MULTIPLE CD-SEM MATCHING FOR 0.18 µm LINES/SPACES AT DIFFERENT EXPOSURE CONDITIONS André Engelen and Ingrid Minnaert-Janssen ASML De Run 1110 5503 LA, Veldhoven The Netherlands This paper was first presented at the SPIE Microlithography Conference in March 1999 Santa Clara Convention Center, Santa Clara, California, USA. MULTIPLE CD-SEM MATCHING FOR 0.18 µm LINES/SPACES AT DIFFERENT EXPOSURE CONDITIONS André Engelen and Ingrid Minnaert-Janssen ASML De Run 1110 5503 LA, Veldhoven The Netherlands ABSTRACT Critical Dimension Scanning Electron Microscopes (CD-SEMs) are used within ASML for evaluating the imaging performance of Stepper and Step & Scan systems. This implies measuring a large number of Focus Exposure Matrices (FEMs) and inter/intra-field CD measurements on different CD-SEMs. Therefore the CD-SEMs in ASML are matched through focus (Best Focus ± 0.3 µm). The matching procedure is done on three steps. First, all CD-SEMs are checked for stability. Then the magnification factor for each of the individual CD-SEMs is checked in order to make sure that the tool is set up correctly. Finally, the CD measurements are matched through focus and multiple exposure energies. In this paper, we will show that the CD-SEMs of different vendors (Applied Materials and Hitachi) can be matched using photoresist features, through focus within 5 nm for 0.18 µm features. This matching includes different orientations (horizontal and vertical structures) and densities (isolated and nested structures) using only one correction offset. 1. INTRODUCTION different feature sizes, line orientations, and line densities. In addition, we are matching CD-SEMs of two vendors, Applied Materials and Hitachi. Critical Dimension Scanning Electron Microscopes (CD-SEMs) are used for evaluating the imaging performance of Stepper and Step & Scan systems. Typical figures of merit derived from these evaluations are Depth of Focus (DoF), Exposure Latitude (EL) and CD Uniformity (CDU). Typically Focus Exposure Matrices (FEMs) involving large defocus and number of energy steps are measured to determine these figures of merit. Currently eight CD-SEMs of various vendors and types are used within ASML’s worldwide equipment set. Therefore, tool-to-tool reproducibility, so called matching, is very important. Additionally, the matching contribution to the multiple tool precision will increase when going to smaller feature sizes. As shown in table 1, the 1997 SIA roadmap predicts feature sizes of less than 100 nm within a decade. Proportional to decreasing feature sizes from 250 nm to 70 nm, the precision of CD-SEMs (or alternative metrology measurement tools) will have to improve at the same time from 4 nm to 1 nm. Table 1: Technology Generation and CD-SEM Precision Goals for the next decade according to the 1997 SIA Roadmap Year of First IC Shipment 1997 1999 2001 2003 2006 2009 2012 Technology Generation (nm) 250 180 150 130 100 70 50 Lithography Budget (in nm, 3σ post etch) 20 14 12 10 7 5 4 Final CD Output Metrology Precision (in nm, 3σ) 4 3 2 2 1.4 1 0.8 One of the major difficulties in submicron CD metrology is the lack of an accepted reference standard. There are no linewidth standards in the nanometer range that are relevant to the kind of features measured (e.g. photoresist lines and spaces on bare silicon wafers or anti reflective layers), and behave similar in CD-SEM inspection (e.g. with respect to charging). Finally, references and calibration standards also need to be stable during use and over time. The matching approach for the evaluation of Stepper and Step & Scan systems differs from the approach of other users because their CD-SEMs are mainly based in semiconductor production environments [2,3]. The CD-SEM usage for system evaluation at ASML requires multiple CD-SEM matching through focus for The most important difference between calibrating and matching is, that instead of using an accepted reference, for matching a reference is determined. This 1 2. EQUIPMENT means that one CD-SEM is chosen to be the reference and that all other CD-SEMs are matched towards this reference tool. As a consequence, accuracy is not an outcome. ASML uses worldwide different types of CD-SEMs: - Automated top down CD-SEMs: - Applied Materials 7830Si (4x) - Hitachi S-8840 (2x) - Hitachi S-8C40 (1x). The terminology used in this work conforms to ISO [4,5]. measurement terminology The overall measurement error is the square root of the sum of all variances, which originate from all sources of variability influencing the measurement. These variances are gathered in two terms: repeatability and reproducibility. Repeatability is the variance on measurements taken under the same conditions and reproducibility is the variance on measurements under different conditions. Long term repeatability is referred to as single-tool stability with