 times the diameter of the coil 36. The infinitely large plane reference surface is provided when the latter has approximately 3 times the diameter of the shielding sleeve 43.

Thus, four reference values are available, x0l, x5l for the coil 18 and x02, x,2 for the coil 36. The signals xl of the coil 18 and x2 of the coil 36 are thus always between the limit values x0l and xsl or x02 and xS2, and are normalized in a known way as follows: X1 - XO1 X2 - X02 for = ~~~~~~ for coil 18, and Xn2 = ~~~~~~~ for the coil 36.

X,1 - X01 Xa2 - X02 The normalized measured variables are represented semi-logarithmically in Figure 3. The respective normalized measured variable is plotted on the abscissa and the logarithm to the base 10 of the layer thickness is plotted on the ordinate. The characteristic denoted by 56 belongs to the inner coil 18, and the characteristic denoted by 57 belongs to the outer coil 36, after calibration on the infinitely large plane. By definition, the respectively normalized measured value is between 0 and 1 and the two curves 56 and 57 approach the limit values 0 and 1 asymptotically. As may easily be seen, the characteristic 57 referring to the outer coil 36 is situated clearly above the characteristic 56 of the inner coil. If, now, the probe 11 is placed on a concavely curved measured object, the characteristics denoted by 58 and 59 result, to which the asymptotes 61 and 62 for the layer thicknesses 0 refer. The characteristic 58 refers to the coil 18, and 59 to the coil 36. In both coils, the asymptote for an infinitely large layer thickness is the ordinate perpendicular to the abscissa value 1, since in the raised state the coils 18, 36 no longer recognize whether a concavely curved or plane measured object was previously present.If it is now desired to establish the size of the systematic influencing variable for a layer thickness t and concave curvature (in the present case a cylinder of diameter 6 mm), then in the case, for example, of an arbitrarily selected layer thickness of 30cm, the corresponding ordinate value is plotted in Figure 3 and cuts the curve 58 at the point 63. The deviation of the plane can be determined by erecting the perpendicular at 63 which cuts the characteristic 56 belonging to the plane measured object at the point 64.

Without correction, the measured value tl corresponding to the point 64 would be indicated. The section 66 is therefore identical to the additional layer thickness simulated by the concave curvature of the measured object (cylinder of diameter 6 mm). The measured value t referring at 64 to the inner measuring coil obeys the relationship t1 = t + c, as is easy to see. The same conceptual approach can be adopted with the characteristics 57, 59, which refer to the outer coil 36. The measured value referring to the same layer thickness cuts the characteristic 59 at the point 67, and the measured value t2 read at the point 68 can be determined on the characteristic 57 referring to the plane measured object.As is evident, the influencing variable c z (Xn) (1) determined by the curvature is, as entirely to be expected, substantially larger than in the case of the coil 18. It is therefore possible to write two equations as follows: tl = t + c (2a) t2 = t + cz (xn) (2b) The factor z(xn) is a virtually constant magnitude to a first approximation, ie. a constant factor dependent on the diameter ratio of the coils 18, 36.It is now possible to eliminate from the two equations the disturbing, systematic influencing variable, which depends on the geometry of the measured object, and to write the following equation: t2 tcorr = tl - ~~~~~~~~~~~~~ (3) z() - 1 tcorr is written instead of t, since in the mathematical sense z (xn) is never exactly constant for all the influencing variables.

In this formula, t1 and t2 denote by definition the measured values which refer in the case of the arbitrary layer thickness t to the same convex or concave curvatures and which were derived by the coil 18 and the coil 36. If measurements are made on the plane, t1 = t2 by definition, and thus the subtrahend in equation (3) = 0. In this special case, the measured layer thickness t corresponds to the measured value tl of the coil 18. The larger the difference t2 - tl, the larger becomes the subtrahend by which, as is easy to see, t1 is reduced, because each convex curvature with respect to the plane simulates an additional layer.This variable is not critical, since the numerical factor z(x) has a value of approximately 12 for a ratio of the coil 36 to the coil 18 of approximately 2. Even in the range 0 < x < 0.6, a fixed numerical value for z permits a 20 - 100-fold reduction in the systematic influence, for a diameter of a cylindrically curved surface which corresponds approximately to the diameter of the coil 36. For a value twice as large, the compensation is possible in a virtually ideal fashion with a constant numerical factor z.

