Methods and systems for processing a device, methods and systems for modeling same and the device

Method and system for processing device, method and system and device of master processing device

Background of the invention 1. Scope of invention

The present invention relates to the field of laser processing methods and systems, and in particular, relates to laser processing methods and systems for processing microstructures formed on substrates. The present invention is particularly applicable but not limited to lasers. Repair redundant semiconductor memory devices.

2. Background skills

When repairing memory integrated circuits such as DRAM and laser setting programs for high-density logic devices, it is difficult to solve the link due to the use of new materials such as aluminum, gold, and copper and the small geometric structure of these components. Remove the problem. New materials are usually metals or highly conductive composites with reflectivity far exceeding 90% in the visible and near infrared wavelength ranges. For example, aluminum reflects more than 90% of the laser energy in the range from UV to near infrared, and gold and copper reflect even stronger reflection in the near infrared and most of the lasers used in the process of repairing memory. .

In addition, because economic benefits and device performance have driven DRAM and logic devices to have extremely small physical sizes, in recent years, devices have not only been small, but also greatly reduced the thickness of wires and links.

The thermal laser processing of the link depends on the differential thermal expansion between the oxide on the link and the link itself. This differential expansion causes the high pressure accumulation of the molten link contained in the oxide, and the oxide on the link requires A long time sufficient to accumulate sufficient pressure to rupture the oxide and explode the link material explosively to contain the link in the molten state. If the pressure is too low, the link cannot be removed cleanly. Alternative laser wavelengths and laser controllers try to increase the "energy window" of the laser without damaging the substrate and the materials adjacent to the link.

For a description of an all-copper double damascene processing technology, please see "The advantages of copper-copper technology that can be used in processing equipment today", NOVELLUSDAMASEUS, December 20, 2001; and "Preventing cross-contamination caused by the diffusion of copper and other sources" , P. Cacouvis, MICRO, July 1999.

Figures 2a and 2b show the laser processing of a multilayer structure in the prior art, in which a target structure is in close proximity to a substrate, and a q-switched pulse 20 from a conventional solid-state laser 21 is irradiated and overflowed. Full one target structure 23. The size of a laser spot area is usually significantly larger than the size of the (target) link, which relaxes the precision positioning requirements. Usually a laser wavelength is selected according to the transmission of the substrate 27 (usually silicon), thereby allowing a higher peak value Laser power or other systems and processing changes. In certain situations, the absorption coefficients of the layers 28 and 25 are controlled (for example, as a transition or protective layer), and/or a wavelength that can avoid damage to the substrate is selected.

Please refer to the following representative U.S. patents and published U.S. patent applications: 4,399,345; 4,532,402; 4,826,785; 4,935,801; 5,059,764; 5,208,437. ; 5,265,114; 5,473,624; 6,057,180; 6,172,325; 6,191,486; 6,239,406; 2002-0003130; and 2002-0005396.

Other representative announcements that can provide background information on link processing of memory circuits or similar laser processing applications include: "Laser adjustment of linear single-body circuits", Litwin and Smart, ICAELO (1983); "Laser available Computer Simulation of Target Link Explosion in the Memory of Setting Program", Scarfone, Chlipala (1986); "Precision Laser Micromachining", Bogard, SPIEVol.61191986); "Laser Machining of Special Application Integrated Circuits (asic)" , SPIE Vol.774, Smart (1987); "Xenon laser repairing liquid crystal displays", Waters, Laser and Optoelectronics (1988); "Laser beam processing and wafer-scale integration", Cohen (1988); "Best Memory Redundant Links Processing", Sun, Harris, Swenson, Hutchens, Vol. SPIE2636 (1995); "Analysis of Laser Metal Cutting Energy Processing Window", Berstein, Lee, Yang, Dahmas, IEEE Trans., On Semicond.Manufact.,Vol13,No.2(2000).

In addition, the following U.S. patent applications and issued patent cases under joint examination are assigned to the assignee of the present invention and are incorporated herein by reference in their entirety:

1. US Patent No. 5,300,756, titled "Method and System for Cutting off the Connection Path of Integrated Circuits by a Phase Plate Adjustable Laser Beam";

2. US Patent No. 6,144,118, titled "High Speed Precision Positioning Device";

3. US Patent No. 6,181,728, titled "Controlling Laser Polarization";

4. US Patent No. 5,998,759, entitled "Laser Processing";

5. US Patent No. 6,281,471, entitled "Energy-efficient laser-based method and system for processing target materials";

6. US Patent No. 6,340,806, entitled "Energy-efficient method and system for processing target materials using amplified wavelength and shifted pulse trains";

7. U.S. Serial No. 09/572,925, filed on May 16, 2000 and published as WO 0187534 A2 in December 2001, entitled "Precision positioning of the waist of a material processing laser beam to process a laser processing part Microstructure methods and systems";

8. US Patent No. 6,300,590, titled "Laser Processing"; and

9. US Patent No. 6,339,604, entitled "Pulse Control in Laser System".

However, please understand that the above list does not mean that any of the above-mentioned reference documents are known skills regulated by the patent law.

The main system of the patent applications and patent cases referred to above is related to the present invention, and the references to the above-listed patent cases and applications are quoted below by numbers.

Summary of the invention

An object of the present invention is to provide an improved method and system for processing a device, a method and system for profiling the device, and the device.

When implementing the above and other objects of the present invention, a method for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multilayer stack. To separate the inner layer of the microstructure and the substrate. The method includes the following steps: a) generating a pulsed laser beam having a predetermined wavelength and including at least one laser pulse, the laser pulse having a predetermined characteristic, and the predetermined characteristic including a transient state At least one of the shape and the spatial shape; b) According to at least one position measurement obtained at a reference position, relatively locate the microstructure and a waist of the laser beam in the three-dimensional space, wherein the position measurement is used to obtain the waist A prediction of a common position between the part and the microstructure; and c) at a time when the beam waist and the microstructure roughly coincide, according to the predicted common position, the microstructure is irradiated with at least one laser pulse, where the microstructure is In the above, the microstructure is cleanly removed with substantially the maximum pulse energy density, and the substrate and the inner layer of the stack are prevented from being undesirably changed.

Moreover, when implementing the above and other objects of the present invention, a system for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multilayer stack. It has an inner layer for separating the microstructure from the substrate. The system includes: a component for generating a pulsed laser beam, the laser beam having a predetermined wavelength and including at least one laser pulse, the laser pulse having a predetermined characteristic, the predetermined characteristic including a transient state At least one of shape and spatial shape. The system includes: a member for relatively positioning the microstructure and a waist of the laser beam in a three-dimensional space based on at least one position measurement obtained at a reference position, and uses this position measurement to obtain the waist and the waist One of the common locations of the microstructure is predicted. The system further includes: irradiating the component of the microstructure with at least one laser pulse at a time when the beam waist approximately coincides with the microstructure according to the predicted common position, wherein the microstructure is cleaned with approximately the largest pulse energy density Remove the microstructures, and avoid undesirable changes in the substrate and the inner layer of the stack.

Furthermore, when implementing the above and other objects of the present invention, a method for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multilayer stack. The stack is It has an inner dielectric layer for separating the microstructure from the substrate. The method includes generating a pulsed laser beam, the laser beam having a predetermined wavelength and including at least one laser pulse. Relative to at least another wavelength, at least the reflection of the laser beam by the stacked layers will significantly reduce the pulse energy density on the substrate. At least one laser pulse is used to process the microstructure, and the pulse energy density on the microstructure is sufficient to remove the microstructure and avoid damage to the substrate and the inner layer of the stack.

Furthermore, when implementing the above and other objects of the present invention, a system for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multi-layer stack. The stack is It has an inner dielectric layer for separating the microstructure from the substrate. The system includes: a component for generating a pulsed laser beam, the laser beam having a predetermined wavelength and including at least one laser pulse, wherein with respect to at least another wavelength beyond the absorption edge, at least this laser beam is subjected to The reflection of the stacked layers will significantly reduce the pulse energy density on the substrate. The system also includes components for processing microstructures with at least one laser pulse, where the pulse energy density on the microstructures is sufficient to remove the microstructures and avoid damaging the substrate and the inner layer of the stack.

Moreover, when implementing the above and other objects of the present invention, a method for profiling the interaction between a pulsed laser beam and a three-dimensional device is provided, the laser beam includes at least one laser pulse, and the The three-dimensional device includes a substrate, a predetermined microstructure formed on the substrate, and a plurality of other structures formed on the substrate. This method includes: providing information about structural materials including predetermined microstructures and substrates. The method also includes: determining the optical propagation characteristics of at least a part of the laser beam that is not absorbed by the predetermined microstructure based on the information. The method further includes: determining at least one characteristic of the at least one pulse according to the optical propagation characteristic.

Furthermore, when implementing the above and other objects of the present invention, a system for profiling the interaction between a pulsed laser beam and a three-dimensional device is provided, the laser beam includes at least one laser pulse, and the The three-dimensional device includes a substrate, a predetermined microstructure formed on the substrate, and a plurality of other structures formed on the substrate. This system includes components for providing information about structural materials including predetermined microstructures and substrates. The system also includes a component for determining the optical propagation characteristics of at least a part of the laser beam that is not absorbed by the predetermined microstructure based on the information. The system further includes: a component for determining at least one characteristic of the at least one pulse according to the optical propagation characteristic.

Furthermore, when implementing the above and other objects of the present invention, a method for profiling the interaction between a pulsed laser beam and a three-dimensional device is provided, the laser beam includes at least one laser pulse, and the The three-dimensional device includes a substrate, a predetermined microstructure formed on the substrate, and a plurality of other structures formed on the substrate. This method includes: providing information about structural materials including predetermined microstructures and substrates. The method also includes: determining the optical propagation characteristics of at least a part of the laser beam that is not absorbed by the predetermined microstructure based on the information. The method further includes: determining at least one characteristic of the at least one pulse according to the optical propagation characteristic, so as to avoid undesirable changes in the electrical or physical characteristics of the substrate and other structures.

Furthermore, when implementing the above and other objects of the present invention, a system for profiling the interaction between a pulsed laser beam and a three-dimensional device is provided, the laser beam includes at least one laser pulse, and the The three-dimensional device includes a substrate, a predetermined microstructure formed on the substrate, and a plurality of other structures formed on the substrate. This system includes components for providing information about the materials and spacing of structures including predetermined microstructures and substrates. The system also includes a component for determining the optical propagation characteristics of at least a part of the laser beam that is not absorbed by the predetermined microstructure based on the information. The system further includes: a member for determining at least one characteristic of at least one pulse according to the optical propagation characteristic, so as to avoid undesirable changes in the electrical or physical characteristics of the substrate and other structures.

