rus
Issue #6/2021
A.Bykova, I.Rod, K.Bitarov, A.Kazachkov, Y.Minayeva, K.Raketov, V.Trushin, A.Frolova, D.Podgornyy, D.Shamiryan
Method for removal of silicon surface microdefects by laser ablation
Method for removal of silicon surface microdefects by laser ablation
DOI: 10.22184/1993-8578.2021.14.6.342.349
One of the limiting factors for continuous yield improvement of micro-sized products at MEMS production lines is a non-zero average level of contamination of the production facilities. This factor impacts appearance of surface defects in the finished goods, which can disrupt the functionality of detectors. This paper proposes a method of post-processing of the manufactured sensitive elements by evaporating silicon defects without violating the integrity of the products in order to transfer the defective products to the category of good ones. Approbation of the proposed method at MEMS production of Mapper LLC showed that the effectiveness of removal of defects by the laser ablation is up to 77% for a batch. The performance indicator can be increased through further process automation.
One of the limiting factors for continuous yield improvement of micro-sized products at MEMS production lines is a non-zero average level of contamination of the production facilities. This factor impacts appearance of surface defects in the finished goods, which can disrupt the functionality of detectors. This paper proposes a method of post-processing of the manufactured sensitive elements by evaporating silicon defects without violating the integrity of the products in order to transfer the defective products to the category of good ones. Approbation of the proposed method at MEMS production of Mapper LLC showed that the effectiveness of removal of defects by the laser ablation is up to 77% for a batch. The performance indicator can be increased through further process automation.
Теги: laser ablation mems performance indicator silicon surface microdefects кремниевые поверхностные микродефекты лазерная абляция мэмс показатель результативности
METHOD FOR REMOVAL OF SILICON SURFACE MICRODEFECTS BY LASER ABLATION
One of the limiting factors for continuous yield improvement of micro-sized products at MEMS production lines is a non-zero average level of contamination of the production facilities. This factor impacts appearance of surface defects in the finished goods, which, in turn, can disrupt the functionality of detectors and sensors. This paper proposes a method of post-processing of the manufactured sensitive elements by evaporating silicon defects without violating the integrity of the products in order to transfer the defective products to the category of good ones. Approbation of the proposed method at MEMS production of Mapper LLC showed that the effectiveness of removal of defects by the precise laser evaporation method is up to 77% for a batch of processed elements. Moreover, this performance indicator can be increased through further process automation.
INTRODUCTION
Since March, 2020, the manufacturers working in microelectronics have faced a rapid shortage of consumables for operating in clean rooms with a strong contamination level control. Deliveries of disposable gloves, masks, protective overshoes, wipes and other things were delayed indefinitely due to an increase in the supply of these products to medical organizations and in order to provide protective means for the society in the period of spread of the new coronavirus infection. Shortage of supply and violation of their delivery schedule forced the industrial enterprises to urgently switch over to alternative unverified consumables, or refuse to use some of them, for example, disposable masks, and this, in turn, caused higher level of contamination concentration (the number of particles in air per volume unit) at the production facilities. It is well known that an average contamination level in microelectronic production impacts directly the number of defects in the finished products and the ratio of defective products [1]. Under these conditions, the search for new effective methods to eliminate defects in order to increase the yield of usable products is very important.
In this paper, we proposed the developed method for rapid detection of the silicon surface defects on sensitive MEMS-elements and their removal by laser ablation (evaporation). Applying of this methodology for MEMS chips in production makes it possible to increase the yield of usable products by 7%.
DETECTION OF SURFACE DEFECTS
Earlier, the general principles of automated surface defect inspection in MEMS elements were presented [2, 3]. The main processes to detect defects are the automated optical inspection of the surface of a wafer and/or element, which allows of obtaining high-resolution images, and fast machinery processing of the received images.
Nowadays, the best methods of image processing to determine defects are based on neural network operation [4]. The main advantage of such software is their self-learning ability, however, for their correct work it is necessary to train this software using a large amount of data. For the existing industry it makes no difficulty to collect the required amount of data.
Figure 1 demonstrates the steps and processes necessary for detecting and removing defects according to the proposed methodology. After detecting defects, which is the step of an automated defect-inspection realized by the software "Axalit" [2], we get information about the location of surface micro-defects on the inspected wafer or element in the given coordinate system. Afterwards the computer code is run, which makes it possible to create, first, a defect map, i.e. an optical image on which the detected defects are marked, and secondly, a machine recipe for automated measurement of defects with a Bruker Contour GT-K optical profilometer.
