This invention relates to the design, layout, testing and manufacture of microelectronic circuits and systems, and more particularly to apparatus, methods and computer program products for verifying microelectronic circuits and systems prior to manufacture.
It is generally known that in current state-of-the-art integrated circuit technologies, fabricated linewidths are frequently smaller than drawn linewidths. For example, it is generally known that in CMOS technologies, the linewidths of fabricated polysilicon gate electrodes may be considerable smaller than their respective drawn linewidths. This gate length reduction is particularly noticeable in dense areas of an integrated circuit where the micro-loading or proximity effects from etch, photo and mask making steps are significant. As will be understood by those skilled in the art, these effects translate into reductions in MOSFET gate lengths in logic circuits such as NAND, NOR and MUX where polysilicon gate electrodes are typically laid out side-by-side at closely adjacent locations (i.e., in a dense manner) as multi-gate field effect transistors. These effects may also become more pronounced and have more impact on circuit speeds and driving capability as CMOS and other technologies continue to be scaled downward. Traditional techniques to account for these effects have included Optical Proximity Correction (OPC) and various schemes have been proposed to physically compensate for the linewidth reductions caused by microloading effects. Such schemes are described in articles by B. D. Cook, entitled “Dose, Shape, and Hybrid Modifications for Pyramid in Electron Beam Proximity Effect Correction”, J. Vac. Sci. Technol. B , Vol. 15, No. 6, pp. 2303-2308, November/December (1997) and C. Dolainsky et al., entitled “Evaluation of Resist Models for Fast Optical Proximity Correction”, Proc. SPIE, Vol. 3236, pp. 202-207 (1998).
Unfortunately, micro-loading effects typically cannot be completely corrected for because severe linewidth reductions may occur during subsequent fabrication and back-end processing steps (i.e., after OPC). Thus, notwithstanding conventional attempts to account for micro-loading or proximity effects, there continues to be a need for improved methods of determining linewidth reductions (e.g., gate length reductions) that can address the additional causes of micro-loading.
It is therefore an object of the present invention to provide improved methods, apparatus and computer program products for determining the extent to which drawn linewidths of an integrated circuit schematic are reduced when the integrated circuit is fabricated.
It is also an object of the present invention to provide methods, apparatus and computer program products for more accurately modeling multi-gate field effect transistors.
These and other objects, features and advantages can be provided by methods, apparatus and computer program products for modeling integrated circuits having dense devices therein that experience linewidth (e.g., gate electrode) reductions during fabrication. In particular, for dense devices having electrical paths therein and first and second gate electrodes that overlie the electrical path, preferred operations include determining an electrical gate length of the first gate electrode by evaluating a change in current through the electrical path relative to a change in gate length of the second gate electrode. This operation to determine an electrical gate length of the first gate electrode preferably comprises determining an electrical gate length of the first gate electrode (L1) by evaluating a change in simulated drain-to-source current (IDSsim) through the electrical path relative to a change in an electrical gate length of the second gate electrode (L2). For example, determining L1 preferably comprises determining ∂IDSsim/∂L2 and ∂IDSsim/∂L1 from the simulated drain-to-source current, where “∂” denotes a partial derivative.
The preferred operations also include determining I-V characteristics of the device based on measured drain-to-source current (IDSmeasured) through the electrical path and determining L1 by solving x1=(∂IDSsim/∂L2)/(∂IDSsim/∂L1) and determining when ∂E(ΔL1, ΔL2)/∂ΔL1=0, where E(ΔL1, ΔL2) is a square error function that equals Σ(ΔL1+x1kΔL2−yk)2, for all integers k ranging from 1 to N, where N is the number of data points in the simulated drain-to-source current IDSsim, yk=(IDSmeasured−IDSsim)/(∂IDSsim/∂L1)|k, and ΔL1 and ΔL2 are gate length reductions for the first and second gate electrodes, respectively. Measurements of actual drain-to-source current can be achieved by measuring dense devices formed on modeling or test chips that can be fabricated side-by-side on a wafer with functional integrated circuit chips containing identical dense devices. By providing actual dense devices that can be measured and modeled individually after fabrication (or at least after most fabrication steps have been performed), more complete assessments of the causes of microloading can be achieved and the models derived therefrom can be used to more accurately represent the performance of active devices within functional integrated circuit chips.