Furthermore, when implementing the above and other objects of the present invention, a method for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multilayer stack. The stack is It has an inner layer for separating the microstructure from the substrate. The method includes: generating a pulsed laser beam having a predetermined wavelength and including at least one laser pulse, the laser pulse having a predetermined characteristic, and wherein: a) the predetermined wavelength is lower than that of the substrate An absorbing edge, and b) at least one pulse has a time length of less than about 10 nanoseconds and a repetition rate of 10 kHz or more. The method also includes relatively positioning the microstructure and a waist of the laser beam in a three-dimensional space based on at least one position measurement obtained at a reference position, and using this position measurement to obtain a common position of the microstructure and the beam waist One prediction. The method further includes irradiating the microstructure with at least one laser pulse according to the predicted common position at a time when the beam waist approximately coincides with the microstructure, and removing the microstructure cleanly with approximately the maximum pulse energy density. Microstructure, and avoid undesirable changes in the inner layer of the stack and the substrate.

Furthermore, when implementing the above and other objects of the present invention, a system for laser processing a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multi-layer stack. The stack is It has an inner layer for separating the microstructure from the substrate. The system includes: a component for generating a pulsed laser beam, the laser beam having a predetermined wavelength and including at least one laser pulse, the laser pulse having a predetermined characteristic, wherein: a) the predetermined wavelength is lower than An absorbing edge of the substrate, and b) at least one pulse has a time length of less than about 10 nanoseconds and a repetition rate of 10 kHz or more. The system also includes a component for relatively positioning the microstructure and a waist of the laser beam in a three-dimensional space based on at least one position measurement obtained at a reference position, and the position measurement is used to obtain the relationship between the microstructure and the waist. A prediction of a common location. The system further includes: irradiating the component of the microstructure with at least one laser pulse at a time when the beam waist and the microstructure approximately coincide according to the predicted common position, and clean the microstructure with approximately the largest pulse energy density Remove the microstructure carefully, and avoid undesirable changes in the inner layer of the stack and the substrate.

In addition, when implementing the above and other objects of the present invention, a multi-level multi-material device is provided. The device includes a substrate, a microstructure, and a multilayer stack. The stack is provided for separating the microstructure and the substrate. The inner layer of the material. At least one inner layer has a predetermined physical parameter based on the interaction between a pulsed laser and the stack, and the laser beam has a predetermined wavelength, wherein the stacking of the inner layer and the stack is avoided when the pulsed laser beam is used for laser processing. The substrate has undergone an undesirable change.

When implementing the above-mentioned and other objects of the present invention, a method for thermally laser processing a multi-material device is provided. The device includes a substrate and at least one microstructure. In a single pass operation controlled by a positioning subsystem of a thermal processing system, processing is generated by multiple pulses. The positioning subsystem initiates a relative movement between the device and the waist of the laser beam. This process removes at least one microstructure without damaging the substrate. The method includes generating a first pulse having a first predetermined characteristic, and irradiating at least one microstructure with the first pulse, wherein a first beam waist and at least one microstructure associated with the first pulse are approximately coincide. The irradiation step starts at least the processing of at least one microstructure. The method also includes generating a second pulse having a second predetermined characteristic, the second pulse being delayed by a predetermined time relative to the first pulse. The method further includes: irradiating at least one microstructure with a second pulse, wherein a second beam waist associated with the second pulse and the at least one microstructure substantially coincide. The step of irradiating at least one microstructure with the second pulse is to further process at least one microstructure, wherein the effect of processing at least one microstructure with the first and second pulses occurs at least one microstructure and the beam waist during a single pass During the relative operation period, the output rate of the thermal processing system can be significantly improved.

In addition, when implementing the above and other objects of the present invention, a system for thermal laser processing a multi-material device is provided. The device includes a substrate and at least one microstructure. In a single pass operation controlled by a positioning subsystem, multiple pulses are generated and processed. The positioning subsystem induces relative motion between the device and the waist of the laser beam. The processing involves removing at least one microstructure. Does not damage the substrate. This system includes: a member for generating a first pulse having a first predetermined characteristic, and a member for irradiating at least one microstructure with the first pulse, wherein a first beam waist associated with the first pulse and The at least one microstructure is substantially coincident, and the first pulse at least initiates the processing of the at least one microstructure. The system also includes: a component for generating a second pulse having a second predetermined characteristic, the second pulse being delayed by a predetermined time relative to the first pulse. The system further includes: a member for irradiating at least one microstructure with a second pulse, wherein a second beam waist associated with the second pulse and the at least one microstructure substantially coincide. The second pulse is to further process at least one microstructure, wherein the effect of processing at least one microstructure with the first and second pulses occurs during the relative movement of at least one microstructure and the waist during a single pass, thereby significantly improving this The output rate of the system.

Furthermore, when implementing the above-mentioned and other objects of the present invention, a method for thermally laser processing a multi-material device is provided, which includes a substrate and a microstructure. The method includes generating at least one laser pulse, the laser pulse having at least one predetermined characteristic according to a differential thermal property of the device material. The method also includes: irradiating the microstructure with at least one laser pulse, wherein at least one pulse of a first part increases the temperature difference between the substrate and the microstructure, and at least one pulse of a second part The system further increases the temperature difference between the substrate and the microstructure, thereby processing multi-material devices without damaging the substrate.

In addition, when implementing the above and other objects of the present invention, a system for thermal laser processing a multi-material device is provided. The device includes a substrate and a microstructure. The system includes: a component for generating at least one pulse with at least one predetermined characteristic according to a differential thermal property of the device material. The system also includes: a member for irradiating the microstructure with at least one laser pulse, wherein at least one pulse of a first part increases the temperature difference between the substrate and the microstructure, and a second part The at least one pulse of is to further increase the temperature difference between the substrate and the microstructure, thereby processing the multi-material device without damaging the substrate.

When implementing the above-mentioned and other objects of the present invention, a method for accurately positioning a waist of a pulsed laser beam to compensate for the microscopic position change of a predetermined target is provided. The first material for injection processing. The method includes: measuring the position of at least one alignment target formed at a predetermined measurement position to obtain a measurement. At least one alignment target is covered by at least one layer of the second material. The measurement step includes: selectively irradiating a part of an area to be measured with a radiation cleaning beam to remove debris from this area, thereby compensating for changes in reflectivity and reducing multiplicative noise and related signals of the detector Variety. The method also includes: predicting the relative position of the predetermined target and the waist of the laser beam based on the measurement to obtain a predicted relative position. The method further includes: triggering a relative movement between the predetermined target and the waist according to the predicted relative position. The method further includes: generating a laser beam including at least one pulse, and irradiating the at least one pulse into a spot area of the predetermined target, wherein the at least one pulse is sufficient to process the predetermined target.

In addition, when implementing the above and other objects of the present invention, a system for accurately positioning a waist of a pulsed laser beam to compensate for the microscopic position change of a predetermined target is provided. The predetermined target has a waiting The first material for laser processing. The system includes: measuring the position of at least one alignment target formed at a predetermined measurement position to obtain a measurement. At least one alignment target is covered by at least one layer of the second material. The measurement member includes: a member for selectively irradiating a portion of a region to be measured with a radiation cleaning beam, thereby removing debris from this region to compensate for reflectance changes and reduce multiplicative noise and detection The related signal changes of the detector. The system also includes a component for predicting the relative position of the predetermined target and the waist of the laser beam based on the measurement, so as to obtain a predicted relative position. The system further includes: a component for inducing a relative movement between the predetermined target and the waist according to the predicted relative position. The system also includes: a component for generating a laser beam including at least one pulse, and a component for irradiating at least one pulse to a point area of a predetermined target, wherein at least one pulse is sufficient to process the predetermined target.

Furthermore, when implementing the above and other objects of the present invention, in a system for laser processing a target structure of a multilayer multi-material device, a method for controlling the energy delivered to the target structure is provided. This method includes obtaining at least one measurement for each of at least two predetermined wavelengths. The method also includes: determining the thickness of at least one layer of the device based on the measurement. The method further includes: controlling the energy delivered to the target structure according to the determined thickness, so as to compensate for the change in energy required to process the target structure due to the interference effect of at least one layer.

In a system for laser processing a target structure of a multilayer and multi-material device, a control system for controlling the energy delivered to the target structure is provided. The control system includes means for obtaining at least one measurement for each of at least two predetermined wavelengths. The control system also includes: a member for determining the thickness of at least one layer of the device based on the measurement. The control system further includes: a member for controlling the energy delivered to the target structure according to the determined thickness, so as to compensate for the change in energy required to process the target structure due to the interference effect of at least one layer.

When implementing the above and other objects of the present invention, a method for accurately positioning a waist of a pulsed laser beam to compensate for the microscopic position change of a predetermined target is provided. The predetermined target has a mine waiting The first material for injection processing. The method includes: measuring the position of at least one alignment target formed at a predetermined measurement position to obtain at least one measurement. The method also includes predicting the relative position of a predetermined target and a laser beam based on at least one measurement to obtain a predicted relative position. The method further includes: generating a laser beam including at least one pulse, and triggering a relative motion between the predetermined target and the laser beam according to the predicted relative position. The method further includes: updating the predicted relative position during the relative movement according to the updated position information, and obtaining the updated position information during the relative movement. The comparison method further includes: irradiating at least one pulse to a point area of the predetermined target, so as to process the predetermined target according to the updated position information.

In addition, when implementing the above and other objects of the present invention, a system for accurately positioning a waist of a pulsed laser beam to compensate for the microscopic position change of a predetermined target is provided. The predetermined target has a The first material to be processed by laser. The system includes: a member for measuring at least one alignment target formed at a predetermined measurement position to obtain at least one measurement. The system also includes a component for predicting the relative position of a predetermined target and a laser beam based on at least one measurement, so as to obtain a predicted relative position. The system further includes: a component for generating a laser beam including at least one pulse, and a component for inducing a relative motion between the predetermined target and the laser beam according to the predicted relative position. The system further includes: a component for updating the predicted relative position during the relative movement according to the updated position information, and obtaining the updated position information during the relative movement. The system further includes: a component for irradiating at least one pulse to a point area of the predetermined target, so as to process the predetermined target according to the updated position information.

The above-mentioned objects and other objects, features and advantages of the present invention can be more clearly understood by referring to the detailed description of the best implementation mode of the present invention below and referring to the drawings.

Figure 1a is a block diagram of a laser system. The laser system generates a laser pulse in response to a trigger signal obtained from a control system. The pulse has a transient state including a rapid rise and fall time Shape, and a selected time course for the material processing application of the present invention; Figures 1b and 1c are partial cutouts showing a multilayer and multi-material device, in which a laser pulse with predetermined transient and spatial characteristics The device is irradiated; Figure 1b is the first side cross-sectional view of a part of the device, which shows a target structure with a rectangular cross-section, one of which has a high numerical aperture with a non-unity size ratio The laser beam is incident on a target structure with multiple layers to form a stack; Figure 1c is a second side cross-sectional view of a part of the device orthogonal to the first side cross-sectional view, in which a rectangular target structure is shown , And one of the high numerical aperture laser beams with non-unity size ratio is incident on the target structure; Figure 2a is a block diagram of a conventional laser system, which shows a conventional laser beam Q-switched or Gaussian pulse; Figure 2b is a diagram of a conventional multilayer structure, this multilayer structure has a single oxide layer between the link and the substrate so that it is adjacent to a substrate, and a conventional The known q-switched laser pulse is irradiated and roughly fills the narrow size of the target structure; Figure 3 is a graph of a function of the wavelength of a multilayer stack with 14 pairs of 28 layers. This stack is Represents a device processed by a method and system of the present invention; Figures 4a and 4b are a top view and related icons showing the effect of irradiating a target structure with a laser beam profile having a variable size relative to the target structure; Figures 4a and 4b show the censored results of a representative non-uniform Gaussian laser space profile, in which the energy system enclosed by the target structure is strongly affected, the energy at the edge of the target changes, and potentially diffuse radiation The effect comes from the unabsorbed energy of the target structure. Figure 4c is a schematic side view of a plurality of microstructures formed on a layer, which shows that in order to reduce the spacing (spacing), internal reflection and diffusion energy cause the irradiation of adjacent target structures; Figures 5a and 5b are Graphical representation of the reduced irradiation on the display device as a function of the depth caused by the precise position control of a high numerical aperture beam (on the top surface), where the focus depth and position of the beam can be used to process the target structure Without causing undesirable changes in other materials; in particular, Figure 5a shows the difference in the area of the dot area for various spherical and elliptical Gaussian radiation distributions in a representative multilayer stack used in copper memory processing.

