Design-based inspection using repeating structures

Abstract

Systems and methods for design-based inspection using repeating structures are provided.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention generally relates to systems and methods for design-based inspection of wafers using repeating structures.

2. Description of the Related Art

The following description and examples are not admitted to be prior art by virtue of their inclusion in this section.

Inspection processes are used at various steps during a semiconductor manufacturing process to detect defects on wafers to promote higher yield in the manufacturing process and thus higher profits. Inspection has always been an important part of fabricating semiconductor devices. However, as the dimensions of semiconductor devices decrease, inspection becomes even more important to the successful manufacture of acceptable semiconductor devices because smaller defects can cause the devices to fail.

Inspection processes can be limited by various noise sources on the wafer. For example, one common inspection method is the die-to-die method that involves comparing output of an inspection system generated for corresponding positions in different dies formed on the wafer. In this manner, output generated for similar structures in multiple dies can be compared and the results of the comparison can be used to detect defects in those structures. However, due to process variations across the wafer, corresponding positions in different dies may have different characteristics such as film thickness and color variation that, while not actually defects, can be misidentified as defects in the die-to-die methods. The variations between dies can be accommodated by increasing the thresholds that are used to detect defects. However, increasing the threshold will obviously eliminate the detection of the smallest defects on the wafers.

Some inspection methods and/or systems detect defects by comparing output generated for multiple locations within a single die. There are generally two types of methods that utilize this concept in current inspection systems. However, their applicability is limited. For example, for array areas such as SRAM or DRAM areas of a die, by knowing the cell size within the array, a cell-to-cell inspection can be performed on wafer inspectors. There are also some existing systems that analyze the image stream during inspection and look for repeating structures by performing auto-correlation analysis and looking for periodic patterns. However, both of these methods are limited to layouts that have some periodicity (e.g., in the x direction) with a certain period.

Accordingly, it would be advantageous to develop inspection systems and/or methods that do not have one or more of the disadvantages described above.

SUMMARY OF THE INVENTION

The following description of various embodiments is not to be construed in any way as limiting the subject matter of the appended claims.

One embodiment relates to a computer-implemented method for inspecting a wafer. The method includes identifying multiple instances of structures in a design for a wafer. The structures have the same or substantially the same geometrical characteristics. Identifying the multiple instances is performed using design data for the design. The method also includes comparing output of an inspection system generated for two or more of the multiple instances formed on the wafer to each other. The two or more of the multiple instances are located within the same die on the wafer. In addition, the method includes detecting defects on the wafer based on results of the comparing. The identifying, comparing, and detecting steps are performed using a computer system.

Each of the steps of the method described above may be further performed as described herein. In addition, each of the steps of the method may be performed using any of the system(s) described herein. Furthermore, the method may include any other step(s) described herein.

Another embodiment relates to a non-transitory computer-readable medium storing program instructions executable on a computer system for performing a computer-implemented method for inspecting a wafer. The computer-implemented method executable by the program instructions includes the steps of the above-described computer-implemented method. The computer-readable medium may be further configured as described herein.

An additional embodiment relates to a system configured to inspect a wafer. The system includes an inspection subsystem configured to generate output for a wafer. The system also includes a computer subsystem configured for performing the steps of the computer-implemented method described above. The system may be further configured as described herein.

BRIEF DESCRIPTION OF THE DRAWINGS

Other objects and advantages of the invention will become apparent upon reading the following detailed description and upon reference to the accompanying drawings in which:

FIG. 1 is a schematic diagram illustrating one embodiment of cell instantiation analysis;

FIG. 2 is a schematic diagram illustrating one example of similar structures that are self-evident from design “layer” intent itself;

FIG. 3 is a schematic diagram illustrating one embodiment of how an iterative algorithm can rank and detect repeating geometries;

FIG. 4 is a block diagram illustrating one embodiment of a non-transitory computer-readable medium; and

FIG. 5 is a block diagram illustrating one embodiment of a system.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Turning now to the drawings, it is noted that the figures are not drawn to scale. In particular, the scale of some of the elements of the figures is greatly exaggerated to emphasize characteristics of the elements. It is also noted that the figures are not drawn to the same scale. Elements shown in more than one figure that may be similarly configured have been indicated using the same reference numerals.

