DE102005031216A1 - Electronic component surface shape recording procedure uses white light interferometer with in pixel processing using incoherent demodulation - Google Patents

Abstract

Electronic component surface shape recording procedure uses a white light interferometer (15) to create intensity modulation by movement of component (11) and a reference mirror (9), creates the position or speed linked envelope using analogue signal processing with incoherent demodulation for each pixel, determines the in pixel (19) maximum position, stores and reads this using classical camera techniques. Independent claims are included for equipment using the procedure.

Description

The The invention relates to a method and an arrangement for fast Capturing the surface shape of objects, in particular of electronic components during the Assembly process, using a white light interferometer and an image sensor as a detector.

The growing efficiency and integration density in semiconductor technology has and connection technology a miniaturization with ever smaller Grid spacing and Connection dimensions result. Ever tighter tolerances regarding the dimensional stability like twisting, warping and coplanarity, will be supported. This mainly affects the manufacture and Assembly of the most modern components, the so-called "Advanced Packages "Advanced Packages are precision components for the electrical, optical and mechanical coupling of highly integrated semiconductors, Microsystems and nanostructures to the environment. Will be a single Connection mounted incorrectly, the assembly, on the most part Several Advanced Packages are mounted, a complex repair process be subjected or even sorted out.

at the assembly of components in electronics production play mold tolerances a crucial role. To achieve robust manufacturing processes and a reliable one Contacting the connections to ensure is an active control by real-time measurement and evaluation the joining partner while necessary for the assembly process. Geometric deviations can do so be actively compensated within certain limits. To the exact one Positioning surface mountable Components on a wiring carrier today is a lateral Position determination of the gripped by the automatic handling device usual for the carrier.

Next the precise one lateral positioning, requires the installation of the most modern Additional components the possibility a vertical measurement (measurement of the height profile) to control the positioning and assessment of dimensional stability.

Of the current State of the art allows the use of optical 3D measuring methods (method under the use from light) for the shape detection. Compared to radiographic methods with X-rays These methods are less expensive. In the last two For decades, many new measurement techniques have been explored. There is more than thirty different optical sensor types with different properties. Through the advances in the development of digital camera technology as well as the use of modern laser systems make optical, area Measuring method increasingly for the 3D measurement of technical surfaces, i.a. also for Measurements under industrial or "harsh" environmental conditions (A. Koch, "Optomechatronic Measuring Systems ", Scriptum Chair for Measuring System and Sensor Technology, Technische Universität München, 2003). The white light interferometry can be as one-dimensional or two-dimensional scanning, optical 3D method can be realized.

White light interferometer are available in different versions. While For a long time, only microscope systems were available telecentric systems with larger measuring fields offered. The two variants differ in the lateral Resolution, the measuring field size and the vertical travel path. While Interferometer with microscopic structure a lateral resolution of 1 μm and smaller, they have a vertical measuring range of only a few millimeters or less. In addition there is the rather small measuring field size in the millimeter range. For capturing larger objects have to therefore individual measurements are lined up (gestichcht). at Interferometers with telecentric structure become typical lateral resolutions between 10 μm and 50 μm and measuring field sizes in the centimeter range reached. The vertical travel is several centimeters.

Together Both bodies have a very high vertical resolution. conditioned through the short interference length the light source are resolutions typical of a few nanometers for white light interferometers (J. Klattenhoff, "White Light Interferometry", Guide to Basics and Applications of 3-D Optical Measurement, Erlangen 2003, p. 15-16, Fraunhofer Alliance Vision, Erlangen).

This in 1 (Prior art) illustrated white light interferometer is an interferometric measuring method according to the principle of the Michelson interferometer. Unlike phase-measuring interferometry, the interference contrast is evaluated here. Measuring methods which operate on this principle are known by the names "coherence radar", "coherence microscopy" or "correlation microscopy".

The white light interferometer consists of a measuring arm 1 and a reference arm 2 , The required mirror in the measuring arm is replaced by the object itself. At the beam splitter 3 shares that of a light source 4 incoming light beam in a reference and a measuring beam. The reference beam is at a reference object, eg a mirror 5 , in the beam splitter 3 thrown back and thus goes through a defined route. The measurement ray is on the object 6 scattered. Part of the light is thrown back into the measuring arm and at the beam splitter 3 merged with the reference beam. In the detector area 7 then overlap both partial beams.