Detailed description of the preferred embodiment

In one aspect of the present invention, a microscopic target structure that is a part of a multi-layer multi-material device is removed, in which the laser energy is incident on several materials with different optical and thermal properties. One of the applications is memory repair. A new process (Damascene) includes a copper target structure, a "stacked" multiple dielectric layer, and functional circuits arranged on the dielectric layer. The target structure and layer are usually formed on a silicon substrate, which is shown in Figures 1b and 1c and corresponds to a device processed by an embodiment of the present invention. This is called "multilevel )"program.

The use of finer scales (for example, less than one wavelength of light) and more complex structures improve the reliable operation of laser processing, thereby meeting the high-yield standards of the semiconductor industry.

The type of the present invention includes the operation method and subsystem of the laser processing system. On the microscopic scale, the waist of the laser beam diverges rapidly due to the size of the small spot area and the depth of focus, and the material in the 3D beam position may include functional circuits. In an automatic system, a powerful measurement of the target position is used together with database information to locate a laser beam in three dimensions at a high speed. The interaction of a laser beam in a multi-level device will affect the yield. Thermal interaction can be used to learn and predict the performance of the thermal processing system. But in terms of the microscopic scale, it is also beneficial to understand the interaction system based on physical optical devices in more detail.

The following paragraphs disclose the detailed types of spatial and transient pulse shaping, three-dimensional measurement and prediction, device profiling and programming, and the focus is on solving the problem of cleanly removing the links on a multi-level device. The damage to the inner layer and the functional circuit between a link and the substrate can be avoided. However, various methods, subsystems, and experimental structures can also be applied to the processing of conventional links of a single inner layer device, and are generally applicable to processing microstructures surrounded by materials with different thermal or optical properties.

Link processing on a multi-level device

A pulsed laser beam is used to cleanly remove at least part of a target structure, the laser beam having predetermined features for processing microstructures. One application of the method and system of the present invention is to cut off a highly reflective copper link that is a part of a high-speed semiconductor memory device. The method and system of the present invention are particularly advantageous for processing objects with sub-micron size, including objects with a size smaller than the wavelength of the laser beam. The target is separated from a semiconductor substrate by a multilayer stack that may have several dielectric layers. In addition, it is possible to select or control the transient and spatial characteristics of the pulse based on the thermal and optical properties of the microscopic target, the material of the covering layer, and the three-dimensional layout of the device structure including the functional inner conductor layer and the interval between the target structure. .

Figures 1a to 1c generally show an embodiment of the present invention. A laser pulse 3 irradiates a rectangular target structure or microstructure 10, and the side view thereof is shown in Figures 1b and 1c. In a preferred embodiment, an output from the short-pulse amplifying laser system 1 is generated to generate pulse 3, which has a fast enough rise to efficiently couple energy into a highly reflective target structure Time 4. Time course 5 is sufficient to process the target structure, in which at least a part of the structure is cleanly removed without leaving residues, slag or other debris, and time course 6 is preferably fast enough to avoid formation of defects on the layer or substrate damage.

The transient pulse shape is selected partly based on the physical properties of the target microstructure 10 (such as thickness, optical absorption, thermal conductivity, or a combination of the above). In an advantageous embodiment of the present invention, the shape of the transient pulse is relatively small. The selected pulse duration of microseconds is processed by a single pulse with a fast edge leading. In an alternative embodiment, the laser output may be a series of narrow q-switched or rectangular pulses with a fast rise time, for example, an 800ps pulse representing the output of a commercially available q-switched microlaser. The pulses may be delayed relative to each other to provide a hair-like pulse to irradiate the target structure. The laser output may be generated by a combination of a high-bandwidth seed laser diode with a Raman shift and a fiber amplifier, or by a waveguide amplifier system. Alternatively, an ideal pulse characteristic may be equipped with various modified q-switched systems, or use high-speed electro-optical modulators. Other pulse shapes may be selected according to the requirements of material processing. For example, Reference Document 5 discloses a series of closely spaced pulses with a time course ranging from a few picoseconds to a few nanoseconds.

In one embodiment, a high-bandwidth MOPA structure is used to amplify the laser output of a high-speed semiconductor diode. It is an advantageous way to directly modulate the diode to generate various pulse shapes and time durations. The limitation is that any influence associated with the variable amplitude drive waveform will not affect the overall performance. For further details of the various types of pulse generation and amplification, please refer to References 5 and 6 (for example, Figure 5 and columns 14 to 16 of No. '471 (Reference 5)).

As mentioned above, embodiments of the laser system may include a fiber amplifier to amplify the preferred square pulse shape generated by a sub-laser. The seed laser may be a high-speed semiconductor diode or the finalized output of a modified q-switched system, as disclosed in reference documents 4 and 6 (for example, in reference document 6 in Figures 12 to 13 and column 14 Line 57 and column 19, line 3), the amplified output may be wavelength-matched to the input or after Raman shift. The wavelength shift of a short pulse q-switched laser output is generally disclosed in Reference Document 4 of '759.

In an alternative configuration, the seed laser is a semiconductor diode and the optical amplifier is a waveguide amplifier. Compared with the optical fiber system, the advantages of an embodiment with a waveguide amplifier include: avoiding Raman shift, operating speed with lower pulse distortion, and having minimal thermal lensing through proper design . A sophisticated anamorphic optical system is used to optimize the interaction between the seed and the amplifier. For the basic description of the waveguide amplitude and laser, please refer to the product documentation provided by Maxio, Inc.s and Beach. "CW and passive Q-switched cladding pumping planar waveguide laser" by et al. Other amplifier systems including the 28DB planar waveguide amplifier used at 1.064 micron wavelength were developed by the University of Southampton and described in "Diode Pumped High Gain Planar Waveguide, Nd: Y3A15O12 Amplifier".

In an alternative configuration, in order to generate a fast rising pulse or other desired shape, a plurality of q-switched microlasers can be used. The module generates a q-switched waveform with a pulse duration of about 1 nanosecond or less, such as 800ps to 2ns of a commercially available unit. An example of a commercially available laser is the AOT-YVO-1Q from Advanced Optical Technology (AOTLasers.com), which can be triggered by a variable repetition rate and a TTL pulse to trigger these recently developed short circuits. Pulse active q-switching laser, while keeping the specified sub-nanomicron timing shaking. Generally speaking, when the repetition rate is close to the maximum rate, the shape of the pulse incident on the target microstructure will change significantly. Reference Document 9 discloses a method for maintaining a fixed pulse shape regardless of the transient interval of the pulse incident on a target (for example, please refer to the diagram and related specifications). AOT provides a pulse width of 2 nanoseconds that can be used at a repetition rate of 20 kilohertz, and it can also have a frequency multiplier version (532 nanometers). The US IMRA proposes to use 800ps pulses of the PicoLite system, in which high peak power is obtained by optical fiber amplification at a repetition rate of up to 10 kHz. In the case of a slower repetition rate, a shorter pulse width of, for example, about lns or less can be used.

As known in this art and as shown in Reference Document 5 (for example, Figures 1c and 2), the q-switched waveform may approximate (at least the first order) a symmetric Gaussian shape, or one with an exponent, depending on the stored energy. The fast rising pulse of the sexual tail. Referring to Figures 15a to 15c, a series of devices with appropriate delays introduced by a plurality of trigger signals or a delay of a trigger signal with a delay line are used to generate a series of separated pulses. The optical output is preferably combined with an appropriate volumetric optical device (polarization sensitivity), optical fiber or waveguide to form a single output beam. The generated additional q-switched waveform produces a fast rise time characteristic and shorter time. To increase the output power, an optical amplifier 122 can be used as needed.

Figure 15a shows a schematic diagram of a basic embodiment with a volumetric optical device, in which a beam combiner 123 is used to deliver the output of two lasers 120, 121 to an amplifier 122. A programmable delay circuit 126 controls the trigger, and uses polarized optical devices 127 and 128 to provide appropriate inputs to the beam combiner. In one configuration, the pulses are separated and appear to exhibit a high frequency burst 124. In a second configuration, the second pulse is triggered at a slightly delayed (but controlled) position, resulting in an approximately square pulse shape 125 characteristic. In the latter configuration, the controlled delay is approximately 50% of the FWHM. Those familiar with this technique will understand that alternative configurations with multiple amplifiers, combiners and bulks, optical fibers, or integrated optical configurations can be used.

The generation of multiple pulse waveforms may also include: some forms of active q-switching of two separate microlasers, or detection of a first pulse from a passive q-switched laser, and subsequent triggering of an active q relative to the first pulse Switching laser or MOPA.

Figure 15b is a basic schematic diagram showing the use of a single laser 140, in which the laser output is divided by the beam splitter 142, so that a part of the beam propagates along a path 141, and then it can be a half wave. The rotator 146 of the plate is combined by a combiner 143 after polarization adjustment. A selective optical amplifier 145 can then be used to generate higher output power.

In a configuration using a single laser and an optical delay line, the optical system is preferably stable and easily aligned. Figure 15c shows an exemplary embodiment in which opposing corner cube retroreflectors 130 are used to make the installation process insensitive to the tilt of the folding element. The angular alignment of the delayed beam paths 131, 132 is very stable in a high vibration environment. At first, the corner cube in each pair of retroreflectors 130 is adjusted in the X/Y translation and Z rotation directions to set the lateral position of the delayed beam path at the center. The λ/2 blocker 133 in the main beam path is adjusted so that the vertically or horizontally polarized light will rotate the polarization by 45 degrees. The λ/2 blocker 133 in the second delay loop is adjusted to make the vertical or horizontal polarization. The light rotates the polarization by 90 degrees, causing the delayed pulse in the second loop to circulate twice before leaving. The peak-to-peak interval of the output waveform 135 (for example, 4 combined pulses) is controlled by the length of the delay loop. If unequal amplitudes are required for the delayed pulses, the λ/2 blocker 133 in the main beam can be set for polarizations other than 45 degrees. Likewise, the shape of the pulse can be changed when a system is installed or possibly during manual or automatic control interval operations. Those who are familiar with the laser pulse generation and shaping skills can understand the advantages of a modular configuration that does not occupy a volume for short pulses with typical delays ranging from a few nanoseconds to tens of nanoseconds. For example, U.S. Patent No. 5,293,389 issued to Hitachi describes a polarization-based fiber delay line that is used to generate laser pulses with reduced amplitude to generate longer pulses such as 100 ns or longer.