a typical time period being one month. - Tiltable process diagnostic CD-SEM: - Hitachi S-7800H (1x). The CD-SEMs that are used in this paper are named according to Table 2. Table 2: Overview of ASML’s CD-SEM equipment used in this paper single tool reproducibility or stability tool-to-tool reproducibility or matching repeatability or single tool precision I-12531.ILL In this study, the main contributors to the measurement error were determined as: dynamic repeatability, and tool-to-tool reproducibility or matching (Figure 1). TYPE LOCATION AMAT #1 Applied Materials 7830Si Veldhoven AMAT #2 Applied Materials 7830Si Veldhoven AMAT #3 Applied Materials 7830Si Veldhoven AMAT #4 Applied Materials 7830Si Tempe S-7800H Hitachi S-7800H Veldhoven S-8C40 Hitachi S-8C40 Veldhoven To match all CD-SEMs, a reference CD-SEM has to be determined. For this matching procedure, AMAT #1 was selected to be the reference CD-SEM for the following reasons: multiple tool precision Figure 1: NAME - This CD-SEM is used to optimize all algorithms using a cross-section SEM for top, sidewall and foot verification. Measurement terminology. In a stable environment, multiple tool precision has been determined from repeatability and tool-to-tool reproducibility - Half of the equipment set, are identical Applied Materials systems which are used throughout ASML. [1], According to the 1997 SIA Roadmap the target for total multiple tool precision is 3 nm for 0.18 µm technology. The Advanced Metrology Advisory Group (AMAG) divided this budget for the total multiple tool precision equally between single-tool precision (dynamic repeatability) and tool-to-tool reproducibility (matching) [4]. This indicates a budget of 2.1 nm for each variable. These budgets are very close to the resolution and performance limits of the state of the art CD-SEMs as used within ASML. Evaluating imaging performance implies measuring FEMs using large defocus values (typically 1.0 µm) and large energy ranges (typically 3 mJ for APEX-E). The recipes, which are used on the CD-SEMs, can be divided into three groups: 1) 2) 2 The FEM recipes, with which a complete FEM is measured (up to 1176 data points). This measurement results in a value for Best Energy (BE) and Best Focus (BF) of the tested Stepper or Step & Scan system. The Usable Depth of Focus (UDoF) measurements and 5-bar measurements. To measure UDoF, nine FEMs divided over the Stepper/Step&Scan exposure field are 3) measured through focus and at best energy. The 5-bar measurements are very similar to the UDoF measurements but put extra attention to the differences in CD-value between the first, third and fifth line of a 5 bar structure in order to determine lens aberrations and proximity effects. The CD Uniformity test, which measures the CD range for both vertical and horizontal lines over a focus range of BF ± 0.3 µm in 6 fields across the wafer. This implies that an offset between vertical and horizontal measurements (HV-differences) of the CD-SEM influences the result of the CD-Uniformity test. of the CD measurements. In contrast with the first two steps, this step requires unique wafers which have to be circulated among the several CD-SEMs. This is critical since exactly the same features have to be measured. These three steps for matching will now be discussed individually in more detail: Step 1: CD-SEM Stability Prior to matching, it has to be assured that the stability of the CD-SEMs which are included in the matching procedure is within the control limits. Applying correction offsets to tools that are not stable in time, can result in incorrect adjustments of the CD measurements. The stability of the individual CD-SEMs is checked daily by measuring an etched polysilicon reference wafer. On this wafer, both vertical and horizontal lines are measured at five sites across the wafer. In this way, both the separate CD values for vertical and horizontal lines can be monitored as well as the horizontal-vertical differences (HV-differences). It should also be noted that with the advent of Step & Scan systems, the metrology requirements on accurate H-V control is essential in assessing the ‘scanning’ performance of advanced lithography exposure tools [6]. The currently allowed range for the 0.30 µm CD values is 6 nm, while the range for the HV-difference is set to 3 nm. A part of the CD range over time can be attributed to charging and contamination effects of the measurement sample. These three groups of recipes, FEM, UDoF and CD-Uniformity measurements, show the importance of CD-SEM matching through focus. ASML’s system evaluations are performed on bare silicon photoresist wafers using a BARC. This means that next to matching over a large focus range, matching for different photoresist types is also very important. Various photoresist processes have different photoresist profiles and different charging characteristics in the CD-SEM. Therefore CD-SEM algorithms need to be optimized in order to measure correctly. This optimization needs to be done for each different type of CD-SEM separately, since each type can respond different to the same photoresist profiles due to different hard- and software. 