X-RAY BASED MEASUREMENTS IN PATTERNED STRUCTURE

Abstract

A method and system are presented for use in X-ray based measurements on patterned structures. The method comprises: processing data indicative of measured signals corresponding to detected radiation response of a patterned structure to incident X-ray radiation, and subtracting from said data an effective measured signals substantially free of background noise, said effective measured signals being formed of radiation components of reflected diffraction orders such that model based interpretation of the effective measured signals enables determination of one or more parameters of the patterned structure, wherein said processing comprises: analyzing the measured signals and extracting therefrom a background signal corresponding to the background noise; and applying a filtering procedure to the measured signals to subtract therefrom signal corresponding to the background signal, resulting in the effective measured signal.

Claims

1. A method for use in X-ray based measurements on patterned structures, the method being carried out by a computer system and comprising: processing, by a processor utility of the computer system, data indicative of measured signals corresponding to a detected radiation response of a patterned structure to incident X-ray radiation, wherein the processing comprising image processing of measured signals indicative of a) an angular span of diffraction orders that were scattered from the patterned structure, and (b) background signals; and applying a filtering procedure to the measured signals, the applying comprises subtracting from measured signals within the angular span, the background signals to provide effective measured signals that are substantially free of the background signals.

2. A measurement system for use in X-ray based measurements on patterned structures, the system comprising: an illumination unit configured and operable to define an illumination channel for directing illuminating X-Ray radiation onto a measurement plane for interacting with a structure being measured; a detection unit for detecting radiation response from the structure propagating along a collection channel; a control unit, a memory utility, an image processor and a processor utility, the processor utility being configured and operable to process imaging data indicative of measured signals corresponding to detected radiation response of a patterned structure to incident X-ray radiation, wherein the processor utility being configured and operable to deduce background signals, based on background signals located outside the angular span; and apply a filtering procedure to the measured signals to subtract from measured signals to provide effective measured signals that are, within the angular span, substantially free of the background signals.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawings will be provided by the Office upon request and payment of the necessary fee.

[0027] In order to better understand the subject matter that is disclosed herein and to exemplify how it may be carried out in practice, embodiments will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:

[0028] FIG. 1 is a picture of the portion of the surface of a patterned structure illustrating typical surface roughness on the nanometric scale;

[0029] FIG. 2 schematically illustrates the principles of X-ray based measurement scheme according to the present invention;

[0030] FIGS. 3A-3E schematically illustrate an example of data processing and analysis according to the technique of the invention;

[0031] FIG. 4 schematically illustrates the radiation propagation scheme according to the measurement scheme of the invention;

[0032] FIG. 5A is a block diagram of a measurement system of the invention for implementing the measurement scheme exemplified in FIG. 4; and

[0033] FIG. 5B is a flow diagram of the data analysis method of the invention.

DETAILED DESCRIPTION OF EMBODIMENTS

[0034] The present invention relates to measurements of periodic patterned structures, such as semiconductor wafers, in which the pattern typically includes some periodicity of unit cells along one or two directions. Each unit cell includes a pattern of spaced-apart regions of different optical properties (e.g. lines and spaces). The invention technique is aimed at solving the above-described problems of X-ray measurements associated with effects of roughness and background noise. FIG. 1 exemplifies the structure's surface having surface roughness on the nanometric scale.

[0035] Reference is made to FIG. 2 schematically illustrating the principles of X-ray based (such as GI-SAXS, as well as XRS and T-SAXS) measurement scheme 10 of the present invention. As shown in the figure, a sample 12 is irradiated by X-ray radiation including a range of illumination angles within a solid angle 14 of incident radiation, and such X-ray interaction with the sample 12 causes a radiation response 16 of the sample in the form of reflection/diffraction orders including zero-order diffraction and high diffraction orders. The radiation response 16 is detected by a 2D sensor matrix 18 having a radiation sensitive surface, where different diffraction orders are detected by different regions (sensor elements/pixels or groups of sensor elements/pixels) of the sensor matrix 18. Analysis of these diffraction orders allows characterization of the measured structure 12.