According to the developed defect inspection methodology, the measurement if the defect sizes is performed using a Mirau interferometer objective with magnification of 50x and Bruker ContourGT-K equipment with motor-operated table and a turrethead. According to our estimations, the error of defect coordinates detection in a given coordinate system does not exceed 3 µm which is sufficient for the subsequent automated positioning of the objective with a field of view of 126 × 95 µm and for defect localization.
The example of an 3D-defect inspection image is shown in Fig.2.
Automation of the defect inspection has significantly reduced its time: the manual defect inspection of a 100 mm wafer can be done in 13–50 minutes, depending on the number of defects on a wafer, whereas the automated defect inspection takes in average 4 minutes per wafer containing dozens or even hundreds of chips. As it was established, the height of the defect in Fig.2 reaches 22.7 µm, and visually the defect shape resembles a pillar. Such defects have been found to be critical to the final product just because of the rather large height index.
Any MEMS sensor has so-called forbidden zones – the zones wherein a defect of certain dimensions is critical (killer defect), since it leads to malfunctioning of the sensor, and the forbidden zone can be the entire surface of the element, as well as certain areas on it, depending on the requirements. An element or chip that has a defect in the forbidden zone is considered to be defective. The number of defective products has a direct impact on the economic benefit of manufacturing, so reducing the number defective products by reducing defects in products is one of the main production objectives.
LASER ABLATION OF SURFACE DEFECTS
One method of reducing pillar-type defects is high precision ablation of silicon pillars using laser irradiation. This method is precise because it is not the entire substrate that is treated, but only the defect itself. Moreover, it is the only method that affects not the whole wafer, but a single micro-sensor separated from the wafer, which significantly reduces the final number of defective elements. In this way, in case of bad processing, only one sensor will be damaged and not the entire wafer, as can happen with wet etching, for example.
The process is quite simple: the holder with the chip is placed under a so-called laser gun. Using the optical microscope built into the machine, the operator points the laser beam precisely at the point where the defect is. By aiming, the operator "shoots" at the defect, starting from the edge areas towards the centre in order not to overlap the silicon crumbles in the area that has not yet been processed. As a rule, one shot is capable of removing up to 5 µm of silicon to depth, but this can sometimes vary. The unit also includes a 10 µm depth treatment if required. After treatment the defect area looks like a crater with protruding edges. A 3D defect profile before and after laser treatment is shown in Fig.3.
As a result of six shots on the defective area, the defect height was reduced from 20 µm to less than 2 µm (Fig.3). A block diagram of the laser ablation setup is shown in Fig.4.
The unit removes discrete masses from the silicon sensor by thermal evaporation of the silicon using a laser. The treated area remains in the form of a hole of a diameter no more than 15 µm.
The stage positioning error is stand in the range of 5 µm. The focusing device is based on an analytical microscope and has an additional television channel for observation and guidance. An impulse laser with a wavelength of 0.53 μm produces spots in the treatment area between 5 μm and 20 μm. The laser power varies from 0.1W to 0.3W at a constant frequency of 10 Hz. Figure 5 shows the cases of the most characteristic types of treatment that have removed the defect wholly or partially, and, in the first case, upgraded the element to a good product. On the left there are the defect profiles before treatment, and on the right – the treated area profile.
Figures 5a and 5b show a defect that has been successfully treated. As can be seen, the height of the defect became less than 2 µm and the depth of the hole did not exceed 10 µm (this depth is critical according to the requirements to the given sensor). Figure 5c shows the high pillar defect and Fig.5d shows a profile of its treated area, where the depth of the indentation was
more than 10 µm, resulting in defective products. It may also be difficult to treat defects that are too high, when laser power is not sufficient to completely remove the height of the defect column, which will not make the element fit (Fig.5e and Fig.5f). It can also be a problem to completely treat the defect area with this laser beam, if the original critical size is too large. For example, Fig.5g shows a wide defect with a height of more than 20 µm (before removal). It can be seen that only a part of the defect area has been treated, while the other part of the defect has hardly changed in height. This is due to the fact that the laser ablation is performed by the operator's decision based on a visual information, that sometimes cannot guarantee accurate results.
