Patent Application
20230377970
This application is a Continuation Patent Application of PCT Patent Application No. PCT/EP2022/050305 having International filing date of Jan. 10, 2022, which claims the benefit of priority of Germany Patent Application No. 10 2021 100 675.9 filed on Jan. 14, 2021. The contents of the above applications are all incorporated by reference as if fully set forth herein in their entirety.
The invention relates to a method for dividing a transparent workpiece by pulsed laser radiation by creating a beam convergence zone in the volume of the workpiece in which the intensity of the laser radiation exceeds a threshold value for non-linear absorption, wherein the beam convergence zone and the workpiece are moved relative to each other, thereby creating a two-dimensional weakening in the workpiece extending along a predetermined separating line, and wherein the workpiece is subsequently divided along the separating line.
The dividing of wafer substrates into chips, i.e. the so-called dicing of wafers, plays an important role in the production of semiconductor devices, which are becoming smaller and more complex. The classical methods of dicing are based on the use of a diamond saw for wafers thicker than 100 μm. For thinner wafers, laser-based methods are increasingly used.
WO 2016/059449 A1 describes a dicing method in which pulsed laser radiation is used to create a beam convergence zone in the volume of the workpiece, i.e. the semiconductor substrate, in which the intensity of the laser radiation locally exceeds a threshold value for non-linear absorption. In the beam convergence zone, multiphoton processes occur accordingly, e.g. in the form of multiphoton ionization or avalanche ionization, which lead to the formation of a plasma. The plasma formation rate increases sharply above a threshold that depends on the material of the workpiece and the parameters of the laser radiation. This is therefore also referred to as “optical breakdown”. The resulting modification and thus processing of the workpiece exhibits high precision, since spatially localized, reproducibly small amounts of energy are introduced into the material. The good spatial localization is primarily achieved by focusing the laser radiation using a high numerical aperture optical probe that is as aberration-free as possible. A beam convergence zone is created in the form of an extended, in the cited print “spiky” focus volume in the direction of the laser beam axis. This beam convergence zone is moved relative to the workpiece to create, as a result, a two-dimensional weakening in the workpiece extending along a predetermined separating line. Possible weakening mechanisms due to the introduced modifications are void and/or crack formation, structural changes of the material of the workpiece, cracks coupled to the modification area in each case, transient or permanent stresses, thermomechanical stresses, stresses due to local volume increases or decreases, solidification cracks, etc. . . . Finally, the actual parting of the workpiece occurs by the application of a small mechanical force or stress, which causes the workpiece to break in the area of weakening, i.e. along the parting line.
Of crucial importance is a uniform modification of the material over and into a given depth of the workpiece. This improves the dividability, production defects such as chipping or material distortion are minimized, and higher edge strengths are achieved. In the known method, the modifications in the workpiece are generated with laser pulses of a pulse duration in the range of 100-15000 fs at a wavelength of 500 nm to 2000 nm wavelength and 10 kHz to 2 MHz repetition rate. In the known method, the beam shaping for creating the beam convergence zone is designed on the basis of undisturbed linear propagation of the laser radiation in the volume of the workpiece. However, at the required fluence to generate the weakening of the material enabling fragmentation, the propagation of the laser radiation within the workpiece is subject to non-linear effects in the aforementioned pulse duration range. At low pulse duration and high energy density, the propagation of the laser radiation in the volume of the workpiece is so strongly disturbed by non-linear effects (e.g. self-focusing as well as two-photon absorption already outside the beam convergence zone) that effective energy coupling into the desired area of the beam convergence zone is prevented to a considerable extent. A defined localization of the energy deposition and the resulting modification of the material of the workpiece cannot be achieved at high peak intensity of the radiation, as it is present at short pulse duration.
With long pulse duration (e.g. >1 ns) and thus lower peak intensity, the material can be modified, but since typically a higher energy is required and diffusion effects become effective, greater damage due to an increased thermally loaded volume is accepted. Accordingly, the result of the separation process is not satisfactory with regard to the quality of the break line.