If the path difference between reference and measuring beam is smaller than the coherence length of the light source, interference phenomena can be observed. In order to keep this path difference as small as possible, radiation with the shortest possible coherence length and thus relatively large spectral width, ie white light, is used. If you move the reference mirror 5 or the object 6 in the z direction, an interference phenomenon in the form of an intensity modulation (formula 1) as a temporal function f (t) or f (z) can be observed at the detector whose maximum lies exactly at the point of the same optical paths of reference and measuring beam ,

Do you wear the intensity modulation dependent on from the location of the reference or the object, one obtains a so-called correlogram or a visibility curve. The height information of the object is included in the correlogram. If you determine the location of the maximum Modulation depth, you get the desired height or distance information.

The maximum modulation depth of the correlogram can vary with different Evaluation algorithms are determined, a simple and real-time method is the envelope modulation. By rectifying the correlogram according to the method of incoherent demodulation or envelope modulation can first the envelope of the correlogram and then evaluated their maximum.

In order to measure several measuring points or even the entire object surface with one stroke of the reference mirror or of the object, many detectors are arranged next to one another. The surface is laterally subdivided into image areas, which are measured and evaluated separately during the stroke, but simultaneously. For this purpose, planar image sensors such as CCD cameras are used. The camera captures a sequence of images during a stroke (this is also known as image stacks). Because pictures can only be taken with this technique at discrete times, the location is also scanned discretely. To determine the interference contrast in the correlogram, a complete scanning of the carrier signal is necessary. Since the period length in the correlogram is λ / 2, z a ≤ λ / 4 (1) must according to the sampling theorem the measured values in a max. Distance from z _{a be} detected (1). at ż ≤ z a · f a (2) given frame rate f _{a of} the camera, the speed ż of the reference mirror according to (2) is limited. If, for example, a CCD camera is used as the detector, which scans according to the television standard at 25 Hz / frame, the maximum speed is about 5 μm / s.

• – Aufzeichnen mit der hohen Abtastfrequenz einer Hochgeschwindigkeitskamera,
• – Synchrone Unterabtastung (Sub-Nyquist Korrelogramm),
• – Synchrone Modulation der Lichtquelle bzw. der Kamera,
• – Korelogrammaufzeichnung mit nicht linearer Signalaufnahme und Mittelwertbildung.

With the conventional technique, only a very slow recording of the correlogram by precise positioning in the nanometer range or a quasi-static method of reference is possible. For quick but extensive recording of the correlogram various approaches are known:

• Recording at the high sampling frequency of a high-speed camera,
• Synchronous sub-sampling (Sub-Nyquist Correlogram),
• - Synchronous modulation of the light source or the camera,
• - Korelogram recording with non-linear signal acquisition and averaging.

at the complete Scanning the correlogram with a high-speed camera must be enormous Data volumes in the shortest possible time Cached or evaluated time. Even high frame rates limit the scanning speed significantly. With growing lateral resolution the camera sinks the possible Sampling frequencies drastically. Because of the greatly reduced integration time to collect photons high light intensities be present in the detector area. The flat illumination of the whole Field of view with high light intensity is to be adjusted by the Apertures, in particular for resolution of speckles, limited.

By defined synchronous subsampling can also be the maximum of Modulation depth are determined. This procedure allows a significantly reduced sampling rate. This can make the object faster be moved continuously. The method is very sensitive against smallest phase shifts of the sample with respect to the interference modulation and thus very prone to failure synchronous method of reference. A detailed description of the procedure is found in P. de Groot, L. Deck, "Three-dimensional imaging by sub-Nyquist sampling of the white-light interferograms ", Optics Letters / Vol. No. 18 17, 01.09.1993.

By synchronous modulation of the light source for intensity modulation of the correlogram can be obtained together with image sensors having a low-pass behavior, directly the envelope of the interference modulation, from which the maximum of the modulation depth can be easily determined can. With this method, the sampling rate can be significantly reduced and the object can be moved faster faster. Alternatively, the incident on the detector light intensity or the sensitivity of the detector can be modulated. This method, like the synchronous subsampling method, is sensitive to phase shifts between light and interference modulation.

Accurate tuning of the modulated light source with the displacement unit is thus required. A detailed description of the method can be found in the DE 19 528 513 A1 ,

extends the integration time in the correlation recording with a Camera over more than half a period of the carrier frequency, that disappears modulated coherence signal because of the integral behavior of the camera. By non-linear recording the intensity modulation with corresponding detector elements (e.g., logarithmic sensitivity) this results in a mean value deviating from the basic intensity. The sampling frequency will no longer be determined by the carrier, but through the envelope certainly. A description of the method can be found in (T. Seifert, "Schnelle Signal preprocessing in white light interferometry nonlinear signal recording "DgaO-Proceedings 2004).