PRODUCTIVITY OF SURFACE DEFECTS REMOVING METHOD
The method described above proved to be effective both from the production and economic points of view – the yield of good products was increased by about 7% in the production of three batches of chips, each batch consisting of 120 chips. Fig.6 shows a histogram of the increase in the yield of good elements as a function of the carried out treatments in the individual batches. As can be seen, in 4 treatments 45–77% of the processed (defective) chips were transferred from the category of discarded to the category of good chips.
CONCLUSIONS
The proposed above method for elimination of surface defects in MEMS elements has been successfully tested at the MEMS produced by Mapper LLC. The effectiveness of the method of defect removal by precision laser evaporation of material reaches 77% for a batch of treated elements. The efficiency can be increased by automating the defect removal process which can be achieved by the following changes in the laser processing system: motorization of the stage, improvement of its positioning accuracy, automation of positioning of the laser beam source and adding a possibility of adjusting the size of the laser beam and its power. ■
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
One of the limiting factors for continuous yield improvement of micro-sized products at MEMS production lines is a non-zero average level of contamination of the production facilities. This factor impacts appearance of surface defects in the finished goods, which, in turn, can disrupt the functionality of detectors and sensors. This paper proposes a method of post-processing of the manufactured sensitive elements by evaporating silicon defects without violating the integrity of the products in order to transfer the defective products to the category of good ones. Approbation of the proposed method at MEMS production of Mapper LLC showed that the effectiveness of removal of defects by the precise laser evaporation method is up to 77% for a batch of processed elements. Moreover, this performance indicator can be increased through further process automation.
INTRODUCTION
Since March, 2020, the manufacturers working in microelectronics have faced a rapid shortage of consumables for operating in clean rooms with a strong contamination level control. Deliveries of disposable gloves, masks, protective overshoes, wipes and other things were delayed indefinitely due to an increase in the supply of these products to medical organizations and in order to provide protective means for the society in the period of spread of the new coronavirus infection. Shortage of supply and violation of their delivery schedule forced the industrial enterprises to urgently switch over to alternative unverified consumables, or refuse to use some of them, for example, disposable masks, and this, in turn, caused higher level of contamination concentration (the number of particles in air per volume unit) at the production facilities. It is well known that an average contamination level in microelectronic production impacts directly the number of defects in the finished products and the ratio of defective products [1]. Under these conditions, the search for new effective methods to eliminate defects in order to increase the yield of usable products is very important.
In this paper, we proposed the developed method for rapid detection of the silicon surface defects on sensitive MEMS-elements and their removal by laser ablation (evaporation). Applying of this methodology for MEMS chips in production makes it possible to increase the yield of usable products by 7%.
DETECTION OF SURFACE DEFECTS
Earlier, the general principles of automated surface defect inspection in MEMS elements were presented [2, 3]. The main processes to detect defects are the automated optical inspection of the surface of a wafer and/or element, which allows of obtaining high-resolution images, and fast machinery processing of the received images.
Nowadays, the best methods of image processing to determine defects are based on neural network operation [4]. The main advantage of such software is their self-learning ability, however, for their correct work it is necessary to train this software using a large amount of data. For the existing industry it makes no difficulty to collect the required amount of data.
Figure 1 demonstrates the steps and processes necessary for detecting and removing defects according to the proposed methodology. After detecting defects, which is the step of an automated defect-inspection realized by the software "Axalit" [2], we get information about the location of surface micro-defects on the inspected wafer or element in the given coordinate system. Afterwards the computer code is run, which makes it possible to create, first, a defect map, i.e. an optical image on which the detected defects are marked, and secondly, a machine recipe for automated measurement of defects with a Bruker Contour GT-K optical profilometer.
According to the developed defect inspection methodology, the measurement if the defect sizes is performed using a Mirau interferometer objective with magnification of 50x and Bruker ContourGT-K equipment with motor-operated table and a turrethead. According to our estimations, the error of defect coordinates detection in a given coordinate system does not exceed 3 µm which is sufficient for the subsequent automated positioning of the objective with a field of view of 126 × 95 µm and for defect localization.
The example of an 3D-defect inspection image is shown in Fig.2.
Automation of the defect inspection has significantly reduced its time: the manual defect inspection of a 100 mm wafer can be done in 13–50 minutes, depending on the number of defects on a wafer, whereas the automated defect inspection takes in average 4 minutes per wafer containing dozens or even hundreds of chips. As it was established, the height of the defect in Fig.2 reaches 22.7 µm, and visually the defect shape resembles a pillar. Such defects have been found to be critical to the final product just because of the rather large height index.
