Patent Application
20130027543
The present invention relates to a method for inspection of light emitting semiconductor devices during and after a production process. The light emitting semiconductor devices can be LEDs.
The present invention also relates to an apparatus for inspection of light emitting semiconductor devices on a substrate.
Solid state lighting (SSL) has several advantages compared to conventional lighting. Main advantages are low power consumption, long lifetime, and small form factor. An important element of SSL is the LED (Light-Emitting Diode) die/chip. Basis for LEDs is a semiconductor material that is undergoing a complex production process in order to obtain a LED. Several metrology and inspection steps are done during and after that production process.
Measuring the output power of an LED is typically done using a probing system. In this system, electrical contacts are made to each LED die and a measurement is done of a.o. the generated light output power and optionally the wavelength.
The international patent application WO98/11425 discloses a method and apparatus for detecting defects in a semiconductor or silicon structure at room temperature, and in an efficient time, using photoluminescence. The invention employs the use of a high intensity beam of light preferably having a spot size between 0.1 mm-0.5 μm and a peak or average power density of 104-109 W/cm2 with a view to generating a high concentration of charge carriers, which charge characters detect defects in a semiconductor by interacting with same. These defects are visible by producing a photoluminescence image of the semiconductor. Several wavelenghts may be selected to identify defects at a selective depth as well as confocal optics are used. This method performs probing of a very small volume of the material with one or more laser beams having very small spot size.
Another method is described in U.S. Pat. No. 7,504,642 B2 where one or more images are created using filtering and image computation to selectively create a defect image of one selected layer of a wafer, while trying to eliminate unwanted contributions of other layers of the same wafer. The method uses photoluminescence to identify defects in one or more specified material layers of a sample. One or more filtering elements are used to filter out predetermined wavelengths of return light emitted from a sample. The predetermined wavelengths are selected such that only return light emitted from one or more specified material layers of the sample is detected. Additionally or alternatively, the wavelength of incident light directed into the sample may be selected to penetrate the sample to a given depth, or to excite only one or more selected material layers in the sample. Accordingly, defect data characteristic of primarily only the one or more specified material layers is generated.
The international patent application WO 2007/128060 A1 describes a photoluminescence based method for testing of indirect bandgap (e.g. Si) semiconductor materials, based on comparison of several regions in two or more images. The method is suitable for identifying or determining spatially resolved properties in indirect bandgap semiconductor devices such as solar cells. In one embodiment, spatially resolved properties of an indirect bandgap semiconductor device are determined by externally exciting the indirect bandgap semiconductor device to cause the indirect bandgap semiconductor device to emit luminescence, capturing images of luminescence emitted from the indirect bandgap semiconductor device in response to the external excitation, and determining spatially resolved properties of the indirect bandgap semiconductor device based on a comparison of relative intensities of regions in one or more of the luminescence images.
Quality control of LED is becoming more and more crucial since LEDs are used for illumination. It is important that, e.g., the LEDs used for the back illumination of a TV set are of similar intensities. Therefore a quality control of the light output power of LEDs needs to be done. Such quality control was until now done by electrically contacting the LED (probing) and then measuring the emitted light output power. This has several disadvantages: LEDs may get damaged during probing, probing is slow and requires an additional tool.
It is an object of the invention to provide a method for quickly and reliably measure the light power emitted by an LED during a production process. Furthermore, the method should be easy to use and should not influence or destroy the LED to be measured.
The object is achieved by a method for inspection of light emitting semiconductor devices, comprising the following steps:
It is a further object of the invention to provide an apparatus the light power emitted by an LED during a production process. Furthermore, the apparatus should be easy to use and should not influence or destroy the LED to be measured.
The object is achieved by an apparatus for inspection of light emitting semiconductor devices on a substrate, comprising:
For defect inspection the photoluminescence effect is used as a kind of back light. This illumination effect enables the finding of defects that are buried or at least are not visible in normal inspection. With the inventive setup it is as well possible to find cuts (finger cuts) or interrupts in the metallization layer. Additionally, the invention allows the detection of non-homogeneities of the LEDs. The light emitted from the LED is subjected to a spatial grey value analysis. Some LEDs only emit light in some parts while no light is emitted in other parts (for example: a dark edge at the corner of the LED does not light up).
According to an embodiment of the invention the light emitting semiconductor devices are illuminated with a light source configured as a ring light. The ring light has a plurality of LEDs. A second filter can be positioned in the detection beam path. The second filter in the detection beam path prohibits the reflections of the incident light to reach the sensor and at least a wavelength of λC passes the second filter.
A further embodiment of the invention is that a first filter is positioned in an illumination beam path of the light source and is designed to pass a waveband λA+λB. The second filter is in the detection beam path and prohibits the reflections of the incident light to reach the sensor and at least a wavelength of λC passes the second filter. In this case the objective lens defines the illumination beam path and the detection beam path. The light source is a coaxial light source.