Another component used to generate a shaped pulse is to use a modulator to cut off the leading edge or tail of the pulse, so that it has a two-stage or shaped modulated voltage pulse. For example: for a 10ns q switching pulse, the modulator can have 100% transmission in the first 1 to 5ns, and then the remaining pulses have 25% transmission. Koechner (US Patent No. 3,747,019) and Smit ( The early pioneering work of Smart) (US Patent No. 4,483,005) showed an exemplary amplitude and pulse shape control method using electro-optical modulators.

The multiple pulse systems shown in Figures 15a to 15c may or may not have the same wavelength, and the transient shape of a pulse may change according to specific requirements. For example, in certain embodiments, an output may be a q-switched pulse with short duration and high peak power and a lower power square pulse shape.

Referring to Figures 1a and 1b, during the operation of the memory repair system, the position information obtained by a precision measurement system is used to relatively locate the focused beam waist of the pulsed laser beam at one of the positions 7, 8, and 9 in space. , So that the target 10 and the three-dimensional coordinates (X link, Y link, Z link) roughly coincide. A trigger pulse 2 generated when the waist of the laser beam roughly coincides with the target position is operated together with the laser and the related control circuit in the laser subsystem 1 to generate an output pulse.

Reference documents 2 and 7 describe the details of a precision positioning method and system (including three-dimensional waist positioning). Reference Document 7 describes a preferred embodiment that can generate an approximate diffraction limited spot size with an adjustment range of the spot size (for example, Figures 7 to 9 of WO 0187534 (No. '534) and related specifications), and A better method and system for three-dimensional positioning of the waist. For example, three-dimensional (height) information is obtained by focusing detection, which is used to estimate a surface and generate a trajectory (for example, Figures 2 to 5 of No. 534 and related specifications). The laser system is pulse-shaped at a position roughly corresponding to the three-dimensional position of the link (X link, Y link, Z link) (for example, Figures 10a to b of No. 534 and related specifications).

In fact, three-dimensional measurement and positioning are used to compensate for topological changes on the wafer surface or other position changes (misalignment) introduced into the system. These changes generally depend on the system or application and may exceed a few microns, thus exceeding the focal depth of the focused laser beam. In some micro-machining applications, if certain tolerances are maintained or if external hardware is used to manipulate the position of the device, it may have looser system positioning requirements. The device may include a micro-portion (such as a single die) positioned at a predetermined reference position by an external micro-positioning subsystem. Similarly, if a miniature part has a predetermined tolerance, the positioning may be based on a single measurement of a reference position, or it may be a single depth measurement combined with lateral (X, Y) measurements. In order to process multi-level devices (such as 300-pin) on a wafer at high speed, especially when the size of the link is reduced, it is expected that the performance can be improved by densely sampled three-dimensional information.

In applications that require extremely high-speed operations on a large surface (e.g., 300 male wafers), an alternative method is to combine the information that may be predetermined (e.g., the plane of a wafer chuck relative to the measurement during a calibration procedure). One beam of the locator action plane) is combined with the size information obtained from each component to be processed. For example, in No. 534, figures 1 to 2, the inclination of a certain proportion of area 28 may be related to treatment. For example, these steps may include (a) obtaining information on microstructures designated for removal, (b) measuring a first set of reference positions to obtain three-dimensional reference data, and (c) generating based on at least three-dimensional reference data A trajectory to obtain a prediction of the position of the waist and the surface of the microstructure, (d) update the prediction during the relative motion based on the updated position information, the updated position information is obtained from a position sensor (such as Encoder) and/or from data obtained during the relative motion. The additional data may be measurement data obtained at additional alignment targets or other positions suitable for an optical measurement (such as dynamic focus). Reference 2 describes a system in which a precision wafer stage is used to position a wafer at high speed. A method for obtaining feedback information with a resolution of less than 1 nanometer by using an interferometric encoder is disclosed, and this high-precision method is a better method. It is noted in Reference 2 that other conventional laser interferometers can also be used. Figures 9 to 11 and columns 5 to 6 of Reference 2 describe the types of precision measurement subsystems related to precision positioning devices. In addition, a designated reference position on the workpiece (such as a wafer) can be used for various applications, and the designated reference position may be an x, y alignment target or an area suitable for three-dimensional measurement. It should also be noted that Nikoonhad et al. proposed an agreement in Optical engineering, Vol.34, No.10, "On-site height correction for laser scanning of semiconductor wafers" in October 1995. With a high accuracy of 0.1 micron, one of the optical position sensors obtains the average height data for the area at a high speed. Similarly, it is possible to use a dynamic focus sensor (such as the astigmatism system used for disc tracking and control) to obtain altitude information. The limitation is that the data rate is fast enough to support "on the fly". Measurement.

Various combinations of the above technologies can be used according to the needs of the application. A combination may depend on the typical distribution and number of microstructures on a device designated for removal. When a large number of repair parts are distributed in a device, it is possible to maximize the output rate by "on the fly" update.

In one application of the present invention, the target structure 10 is configured as a part of a multi-material multilayer structure (such as a redundant memory device). The multilayer stack with dielectric layers 14, 15 provides a space between the link and a coated substrate 17. In a type 1 multilayer memory device, the alternating layers of silicon dioxide 15 and silicon nitride 14 may be arranged between a copper link target structure 10 and a silicon substrate 17, and the copper target structure is generally adjacent to other similar structures. Forms the focal point of one of the 1-dimensional or 2-dimensional arrays designated for removal. In addition to the copper link structure, the covered conductor 16 configured as part of the functional device circuit may be adjacent to the link structure and placed on the thinner (usually <0.1 micron) silicon nitride 14 and the thicker (usually <0.1 micron) silicon nitride 14 1 micron) silicon dioxide 15 material covered in a sequence pattern.

The radiation distribution on the link may roughly conform to a diffraction-restricted circular Gaussian profile. In another effective embodiment, the beam system has an approximately elliptical Gaussian irradiation profile that may be produced by an anamorphic optical system or by a non-circular laser output beam. In one embodiment, as also shown in Figure 4b, the incident beam has an uneven size ratio of 12, 11 (for example, 3:1). Alternatively, it is possible to implement a rectangle or another optional spatial profile in the lateral dimension. For example, Reference Document 1 discloses various advantageous methods and optical systems for "non-Gaussian" spatial shaping laser beams in memory repair applications.

In the case of a diffraction-restricted elliptical Gaussian beam, the preferred minimum beam waist size at position 11 is approximately the size of the narrow target 10 in Figure 1b, thereby generating a high pulse energy density on the link. Moreover, in this method, a high proportion of laser energy is coupled to the link, and the background radiation is reduced.

A typical copper link used in this memory has a width and thickness of about 1 micrometer or less (for example, 0.6 micrometer) and a length of about 5 micrometers. Future memory requirements are expected to further reduce the target size. The minimum waist width Wyo at 11 usually overflows the micron link to some extent, and the size along the link is larger than Wxo/Wyo12, 11 (where Wxo is several microns), which facilitates clean removal of the link. In addition, due to the defocusing of the high numerical aperture beam portion 11, the energy density on the layers 14, 15 and the substrate 17 can be rapidly reduced.

Figures 5a and 5b show the defocus estimates for various size ratios of a circular Gaussian and an elliptical beam with the best focus. Figure 5a shows a very fast drop of a 1.6 micron circular Gaussian (0.002 mm numerical unit = 2 micron). Figure 5b shows the results of the normalization of the scale of the energy density with the best focus for different spot shapes. These results show that since the depth of the beam is precisely positioned and the target site has the largest power density, for a memory One of the exemplary multilayer stacks of the copper base process produced has a relative energy density reduction greater than a power series of 10 at the base material level. In addition, the fast defocusing system for the waist Wyo is beneficial to avoid damage to the inner layer, and the restriction condition is that the "tail end" of the incident beam irradiates the functional inner layer 16 (such as copper) with a low value.

In an embodiment for processing a multi-level device, copper link removal is initiated by applying a fast rise time pulse, which has a nominal value in the preferred range of less than 1 nanosecond to about 2 nanoseconds 10 to 90% rise time4. Preferably, a pulse duration 5 of about 2 nanoseconds to 10 nanoseconds is used to cut off the link while limiting thermal diffusion. The range of about 0.1 microjoules (μj) to 3 microjoules is the effective pulse energy, and a preferred typical range of about 0.1 to 5 microjoules is regarded as a sufficient boundary between the shape of the spot area and the change of the program. It is possible to select a better pulse duration based on the nominal thickness of the link, or based on one of the different thermal and optical properties of adjacent materials. During the pulse duration, the thermal shock of the top layer 13 and the thermal expansion of the target 10 cause the broken links of the top oxide layer 13 to explode, thereby reducing the stress at the lower corners of the link structure adjacent to the layer 14. The laser pulse is preferably terminated quickly within a few nanoseconds of fall time 6 after the explosion, which is just after the thin link is cleanly cut and before the lower corner of the link causes at least layer 14 to rupture. Further details and results of the interaction of a laser pulse with a metal link and overlay are disclosed in References 4 and 5. The '471 patent and related specifications describe the interaction procedure (for example, No. 1a, 1b). , Figures 11a, 11b and column 18).

Therefore, a combination of spatial characteristics (such as waist shape and position) and transient state (such as rise time 4 and time duration 5) pulse characteristics are to avoid undesirable rupture of the lower layers 14, 15 and significant pulses with the inner conductor 16 Interaction and limit the heating of the substrate 17. Therefore, although the copper link has high reflectivity at visible light and near-infrared wavelengths and is expected to be incompletely removed and damage the surrounding structure and substrate in the prior art, the target structure is processed without adversely damaging it. Other structures. It is also known that copper not only has a near-maximum reflectance in the near infrared, but also has higher reflectivity than other link materials (such as aluminum and platinum). However, due to the optical interaction between the near-infrared beam and the target and the optical and thermal properties of the adjacent (layup) layer, better copper materials can be processed.

In addition, the near infrared (IR) wavelength has traditionally corresponded to the wavelength of the available high-bandwidth laser diodes, and corresponds to the optically amplified spectral range of the pulsed laser beam that can be efficiently generated by optical fibers and waveguide amplifiers. Those familiar with the art understand that an amplified laser diode output with a desired transient pulse shape can also multiply the frequency to produce a visible laser output when needed. The fast rise time of semiconductor diodes is particularly conducive to generating a fast rise time, square pulse characteristics. The future development of visible diode and optical amplifier technology may support direct pulse amplification in the visible range.

In a preferred system for copper link blowing, for this processing technology, the link width is a very small part of 1 micron and the link spacing (spacing) is several microns. The width of the link may generally correspond to the wavelength of visible light, and, in terms of microscopic operating scales, the lateral and/or thickness dimensions of the material in Figures 1b and 1c are about the laser wavelength, and the refraction of the stacked materials The rate and thickness will significantly affect the overall optical characteristics of the stack.