In order to eliminate an error of second order, z can be calculated as a function of Xni or x < . It is expedient to select xnl. Consequently, only the notation xn will be used below. As already described, this follows from the notation z(xn). The function z(rh) can appear as follows, for example: z(x) = z0 (1 - axnf) 2 < p < S ; 0.2 < &alpha; < 0.6 (4) Since z(xn) > 1, the following approximation can be written after differentiating equation (3): t2 - t1 A Z(Xn) # t = ~~~~~~~ . ~~~~~~~ (5) z(xn) z(xn) As stated, the relative change # z(xn) / z(xn) is small in dependence on xn and, as follows from equation (4), thus leads to an error at, which is negligible in practical terms, in the range of x = 0 to 0.6, which is of interest for measurement technology.Since the difference t2 - t is reduced both by z(xn) and by the relative deviation, taking account of equation (4) an- outstanding curvature compensation is likewise possible in the less important range of xn > 0.6, since the product of the two quotients determines the residual error of second order, as may be seen from equation (5).

The characteristics shown in Figure 3 are based on a diameter of the inner coil 18 of approximately 1.5 mm and of the outer coil 36 of approximately 3 mm. Layer thicknesses of up to 1500 zm can easily be measured with curvature compensation using such a probe 11. If the system is enlarged in a linear fashion, the measuring range is also correspondingly larger. However, the extent to which the curvature can be compensated for is correspondingly smaller. Likewise the measuring sensitivity for thin layers.

It is clear from Figures 5 and 6, which relate to the characteristics in Figure 3, that splendid results are obtained using the arrangement described in accordance with the invention. With the present state of digital technology, it is easily possible by making use of microprocessors to process the mathematical algorithm in fractions of a second, so that the formation of the measured value is not encumbered by additional algorithms. The arrangement according to the invention comes very close to the ideal of a curvature-independent measurement after calibration on the plane. Since, in most cases, the measured object is present only in coated form, and calibration is easily possible on a plane reference material for checking purposes, the user provided with a substantial simplification of the method of measurement.

The desired layer thickness, ie. the actual layer thickness, is given in pm on the left in Figure 5. If only the coil 18 were to hand, layer thicknesses of 79.3, 103.2 etc pm would be measured. Measured values would be indicated which were larger by the factor 3...2 than the true measured value, referred to a cylinder of diameter 6 mm. The next column represents the layer thickness of 26.4, 49.5 etc. which can be determined by the invention.

The difference between the corrected layer thickness and the desired layer thickness in pm is represented in the column at far right. As is evident, the invention yields a virtually complete compensation for the influence of the curvature. While Figure 5 shows the case in which a measured object with a constant diameter, that is to say a diameter of 6 mm, is coated with different layer thicknesses, Figure 6 shows the variants of different diameters in pm with a coating of the same unchanged thickness of 49 pm. Here, too, it is evident that substantial improvement can be achieved by the invention.

Figure 4 shows a flow diagram for calculating the corrected layer thickness to from the measured values xl and x2 of the coils 18 and 36. The two boxes drawn with round corners signify program memories in which are stored - left - the measured value x0l for the plane, uncoated measured object and the measured value for the saturation layer thickness x5l, specifically for the coil 18. The same is given on the right for the coil 36. This computational formula is given in the description. xnl and are thus obtained, so that the measured variable can vary only between 0 and 1, as shown in Figure 3.The normalized measured variable Xni is now combined on the left with the parameters of the coil 1, that is to say the response of the curve 56 is located in this program memory. The combination yields the layer thickness tl, which is also to be found in Figure 3.

The same takes place analogously on the right, i.e. the response of the curve 58 is stored in the program memory "Parameter Coil 2". The difference tl - t2 is then formed on the right. The parameter for z is in a program memory at the same level on the left. This parameter is, after all, largely constant, eg. equal to the number 12 and has a noticeable deviation only in the upper region of the normalized measured variable. The further combinations can easily be understood with reference to the formulae. As is evident from equation (3), it is still necessary to combine t1 with the quotients written therein, so that the corrected layer thickness tc is obtained, which is equal to t'in equation (3) and is also to be found in Figures 5 and 6 in the second column from the right.

The rectangular boxes in Figure 4 signify measured values or results of computations. Boxes with round corners signify parameters which are stored in the program memory. These parameters have either been preset to fixed values or have been determined at an earlier time by measurement. Circles denote a computational operation.

The slight correction of the computational factor z () can be performed both as a function of Xni (as illustrated in Figure 3) and as a function of x < . Since the layer thickness and, in particular, the corrected layer thickness are a function of , it is expedient to perform a correction of z(x=) as a function of xul.

The invention is capable of numerous variations.