Any MEMS sensor has so-called forbidden zones – the zones wherein a defect of certain dimensions is critical (killer defect), since it leads to malfunctioning of the sensor, and the forbidden zone can be the entire surface of the element, as well as certain areas on it, depending on the requirements. An element or chip that has a defect in the forbidden zone is considered to be defective. The number of defective products has a direct impact on the economic benefit of manufacturing, so reducing the number defective products by reducing defects in products is one of the main production objectives.
LASER ABLATION OF SURFACE DEFECTS
One method of reducing pillar-type defects is high precision ablation of silicon pillars using laser irradiation. This method is precise because it is not the entire substrate that is treated, but only the defect itself. Moreover, it is the only method that affects not the whole wafer, but a single micro-sensor separated from the wafer, which significantly reduces the final number of defective elements. In this way, in case of bad processing, only one sensor will be damaged and not the entire wafer, as can happen with wet etching, for example.
The process is quite simple: the holder with the chip is placed under a so-called laser gun. Using the optical microscope built into the machine, the operator points the laser beam precisely at the point where the defect is. By aiming, the operator "shoots" at the defect, starting from the edge areas towards the centre in order not to overlap the silicon crumbles in the area that has not yet been processed. As a rule, one shot is capable of removing up to 5 µm of silicon to depth, but this can sometimes vary. The unit also includes a 10 µm depth treatment if required. After treatment the defect area looks like a crater with protruding edges. A 3D defect profile before and after laser treatment is shown in Fig.3.
As a result of six shots on the defective area, the defect height was reduced from 20 µm to less than 2 µm (Fig.3). A block diagram of the laser ablation setup is shown in Fig.4.
The unit removes discrete masses from the silicon sensor by thermal evaporation of the silicon using a laser. The treated area remains in the form of a hole of a diameter no more than 15 µm.
The stage positioning error is stand in the range of 5 µm. The focusing device is based on an analytical microscope and has an additional television channel for observation and guidance. An impulse laser with a wavelength of 0.53 μm produces spots in the treatment area between 5 μm and 20 μm. The laser power varies from 0.1W to 0.3W at a constant frequency of 10 Hz. Figure 5 shows the cases of the most characteristic types of treatment that have removed the defect wholly or partially, and, in the first case, upgraded the element to a good product. On the left there are the defect profiles before treatment, and on the right – the treated area profile.
Figures 5a and 5b show a defect that has been successfully treated. As can be seen, the height of the defect became less than 2 µm and the depth of the hole did not exceed 10 µm (this depth is critical according to the requirements to the given sensor). Figure 5c shows the high pillar defect and Fig.5d shows a profile of its treated area, where the depth of the indentation was
more than 10 µm, resulting in defective products. It may also be difficult to treat defects that are too high, when laser power is not sufficient to completely remove the height of the defect column, which will not make the element fit (Fig.5e and Fig.5f). It can also be a problem to completely treat the defect area with this laser beam, if the original critical size is too large. For example, Fig.5g shows a wide defect with a height of more than 20 µm (before removal). It can be seen that only a part of the defect area has been treated, while the other part of the defect has hardly changed in height. This is due to the fact that the laser ablation is performed by the operator's decision based on a visual information, that sometimes cannot guarantee accurate results.
PRODUCTIVITY OF SURFACE DEFECTS REMOVING METHOD
The method described above proved to be effective both from the production and economic points of view – the yield of good products was increased by about 7% in the production of three batches of chips, each batch consisting of 120 chips. Fig.6 shows a histogram of the increase in the yield of good elements as a function of the carried out treatments in the individual batches. As can be seen, in 4 treatments 45–77% of the processed (defective) chips were transferred from the category of discarded to the category of good chips.
CONCLUSIONS
The proposed above method for elimination of surface defects in MEMS elements has been successfully tested at the MEMS produced by Mapper LLC. The effectiveness of the method of defect removal by precision laser evaporation of material reaches 77% for a batch of treated elements. The efficiency can be increased by automating the defect removal process which can be achieved by the following changes in the laser processing system: motorization of the stage, improvement of its positioning accuracy, automation of positioning of the laser beam source and adding a possibility of adjusting the size of the laser beam and its power. ■
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Readers feedback