Of the Invention is based on the object, a method and an arrangement for rapid detection of the surface shape of electronic components with in-situ measurements, with which the disadvantages of Prior art can be avoided.

The The object underlying the invention is characterized by the characterizing Features of method claim 1 and independent claim 11 solved. Further embodiments of the invention will become apparent from the respective dependent claims.

core the invention is that contrary to the usual application of image sensors, where individual intensity values reproduced according to the invention in each pixel a separate evaluation of the interference contrast itself is performed. Through an analog signal preprocessing can be on a discrete Scanning be waived. By comparison with a reference signal for path coding (distance coding) can in each pixel the Distance to the object to be saved.

By the type of information processing results in an image sensor with completely new properties. The structure of each pixel allows one very fast evaluation of the intensity modulations. The image sensor allows at areal (laterally extended scan) a direct evaluation of the correlograms and thus a very fast detection of the surface shape of the object, for example the surface of a semiconductor component.

Of the Reference mirror or the object can be at much higher speeds be moved in the multi-digit mm / s range while using the conventional Technique only speeds of fractions of millimeters per Second can be achieved. Since the caching of a picture stack is omitted can, a huge reduction of the data to be recorded is achieved. Unlike traditional ones Cameras are the value of the pixel not the light intensity, but around the distance information to the associated object point. The "picture" of the sensor is thus already the height map the object surface itself. An exact synchronization to the interference modulation can omitted. The new method is thus robust to phase shifts, whereupon other methods are very sensitive.

following the invention is intended to exemplary embodiments be explained in more detail. In the associated Drawings show:

1 : Schematic representation of a white light interferometer (prior art);

2 a schematic representation of an arrangement for the rapid evaluation of interference modulations in an image sensor;

• a) klassischer Methode und
• b) im Bildsensor integrierte Korrelogrammauswertung;

3 : a comparison of the signal processing structure between

• a) classical method and
• b) Correlogram evaluation integrated in the image sensor;

4 : an integration of the sensor in a white light interferometer;

5 : an example of pixel-internal signal preprocessing for obtaining the coherence information S _k (t) in the form of an envelope;

6 : Evaluating the Correlogram Envelope with Simple Threshold Operation;

7 : a signal evaluation with simple thresholding operation;

8th : the processing of the coherence signal for the evaluation with threshold hysteresis;

9 : the evaluation with threshold hysteresis using the speed signal; and

10 : the signal evaluation with threshold hysteresis using a path signal.

In the 2 illustrated arrangement similar to those in 1 shown arrangement works), consists of a reference arm 8th with associated reference mirror 9 and a measuring arm 10 at which the measurement object 11 is arranged, as well as one at the intersection of reference arm 8th and measuring arm 10 arranged beam splitter 12 , Opposite the reference mirror 9 there is a detector 13 , eg a CMOS image sensor, with associated optics 14 , as well as a Michelson interferometer 15 , Furthermore, opposite the object 11 a white light source 16 arranged.

For path measurement of the reference mirror 9 is this on a manipulator 17 with associated incremental scale 18 assembled. Instead of path measurement with incremental scale 18 can also do a triangulation measurement with a triangulation measuring device 22 , For example, a PSD element, interferometric, capacitive, or be implemented in other suitable ways.

In order to avoid the high sampling rates required per se, a discrete sampling is performed on the detector 13 picture omitted. Every single pixel of the detector 13 therefore evaluates the incoming interference modulations themselves. The interference signal is detected by the image sensor and preprocessed analogously in each pixel, so that the respective envelope is present. The time at which the max. Modulation depth and thus the max. Interference contrast is observed, for example, by a difference and threshold operation can be mrkiert. That way, every pixel gets it 19 individually generates a correlation signal to mark the maximum. The value of a pixel 19 is determined by the correlation signal and a path signal and in a memory element of the pixel 19 stored. The path signal is equally accessible to all pixels and is fed in centrally. The path signal represents the current location or the current speed of the reference mirror 9 or the object 11 , It can be fed from the outside, or on the detector 13 be generated.