The computation of the corrected layer thickness tcorr need not necessarily be performed via a normalization. Rather, the curves can be plotted without being normalized and then computed later, and this is not a problem with present day computing devices. The advantage with normalization resides in that it is possible to perform a calibration on a plane non-coated measured object that need be checked only at relatively long time intervals.

Claims (30)

Claims:

1. A method for the non-destructive measurement of thin layers using a probe, a coil device being wound nondisplaceably onto an inner core, the geometrical center of the inner core and the geometrical center of the coil device coinciding, the coil device being led outwards by a line device and an evaluation circuit for computing the layer thickness from output values being connected to the line device, the evaluation circuit determining the line thickness, having the following characteristic features: a) The probe has at least two different coil devices, which each transmit different output values for the same measurement problem.

b) The output values occurring during measurement of the two coil devices are resolved for the layer thickness t in a computing device.

2. The method as claimed in claim 1, wherein the output variable of each coil device is of the type of an equation having two unknowns.

3. The method as claimed in claim 2, wherein the output variable of one coil device is of the type tl = A1 + t1 being the measured value, A1 a constant and t the desired layer thickness, and the output variable of the other coil device being of the type t2 = A2 + t, t2 being its measured variable and a2 its constant.

4. The method as claimed in claim 3, wherein the following is set 1 = c and A2 = c =cz(rh)r z (x), z being a constant and xE a normalized measured variable of the respective coil device.

5. Method as claimed in claim 1, wherein the characteristics of the measured variable/layer thickness for a plane measurement problem are stored in the computing device.

6. Method as claimed in claim 5, wherein the characteristics are stored in a form in which the measured variable is normalized.

7. Method as claimed in claim 6, wherein the measured variable is normalized in a range between 0 and 1 (0 and 1000.

8. Method as claimed in claim 1, wherein the two coil devices have different mean diameters.

9. The device for carrying out the method as claimed in claim 1, wherein it has two different coil devices which see the thin layer and the metal layer situated under it in different ways.

10. The device as claimed in claim 9, wherein the coil devices comprise wire wound coils.

11. The device as claimed in claim 9, wherein the coil devices are super high frequency components, which generate focussed magnetic fields.

12. The device as claimed in claim 9, wherein the coil devices are arranged concentrically with the geometrical central axis of the probe.

13. The device as claimed in claim 9, wherein the coil devices are circular.

14. The device as claimed in claim 10, wherein the coil bodies have a constant elongated rectangular crosssection.

15. The device as claimed in claim 9, wherein, in the case of low-frequency magneto-inductive operation, one coil device is seated on a core of high permeability iron.

16. The device as claimed in claim 15, wherein the iron is silicon/iron.

17. The device as claimed in claim 9, wherein, in the case of high frequency operation using the eddy-current principle, one coil device is seated on a core of ferritic material.

18. The device as claimed in claim 9, wherein, in the case of super high frequency operation, one coil device is seated on a non-conductive material (ceramic, plastic).

19. The device as claimed in one or more of claims 9 to 18, wherein, in the case of low-frequency operation, a shield is provided between the two coil devices, which consists of high permeability iron, in particular silicon/iron.

20. The device as claimed in claim 19, wherein, in the case of high-frequency operation, the shield is of a ferritic material.

21. The device as claimed in claim 19, wherein, in the case of super high frequency operation, the shield is of a highly conductive metal (copper, silver).

22. The device as claimed in claim 21, wherein the shield is displaceable.

23. The device as claimed in one or more of claims 19 to 22, wherein the shield is an annular wall which is concentric with the geometrical center of the probe.

24. The device as claimed in claim 9, wherein a shielding sleeve covering the outer coil device is provided.

25. The device as claimed in claim 9, wherein, for adjusting purposes, at least one part of the probe can be displaced relative to other parts of the probe.

26. The device as claimed in claim 9, wherein at least one inner core and one outer sleeve are in one piece, and the monitoring locating spaces for the coil arrangement are produced from solid material.

27. A device for measuring the thickness of a thin layer comprising a probe provided with at least two coil devices of different characteristics for providing different output values for the layer thickness and means for providing a measurement of said thickness from said output values.

28. A method of measuring the thickness of a thin layer comprising positioning at least two coil devices of different characteristics relative to said layer, obtaining respective outputs from said coil devices and using said outputs to provide a measurement of said thickness.

29. A device for measuring the thickness of a thin layer substantially as herein described with reference to Fig. 1 or Fig. 2 of the accompanying drawings.

30. A method for measuring the thickness of a thin layer substantially as herein described with reference to the accompanying drawings.