Various embodiments for design-based inspection using repeating structures are described herein. One embodiment relates to a computer-implemented method for inspecting a wafer. The method includes identifying multiple instances of structures in a design for a wafer. The structures have the same or substantially the same geometrical characteristics. “Geometrical characteristics” as that term is used herein is intended to mean any characteristics of the structure that are related to the geometry of the structure. Therefore, “geometrical characteristics” may include such characteristics as width, height, side profile, shape of the structure, two-dimensional shape, three-dimensional shape, and the like. Therefore, structures that have the same geometrical characteristics would have the same width, height, side profile, shape of the structure, two-dimensional shape, three-dimensional shape, and the like. The term “substantially the same geometrical characteristics” references geometrical characteristics that are not different to such a degree that the differences themselves would be detected as defects in the methods described herein. In addition, the extent to which different geometrical characteristics can vary and still be considered substantially the same may vary depending upon the geometrical characteristic itself. For example, geometrical characteristics that are different by about 1% may be considered substantially the same for one characteristic while geometrical characteristics that are different by about 5% may be considered substantially the same for another characteristic.

A “structure” as that term is used herein is intended to mean a patterned structure that is formed on a wafer. In other words, the “structures” described herein are not simply films formed on a wafer. In addition, a “structure” as that term is used herein refers to a single continuous structure. In other words, a “structure” as that term is used herein does not refer to a collection of individual structures such as an array of lines, an array of contacts, or any other set of discrete structures. Instead, each of the lines, contacts, or discrete structures included in a set would each be a structure as that term is used herein.

Identifying the multiple instances is performed using design data for the design. For example, the embodiments described herein may use the design layout of a semiconductor layer to determine “similar” structures in relatively close proximity to one another in a die layout so that during inspection, these similar structures can be compared with each other and “outliers” marked as defects, the basic assumption being that such defect locations will be a substantially small fraction of all such similar structures. In one embodiment, identifying the multiple structures is not performed using output of an inspection system generated for a wafer. Instead, as described above, the repeating structures may be identified using the design data and in some cases only the design data. The design data may not include or be data or information acquired using a physical wafer. In other words, design data is not output generated by scanning a physical wafer or information or data generated based on such output. Since the repeating structures are identified in the embodiments described herein from analysis of design data, there is no image noise introduced into the algorithm. The embodiments described herein exploit the fact that some image computer systems can read in a large swath of data across a die and one can use the embodiments described herein to determine all locations of certain groups of geometries within that image buffer.

Several techniques may be used to find repeating structures in the design. For example, in one embodiment, identifying the multiple structures is performed without using periodicity of the structures. In another embodiment, the multiple instances do not occur in the design with any periodicity. For example, the embodiments described herein may look for similar patterns even if they are not periodic.

In some embodiments, identifying the multiple structures is not performed based on cell size of the design. In this manner, the embodiments described herein may be independent of cell size, which is different than some methods used for array inspection. For example, in some currently used methods for array areas such as SRAM or DRAM areas of a die, by knowing the cell size within the array, a cell-to-cell inspection is performed on wafer inspectors. However, these methods are limited to layouts that have some periodicity (e.g., in the x or scanning direction) with a certain period. In contrast, the embodiments described herein can utilize the repeating nature of certain structures within the design regardless of the cell size, periodicity, frequency, and any other cell characteristics. Therefore, the embodiments described herein may be much more flexible in the types of structures that can be used as repeating structures, the algorithms that are used to detect defects in those repeating structures, and the way defects in different repeating structures are classified.

In some embodiments, each of the multiple instances includes only one structure in the design. For example, each repeating structure itself may be identified and used as a single instance of the structure. In this manner, for defect detection purposes, the output for one repeating structure may be used for comparison to output for at least one other of the same or substantially the same repeating structure.

In another embodiment, each of the multiple instances includes multiple structures in the design. For example, the method may find groups of geometries that occur in various places in the layout. The groups of geometries do not have to occur with any fixed periodicity. Such groups of geometries may be identified as described further herein.