One or several LED die/chips are illuminated with a light source with wavelengths that can generate electron hole pairs in the LED. The light emitted by the LED (caused by the electron hole pair and following recombination process) is captured with a sensor/camera that is sensitive to the wavelengths of the emitted light. The sensor response (gray value) is a measure for the power of the light output of the LED and can for example be used to classify the LEDs according to their light output power.
The waveband of light λA±λB for the illumination of light emitting semiconductor devices or the LED die/chips is generated by inserting a first filter prior to the objective lens in an illumination beam path. A second filter is positioned in a detection beam path after the objective lens, so that only the light emitted by the light emitting semiconductor reaches the sensor of the camera. The image acquisition setup, especially the objective lens comprises microscope optics. Various types of illumination can be used in the apparatus for the illumination of light emitting semiconductor devices or the LED die/chips. The light source could be a coaxial light source or a ring light. The illumination light is provided with a plurality of LEDs.
The inventive method is applied to LED die/chips, which are structures on a substrate or wafer. The inspection result is then a measure for a light output power of a LED or the LED die/chips detected by the sensor of the camera. The output of the sensor is at least one gray value of a matrix of pixels. A range of the gray value establishes function of the light output power per LED in the LED die/chips or in the light emitting semiconductor devices.
The inventive apparatus has a stage, which moves the substrate with the LED die/chips in a X/Y direction. The movement is controlled by the computer system. With the relative movement between the camera and the substrate, the sensor of the camera captures an image of the entire surface of the substrate. The data from the sensor is sent to the computer system which calculates a wafermap of the surface with the LED die/chips. The wafermap is shown on a display of the computer system, wherein to each class of gray value a separate color code is assigned.
The function of the light output power per LED is implemented as a look up table. A further embodiment is that the function is implemented as a polynomial. A calibration of look up table or the polynomial is done by measuring the light output power of an LED sample by connecting it to an electrical prober.
The inspection result generated by the sensor is at least one gray value per LED. The inspection results of the LEDs are sorted in at least two bins according to their registered gray value. In a further form, the inspection result generated by the sensor is at least two gray values per LED die/chip. The variations or differences in gray value of one LED die/chip are used as a quality measure of the LED die/chip.
The inspection result is at least one gray value per LED die/chip and the method comprises the steps of:
As mentioned before, the light emitting semiconductor devices are LED die/chips and the emitted light of the LED is caused by a recombination process of electron hole pairs that are generated by the illumination in an active layer of the LED. The emitted wavelength or waveband has a similar wavelength or waveband as if a forward voltage would be applied to the LED.
Due to fluctuations in the production process, the LED chips are sorted according to several criteria including center wavelength of the emitted light, power of the emitted light, etc.
This invention would allow for fast and contactless inspection on an inspection tool which is widely used by LED manufactures for other inspection tasks.
The nature and mode of operation of the present invention will now be more fully described in the following detailed description of the invention taken with the accompanying drawing figures, in which
a is an image of the surface of a wafer with LED die/chips which are illuminated with normal illumination;
b is an image of the InGaN—layer below the surface of the LED die/chips on a wafer which is illuminated with photo luminescence illumination; and
Same reference numerals refer to same elements throughout the various figures. Furthermore, only reference numerals necessary for the description of the respective figure are shown in the figures. The shown embodiments represent only examples of how the invention can be carried out. This should not be regarded as limiting the invention. The description below refers to LED die/chips which should not be regarded as a limitation of the invention. It is evident for any person skilled in the art that the present invention is applicable to light emitting semiconductor material in general.
The light source 7 is a white-light broadband spectrum light source, is used for illuminating the substrate 3 with the LED die/chips 5. The light from the light source 7 supplied to a microscope 6 via a light guide 8. The microscope 6 defines an illumination beam path 11. A beam splitter 12 directs the illumination beam path 11 via an objective lens 14 onto the LED die/chips 5 on the substrate 3. By providing means (not shown) for inserting a respective first filter 15 in the illumination beam path 11, a certain part of the broadband spectrum that is generated by a light source 7, is selected. The light is transmitted through the objective lens 14 (incident light beam) excites the semiconductor material in the LED die/chips 5 on the substrate 3. This could be for example a direct band gap material such as a III-V semiconductor material as used for LED fabrication. The semiconductor material will emit light at a known wavelength and this light is collected in the same objective lens 14. The objective lens 14 defines as well a detection beam path 21.