Against this background, the object underlying the invention is to provide an improved method for dividing a transparent workpiece. The above-mentioned mentioned disadvantages of known methods are to be avoided.
The invention achieves this object on the basis of a method of the type indicated at the beginning by suppressing non-linear propagation of the laser radiation in the volume of the workpiece outside the beam convergence zone by selecting the duration of the energy input generated by the non-linear absorption of the pulsed laser radiation in the beam convergence zone and/or by spatial beam shaping.
Details and advantages of the invention are apparent from the wording of the claims and from the following description of exemplary embodiments based on the drawings. The drawings show in:
An aspect of the invention is the consideration of the non-linear propagation characteristics for the introduction of the weakening into the workpiece. In order to suppress, i.e. significantly reduce, non-linear propagation of the laser radiation outside the beam convergence zone, the most optimal possible process parameters are defined with respect to the duration of energy input and/or beam shaping, thereby enabling improved control of energy deposition. In this context, the local energy density in the volume of the workpiece is ultimately controlled by the duration of energy input and improves energy coupling within the beam convergence zone. Damage in the surrounding volume, i.e. outside the beam convergence zone, is minimized. The invention achieves that the lowest possible interaction outside, in particular in the propagation direction of the laser radiation in front of the beam convergence zone is achieved in order to minimize interfering non-linear effects (e.g. self-focusing, non-linear absorption) or other propagation disturbances. It is also ensured that during the duration of non-linear absorption in the beam convergence zone, modifications or zones of high electron density occurring first in the beam direction do not shield more than 50%, preferably not more than 20%, particularly preferably not more than 10% of the incident energy from the parts of the beam convergence zone reached later in the beam direction.
By minimizing the non-linear propagation, a suitable fluence can be achieved in a targeted and defined manner by means of beam focusing in the beam convergence zone. As a result, the desired modification only occurs in this area. The combination of temporal and spatial beam shaping results in a uniform, tailored energy deposition over the entire specified beam convergence zone. As a result, the dividing process is facilitated. Chipping or material stresses are minimized. The quality of the break line edge is improved compared to the prior art.
By targeted selection of the process parameters (amount and duration of energy input, spatial beam shaping), according to the invention, in other words, it is achieved that
Ideally, the wavelength of the laser radiation should be selected such that the linear absorption of the laser radiation in the material of the workpiece is below 20%, better still below 10%, particularly preferably below 5% over a length of one centimeter in the laser beam direction. In addition, the wavelength of the laser radiation should be selected according to the proviso that the non-linear refractive index in the volume of the workpiece at this wavelength is so low that non-linear effects do not prevent sufficient energy deposition in the beam convergence zone. At the same time, the wavelength should lie in a range in which good focusability is ensured.
According to the invention, the suitable choice of the duration of the energy input by the non-linearly induced absorption of the pulsed laser radiation in the beam convergence zone, i.e. at a certain position of the workpiece along the separating line, is crucial. The duration of the energy input can be specified, for example, by the pulse duration of the pulsed laser radiation. An upper limit of the duration results from the tolerable size of the thermal damage zone due to heat diffusion. Furthermore, an upper limit for the duration is given by the maximum tolerable energy absorbed in the beam convergence zone. The greater the duration, the greater the energy input when the intensity is above the threshold value for non-linear absorption. Too much energy and/or energy introduced over too long a period of time will prevent the local limitation of the weakening to the beam convergence zone. In terms of avoiding nonlinear propagation of the laser radiation outside the beam convergence zone, the lower limit of the energy input duration matters. In particular, the pulse duration should be greater than a critical value, where the critical value is, for example, the quotient of pulse energy and material-specific critical power, above which non-linear propagation, in particular self-focusing, occurs in the volume of the workpiece. This ensures that the energy deposition is not excessively disturbed by non-linear effects, thus ensuring a sufficiently high and localized energy deposition in the beam convergence zone.