In one embodiment, identifying the multiple instances includes identifying multiple instances of the same design cell in the design using a design layout hierarchy for the design. In this manner, identifying the multiple instances may include cell instantiation analysis. For example, in this method, the design layout hierarchy is examined and instances of the same design “cell” in the layout are identified. In particular, as shown in FIG. 1, the method may include defining cells 100 (e.g., Cell A, Cell B, Cell C, etc) in the design. The method may also include using the design hierarchy with cell references (instantiations) 110 to search in design data for die 120 for the various cells (e.g., Cell A, Cell B, etc.). These areas can be assumed to look similar on the scanned die.

In some embodiments, identifying the multiple instances includes searching the design data for any instances of a known pattern of interest by performing a polygon match of the structures that make up the known pattern of interest and the structures in the design data. For example, if a pattern of interest (POI) is known a priori, the design can be searched for all locations of the same pattern by performing a polygon match of the figures that make up the POI.

In an additional embodiment, identifying the multiple instances includes performing Fourier analysis of the design data. For example, Fourier analysis of the design may be performed either by rendering it or by analyzing the polygonal representation and looking for periodicities in the x (comparing) direction.

In one embodiment, identifying the multiple instances is performed using design layer intent for the design data. For example, certain structures that are similar may be self-evident from the design “layer” intent itself. In one such example, a via/contact layer is simply a set of identical small structures and other than the center location and dimensions of each via/contact, no further processing is required to identify “similarity.” In one particular example, as shown in FIG. 2, one layer may be the metal layer with vias 200, and the corresponding via design layer 210 may be used to identify the center location and dimensions of each via/contact.

In a further embodiment, identifying the multiple instances includes analyzing a design layout for the design data at a polygonal level to identify the multiple instances of the structures. In this manner, the design layout can be analyzed at the polygonal level to find similar collections of neighboring geometries that repeat in several places in the design. These can be marked as belonging to one “care area group” and can be compared with each other within a die instead of in a die-to-die comparison. One example of an algorithm that can be used to find repeating structures is described in Gu et al., “Lossless compression algorithms for hierarchical IC layouts,” IEEE Transactions on Semiconductor Manufacturing, Vol. 21, No. 2, May 2008, which is incorporated by reference as if fully set forth herein. Although the algorithm described in this paper is used for the purposes of compressing the design representation, this algorithm, or a similar one, may be used to identify repeating structures for comparison purposes during inspection.

In some embodiments, identifying the multiple instances is performed by identifying a first structure in the design data that can be included in the multiple instances and then iteratively searching the design data for other structures that can also be included in the multiple instances with the first structure. For example, FIG. 3 is taken from the above-referenced paper and shows how an iterative algorithm ranks and detects repeating geometries. In particular, FIG. 3 shows an intracell subcell detection example in which (a) shows the 0th iteration; (b) shows the first iteration; (c) shows the second iteration; and (d) shows the final result. In this example, as shown in the 0th iteration, the algorithm may search the cell for a single selected structure and find 5 instances of the single structure within the cell. Each of the 5 instances has the same geometrical characteristics (e.g., shape, dimensions, and orientation) in the cell. The algorithm may then define the subcell as SC(0) shown in FIG. 3, 5 instances of which were found, each containing one instance of the structure and some space around the structure.

In the next iteration, the algorithm searches the cell for instances of other structures that are near more than one instance of the first structure included in SC(0) (i.e., other structures having the same geometrical characteristics as each other and that are repeating with the first structure). For example, as shown in FIG. 3, the algorithm may search the cell again and identify two other structures that are located near more than one instance of the first “L” shaped structure. Each of the two other identified structures have the same geometrical characteristics as each other (e.g., the same shape as each other and the same spatial relationship with respect to the first structure). The algorithm may then define the subcell SC(1) shown in FIG. 3, 4 instances of which were found, each containing one instance of 3 different structures and some space around the structures.