In an embodiment of the present invention, in the visible light or near-infrared range where a non-absorptive optical property (such as interference or reflection loss) of the layer is fully utilized, a better reduced wavelength is selected. The device structure of Figures 1a and 1b will be damaged and will have significant absorption in the lower layer. This damage is unacceptable because of adjacent circuits. This is different from the link processing of the prior art system shown in Figure 2b, in which damage to the inner layer usually does not detract from the performance of the overall device.

US Patent No. 6,300,690 (Ref. 8) describes a system and method for evaporating a target structure on a substrate. The method includes providing a laser system that can produce a wavelength with an absorption edge lower than that of the substrate. The laser output. In addition, Reference 4 discloses the benefits of a wavelength less than 1.2 microns for processing links on memory devices, in which the substrate is silicon, that is, it has a smaller spot size and a shorter laser pulse width. . According to the present invention, the non-absorptive stacking properties of the selected wavelength can be fully utilized to achieve improved performance. Moreover, at least one of the precise positioning of a high numerical aperture beam, the spatial shaping of the spot area, or the transient pulse shaping will also provide reduced energy on the substrate. The result corresponds to a lower value of the energy expected to be deposited on the substrate, and is independent of an incident beam energy required to be deposited in the target structure sufficient to evaporate the unit energy of the target structure.

The influencing factor for the energy deposited in the substrate is actually multiplicative. Similarly, at short visible wavelengths, copper is absorptive (for example, compared to 2% of 1.064 microns, it is 50 at about 500 nanometers). %, 70% at near UV), so clean removal requires at least an order of magnitude smaller energy. The better-recognized wavelength corresponding to a lower energy value expected to be deposited in the substrate is in a visible or near-infrared region of the spectrum. It is possible to use a profiling-based method to estimate the shortest wavelength with sufficient boundaries for a specified dielectric stack, spot location, tolerance, transient and three-dimensional spatial pulse characteristics.

In order to process the links of multi-level devices with silicon substrates, the limiting wavelength corresponding to the lower energy value expected to be deposited in the substrate (for example, lower than the image critical value) may be located in the green or near the spectrum. In the ultraviolet region, however, this usage method may be a system parameter that needs to be closely controlled, including possible control of the thickness or refractive index of the stacked layer.

By selecting the wavelength according to the present invention and in which the internal transmission and preferred reflection of the stack are at or near the maximum, damage to the stacked layers can be avoided. Moreover, it is a better way to reduce the radiation of the substrate while providing a reduced spot size (at or near the diffraction limit) for the removal of the link. The limitation is that the radiation of the functional inner layer is located in the Within the acceptance limit. Generally, the spectral transmission system of a larger band gap dielectric material shows that the transmission will be slightly reduced at the ultraviolet wavelength. For example, in the manual of laser science and technology, the transmission range of silicon dioxide is specified as a wavelength greater than 0.15 microns, and the absorption coefficients of silicon nitride and silicon dioxide remain low in the visible light range (>400 nanometers). Gradually increase in the ultraviolet range.

Figure 3 shows the estimated back reflection produced by a representative multilayer stack of 14 pairs of silicon dioxide 15 and silicon nitride 14 in the near-infrared wavelength range, where the layer thicknesses are approximately 1 μm and 0.07 μm, respectively. According to the present invention, a large number of layers can be accommodated and it may be 4 to 28 layers depending on the procedure (for example, a functional conductor layer may sometimes be separated by multiple layers).

The example shows that significant reflection occurs over a wide wavelength range. A single layer configured as an inner layer 14 usually reflects about 2% on each surface in the visible and near-infrared spectrum. It is well known in the semiconductor processing and linking technology that the silicon absorption system changes with the order of magnitude in the near-infrared spectral range. In addition, research on silicon material processing has shown that as the laser power increases and the substrate is heated at a wavelength close to the absorption edge, the absorption becomes unstable and nonlinear, and shorter wavelengths are better to produce The smaller dot area (refer to documents 4 to 6 and 8) and the higher energy concentration of the link position.

According to the present invention, full utilization of layer reflection by wavelength will further enhance system performance, and supplement related advantages for transient and spatial control of pulses in a preferred short wavelength range. For the wavelength that would greatly increase the absorption of the substrate, this choice of wavelength is considered to be particularly advantageous, and a significant boundary can be obtained when the number of layers 14, 15 arranged between the link and the substrate substantially exceeds the number of overlay layers 13. . A preferred structure for processing includes a large number of layers, where a predetermined short wavelength has a large reflection coefficient, and this wavelength is well matched to produce a better fast square transient pulse shape.

The standard laser wavelengths in the range of Figure 3 include 1.047 microns and 1.064 microns, the latter being a standard wavelength of semiconductor diodes. And, the usual wavelength includes 1.08 microns, and other wavelengths are generated with Raman shift. Those familiar with this art understand that frequency multiplication of near-infrared wavelengths can be used to generate short wavelengths, and multiple wavelengths can be provided in a single system through appropriate design. For example, by doubling the frequency of a near-infrared laser, a better transient pulse shape with a fast rise time can be generated in the green part of the visible spectrum.

In an alternative embodiment, wavelength adjustment is used to match the wavelength to the approximate peak reflection coefficient of the stack. This configuration may be particularly advantageous for adjusting one of the laser wavelengths at the edge of the reflection coefficient range (ie, the "cut-off" range) on a limited wavelength range, so as to avoid sensitivity to tolerances of refractive index and material thickness. As mentioned above, please refer to References 4 to 6 for further description of the laser amplification system and the application of other link structure.

The generation of a pulsed laser beam may include the step of shifting the laser beam from a first wavelength to a predetermined wavelength. The predetermined wavelength may be based on material characteristics including at least one of the following: (1) Coupling microstructures Features, (2) multilayer interference, and (3) substrate reflectivity.

Experimental results have shown that at a wavelength of 1.047 microns with silicon absorption on the order of 1.2 microns, the stacking feature in Figure 3 and a short q-switching (standard) pulse prevent damage to the substrate. However, the result of a standard laser with a q-switched transient pulse shape shows a crack of an oxide layer 14 under the link. According to the experimental results, the slower-rising q-switched pulse shape, which is a significant proportion of Gaussian approximation, is regarded as a limiting factor for link removal without causing cracks in the inner layer. However, according to the principles of the prior art, because the absorption system is several orders of magnitude higher than a wavelength corresponding to the maximum transmission, serious damage to the silicon substrate is expected to occur at a wavelength of 1.047 microns. According to the principles of the present invention, spatial pulse characteristics and stacked reflection are important considerations for avoiding damage to the inner layer and substrate and short operating wavelength (which also provides smaller spot size and higher energy concentration on the link) factor. Moreover, according to the present invention, a predetermined square pulse shape generated at a laser wavelength of 1.047 microns is expected to produce clean removal without undesirably changing the stack and the substrate.

Sub-micron laser processing and processing design

Also, in an exemplary advantageous embodiment of short-wavelength processing of reflective microstructures, specifications for multi-layer stacking can be considered in program design. For example, alternating dielectrics or other suitable configurations with too different refractive index and high transmission within each layer, quarter-wave stacking designates an optional wavelength. It shows that a high reflection coefficient can be achieved, and it is easy to calculate the quarter-wave stack in a closed form and method. Therefore, the method and system of the present invention can be effectively used by other types of programming. The method and system of the present invention are beneficial to the situation where the deep buried layer and the substrate have higher absorption, or the width of a target structure Significantly lower than the laser wavelength.

The design of the device structure may have specific restrictions on the circuit layout. Therefore, a specific thickness and material may be defined for a specific layer, such as an insulator that is located in a conductor plane and has an approximate conductor thickness or is related to the conductor thickness. It is possible to choose a material with a refractive index different from that of the specified layer. The specified thickness may be based on the estimated reflection value of a favorable laser wavelength. It is required for special laser equipment operating at "exotic" wavelengths. It is possible to use a kind of profiling process to estimate the reflection, where the thickness is a variable and the estimation is made to maximize the reflection as much as possible and is limited by other devices.

The thickness of the layer can be adjusted to a wavelength, as long as the wavelength (or angle) can be adjusted according to the layer. The refractive index can be used for fine-tuning in a limited range, but this range may not be obvious for small changes in the refractive index. Even if all thicknesses are fixed by the program, the addition of one or more variable thickness adjustment layers with a predetermined thickness can still be used to significantly affect the reflectivity of the overall stack. For example, a layer that is not restricted by the requirement of metallization can be used as a precision spacer between an upper and a lower stack, which can become a very powerful tool that can adjust the processing by adjusting only one layer.

Laser processing and physical optics of multi-level device

It is possible to make full use of other controllable laser features by adding or extracting wavelength selection to further improve the processing energy window. Reference 3 describes an advantageous method and system for polarization control, including dynamic polarization selection and computer control to align the polarization with the orientation of a link (for example, the details shown in Figure 4 and the related description of the reference file) . The polarization can be selected based on the target coupling characteristics, reflection coefficient, or a combination of the above.

With the link size smaller than the dot area size, effects such as diffraction, scattering, and edge reflection should be regarded as physical phenomena. These physical phenomena have beneficial or harmful results depending on the device geometry and beam characteristics. Likewise, a high energy density, non-linear absorption may affect the results, especially the damage related to semiconductor materials.

A particularly important consideration of the fine spacing (spacing) of adjacent links and circuits is the accompanying damage. In addition, the functional circuits located in the first layer of the plane cannot be damaged. Due to the development trend towards fine pitch and high-density memory, the three-dimensional structure of the device should be considered and it may affect the choice of beam spatial and transient characteristics. In an example, Figures 4a to 4c show the diffraction and reflection effects of the Gaussian beam 11 of the censored 43 and 44 caused by the sub-micron width link 10, in which the size of the dot area (measured at the 13.5% point) is Various degrees are wider than the link. The depiction in the figure represents a diffraction-restricted beam waist at a near-infrared wavelength, and the central leaflet is clamped by the link, which seems to show a near field mask (near field). obscuration) resulting in punctured transmission beam portions 43, 44. The energy that is not incident on the link member may propagate into the layer 49 at a wide angle. This method may be advantageous from the viewpoint of avoiding damage to the substrate 17 as shown in FIG. 1. In any case, the adjacent irradiation has some correlation with the size of the spot area. The large spot area with a larger focal depth has reduced divergence and the adjacent irradiation may be smaller. The limitation is: link spacing It is large enough so that the incident beam impinging on an adjacent structure has a weak unabsorbed energy 43, corresponding to a value of 44, for example. With a higher numerical aperture and a smaller spot size, the diameter of the reflected beam at the link position 46 is increased. The size 41, 42 has a maximum value for a part of the dot area, and then as the reflected energy in the area increases, the radiation on an adjacent link 48 decreases.