According to another embodiment of the present invention, operations to model a multi-gate field effect transistor include measuring current through an actual drain-to-source path of the multi-gate field effect transistor at a plurality of actual drain-to-source voltages and simulating current through a simulated drain-to-source path of the multi-gate field effect transistor at a plurality of simulated drain-to-source voltages. Operations are also performed to determine a first derivative of the simulated current with respect to a gate length of a first gate electrode of the multi-gate field effect transistor and determine a second derivative of the simulated current with respect to a gate length of a second gate electrode of the multi-gate field effect transistor. Then the electrical gate lengths of the first and second gate electrodes are found by determining, based on the first derivative and the second derivative, a least squares fit between the measured current and the simulated current. These electrical gate lengths accurately account for the process-induced gate line reductions that occur during fabrication of an integrated circuit.
According to preferred aspects of this embodiment, the operations to measure current comprise operations to measure the I-V characteristics of the multi-gate field effect transistor with all but a first of the gate electrodes therein held at respective fixed control voltages. To achieve higher degrees of accuracy, the operations to simulate current also preferably include simulating current through a simulated drain-to-source path of the multi-gate field effect transistor at a plurality of simulated drain-to-source voltages of a first polarity and also at a plurality of simulated drain-to-source voltages of a second polarity opposite the first polarity.
According to yet another embodiment of the present invention, operations to model a multi-gate field effect transistor include determining a square error function E(ΔL1, ΔL2, . . . , ΔLn), where n is the number of gate electrodes in the multi-gate field effect transistor and ΔLn is the gate length reduction associated with the nth gate electrode. Gate length reductions can then be determined from the square error function by solving for when ∂E/∂ΔL1, ∂E/∂ΔL2, . . . , and ∂E/∂ΔLn all equal zero. According to this embodiment of the present invention, the square error function for a field effect transistor having n gate electrodes can be expressed as:
In the case where n=3, the operations to determine the gate length reductions comprise determining when the following equations are simultaneously satisfied:
The present invention now will be described more fully hereinafter with reference to the accompanying drawings, in which preferred embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. Like numbers refer to like elements throughout.
Referring now to
A verification system 400 is also preferably provided for performing an independent verification of the physical layout to ensure compliance with the requirements of the functional specification and logic synthesis system 200 as well as the manufacturing system 500. Accordingly, the verification system 400 is typically referred to as a “post-layout” verification system and is typically employed near the end of the design process. In addition to acting as an independent verification of the operability and correctness of the layout of the circuit design, the verification system 400 may provide means by which changes and optimizations can be performed. As will be understood by those skilled in the art, various other types of analyses such as timing analysis and circuit/logic simulation may be performed to check whether the specifications and requirements of the first two subsystems 200 and 300 are satisfied. After verification, the physical layout is forwarded to the manufacturing system 500 to produce the integrated circuit. The microelectronic circuit manufacturing system 500 may generate the required masks, and may control the manufacturing tools necessary to fabricate the integrated circuit on a semiconductor wafer, for example. The manufacturing system 500 may also generate a “modeling chip” from which critical modeling parameters can be derived. These additional modeling parameters can then be used by the functional specification and logic synthesis system 200 and post-layout verification system 400, as described more fully hereinbelow.
It will be understood by those having skill in the art that the integrated circuit functional specification and logic synthesis system 200, microelectronic circuit layout system 300 and various parts of the microelectronic integrated circuit manufacturing system 500 may be realized in whole or in part as software modules running on a computer system. Alternatively, a dedicated stand-alone system with application specific integrated circuits for performing the above described functions may be provided. The general design and operation of the functional specification and logic synthesis system 200, layout system 300 and manufacturing system 500 are well known to those having a skill in the art and need not be described further herein.