The algorithm may perform any additional number of iterations. For example, as shown in FIG. 3, the algorithm may search the cell again and identify one other structure that is located near more than one instance of the first “L” shaped structure. Each instance of this other identified structure has the same geometrical characteristics as each other (e.g., the same shape and the same spatial relationship with respect to the first structure). The algorithm may then define the subcell as SC(2) shown in FIG. 3, 4 instances of which were found, each containing one instance of 4 different structures and some space around the structures. As also shown in this example, the algorithm may search the cell in the space in and around one of the subcells and identify another structure that is near the first structure. However, upon searching the other subcells the corresponding areas in which that additional structure was found in the first instance, that other structure was not found in the other subcells. In this manner, there is only one instance of the structure and it is not repeating. Therefore, this other structure is not included in the subcell results (i.e., SC(2)), and this subcell is not included in the results (e.g., it is not another repeating subcell). Therefore, as shown in the final result, the subcell SC(2) defines each of the multiple (4) instances of the repeating structures.

In another embodiment, the multiple instances include at least three instances of the structures. For example, different inspection methods may require a different minimum number of repeats of a structure. In one such example, in a die-to-die type method, defect arbitration is performed by comparing three adjacent die locations by doing two die-pair comparisons, where each comparison is preceded by an alignment/interpolation step to align the pixels to a sub-pixel (typically less than 0.05 pixels) accuracy. The method performs interpolations of a sub-frame (256 pixels×256 pixels) at a time in bright field (BF) tools, the alignment offset (delta x, delta y) being provided by a real time alignment (RTA) system. One could therefore argue that the minimum number of repeats of a structure required to arbitrate a defect location is three. Therefore, to perform a similar defect arbitration using the multiple instances described herein, at least three instances of the structures would be needed. However, having many more occurrences of the structure within an image frame allows one to use statistics to flag outliers in a more robust manner. For example, in one embodiment, the multiple instances include all instances of the structures in the design data.

The method also includes comparing output of an inspection system generated for two or more of the multiple instances formed on the wafer to each other. The two or more of the multiple instances are located within the same die on the wafer. In this manner, the embodiments described herein bypass the die-to-die method of random mode defect detection and in some way mimic the array-mode detection algorithm. The advantage of the embodiments described herein over die-to-die random mode is that the die-to-die noise sources are removed and one can potentially lower the detection threshold thereby enabling more sensitive inspection.

In one embodiment, prior to the comparing step, the method includes interpolating the multiple instances of the structures to a common pixel grid. For example, all occurrences of the structure may be interpolated to a common pixel grid before performing the comparisons (differencing their gray level values from corresponding pixel locations). Clearly, interpolation errors will introduce a noise source. However, for a properly sampled image, this noise source may be much less than the die-to-die noise, particularly in electron beam based inspection systems.

In a further embodiment, during inspection and prior to the comparing step, the method includes aligning the output generated for the two or more of the multiple instances in corresponding pixel streams of the output. For example, the embodiments described herein assume one can accurately align the similar areas (detected from analyzing the design) with the corresponding pixel stream during inspection.

The method further includes detecting defects on the wafer based on results of the comparing. For example, detecting the defects may include applying some threshold to the results of the comparing step, and any output determined to be above the threshold may be identified as corresponding to a defect. The threshold may include any suitable threshold known in the art.

The identifying, comparing, and detecting steps are performed using a computer system. The computer system may be further configured as described herein.

The embodiments described herein have a number of advantages over other inspection systems and methods. For example, by comparing two structures within the same die, rather than die-to-die comparisons as is done for inspecting logic areas, the die-to-die noise such as changes in film thickness, focus, etc. are eliminated. In addition, the embodiments described herein may find all repeats of a certain set of geometries within a die, while the current methods look for periodicity (in the scanning (and comparison) direction). Thus, the embodiments described herein are more general than the current methods. Furthermore, the embodiments described herein use design data rather than trying to infer from the wafer image where the repeating structures are located. Using the design has the advantage that it is the ideal representation (intent) of what the wafer image should look like, uncorrupted by image noise sources in the image acquisition system.

All of the methods described herein may include storing results of one or more steps of the methods in a storage medium. The results may include any of the results described herein and may be stored in any manner known in the art. The storage medium may include any suitable computer-readable storage medium known in the art. After the results have been stored, the results can be accessed in the storage medium and used by any of the method or system embodiments described herein, formatted for display to a user, used by another software module, method, or system, etc. Furthermore, the results may be stored “permanently,” “semi-permanently,” temporarily, or for some period of time.