In the detection beam path 21 a second filter 16 is positionable in order to make sure that only the light emitted by the LED die/chips 5 on the substrate 3 reach the camera 9 and the sensor 10. The second filter 16 prohibits the reflections of the incident light to reach the camera 9 or the sensor 10. The image data collected by the sensor 10 of camera 9 are fed to a computer system 17 which uses an image processing software to derive an average intensity for each LED 4 on the substrate 3. The computer system 17 computes a wafermap (see
The light source 7 is coaxial light source. It is advantageous if the light source 7 is ring light. The illumination light is provided by a plurality of LEDs. The light source 7 is configured as a pulsed light source or a continuous light source. The waveband constraint (λA+λB)<λC is implemented using optical high pass and/or low pass and/or band pass filters. The sensor 10 is a line sensor. The camera 9 is configured as a TDI (time delay integration) line scan camera. The sensor 10 could be as well a 2-dimensional sensor so that an area scan camera results.
The emitted light of the LED die/chips 5 or LED 4 is caused by the recombination process of electron hole pairs that are generated by the illumination. The emitted light of the LED die/chips 5 or LED 4 is caused by the recombination process of electron hole pairs that are generated by the illumination which has a similar wavelength as if a forward voltage would be applied to the LED die/chips 5 or LED 4. The recombination process takes place in the active layer of the LED die/chips 5 or LED 4. In case of a blue LED an example implementation would be λA≈380 nm, λB≈20 nm and λC≈440 nm.
A calibration is done to correlate the measured average intensity of the LED material to an output power (density) number. The inventive apparatus 1 uses a white-light source with area illumination. Prior art devices instead use a commonly laser beam source with small spot size, and a camera as a detector. The computer system 17 controls as well a X/Y-stage 19. The X/Y-stage 19 moves the substrate 3 in a controlled manner so that the entire surface of the substrate is imaged by the objective lens 14 onto the sensor 10 of camera 9. The position of the X/Y-stage 19 is recorded in order to correlate the visually captured data with the position data on the substrate 3 and to generate the wafermap 30.
The ring light source defines an illumination 11 by which a certain area on the LED die/chips 5 on a substrate 3 illuminated. The embodiment shown in
a is an image of the surface 3a of a substrate 3 (wafer) with the LED die/chips 5 which is illuminated with normal illumination (white light). The image of the surface 3a of a substrate 3 (wafer) with the LED die/chips 5 is taken using standard illumination. With this illumination all LED die/chips 5 appear to be identical.
The inventive method is suitable for inspecting at least one LED die/chip 5 or more general a light emitting semiconductor material, which is structured on a substrate 3 or wafer. At least the area of one LED die/chip 5 is illuminated with a waveband (λA±λB) that can generate electron-hole pairs in the LED die/chip 5 to be inspected. The waveband is obtained with the first filter 15 in the illumination beam path 11. At least a part of the light emitted by the LED die/chip 5 is captured with the sensor 10 of the camera 9. The second filter 16 is positionable in order to make sure that only the light emitted by the LED die/chips 5 on the substrate 3 reach the camera 9 and the sensor 10, so that sensor 10 is sensitive to wavelengths (λC+λD) of the emitted light. The wavelength λC is larger than the wavelengths (λA+λB). The inspection result is the output of the sensor 10 which is fed to the computer system 17.
The inspection result is a measure for the light output power of an LED or a LED die/chip 5. The output of the sensor 10 is at least one gray value of at least one pixel. Usually, the gray value is represented by a matrix of pixels. The range of the grey value, e.g. for an 8 bit computer system 17 is between 0-255 per LED die/chip 5. The output power is a function of the measured grey values. The function can be implemented as a look up table or as a polynomial. The calibration of the look up table or the polynomial is done by measuring light output power of a LED sample when connected to an electrical prober.
The inspection result is at least one gray value per LED die/chip 5. The LEDs are sorted in at least two bins according to their gray value (at least one threshold value) here the inspection result is at least two gray values per LED die/chip 5, the variations/differences in gray value of one LED die/chip 5 are used as a quality measure of the LED die/chip 5.
The inspection result is at least one gray value per LED die/chip 5, of each LED die/chip 5 multiple inspection images (at least two) are taken in order to detect stability and deviations on the emitted light. All images can be taken under same conditions or mages can be are taken under varying illumination intensity and/or exposure settings. That means the first inspection image is taken under condition A, the second under condition B, where conditions A, B, and so on are configurable. Calibration of parameters may be done using the result from electrical prober. A histogram of the gray values is generated for each LED die/chip 5 and a classification into pass/fail is done by analyzing the histogram distribution. Examples: If the histogram distribution is bi-modal then fail. If the histogram distribution uni-modal and has a low gray value then fail. If the histogram distribution is uni-modal and has a large gray value then pass. One of the methods above can be used as pre/post check for the electrical prober.
The invention has been described with reference to specific embodiments. It is obvious to a person skilled in the art, however, that alterations and modifications can be made without leaving the scope of the subsequent claims.
This patent application claims priority of U.S. provisional patent application No. 61/500,987 filed Jun. 24, 2011, the application is incorporated herein by reference.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/IB2012/053052 | 6/18/2012 | WO | 00 | 8/7/2012 |
| Number | Date | Country | |
|---|---|---|---|
| 61500987 | Jun 2011 | US |