[0036] Thus, the X-ray scatterometry in general involves three main functional parts: illumination, the patterned structure being measured, and detection of the radiation response. The illumination mode/configuration provides for directing a beam of monochromatic X-ray radiation 14 towards a sample/structure 12. This beam 14 can be collimated (as used in T-SAXS), but for a most general case, it has an angular span as shown in FIG. 2. The illuminating beam 14 is focused onto a structure (for a collimated beam no focus is needed), and is reflected/returned from the structure. For metrology purposes, the measured structure 12 has a patterned surface 12A, where the pattern typically includes a periodic set of features, i.e. the elements under study (i.e. an array of transistors, memory cells, etc.). Upon reflection from this structure, the X-ray is scattered both in the specular direction (zero-order reflection/diffraction), and in the directions corresponding to high reflection/diffraction orders from the structure. The reflected radiation 16 is detected/measured by the 2D matrix 18 of the detector, creating an image 20 of reflection per angle of incidence and azimuth of the illumination 14.

[0037] Different SAXS implementations involve different sequence of data acquisition (for example, T-SAXS involves rotating the sample and sequentially acquiring a set of images, from which the full 2D dependence of the reflected signal is deduced). For simplicity, in the description below a measurement sequence is referred to as representative of GI-SAXS, in which multiple angles of incidence are probed in each measured image. It should, however, be understood that the invention is not limited to any specific type X-ray measurement technique, and therefore the invention should be interpreted broadly in this respect.

[0038] The present invention is based on the fact that the reflected diffraction orders from the structure being illuminated by angular span can be separated from each other, leaving regions in the image where only non-nominal signals arrive The present invention is based on the fact that the reflected diffraction orders from the structure being illuminated by angular span can be separated from each other. This means that the angular span of adjacent diffraction orders is smaller than their separation. This way, between adjacent diffraction orders there is a region where nominally no radiation is expected, and only non-nominal signal is measured. Examples for such ‘non-nominal’ signal sources would be stray light, roughness-related reflections, reflections from defects/pattern irregularities, etc. The signals obtained in these regions are thus entirely related to ‘background’ contributions, which in many cases is highly desired to remove. Such separation can be controlled by appropriate choice of the illumination angular span, which in selected in accordance with the structure being measured and its azimuthal orientation. This will be described more specifically further below. Once the diffraction orders are separated, it is possible to isolate the contribution from direct reflections, i.e. reflections associated with diffraction orders from the periodic pattern, and background contribution.

[0039] As described above, the background reflection components (being background noise) are characterized by smooth-varying intensity distributions added onto the X-ray signal being measured, contrary to the main diffracted signal which typically includes sharp variations in reflected intensity. This allows an algorithmic approach for deducing the contribution from these components at the regions where ordinary reflection resides. Once this contribution is obtained, it can be subtracted from the ordinary reflection regions, allowing straightforward analysis and interpretation of the measured signals.

[0040] Reference is made to FIGS. 3A-3E, schematically exemplifying the analysis methodology of the invention. FIG. 3A exemplifies a detected/measured signal MS (raw signals/image) as a function of illumination angles Theta and Azimuth of illumination. The measured signal MS is formed by reflected orders RO and background signature BS associated with the reflection from the pattern regions PR. These regions of the pattern features PR, creating the background signature BS (background noise) are to be isolated. As will be described below, measurement is set so that diffraction orders RO are spatially separated on the radiation sensitive surface (2D matrix) of the detector. As shown in FIG. 3B, once the diffraction orders RO are removed, background-only signal BS can be identified. A dedicated algorithm van be used to expand the background signal. In this example, a Lorentzian fit is used (FIG. 3C). The background signal BS then undergoes interpolation\extrapolation, resulting in an image representing the background across the entire image span, as shown in FIG. 3D. The background signal BS can then be subtracted from the measured signal MS, providing a clean image of the reflection orders RO.

[0041] Preferably, this approach utilizes a correct choice of the illumination angular range 14. This choice is based on the requirement of having different diffraction orders spatially separated on the 2D sensing matrix of the detector. In the present example, the measurement setup schematically illustrated in FIG. 2 is considered as reference, and the required illumination angular range suitable for this case is analyzed. It should be understood that a similar analysis is straightforward for other geometries.