Referring now to
Referring now to
The third operation 430 is performed by a layout parameter extraction (LPE) tool. This tool may, among other things, perform an initial operation of modeling each of a plurality of interconnect nets within the integrated circuit layout in order to obtain estimates of the parasitic resistance and capacitance of the nets. Once this layout parameter extraction operation 430 is complete, modeling and simulation operations 440 are performed as described more fully hereinbelow with respect to
The operations of the present invention, as described more fully hereinbelow, may be performed by an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product on a computer-readable storage medium having computer-readable program code embodied in the medium. Any suitable computer-readable medium may be utilized including hard disks, CD-ROMs or other optical or magnetic storage devices.
Additional operations performed by various aspects of the present invention are illustrated in more detail in
Referring now to
Referring next to Block 442, models for the isolated and dense devices are then generated. As described more fully hereinbelow, these models are developed to accurately account for process-induced linewidth variations (e.g., gate length reductions). The models are then verified in a preferred manner, Block 443, and then the verified models are stored in a model library, Block 444. This model library may be used by the above-described specification and logic synthesis system 200 and layout parameter extraction tool 430, for example.
Referring again to Block 441 of
The I-V characteristics of a plurality of isolated and essentially “micro-loading free” devices (e.g., MOSFETs) having varying gate lengths, widths, orientation, etc., are then measured on the modeling chip. From these measurements, accurate models (e.g., SPICE models) of the isolated devices can be extracted using conventional parametric extraction techniques well known to those skilled in the art. The accuracy of these SPICE models can then be confirmed for each isolated device. For example,
The I-V characteristics for each of the dense devices are then measured by, among other things, applying a constant voltage (e.g., power supply voltage Vdd) to the designated control gate electrode (CG) and applying various gate voltages to the independent gate electrode, such as G1 in the dense device of FIG. 5. The I-V characteristics are measured in order to obtain relationships between IDSmeasured v. VDSmeasured for each of a plurality of gate electrode voltages VG1 (i.e., a family of characteristic I-V curves). Referring now to
An extracted SPICE model is next used to simulate the I-V characteristics of the equivalent circuit of
Referring now to Block 442 of
R1=RSH(S1+ΔL1/2+ΔL2/2)/W
R2=RSH(S2+ΔL1/2+ΔL3/2)/W (1)
R3=RSH(d1+ΔL2/2)/W
R4=RSH(d2+ΔL3/2)/W
where RSH is the source/drain sheet resistance and ΔLi (i=1,2,3 . . . ) are the gate length reductions for transistors T1, T2 and T3 in the equivalent circuit. A simulation (e.g., SPICE simulation) of the drain-to-source current (IDSsim) as a function of, among other things, the electrical gate lengths Li (i=1,2,3 . . . ) is then determined along with a calculation of a first order derivative of IDSsim with respect to Li (i.e., ∂IDSsim/∂Li). As will be understood by those skilled in the art, values of IDSsim are also a function of the gate voltages and the drain-to-source voltages (VDSsim).
Referring still to
IDSmeas−IDSsim=(∂IDSsim/∂L1)ΔL1+(∂IDSsim/∂L2)ΔL2+(∂IDSsim/∂L3)ΔL3 (2)
Dividing both sides of equation (2) by ∂IDSsim/∂L1 and defining “y” as (IDSmeas−IDSsim)/(∂IDSsim/∂L1), x1 as (∂IDSsim/∂L2)/(∂IDSsim/∂L1) and x2 as (∂IDSsim/∂L3)/(∂IDSsim/∂L1), yields the following relationship:
y=ΔL1+ΔL2x1+ΔL3x2 (3)
Based on equation (3), a square error function can be written as:
where N is the number of data points in the IDS v. VDS family of curves that are used in the linear least square fit operation. In order for the square error function to have a minimum, the following relationships should be satisfied:
These relationships should allow ΔL1, ΔL2 and ΔL3 to be uniquely determined. Referring still to
Li(m)=Li(m-1)+ΔLi (6)
where Li(m-1) is the gate length recorded after the (m-1)th iteration. As illustrated by
As illustrated by FIG. 6C and Block 443 of
In order to improve the accuracy of the models of the dense devices even further, the above-described operations should be performed again with the polarity of VDS switched. Performing the operations again with a switched VDS is advantageous because the influence of T2 (or T3) in
Using a switched VDS polarity, equation (3) above can be rewritten as follows:
y′=ΔL1+ΔL2x1′+ΔL3x2′ (7)
By adding equations (3) and (7) together and defining {overscore (x)}1=(x1+x1′)/2, {overscore (x)}2=(x2+x2′)/2 and {overscore (y)}=(y+y′)/2, the above equations (3)-(5.3) can be rewritten as follows:
Accordingly, by switching the polarity of VDS, the relationships of equations (10.1)-(10.3) allow ΔL1, ΔL2 and ΔL3 to be more accurately determined through iteration.