To generate a signal on the detector 13 becomes the Michelson interferometer 15 used, which can be easily integrated into the white light interferometer. With the Michelson interferometer 15 becomes the distance of the reference mirror 9 measured. Because of this, are reference arm 8th and measuring arm 10 switched to white light interferometer. Part of the white light and a coherent light beam (laser beam) of a laser 20 be object-side to a fixed mirror 21 reflected at the edge of the field of view of the CMOS sensor 19 is arranged. About the beam splitter 12 the beams are on photodetectors at the edge of the detector 13 displayed. The interference modulation of the coherent light beam can be used for displacement measurement of the reference mirror 9 and the maximum of the intensity modulation of the white light source 16 be used as a reference signal for an absolute displacement measurement.

Depending on the type of path encoding, the memory element of each pixel of the CMOS sensor 19 be executed as a counter, bit memory or as a latch. The design as a holding member requires a low circuit complexity. The "image" is then read in classic camera technology via central A / D converters.

Integrated correlation analysis on flat sensors:
In order to achieve the required high sampling rates, the discrete scanning of entire images and output of the correlograms in the form of a picture stack should be dispensed with. It is more advantageous to evaluate the correlograms on the sensor itself. In each individual pixel of an image sensor is a signal of the photodetector available that can be sampled at a relatively high rate. However, the effective sampling rate of a single pixel in cameras is limited by the time it takes to transmit an entire image. Due to the limited bandwidth of the transmission path, the sampling rates of images are significantly lower than those of a pixel. This becomes clear with the example of modern CMOS cameras. These provide the ability to turn off some of the existing pixels. The freed-up bandwidth is used to transfer the smaller images faster. Because the evaluation of the correlograms in the case of planar scanning with a camera has hitherto always been carried out behind the transmission path, a substantial increase in the measuring speed can be achieved by shifting the correlationogram evaluation before the transmission path into the sensor itself (see 3 ). This change in structure creates an image sensor with completely new properties. Each individual pixel of the sensor detects and evaluates the correlogram itself.

The information of the speed or the location of the reference must therefore be equally available to each pixel. 4 shows the integration of the sensor in a white light interfero meter. In the pixel itself, the interference signal is detected by the photodetector. The type of signal preprocessing depends on the subsequent evaluation method. In general, however, should the coherence information S _k (t) in the form of the envelope of the correlogram such as in 6a be won. This can generally be done by signal preprocessing 5 be achieved. Due to the limited dimensions of a pixel (even taking into account the filling factor) but only very simple circuits can be realized within the pixel structure. For example, simple analog filters can be constructed from a few components for signal preprocessing. Also for subsequent pixel-internal evaluation and storage of the distance information only those methods can be implemented that can be realized with little circuit complexity.

Signal evaluation with symmetrical threshold edges:
To further reduce samples, a threshold operation may be applied to the signal of the envelope. Due to the symmetrical position of the Gaussian function to the object surface, the edges of the threshold signal S ₁ (t) are also symmetrical to the object surface ( 6b ).

man die Distanz Δz von dieser Referenzposition zur Objektoberfläche (18). An der Objektoberfläche selber steht jedoch kein Signal zum Beenden der Integration zur Verfügung. Wird das Signal S_ż(t)

zwischen den Schwellwertflanken t₁ und t₂ mit einem Teiler halbiert, entspricht das integrierte Signal genau dem Ort z₀ weil auch die Objektoberfläche (bzw. der Zeitpunkt t₀ an dem z₀ erreicht wird) in der Mitte zwischen den Schwellwertflanken t₁ und t₂ liegt. Dieses Verhalten gilt jedoch nur für konstante Geschwindigkeiten und ist in 6c und 6d für eine konstante Referenzgeschwindigkeit k dargestellt.The measurement and storage of the surface distance may be, for example, as in 7 be realized by an integrator. Integrating a signal S _ż (t) proportional to the speed ż of the reference from a reference position z _ref to the object surface z ₀ is obtained

the distance Δz from this reference position to the object surface ( 18 ). On the object surface itself, however, no signal is available for terminating the integration. The signal S _z (t)

halved between the threshold edges t ₁ and t ₂ with a divider, the integrated signal corresponds exactly to the location z ₀ because the object surface (or the time t _{0 is reached} at the z ₀ ) in the middle between the threshold edges t ₁ and t ₂ lies. However, this behavior only applies to constant speeds and is in 6c and 6d for a constant reference speed k.

For the analog implementation of the threshold value concept, the signal S _ż (t) can be represented, for example, by a voltage u _v (t). A voltage divider bisects the signal. The voltages are supplied via a threshold-controlled switch to the integrator, which also serves as a storage element.