[0042] The following is the description of a derivation for the angular separation between diffraction orders. This enables to set the allowed illumination angular range such that no overlap exists between the different reflection orders.

[0043] In this connection, reference is made to FIG. 4 schematically illustrating the radiation propagation scheme with respect to the pattern orientation on the sample. As shown in this example, the pattern P is configured as a grating formed by parallel features (lines) L extending along X-axis (defining the grating axis) and arranged in a spaced-apart relationship along Y-axis. Illuminating radiation 14 (angular span) is incident upon the patterned surface of the sample 12 at angle θ.sub.ILL (with respect to the normal to the sample), and azimuthal angle ϕ.sub.ILL. In this example, azimuth is defined with respect to the grating axis, but this is not a necessary restriction.

[0044] Considering radiation of wavelength λ and wavenumber k.sub.0=2π/λ, the transverse wavenumber k.sub.ILL (i.e. wavenumber on the sample plane) is given by

k.sub.ILL=k.sub.0(sin(θ.sub.ILL)cos(ϕ.sub.ILL),sin(θ.sub.ILL)sin(ϕ.sub.ILL))

[0045] Interaction of the illuminating radiation 14 with the grating P results in a radiation response 16 formed by radiation reflected in a set of discrete directions, corresponding to the diffraction orders from the grating. Generally, although not specifically shown, the grating P can have periodicity of features L in two directions, with pitch P.sub.x in the x direction and P.sub.y in the y direction (either of which can be zero if periodicity exists only in one dimension). Diffraction orders will be denoted (n.sub.x,n.sub.y) for the n.sub.x order arising from the periodicity in direction x and n.sub.y order arising from periodicity in they direction.

[0046] The (n.sub.x,n.sub.y) diffraction order will have transverse wavenumber given by:

[00001]$k_{COL}=\left(k_0\sin \left({\theta }_{ILL}\right)\cos \left({\varphi }_{ILL}\right)+\frac{2\pi }{P_x}{n}_x,k_0\sin \left({\theta }_{ILL}\right)\sin \left({\varphi }_{ILL}\right)+\frac{2\pi }{P_y}{n}_y\right)$

[0047] The reflection direction of this order is given by:

[00002]${\varphi }_{COL}=\mathrm{atan}\left(\frac{k_{COL}^y}{k_{COL}^x}\right)=\mathrm{atan}\left(\frac{\sin \left({\theta }_{\mathrm{ILL}}\right)\sin \left({\varphi }_{\mathrm{ILL}}\right)+\frac{\lambda }{P_y}{n}_y}{\sin \left({\theta }_{\mathrm{ILL}}\right)\cos \left({\varphi }_{\mathrm{ILL}}\right)+\frac{\lambda }{P_x}{n}_x}\right),\mathrm{and}$${\theta }_{COL}=\mathrm{asin}\left(\frac{\frac{\lambda }{P_x}{n}_x}{\cos \left({\varphi }_{\mathrm{ILL}}\right)}\right).$

[0048] By these relations, it is straightforward to relate any illumination angular span to the angular span of all reflected orders. Specifically, it is possible to check whether overlaps are expected between the reflected orders.

[0049] Turning back to FIGS. 1 and 4, the invention provides that, for a given wavelength λ of illumination, the angular span θ.sub.ILL of incident radiation 14 and its azimuth orientation ϕ.sub.ILL. with respect to the pattern P on the structure being measured, can be controlled/selected to provide a required angular span ϕ.sub.COL of the reflection orders 16. This required angular span ϕ.sub.COL of the reflection orders 16 is such that radiation components of different reflection orders interact with different regions R on radiation sensitive surface 18 of the detector spatially separated from one another by gap(s) G. It should be understood that the value(s) of G can also be controlled by the length of the collection channel.

[0050] Several approaches are possible to implement this limitation/selection of the illumination range of angular span. Some examples include the use is a single aperture with controllable dimensions placed in the optical path, or an interchangeable set of apertures placed in the optical path. These could be exchanged or modified according to the measured sample pitch. Further flexibility can be obtained by allowing different shapes of apertures. It should be noted that when the pattern on the measured structure is periodic in two directions, such approach can guarantee separation of high diffraction orders in both directions.