Notwithstanding the high level of accuracy of the above described operations, equations (10.1)-(10.3) may on some limited occasions provide erroneous results for ΔL2 and ΔL3 (although ΔL1 is typically still correct), even though the above equations are mathematically correct. This possibility of error is due to the fact that the illustrated least squares technique requires the function y of equation (8) to be sufficiently sensitive to the variations of ΔL1, ΔL2 and ΔL3. However, because ΔL2 and ΔL3 may frequently be about equal because they experience similar micro-loading effects, the variables x1 and x2 may also be about equal and, accordingly, equations (10.2) and (10.3) may become similar. In fact, because ΔL3 equals ΔL2+s, where s is the difference between ΔL3 and ΔL2 and is very small, the least squares technique can become essentially a technique to find ΔL1, ΔL2 and s. But, because the function y of equation (8) may not be sufficiently sensitive to s, the use of equations (10.1)-(10.3) may fail to yield sufficiently accurate results in some limited circumstances.
To address this potential limitation on accuracy, equation (3) is preferably subtracted, from equation (7) and the following relationships are defined: Δy=y−y′, Δx1=x1−x1′, Δx2=x2−x2′. Using these relationships, the following equation can be derived:
Δy=Δx1ΔL2+Δx2ΔL3 (11)
Because this equation (11) can be used to replace equation (10.3), the following set of equations can be used to more accurately determine ΔL1, ΔL2 and ΔL3 in situations where s is small:
A solution of equations (12.1)-(12.3) to determine ΔL1, ΔL2 and ΔL3 may be best illustrated by performing the operations of
Using the results of TABLE 1 and rewriting equations (12.1)-(12.3) in matrix form yields:
The solution of the matrix equation (13) provides ΔL1=−0.033805, ΔL2=−0.004301 and ΔL3=−0.010415. Therefore, based on the initial estimates, the electrical gate lengths are: L1=0.26−0.033805, L2=0.27−0.004301 and L3=0.28-0.010415.
As will be understood by those skilled in the art, the above described operations may be applied to numerous other dense devices, such as those illustrated by
and then determining when the square error function of equation (15) has a minimum.
Referring again to
The above-described model verification operations may also include the use of simple delay circuits formed on the modeling chip. These delay circuits are preferably constructed to have dense poly gate patterns and can be used to verify the gate length reductions extracted from the dense devices. For example, as illustrated by
In the drawings and specification, there have been disclosed typical preferred embodiments of the invention and, although specific terms are employed, they are used in a generic and descriptive sense only and not for purposes of limitation, the scope of the invention being set forth in the following claims.
Number | Name | Date | Kind |
---|---|---|---|
4658367 | Potter | Apr 1987 | A |
5481485 | Takeuchi | Jan 1996 | A |
5693178 | Chan et al. | Dec 1997 | A |
5698902 | Uehara et al. | Dec 1997 | A |
5783101 | Ma et al. | Jul 1998 | A |
5843847 | Pu et al. | Dec 1998 | A |
5883007 | Abraham et al. | Mar 1999 | A |
5917205 | Mitsui et al. | Jun 1999 | A |
5930677 | Zheng et al. | Jul 1999 | A |
5933356 | Rostoker et al. | Aug 1999 | A |
6174741 | Hansch et al. | Jan 2001 | B1 |
6246973 | Sekine | Jun 2001 | B1 |
6278964 | Fang et al. | Aug 2001 | B1 |
6553340 | Kumashiro | Apr 2003 | B1 |