Appropriately, the duration of energy input (e.g., pulse duration) and the amount of energy input (e.g., pulse energy) in a pulse event should be selected on the proviso that damage, i.e., a desired modification in the volume of the workpiece within the beam convergence zone, occurs by a single laser pulse or a laser pulse burst consisting of a sequence of a predetermined number of laser pulses. A suitable single pulse may be, for example, a laser pulse with a Gaussian temporal profile of specific pulse duration. A burst comprises a predetermined number of laser pulses with a small temporal interval (pulse repetition frequency in the GHz or THz range). If the temporal interval of the bursts from each other is at least 100 times greater than the duration of the individual burst, such a burst is also considered a pulse event. The modification at a particular location on the workpiece should be completely generated during a single such “pulse event” (single pulse or burst). Accordingly, energy input in the context of the invention refers to the energy input during a single pulse event by non-linear induced absorption in the beam convergence zone. The duration of the energy input results accordingly from the pulse duration or the burst duration. It should be noted that the laser pulses do not necessarily have to have a Gaussian shape, a “flat top” shape or any other common shape. Any pulse shape is conceivable. What is decisive is the effective duration of the energy input. In the method according to the invention, this is preferably 20-500 ps.
The two-dimensional weakening is then generated repetitively by moving the workpiece relative to the beam convergence zone incrementally along the parting line from pulse event to pulse event. If possible, the weakening is generated during a single motion event along the parting line. In this way, a high process speed can be achieved and the workpiece breaks reliably with high quality of the dividing edge along the separating line.
In an embodiment, the shortest possible duration of energy input is determined at which the desired modification occurs with a probability of at least 80%, preferably at least 90%, more preferably at least 95%. The duration of energy input when the weakening is introduced is then chosen to be greater than or equal to this determined shortest possible value, which may depend on numerous factors (material, size and geometry of the beam convergence zone, wavelength, pulse shape, etc.). E.g., the selected duration of the energy input can be above the determined shortest possible value by a factor of 10, preferably by a factor of 5, particularly preferably by a factor of 2. Ideally, the factor is in the range of 1.1 to 5. This provides good control of the energy input for the desired generation of the weakening.
The beam convergence zone may have an elongated shape perpendicular to the workpiece surface. Similar to WO 2016/059449 A1 cited above, the beam convergence zone should be elongated in the beam direction and thus extend over most of the full thickness of the workpiece to ensure suitable weakening. For example, the length of the beam convergence zone in the beam direction may be greater than the extent of the beam convergence zone perpendicular to the desired two-dimensional weakening by at least a factor of 10, preferably by at least a factor of 50, particularly preferably by at least a factor of 100. The laser beam can initially have a Gaussian profile or any other feasible input beam shape. Particularly suitable are Gauss-Bessel beams or other beam shapes that can substantially be described as non-diffracting beams. A tailored spatial intensity distribution in the beam convergence zone is expediently achieved by means of suitable optical components, such as focusing optics in combination with beam shaping optics, also with adaptive beam shaping components, such as spatial light modulators (SLM) or piezo mirrors. Improvement can be achieved by depth-dependent aberration correction. The goal of spatial beam shaping is to achieve the most undisturbed energy input possible into the desired beam convergence zone, from the workpiece surface (beam entrance surface) to the depth into the workpiece required for the separation process.
At the same time, the extent of the beam convergence zone transverse to the beam axis should be greater in the direction parallel to the weakening plane than perpendicular to it. This minimizes the portion of the laser radiation that is influenced (shielded) by the above-mentioned introduced modifications during repetitive introduction of the modifications with relative movement of the beam convergence zone and the workpiece.
As mentioned above, the invention aims at achieving the lowest possible interaction outside, in particular in the propagation direction of the laser radiation in front of the beam convergence zone in order to minimize disturbing nonlinear effects. In addition, it should be ensured that during a pulse event in the beam convergence zone, modifications or zones of high electron density that arise first shield only as small a portion as possible of the incident energy from the parts of the beam convergence zone that are reached later. Suitable beam shaping can contribute to this. Beam shaping can advantageously be carried out in such a way that those beam components (consisting of individual beams or bundles of individual beams) of the laser radiation which converge closer to the workpiece surface in the volume of the workpiece enclose an equal or smaller angle with the beam axis than those beam components which converge further away from the workpiece surface in the volume of the workpiece. This should apply in any case to the majority of the beam components converging in the beam convergence zone; the deviation of a small portion of the beam components from this geometry can be tolerated in individual cases, provided that nonlinear effects are sufficiently suppressed.