FIG. 4 illustrates one embodiment of non-transitory computer-readable medium 400 storing program instructions 402 executable on computer system 404 for performing a computer-implemented method for inspecting a wafer. The method for which program instructions 402 are executable on computer system 404 may include any step(s) of any method(s) described herein. In some embodiments, computer system 404 may be a computer system of an inspection system as described further herein. In some alternative embodiments, the computer system may be connected to the inspection system by a network. However, in other embodiments, computer system 404 may not be coupled to or included in an inspection system. In some such embodiments, computer system 404 may be configured as a stand alone computer system. Computer-readable medium 400, program instructions 402, and computer system 404 may be further configured as described herein.

Program instructions 402 implementing methods such as those described herein may be stored on computer-readable medium 400. The computer-readable medium may be a storage medium such as a read-only memory, a random access memory, a magnetic or optical disk, a magnetic tape, or other non-transitory computer-readable medium.

The program instructions may be implemented in any of various ways, including procedure-based techniques, component-based techniques, and/or object-oriented techniques, among others. For example, the program instructions may be implemented using ActiveX controls, C++ objects, C#, JavaBeans, Microsoft Foundation Classes (“MFC”), or other technologies or methodologies, as desired.

The computer system may include any suitable computer system known in the art. For example, computer system 404 may take various forms, including a personal computer system, mainframe computer system, workstation, image computer, parallel processor, or any other device known in the art. In general, the term “computer system” may be broadly defined to encompass any device having one or more processors, which executes instructions from a memory medium.

Another embodiment relates to a system configured to inspect a wafer. For example, as shown in FIG. 5, the system includes inspection subsystem 500 configured to generate output for a wafer. The inspection subsystem may include an inspection subsystem of any commercially available wafer inspection system. The inspection subsystem may also be configured for any suitable inspection methods such as bright field inspection, dark field inspection, optical (light-based) inspection, electron beam based inspection, etc., or some combination thereof. The system also includes computer subsystem 502 configured for performing the steps of the methods described above. Computer subsystem 502 may be further configured as described above with respect to computer system 404. The computer subsystem and the system may be further configured as described herein.

Further modifications and alternative embodiments of various aspects of the invention may be apparent to those skilled in the art in view of this description. For example, systems and methods for design-based inspection using repeating structures are provided. Accordingly, this description is to be construed as illustrative only and is for the purpose of teaching those skilled in the art the general manner of carrying out the it) invention. It is to be understood that the forms of the invention shown and described herein are to be taken as the presently preferred embodiments. Elements and materials may be substituted for those illustrated and described herein, parts and processes may be reversed, and certain features of the invention may be utilized independently, all as would be apparent to one skilled in the art after having the benefit of this description of the invention. Changes may be made in the elements described herein without departing from the spirit and scope of the invention as described in the following claims.

Claims

1. A computer-implemented method for inspecting a wafer, comprising: identifying multiple instances of structures in a design for a wafer, wherein the structures have the same or substantially the same geometrical characteristics, wherein each of the structures included in the multiple instances comprises a single continuous structure, and wherein said identifying is performed using design data for the design;comparing output of an inspection system generated for two or more of the multiple instances of the single continuous structures formed on the wafer to each other, wherein the two or more of the multiple instances are located within the same die on the wafer, and wherein the output that is compared for the two or more of the multiple instances comprises all of the output generated for the single continuous structures included in the two or more of the multiple instances; anddetecting defects on the wafer based on results of said comparing, wherein said identifying, said comparing, and said detecting are performed using a computer system.

2. The method of claim 1, wherein said identifying is further performed without using periodicity of the structures.

3. The method of claim 1, wherein said identifying is not performed using output of an inspection system generated for a wafer.

4. The method of claim 1, wherein said identifying is not performed based on cell size of the design.

5. The method of claim 1, wherein the multiple instances do not occur in the design with any periodicity.

6. The method of claim 1, wherein each of the multiple instances comprise only one structure in the design.

7. The method of claim 1, wherein each of the multiple instances comprise multiple structures in the design.

8. The method of claim 1, wherein said identifying comprises identifying multiple instances of the same design cell in the design using a design layout hierarchy for the design.

9. The method of claim 1, wherein said identifying comprises searching the design data for any instances of a known pattern of interest by performing a polygon match of the structures that make up the known pattern of interest and the structures in the design data.