At the same time, the internal reflection has an angular change. Therefore, the thickness of the stacked layer can also affect the irradiation of adjacent structures, including the internal structure of Figure 1. Also, the polarization change of the angle is expected to change. Figures 6a and 6b show examples of geometric ray tracing effects of internal reflections propagating on an extended area.

Similarly, as shown in Figure 4c, if a part of the laser beam 54 incident on the edge of the link 46 is considered, the energy not coupled to the link structure may also be scattered and/or specularly reflected to the adjacent link Piece 48. Since at least the physical edge profile of the link 46 may be slightly curved or inclined, internal reflection 47 occurs.

Another additional consideration is the three-dimensional space between an inner conductor layer 16 in FIG. 1, the waist portion 11, and adjacent links 48 in FIG. 4c. In a preferred situation, the beam waist 11 with a large numerical aperture can produce the smallest dot size on the link, and its divergence and reflection mode can avoid significant interaction with the inner layer 16. Looking at Figures 4a to 4c, it is understood that the reduced spot size with controlled precision 3D waist positioning is expected to reduce the accompanying damage by maximizing the energy coupled into the link. With the high containment energy and a low-strength transmission profile 44 in the link, edge reflections can be minimized. The spatial profile should be selected under the limitation of only low-value negligible interaction between the beam angle distributions 16.

The interaction mechanism related to a part of the three-dimensional device structure is preferably subjected to a profiling process to select at least one spatial pulse characteristic, which may be numerical control and the position of the waist. Preferably, the profiling system includes an estimate of the radiation seen by each adjacent link structure 48, the inner layer 16, and the substrate 17. Although the conventional microscopic inspection can clearly see the damage to the adjacent link structure, the damage to the inner layer 16 and the substrate 17 of the 3D device structure is obviously more difficult to estimate.

For link widths less than 1 micron and pitches of several micrometers, precise, sub-micron alignment is required to compensate for wafer-to-wafer variations, local variations within wafers, and system tolerances (for example, with 25-micron topology changes and 300mm wafers with a manufacturing tolerance of 5 microns). According to the present invention, a precise positioning method and system are used to relatively position the beam waist to provide higher laser energy concentration on the link. In addition, an important consideration for precise positioning is to predict accurate (X link, Y link) location information. Subsequently, during the relative motion of the laser beam and the target 10, a motion control and positioning system uses this prediction to generate a laser output via the trigger 2 of the target coordinates. A preferred embodiment includes a scanning and detection system that is insensitive to polarization as described below, in which an area containing an alignment target is imaged to obtain reference data. The target location is often covered by a dielectric layer of silicon dioxide, silicon nitride or other insulating materials. Experiments have shown that a detection system that is not sensitive to polarization is beneficial to avoid spurious measurements. measurements), the result will lead to the following hypothesis: the birefringence is introduced into the insulating layer due to polishing or other processing operations and becomes prominent due to polarization changes in the reflected beam. These changes reduce the signal-to-noise ratio and seem to cause positional distortion. An 8-parameter minimum square alignment algorithm is used to use the digital output data from each target position to estimate and correct the link that contains the link to be processed. Position information on the wafer affected by offset, angle, scale, orthogonality, and trapezoidal changes.

Due to the change of the received beam at the known target position, it is necessary to worry that the program change may affect the optical properties of the layer close to the target structure. Moreover, it is actually a change in the thickness and reflectivity of the target and layer of a wafer to be processed or wafer lot by lot. The thickness and reflectance measurement system can be effectively used for monitoring and can also be used to determine the adjustment of laser power and wavelength to increase the energy window. For example, any change in the reflectivity of the link will affect the energy required for processing. The following describes a better method and system for switching energy control.

As the size of links and other microstructures continues to shrink rapidly, those who are familiar with this technique have learned the benefits of multi-parameter profiling. A method based on profiling results in the selection and precise control of the spatial and transient characteristics of the laser output, resulting in a controlled three-dimensional interaction between the laser and the complex multi-layer and multi-material structure.

For polarization-insensitive detection and X, Y reference measurement

The commercial laser system of the assignee of the present invention uses a beam splitter to position the laser relative 152 on the alignment target (such as a reference) to remove a part from the work surface (such as a multi-layer memory device) Reflected light. The block diagram of the subsystem is shown in Figure 10. The reflection/transmission (R/T) division of the beam splitter 150 depends on the laser used. In the case where the laser has a low total energy and needs to have sufficient transmission , Is a division of 90% transmission and 10% reflection, so 90% is supplied to the work surface during the entry and 10% of the reflection is discarded, but this only removes 10% of the reflected light, and 90% of the reflected light is The laser path is transmitted back. If possible, make a 70/30 split, which will provide a smaller total energy for the work surface but a higher reflection signal.

Regardless of the R/T division, this specification is: R/T for S polarization = R/T for P polarization (within 5%), which can be achieved by a special dichroic coating that produces good results Achieve this effect. Because any polarization state is regarded as a vector sum of S and P, the beam splitter operates with the correct R/T ratio for any polarization.

Since the polarization is switched to any desired state in a better link processing system to improve the cutting effect of the link, the above effect is very important. For example, in the US Patent Application No. 01/013,956 filed on December 13, 2001, which is a continuation application of US Patent No. 6,181,728 (Reference Document 3) and assigned to the joint examination of the assignee of the present invention, particularly in the point When the area size is reduced, the polarization is perpendicular to the link, which results in an improved processing window. The better polarization controller disclosed in the '728 patent is used to switch states.

When there is an oxide layer above the scanned and measured target, the method and system of the present invention are advantageous. The oxide layer may affect the polarization of the beam, which may occur because the oxide layer bears stress and generates birefringence. But because it has a configuration that is insensitive to polarization, it will not cause problems. No matter how the polarization is changed, the same reflection coefficient and the same signal value can be obtained from the beam splitter. If a more common polarization beam splitter or a simpler coating is used for the beam splitter, the changed polarization will result in a change in the reflected signal. If the stress of the oxide layer changes especially on the target microstructure (which may be stressed by passing on the edge of a target), the polarization may change when the beam scans the target. Similarly, because of the coating, it will not cause problems. In the case of a polarization beam splitter, the reflected signal 151 measured on the detector will change due to the polarization change generated while trying to collect the edge data, so the result is uncontrollable and unpredictable distortion.

This polarization-insensitive technology is regarded as the most powerful method, and is the preferred method for measuring targets covered by at least one oxide layer. However, other imaging and edge location methods may also be used, but more complex measurement algorithms are required to accurately measure the target in the presence of multiplicative image noise.

Measurement of abnormal reflectance changes-using a pulsed laser beam to clean

A typical alignment target 100 is depicted in the schematic diagrams of FIGS. 14a and 14b. The target 100 is usually covered by one or more passivation layers that may correspond to but not limited to the layer 13 of FIGS. 1b and 1c. During the experiment of multi-level removal of the link, a better measurement method that is insensitive to polarization was used to obtain the X, Y target position. However, it is found that the residual solder flux that may come from nearby solder deposits (solder balls) and the debris 1001 located in the target area 100 will significantly affect the reflected signal obtained by a detector and cause the noise profile 101. In the smallest square fit algorithm used to estimate the position, the effect on the measurement is displayed as a large residue. This figure shows the target area as a positive contrast (for example, a higher measurement density) area, but you are familiar with this The artisan understands that reverse contrast is acceptable, but the limitation is that the target 100 and the background have a suitable contrast for measurement.

Using a pulsed beam with low peak power to remove debris, an enhanced exemplary signal profile 102 (such as relatively uniform intensity and substantially debris-free) is obtained by the cleaning operation shown in Figure 14b area). A typical clean energy system is about 0.01 microjoules, such as 0.005 microjoules. This is significantly lower than the damage threshold of the material and significantly lower than the typical energy used to remove the link 12.

In one embodiment, a single linear scan or a plurality of linear scans 104 across the target 100 are used to obtain reflection intensity data. The reflection intensity data is, for example, determined by determining% intensity change or standard deviation for statistical analysis to measure authenticity. In an exemplary embodiment, data is obtained approximately every 0.001" along the line 104. However, the sample space may be made finer or rougher depending on the realness of the obtained signal. If the interval is too fine, additional " Texture noise". If it is too rough, it will reduce an edge contrast 107, or introduce errors due to insufficient sampling. If there are too large changes, start a cleaning operation with a pulsed beam. Preferably, by yourself Acousto-optical modulator (also known as the "energy controller" in Figure 13) is one of the standard components of laser processing equipment to control the laser power. It is used for intensity control and pulse selection in a link blowing system The operation of the modulator is more detailed in U.S. Patent No. 5,998,759 (for example, refer to Column 7 of Document 4 and related drawings). Those familiar with this technique understand that this modulator provides intensity control over a wide dynamic range. For example, 100:1. A simpler user interface can provide operator interaction based on "pass/failure" or other criteria to initiate operations.

In another embodiment, a linear scan can be automatically achieved, and a cleaning operation is performed at each measurement position.

In a preferred embodiment, during the cleaning operation or due to cleaning factors, only an energy adjustment is required, and other system parameters are not changed. Those who are familiar with measurement skills can make various adjustments to system parameters based on the correlation between the results and other program parameters.

In a preferred configuration, the cleaning operation is only applied to the scanning area as needed. In one configuration, this process is iterative and its measurement purpose is to obtain appropriate residues in the minimum square fit algorithm. If the residue is higher than a specified value, a scan of at least one area is obtained and a cleaning effect occurs. In some cases (for example, if it is difficult to clean), it may be necessary to adjust the position of the scan line, and a real degree measurement (for example, contrast, standard deviation) may be used to guide the cleaning operation. Preferably no more than one pass is required.

Understand that a variety of configurations can be used to implement this cleaning invention. For example, an array camera can use different wavelengths of illumination to identify areas with uneven intensity, and these areas can be designated for cleaning. Those familiar with this optical measurement technique can implement this And other configurations, and these configurations are within the scope of the present invention.

Reflectance measurement and power adjustment-Option 1: Single wavelength

The above is about a better measurement method and system for positioning and measuring the X, Y reference position. Another option to further improve the processing energy window is to adjust the laser energy and power measurement according to the material to be processed. And control concept. If it has a high reflectivity, the energy system is increased to compensate for these reflection losses; if it has a low reflection, because more energy is coupled to the workpiece or microstructure, the energy and power system are reduced. There are several ways to adjust this power and energy. The simplest way is to measure the reflection coefficient from the surface and adjust this energy and power control for the best energy coupling.

The light interference system between the metal and the oxide layer will greatly affect the reflection, and therefore also affect the absorption of the metal link (see Figures 11 and 12). Even if process engineers try to optimize absorption in the link by designing the optimal oxide thickness, the required thickness tolerance is still difficult to control. Generally speaking, the thickness of a layer may change by 10%, and there may be several layers of oxide between the top layer and the metal layer to be processed.

If the thickness and refractive index on the link can be determined, the energy required to process the link can be calculated and adjusted accordingly. There are two methods to determine the optical constants of the film, namely ellipsometry and spectral analysis. Ellipsometry uses the polarization change when a beam of light passes through a surface or is reflected. The amount of polarization change determines the refractive index of the material and the thickness of the material that the beam traverses. Spectroscopy measures the reflection from a surface at different wavelengths to determine the same optical constants. Commercial spectrometers use 256 different wavelengths to sense the reflected light and calculate the thickness, refractive index and extinction coefficient of the layer with extremely high accuracy ( Absorption rate).