The method according to the invention is particularly suitable for dicing semiconductor wafers into chips. The temporal and spatial beam shaping proposed by the invention prevents or at least minimizes non-linear propagation of the laser radiation, which is detrimental to the dicing process. As a result, the propagation of the laser radiation in the material of the substrate is undisturbed and an elongated modification zone can be introduced with each pulse event. By repeating the procedure with relative movement of the laser beam and the substrate, a two-dimensional weakening is created along a predetermined separating line. This serves as a predetermined breaking point during fracturing by a subsequently applied tensile stress. Thus, on the one hand thin, but on the other hand even thick semiconductor substrates can be effectively separated with minimized thermally stressed or otherwise impaired breaking point. Due to the small extension of the modification zone transverse to the weakening plane, it is possible to introduce modifications closely adjacent to each other in the feed direction without the beam propagation being significantly disturbed by the above-mentioned introduced modifications. The contour accuracy of the separating line is particularly high. Production errors and rejects are minimized. At the same time, a high process speed can be achieved.
The method according to the invention is also suitable for dividing flat glass products or ceramic and crystalline workpieces.
The core of the invention is the consideration of the non-linear propagation characteristics of the laser radiation 2 for the introduction of the modification zones 5 into the workpiece 1. The propagation of the laser radiation 2 in the volume of the workpiece 1 would be so strongly disturbed by non-linear effects (e.g. self-focusing as well as, two-photon absorption already outside the beam convergence zone) to such an extent that an effective energy coupling, which is limited to the desired area of the beam convergence zone 3 on the one hand, but also fills it as completely as possible on the other hand, is prevented to a considerable extent. In order to suppress, i.e. reduce, such non-linear propagation of the laser radiation 2 outside the beam convergence zone 3, the most optimal possible process parameters are defined in accordance with the invention with regard to the duration of the energy input and beam shaping, thus enabling extensive control of the energy deposition. Damage in the surrounding volume, i.e. outside the beam convergence zone 3, is reduced to a minimum.
In order to achieve the best possible control of the energy deposition, the shortest possible duration of the energy input generated by the non-linear absorption of the pulsed laser radiation, i.e. in this case the shortest possible pulse duration for the desired achievement of the modification in the workpiece 1 is determined. This depends on the parameters already determined in advance, namely material, geometry of the beam convergence zone 3, i.e. beam shape and wavelength. The optimization steps shown in
In a practical implementation of the invention, a 525 μm thick silicon wafer is irradiated with pulsed laser radiation at a wavelength of 1960 nm. A significant reduction of the non-linear propagation and thus the first occurrence of modifications can be observed from 20 ps pulse duration. The determination of the modification probability by a single laser pulse shows that a probability of >95% is reached from 25 ps pulse duration. In this case, the pulse energy is 15 μJ. Thus, modification zones with a diameter of 5 μm and a length of 350 μm in the beam direction can be generated by means of spatial pulse shaping. The modification zones are aligned with an interval of 10 μm by moving the focus of the laser radiation, i.e. the beam convergence zone, relative to the wafer. Subsequently, the wafer can be broken at right angles with a small mechanical force. An exact break line is formed. The heat-affected area along the break line is small. The roughness of the surface is less than 5 μm.
To create a continuous weakening in the material of the workpiece 1, the speed of the relative movement of the laser beam 2 and the workpiece 1 can be lowered to such an extent that there is an overlap of the modification zones 5 in the volume of the workpiece 1. The diagram of
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2021 100 675.9 | Jan 2021 | DE | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/EP2022/050305 | Jan 2022 | US |
| Child | 18221501 | US |