10. The method of claim 1, wherein said identifying comprises performing Fourier analysis of the design data.

11. The method of claim 1, wherein said identifying is further performed using design layer intent for the design data.

12. The method of claim 1, wherein said identifying comprises analyzing a design layout for the design data at a polygonal level to identify the multiple instances of the structures.

13. The method of claim 1, wherein said identifying is performed by identifying a first structure in the design data that can be included in the multiple instances and then iteratively searching the design data for other structures that can also be included in the multiple instances with the first structure.

14. The method of claim 1, wherein the multiple instances comprise at least three instances of the structures.

15. The method of claim 1, wherein the multiple instances comprise all instances of the structures in the design data.

16. The method of claim 1, further comprising prior to said comparing, interpolating the multiple instances of the structures to a common pixel grid.

17. The method of claim 1, further comprising during inspection and prior to said comparing, aligning the output generated for the two or more of the multiple instances in corresponding pixel streams of the output.

18. A non-transitory computer-readable medium storing program instructions executable on a computer system for performing a computer-implemented method for inspecting a wafer, wherein the computer-implemented method comprises: identifying multiple instances of structures in a design for a wafer, wherein the structures have the same or substantially the same geometrical characteristics, wherein each of the structures included in the multiple instances comprises a single continuous structure, and wherein said identifying is performed using design data for the design;comparing output of an inspection system generated for two or more of the multiple instances of the single continuous structures formed on the wafer to each other, wherein the two or more of the multiple instances are located within the same die on the wafer, and wherein the output that is compared for the two or more of the multiple instances comprises all of the output generated for the single continuous structures included in the two or more of the multiple instances; anddetecting defects on the wafer based on results of said comparing.

19. A system configured to inspect a wafer, comprising: an inspection subsystem configured to generate output for a wafer; anda computer subsystem configured for: identifying multiple instances of structures in a design for the wafer, wherein the structures have the same or substantially the same geometrical characteristics, wherein each of the structures included in the multiple instances comprises a single continuous structure, and wherein said identifying is performed using design data for the design;comparing output of the inspection subsystem generated for two or more of the multiple instances of the single continuous structures formed on the wafer to each other, wherein the two or more of the multiple instances are located within the same die on the wafer, and wherein the output that is compared for the two or more of the multiple instances comprises all of the output generated for the single continuous structures included in the two or more of the multiple instances; anddetecting defects on the wafer based on results of said comparing.

20. The system of claim 19, wherein said identifying is further performed without using periodicity of the structures.

21. The system of claim 19, wherein said identifying is not performed using output of the inspection subsystem generated for a wafer.

22. The system of claim 19, wherein said identifying is not performed based on cell size of the design.

23. The system of claim 19, wherein the multiple instances do not occur in the design with any periodicity.

24. The system of claim 19, wherein each of the multiple instances comprise only one structure in the design.

25. The system of claim 19, wherein each of the multiple instances comprise multiple structures in the design.

26. The system of claim 19, wherein said identifying comprises identifying multiple instances of the same design cell in the design using a design layout hierarchy for the design.

27. The system of claim 19, wherein said identifying comprises searching the design data for any instances of a known pattern of interest by performing a polygon match of the structures that make up the known pattern of interest and the structures in the design data.

28. The system of claim 19, wherein said identifying comprises performing Fourier analysis of the design data.

29. The system of claim 19, wherein said identifying is further performed using design layer intent for the design data.

30. The system of claim 19, wherein said identifying comprises analyzing a design layout for the design data at a polygonal level to identify the multiple instances of the structures.

31. The system of claim 19, wherein said identifying is performed by identifying a first structure in the design data that can be included in the multiple instances and then iteratively searching the design data fir other structures that can also be included in the multiple instances with the first structure.

32. The system of claim 19, wherein the multiple instances comprise at least three instances of the structures.

33. The system of claim 19, wherein the multiple instances comprise all instances of the structures in the design data.

34. The system of claim 19, wherein the computer subsystem is further configured for, prior to said comparing, interpolating the multiple instances of the structures to a common pixel grid.

35. The system of claim 19, wherein the computer subsystem is further configured for, during inspection and prior to said comparing, aligning the output generated for the two or more of the multiple instances in corresponding pixel streams of the output.