Another method is to measure the reflection coefficient of two different wavelengths and calculate the oxide thickness. If the refractive index of the oxide used in the device can be measured, the reflection coefficient and the ratio of the laser radiation absorbed on the link can be calculated . If the absorption is known, the optimal laser energy for link removal can be input into the program in the laser beam. This second method is more accurate for thin film trimming systems with thin trimming materials, and part of the energy is transmittedThe film. The film.

The implementation method of thickness measurement and energy control is shown in Figure 13. The laser 60 used to remove the link provides a laser wavelength for thickness measurement. The energy delivered to this element is as shown in Figure 13. The acousto-optic modulator (also known as the "energy controller") 161 controls and reduces the reflection coefficient to be able to measure the reflection coefficient without compromising the value of the component. A red laser diode (that is, a 670 nm diode) 162 added to the optical path as shown in the figure can provide another wavelength for measuring reflectivity. Generally, splitters 166, 167 (ie dichroic mirrors) are used to transmit two wavelengths to the surface of the device and guide the reflected beam to the photodiode detectors 164, 165. Two photodiodes can be used as shown in Figure 13. Body 164, 165 to monitor the reflection coefficient. The thickness of the oxide can be uniquely determined by the intensity of the reflection coefficient of the two photodiodes (that is, the 670 nm diode and the 1047 nm detector) and the refractive index of the oxide layer. Once the thickness and refractive index are known, the absorption in the link material can be calculated and the computer can input the best energy into the program in the acousto-optical energy control device.

In order to have the highest accuracy, the dot area size and the size of the link can be used in the calculation. Refer to Figures 4a and 4b to see that there is a part of the energy falling out of the link, so the reflection that does not fall on the link must be calculated The amount of difference in light. Therefore, two measurements must be made to accommodate the reflected energy not covered by the link. These measurements may be made on each die as needed, and as the oxide thickness on the wafer changes, the energy per pulse will also change. Or, for example, this method can be selectively applied to each wafer one by one for processing monitoring. This technology can reduce the need to use high-value laser processing energy due to the absorption change of the interference effect in the link. .

Reflectance measurement and power adjustment-Option 2: Fine-tuning or adjustable wavelength

In certain situations, it is possible to adjust the wavelength in a range to improve the processing energy window. In this range, the energy coupled to the target can be improved and interference effects can be used to increase the stack reflectance or increase the substrate reflectivity. A special solid-state fine-tunable laser-optical parametric oscillator (OPO), Raman or other fine-tunable lasers can be used. The limitation is that it needs to meet the power and repetition requirements for a known application. For example, a parametric oscillator with a fixed wavelength and using 2 or 3 crystals at the same time can be used. The fine-tunable laser can operate in a specific environment. The published US Patent Application No. 2001-0036206 describes a fine-tunable laser diode developed for use in the telecommunications industry with a range of 40 nanometers (that is, a wavelength of 1.55 microns). Standard OPO lasers provide high power and narrow pulses but generally have extremely slow repetition rates and may be suitable for specific applications. However, the 10 kHz version shows and provides a repetition rate of 20 kHz. U.S. Patent Nos. 6,334,011 and 5,998,759 (Reference Document 4) and U.S. Patent No. 6,340,806 (Reference Document 6) disclose different combinations of offsets, such as 759 As disclosed in the patent number, the Fosterite laser system has a fine-tunable area to specifically span the absorption edge area of silicon, and allows operation beyond and below the absorption edge of silicon. Over time, this method seems inefficient at the current level of technology, because materials and improvements continue to develop in the field of lasers. The present invention includes the use of these devices and corresponding processing advantages. For example, the thickness and reflectance measurement of multiple layers can be expanded to select a wavelength range that can provide an improved energy window.

Application with a single-layer copper link between the base material and the link

It should be noted that the above principle can also be selectively applied to the traditional link structure (see Figure 2B), such as for processing high reflectivity copper links separated from the substrate by a single dielectric layer. The manufacturing trend is no longer favoring polysilicon structures but toward aluminum and copper substrate structures. Therefore, the link processing system has a continuous challenge to avoid reliability problems and improve yield. As mentioned above, many copper-based devices have a multi-layer stack in which damage to the substrate and stack can be avoided by selecting wavelength, spatial beam shaping, or transient shaping according to the above principles. However, some manufacturers etch all the dielectric material under the copper link and build the fuse on a single layer of dielectric material, and there is no SiN layer between the link and the substrate. For conventional laser processing, copper processing requires high power, which increases the possibility of substrate damage.

In certain situations, multiple pulse ("double blast") processing has been used to process metal fuses, but because the current online memory processing system requires two passes, the double blast method generally has a yield rate The problem. The simulation results and experiments show that although the second injection takes longer than the first injection, even if the first injection fails, the second injection can still fully open the link. According to the simulation results, the dual injection system with 50% of the single injection energy is very noteworthy; it is observed that the silicon substrate acts as a heat sink and cools down quickly. As shown in Figure 16, the results show that the silicon substrate 201 only needs 10 to 20ns to stabilize to room temperature, while the recovery of the copper target 202 is much slower, indicating a distinct thermal property. The second pulse is also Remove debris at the cutting site and cause "open circuit". It is estimated that each pulse of "double injection" needs about 60 to 70% of the energy used in "single injection". The pulse energy can vary with each pulse. In this example The pulse delay is 50ns, but it is obviously possible to use a much shorter delay.

In one embodiment, a delay line configuration of Figure 15a may be used to avoid any delay in output. For example, with a preferred positioning system of Patent No. 118 (Reference Document 2), it can move at a fine step speed of about 150 square meters per second. Since there is 30 ns between the two pulses, the link position is only 0.0045 microns, which is a negligible beam position change. In an optical delay line (for example, Figures 15b and 15c), a 9-meter extension path in the beam space will delay the second pulse by 30 ns. Alternatively, as shown in Fig. 15a, the second laser can use a 30ns or other controllable delay between trigger pulses, and a programmable digital delay line may be used to generate the trigger delay. The transient pulse shape may be, for example, a fast-rising, square pulse generated by a sub-laser diode (as used in simulation).

Various options for generating pulse combinations may be implemented based on the principles of this article. For example, at least one pulse may have a time course greater than a few picoseconds to a few nanoseconds, the pulse may be an amplified mode lock pulse, and at least one pulse may generate a pulse. A q-switched microlaser with a pulse width of less than 5 nanoseconds. At least one pulse may propagate along a second optical path, whereby the pulse delay is determined by the difference in optical path length as shown in Figures 15b and c. It is possible to use multiple lasers and/or amplifiers as shown in Figure 15a.

As shown in Figure 18, the generated pulse 275 may have a transient interval approximately equal to or shorter than one of the predetermined delays (for example, a 60MHz mode lock system) and repetition rate, and a modulator is used to select the radiation At least the second pulse of the microstructure or group of pulses 276 is taken. US Patent No. 5,998,759 (for example, reference document 4, column 7 and related drawings) discloses the use of a modulator to allow pulses to irradiate a link as needed. It is better to use an electro-optical modulator for the repetition rate at extremely high speeds.

Additional optical devices may be used, such as spatially shaping at least one delayed pulse before combining. For example, as shown in Figure 17, the first pulse 210 may have an elliptical or circular Gaussian space shape, or a high hat along the length of the link. The second pulse 212 may have a different size ratio, or may be a spatial form of a "cleaning pulse" in which the central area of the spot area is attenuated by a cut-toe filter or is effectively removed by a central mask. In this case, energy will be concentrated on the periphery of the link to remove the debris 211 around the location of the link due to the first pulse processing, thereby completing the processing 123. (For the sake of clarity, distinguish this "flying type" link part cleaning step from the above-mentioned "measurement cleaning" method). Reference 1 provides at least one example of beam shaping for link blowing applications, in which a uniform distribution rather than a Gaussian point area profile is disclosed.

In certain situations, there may be obvious microstructures and relative movements between the laser beams between pulses, such as greater than 25% of the spot size. This may be due to a slower repetition rate (increased pulse energy), Caused by a faster movement speed, a longer predetermined delay or a reduced target area. For example, an ultra-short or other short-pulse laser system with amplified pulses with output energy in the microjoule-millijoule range may have a repetition rate of 100 kHz to 10 MHz. In the former case, a high-speed, small-angle beam deflector can be used to compensate for the action and deflect a delayed pulse to irradiate the first microstructure at approximately a slower repetition rate during the relative action 258.

In an embodiment generally shown in Figure 19, the deflector is operatively coupled to the relative positioning system controller 251 in a closed loop configuration. The deflector is preferably solid and may be a single-axis acousto-optical device with a very fast "retrace"/access time. Alternatively, a higher speed electro-optical deflector (such as a gradient reflector or possibly a digital optical deflector) may be used. The response time can be exchanged with the time bandwidth product (number of dots) for the application. As disclosed in Reference 4 (column 7 and related figures), the deflector is preferably used for intensity control and pulse gate/selection. Alternatively, an electro-optical modulator may be used in conjunction with a separate acousto-optical deflector in a "chirp mode" 252 (for example, a linear sweep different from the random proximity mode) and depends on the positioning system The coordinate system 254 makes it synchronized. The positioning system coordinate system is related to the time when the modulator turns on the laser pulse, so that the time t corresponding to the selected pulse 259 during the relative movement 258₁, T₂, T₃The same single microstructure 256 is irradiated.

In another embodiment, a single laser pulse is used to spray up to two links at the same time (for example, no spray, one spray, or two links). Referring to FIG. 20, the two focused spot areas 306 and 307 spatially divide a single collimated laser beam 310 into two scattered collimated beams 309 to be formed on two links. This art has been known to use acousto-optical devices in material processing applications to split the beam spatially. For example, Patent Abstract No. JP53152662 shows a method that uses a frequency f₁....f_nThe multi-frequency deflector is used to drill the micro-hole configuration.

A laser 300 presents pulses at a predetermined repetition rate, and the laser beam passes through the relay optical device 302 to form an intermediate image of the waist of the laser beam into an acousto-optic modulator (AOM) aperture. It is preferable to use the AOM303 operating in the Braggregime system to controllably generate two slightly divergent collimated first-order diffracted laser beams and control the energy in each beam. AOM is composed of two frequencies f₁And f₂Drive, where f₁=f₀+_△_fAnd f₂=f₀_-_△_f, And Δf is the original RF signal frequency f₀A small proportion of it. The angle between the two beams is approximately equal to f₀Multiply by 2(Δf/f₀) Of Brig's angle. AOM system modulates the two frequency components in the RF signal f₁And f₂The signal amplitude is adjusted and the cross-coupling is adjusted to control the energy in each laser beam.

After the beam leaves the AOM, the beam rotation control module 313 is passed to rotate the beam 90 degrees on the axis, where the link is oriented in X or Y. In an embodiment, although many types of rotation techniques described in the general US application in the cross-reference paragraph of the related application are well known, a scallop is used for this rotation system.