In the digital implementation, the speed-proportional signal S _ż (t) may be represented by a frequency. The divider can then consist of a simple binary divider. A threshold controlled switch connects the signals to a counter which also serves as a storage element.

Signal evaluation with centric threshold edge:
In order to obtain a signal which marks precisely the time at which the reference passes through the object surface, the coherence signal can be differentiated. A zero crossing is then obtained exactly at the surface of the object ( 8a and 8b ). A thresholding operation with hysteresis (Schmitt trigger) can be used to generate a digital signal that has a digital edge at exactly t ₀ = t ₂ . If you replace the simple thresholding operation with this method, you can integrate exactly until t ₀ ( 9 ), and the integration process described there can be simplified. If the location is used directly instead of the speed signal, the integration element can be replaced by a holding element ( 10 ). The holding member scans at time t ₀ from the signal S _ż (t) directly from the associated location z ₀ and holds the distance until further processing. If the path information is present in digital form at the pixel, the holding element can be replaced by a memory into which the corresponding location is written.

1

measuring arm

2

reference arm

3

beamsplitter

4

light source

5

mirror

6

object

7

detector region

8th

reference arm

9

reference mirror

10

measuring arm

11

object

12

beamsplitter

13

detector

14

optics

15

Michelson interferometer

16

White light source

17

manipulator

18

incremental scale

19

CMOS pixel

20

laser

21

mirror

22

Triangulationsmess

Facility

Claims (13)

Method for rapid detection of the surface shape of objects, in particular for in-situ measurement of electronic components during production, using an interferometer and a pixel-containing image sensor as a detector, characterized by illuminating the object ( 11 ) and a reference mirror ( 9 ) with white light, - generating an interference phenomenon in the form of an intensity modulation by superposition of a reference beam with a measuring beam in an interferometer ( 15 ) by moving the object ( 11 ) or the reference mirror ( 9 ) of the interferometer ( 15 ), - parallel detection of the interference modulation together with a position or velocity information of the referential mirror ( 9 ) or of the objet ( 11 ) as a path signal in each pixel of the image sensor, - generating an envelope of the interfering signal by analog signal preprocessing by incoherent demodulation (envelope demodulation) in each pixel, - determining the maximum of the envelope with position signal and velocity signal by analog or digitally by evaluating the threshold edges, - generating the position in the pixel, - analog or digital storage of the position in the pixel and - readout of the stored information according to classical camera technology via central A / D converter with analog position storage. Method according to claim 1, characterized in that the evaluation takes place with symmetrical threshold edges. Method according to claim 2, characterized in that that the evaluation is done with centric threshold edges. Method according to Claims 1 to 3, characterized that the generation of the position in the pixel by means of evaluation of the symmetrical threshold edges with a path signal, which the Reference mirror speed represents. Method according to one of claims 1 to 4, characterized that part of the white light and a coherent one Light beam on a stationary mirror at the edge of the field of view reflected and imaged on the edge of the detector. Method according to Claims 1 and 5, characterized that the interference modulation of the coherent light beam for displacement measurement of the Reference mirror and the maximum of the intensity modulation of the white light source as a reference signal for the absolute displacement measurement is used. Process according to claims 1 to 6, characterized that the path signal is supplied centrally to all pixels simultaneously. Process according to claims 1 to 7, characterized by lateral division of the surface of the object into image areas, which are separately measured and evaluated simultaneously at the hub. Method according to one of claims 1 to 8, characterized that the signal processing is only partially performed in the pixel and a shift of the further signal processing to the outside is made in separate areas of the chip. Method according to claim 9, characterized that the signals for the displacement of the signal processing is sampled and / or buffered become. Arrangement for carrying out the method according to claims 1 to 10, characterized in that a reference arm ( 8th ) with associated reference mirror ( 9 ) and a measuring arm ( 10 ) on which the measurement object 11 is arranged, wherein the reference mirror ( 9 ) on a manipulator ( 17 ) with associated incremental scale ( 18 ) is mounted at the intersection of reference arm ( 8th ) and measuring arm ( 10 ) a beam splitter 12 is arranged that opposite to the reference mirror ( 9 ) a detector ( 13 ), with associated optics ( 14 ) with an upstream Michelson interferometer ( 15 ) and that opposite the object ( 11 ) a white light source ( 16 ) is arranged. Arrangement according to claim 11, characterized in that the detector ( 13 ) is a CMOS image sensor. Arrangement according to claim 12, characterized in that each pixel of the CMOS sensor ( 19 ) is designed as a counter, as a bit memory or as a holding member.