[0051] More generally, it is possible to implement such separation only on one subgroup of the reflection orders. For example, diffraction orders related to the grating pitch in one direction can be separated, while leaving orders related to the pitch in another direction overlapping. Although with this approach some of the background signal might not be removed, providing a degraded result, it may be easier to implement in practice and may be suitable/sufficient in some applications.

[0052] Reference is now made to FIGS. 5A and 5B schematically illustrating a block diagram of a measurement system 100 of the invention (FIG. 5A) and a flow diagram 200 of a method of the invention (FIG. 5B). The system 100 includes an illumination unit 102 defining an illumination channel for directing illuminating radiation 14 onto a measurement plane (sample plane), a detection unit 104 for detecting radiation response from a sample 12 propagating along a collection channel, and a control unit 106 configured to be in data/signal communication with the illumination and detection units (via wires or wireless signal transmission using any known suitable communication techniques and protocols). Also provided in the system 100 is an angular span controller 110 located in the illumination channel.

[0053] The control unit 106 is configured as a computer system including inter alia such functional and structural utilities as input and output utilities 106A and 106B, memory utility 106C, and data processor 106D. The control unit 106 further includes a measurement scheme controller (a so-called orientation controller) 106E, which is configured to receive and analyze data indicative of a pattern P on the structure/sample 12 (e.g. pattern pitch along one or two axes) and its azimuthal orientation with respect to the illumination channel, and generate data indicative of an optimal choice for the illumination angular span for operating the angular span controller 110. It should be noted that the structure under measurements may be located on a stage/support 108 driven for rotation so as to adjust the pattern orientation with respect to the illumination channel, as the case may be. The angular span controller may be configured as described above, namely may include a single aperture with controllably varying dimensions and/or shapes, or an interchangeable set of apertures of different dimensions/shapes. This enable selection of an aperture shape according to the selected measurement scheme.

[0054] Thus, the orientation controller 106E operates the angular span controller 110 and possibly also the stage 108 (via its drive) to select the illumination angular span 14 such that the measured image is formed by distinct regions corresponding to the high reflection orders, with gaps between them, i.e. such that no overlap exists between different orders. Such an image is exemplified in FIG. 3A described above.

[0055] Once this situation is assured, i.e. the desired orientation is provided (step 202 in FIG. 5B), the following measurement scheme is performed in order to remove the effects of background signal/contribution. Measured signal MS is detected by the pixel matrix of the detection unit 104 which generates measured data indicative thereof (step 204) and coveys this data to the control unit 106 where it is processed by the data processor 106D (step 206). The data processor 106D includes an analyzer module 112 configured (preprogrammed) for filtering out signals corresponding to radiation components of reflected diffraction orders (presenting “active signals”), leaving only background signature (presenting “noise”)—step 208. This is illustrated in FIG. 3B described above. Then, the background signature signals are interpolated (and extrapolated) by a fitting module 114 (step 210) to deduce the background contribution to the entire measured image or a selected part of it. To this end, the fitting module 114 utilizes some underlying functional shape/form for the background signal (e.g. approach shown in FIG. 3C according to which each angle of incidence is fitted separately by Lorentzian lineshape; or any other functional form, or a two-dimensional functional form, and using a fitting approach based on some theoretical model). This results in the background signal alone (step 212). The so-obtained image describing the background signal alone may optionally be further processed by any suitable image processor 116 to smooth and correct this image (optional step 214). The initial image (measured data) is then processed by a filter module 118 which operates to subtract the background signal/image (either being processed by image processor or not) from the initial image (step 216) resulting in the final, background-removed signal (step 218). This final signal presents the active/effective measured data that can be used for analysis and interpretation using any known suitable technique, e.g. model-based data interpretation.

[0056] Thus, the present invention provides a simple and effective technique for use in X-ray based scatterometry measurements on patterned structures. According to the invention, a measurement scheme is optimized to enable subtraction of background-noise free effective measured signal, which can then be interpreted to determine the structure parameters using any known suitable data interpretation approach.