Then, the beam passes through a set of optical devices to position the beam waist and set the beam size to fit the zoom optical device and the objective lens 305. Please note that the zoom optical device also modifies the angle between the two beams, so the angle between the two beams leaving the AOM must be adjusted according to the zoom setting, so that the focal plane has the required point area separation. Then, the laser beam enters the objective lens 305 to provide a pair of focused spot areas 306, 307 on the two links. The separation of the two-point zone is approximately equal to the focal length of the lens multiplied by the angle between the two beams. In an exemplary embodiment, an AOM center frequency of 80 MHz with a sweep range of about 2.3 MHz (77.7 to 82.3 MHz) may be used to generate a pair of adjacent links about 3 microns apart. Dot size of about 1.8 microns. As mentioned above, these links may have a size of about a laser wavelength (for example, 1 micron), and precise positions of the laser beam and microstructures are required for extremely high-speed operation.

Summary of some general forms of the present invention

In summary, one aspect of the present invention is a method for selective material processing of a microscopic target structure with a pulsed laser beam. The target structure is combined with a base by forming a plurality of layers of a multilayer structure. Material separation, the target structure, layer and substrate have different thermal and optical properties. The method includes: generating a pulsed laser beam with an energy density; irradiating the target structure with at least one pulse. By selecting at least one pulse characteristic, the stacked structure and the substrate are prevented from undesirable changes.

During the processing of the target structure, a part of the stack may be irradiated by the laser beam, but the layer, the substrate, and the functional circuit in a plane of the inner layer are prevented from causing undesirable losses.

The undesirable damage of the stacked structure includes the rupture of the inner dielectric substance caused by thermal stress, the undesirable damage of the stacked inner conductor includes the thermal damage caused by irradiation, and the undesirable damage of the substrate may be caused by laser irradiation and Caused by thermal diffusion.

The dielectric layer may include silicon nitride or silicon dioxide, and the substrate may be silicon.

The target structure is preferably copper, and may have a thickness or width of less than 1 micrometer, that is, a size of visible light wavelength or less. Alternatively, the target structure may be a metal link, such as aluminum, titanium, platinum, or gold.

One aspect of the present invention is the selection or control of the spatiality of the pulse and the characteristics of the transient beam, so that the target structure can be cleanly processed while avoiding the layer, the substrate, and the functional circuit in a plane of the inner layer. Bad damage.

A transient characteristic of the pulse is the pulse shape. The pulse shape includes: a fast rise time sufficient to efficiently couple the laser energy into the target, a time course sufficient to cleanly remove a part of the target structure, and sufficient One of the quick fall times to avoid undesirable damage caused by subsequent optical transmission. A preferred pulse rise time for link processing is less than 1 nanosecond (ns) to about 2 ns, a preferred duration is less than 10 ns, and a fall time is preferably less than 3 ns. The pulse shape may be roughly square, with approximately ±10% damping oscillations or fluctuations between the rising and falling edges. It is possible to use a single pulse or multiple pulses in the form of rapid bursts. Alternatively, q-switched pulse systems with different output powers and time-separated sequences may be combined as needed to form a pulse shape with a fast leading edge with a high peak power, and then a second pulse with a lower power . In another embodiment of the present invention, the q-switched pulses may have approximately the same output power and combine to produce a substantially square pulse shape.

Another transient pulse characteristic is the pulse power at the leading edge, if the irradiation on the target structure is greater than about 10⁹Watt/cm², the reflectivity of the target structure will be reduced and the coupling of laser energy will be improved.

A fast-rising pulse feature can avoid undesirable damage to the dielectric stack of a memory device with a metal target structure. The upper corner fracture occurs during the pulse duration and reduces the adjacent stacking layer. Stress in the lower corner.

A spatial feature of the beam is the irradiation profile of a controlled beam waist position. The irradiation profile may be similar to a circular Gaussian beam, an elliptical Gaussian beam, a rectangular profile in one direction and a rectangular profile in the orthogonal direction. Gaussian. This beam may be close to diffraction limited. It is possible to select a spatial shape and beam numerical aperture to control the interaction of the pulsed laser beam with the target and the overlay structure of the 3D device structure to avoid undesirable damage. The material interaction may be further controlled by the precise positioning of the waist of the pulsed laser beam, and the numerical aperture and beam shape may be selected so that the spot size and the link size roughly match in at least one size.

One aspect of the present invention is a method for selecting pulse characteristics based on a profiling process of a pulse interaction within a part of a three-dimensional device structure. The three-dimensional device includes a target structure with a different optical property, stacking And substrate. A series of structures are arranged at a predetermined interval to form an array, and at least one structure is not used as a target structure. A specification system may further include information about the spacing and materials of functional circuit elements located in a plane of the stack. The method includes determining the optical propagation characteristics of a part of an incident pulsed laser beam that is not absorbed by the target structure. The method further includes: assigning a laser pulse characteristic to avoid undesirable damage to any non-target structures, stacks, and substrates.

The interaction mechanism used to cause the selection of a pulse feature includes: reflection and internal reflection from the target surface, layer surface, polarization, interference effect, near-field diffraction, scattering and absorption, or a combination of the above. It is possible to use a thermal profiling process in conjunction with an optical profiling process.

The energy in the pulse of a copper link target structure used to process semiconductor memory devices may be approximately in the range of 0.1 to 5 microjoules. This energy density corresponds to the irradiation profile area of the beam waist, which may be The range is less than 20 square microns, and preferably less than 10 square microns.

Another controllable laser pulse characteristic is polarization, which may be controlled or selected based on the relative reflection coefficient of the layer and the laser energy optically coupled to the target structure at a wavelength.

The wavelength of the laser pulse may be selected according to the reflection coefficient (interference effect) of the multilayer stack. The preferred wavelength corresponds to a significant stack reflection such as 60% and a high internal transmission in one layer of the stack, which is approximately one of the maximum values. Spectral region. In order to have the greatest control over the spatial characteristics of the beam, it is better to use a short wavelength (for example, the smallest beam waist that can be achieved by a controllable selection of a larger beam waist and focus depth option). The laser wavelength system may be fixed or may change with wavelength shift or harmonic generation. A thickness or reflection coefficient measurement can be used to select or adjust the wavelength.

In at least one embodiment, the target structure may be substantially reflective at the laser wavelength. The laser wavelength may be lower than the absorption edge of the substrate and corresponds to an absorption or reflection area, and the laser wavelength is higher than the absorption edge of the stacked dielectric material layer and corresponds to a substantially maximum transmission area.

A selected wavelength ranges from less than 0.4 microns to about 1.55 microns and corresponds to the UV, visible light, and near-infrared spectrum. This lower limit may be determined by the absorption of a layer. For silicon substrates, absorption and reflection increase at shorter wavelengths. For silicon dioxide and silicon nitride, the internal transmission and single-surface reflection coefficients are approximately constant throughout the visible and near-infrared range. This upper limit corresponds to one of the wavelength ranges of laser diodes and optical amplifiers, and the output of one amplifier is acceptable. It can be processed by wavelength preservation or Raman shift.

Another aspect of the present invention is a method of aligning a microscopic target structure of a multi-material multilayer device with a pulsed laser beam for selective material processing. The target structure, layer, and substrate have a different thermal and optical properties. The beam has a focused beam waist with a centerline, and includes an alignment at one of a plurality of predetermined measurement positions associated with the device. Pattern, the alignment pattern is covered by at least one layer. The target structure is separated from a substrate by multiple layers used to form a multilayer stack. The method includes: measuring the position of the aligned target in at least one dimension; predicting the relative position of the target structure and the center line based on the measurement; triggering a relative action between the target structure and the center line based on the measurement; generating an energy density Pulsed laser beam; irradiates the target structure with at least one pulse. By selecting a pulse characteristic, the stacked structure and the substrate are prevented from undesirable changes.

The measurement system of a location may include a method and system that is not sensitive to polarization to avoid spurious measurement due to changes in the reflected signal. The signal changes may be caused by optical characteristics (including birefringence) caused by processing.

It is possible to predict the relative position of the target structure, the waist and the center line based on the multi-parameter minimum square fit.

It is possible to remove contaminants that will produce multiplicative changes (reflection noise), and use a cleaning process to enhance the data used for measurement.

May use alignment targets, wafers or other suitable materials to achieve three-dimensional (depth) measurement, may use measurement to predict the relative position of the target structure with respect to the beam waist, the position of the beam waist is along the pulsed laser beamCenterline. It is possible to estimate a surface from three-dimensional measurements, and it is possible to introduce a numerical offset based on the stack thickness to compensate for a depth difference between a measurement position and the target structure.

One aspect of the present invention includes measuring the thickness or reflectivity of a layer at a location, and using this measurement to control a pulse characteristic. The pulse characteristic may be pulse energy, pulse width or wavelength, and this position may be a single position or multiple positions on the device.

Although the best mode of implementation of the present invention has been described in detail, those familiar with the related art of the present invention will understand that various alternative designs and embodiments can be used to implement the present invention as defined by the scope of the following patent applications.

Component label comparison

1 Short pulse amplification laser system 2 Trigger pulse 3 Laser pulse 4 Rise time 5 Pulse duration 6 Fall time 7 Space 8 Space 9 Space 10 Target structure or microstructure 11 Large numerical aperture beam waist (Gaussian beam) 13 Top layer (paving) Cladding) 14 Inner layer (Silicon Nitride) 15 Lower layer (Silicon dioxide) 16 Inner conductor layer (Laying conductor) 17 Silicon substrate 20 q-switched pulse 21 Conventional solid-state laser 23 Target structure 25 Layer 27 Substrate 28 Area 41 Partial spot size 42 Partial spot size 43 Truncated transmission beam portion 44 Truncated transmission beam portion 46 Link 47 Internal reflection 48 Adjacent link 49 Layer 54 For laser beam 60 Laser for link removal 100 Target area 101 Noise profile 102 Exemplary signal profile 104 Linear scan 107 Connection contrast 120 Laser 121 Laser 122 Optical amplifier 123 Beam combiner 124 High frequency emission 125 Square pulse shape 126 Programmable delay circuit 201 Silicon substrate 210 First pulse 127 Polarization optical device 211 Debris 128 Polarization optical device 212 Second pulse 130 Retroreflector 251 Relative positioning system controller 131 Delayed beam path 252 Chirping mode 132 Delayed beam path 254 Positioning system coordinate system 133 λ/2 blocker 256 Single microstructure 135 Output waveform 258 Relative Action 140 Single laser 259 Selected pulse 141 Path 275 Pulse 142 Beam splitter 276 Pulse 143 Combiner 300 Laser 145 Selective optical amplifier 302 Relay optics 146 Rotator 303 AOM150 Beam splitter 305 Objective lens 152 Laser relative positioning 306 Focus Spot area 161 Acousto-optic modulator 307 Focused spot area 162 Red laser diode 309 Dispersed collimated beam 164 Photodiode detector 310 Single collimated laser beam 165 Photodiode detector 313 beam rotation control module 166 beam splitter 1001 debris 167 beam splitter