3. METHOD Step 2: Magnification matching or pitch calibration The CD-SEM matching procedure at ASML consists of three steps: 1) CD-SEM Stability 2) Magnification matching or pitch calibration 3) Matching of Critical Dimension through focus. The magnification factors of all CD-SEMs have to be checked in order to make sure that the tool setup is correct. Magnification matching is a relative simple way to make sure that the basic machine performance is identical on the various CD-SEMs. Since pitch measurements are relatively insensitive to charging, focus, energy and the measurement algorithm, magnification matching is done by measuring pitch linearity over a large pitch range of 0.5 µm to 2.4 µm on an etched polysilicon wafer for both horizontal and vertical features. This pitch range was chosen in order to check linearity over a large CD range. This step 2 matching must be done before step 3 of the matching procedure is started. Adjustment of the magnification factor afterwards, results in a different value for the measured CD. The first two steps involve measurements using an etched polysilicon wafer. Since pitches are determined by the reticle, these measurements can be performed on different wafers as long as they are exposed with the same reticle. These two steps make sure that the operating condition of all individual CD-SEMs is optimal. Matching within tight specification limits can only be done when the stability over time is better than the repeatability of the individual CD-SEM. Additionally, it is obvious that unstable tools and tools with an incorrect magnification factor cannot be included in the matching procedure until these errors are corrected. The last step of matching involves the actual matching 3 Step 3: Matching of Critical Dimension through focus Before the offsets between the several tools can be determined and corrected, the repeatability budget has to be quantified. Since a limited number of data points is used, matching cannot correct beyond the repeatability limits of the individual tools. Quantification of the dynamic repeatability through focus is done by measuring a particular feature ten times where the energy used remains constant at a nominal or best energy while focus is varied in a FEM layout. To correct for charging and contamination effects, for each focus step a linear equation is fitted through the ten data points. The dynamic repeatability is defined as three times the standard deviation of the residual errors from the fit averaged over a focus range of BF ± 0.4 µm. The slope of the linear regression line characterizes the charging and contamination. Because photoresist wafers are measured eight times on the several tools, during the matching procedure, analysis of the matching results will include the correction for charging and contamination to judge on the absolute offset between the CD-SEMs. Different types of CD-SEMs use different hardware (e.g. columns, detectors) and software (e.g. measurement algorithms). As a consequence, they respond differently to photoresist images, sidewall profiles and substrates. Therefore a stable CD-SEM and magnification matching is insufficient. The matching of CD measurements is done by measuring Focus Exposure Matrices (FEMs). These photoresist FEMs are measured for: - feature sizes: - 0.35 µm (365nm I-line) - 0.25 µm, 0.22 µm and 0.18 µm (248nm DUV). - line orientations: - horizontal - vertical. - line densities: - isolated - nested. 4. RESULTS Step 1: CD-SEM Stability Additionally, pitch measurements (vertical and horizontal), and measurements for the first, third and fifth line of a vertical 5-bar structure are included. This so-called 5-bar measurement is only performed on the AMAT systems since these systems can measure the three lines in one single pass with minimal throughput penalty. The stability of the CD-SEMs is monitored by measuring, on a daily basis, horizontal and vertical features on an etched polysilicon wafer. Figure 2 shows a typical example of these measurements over a month for both the CD values and the HV-difference, including the horizontal (HCL) and vertical control limits (VCL). In this figure all measurements are shown, including the out of specification situations. It should be noted that the results are not corrected for charging and contamination, which result in an increase of CD values after some measurements. The charging and contamination can be removed by putting the wafer in an acid bath, after which the measured CD value returns to its lower, pre-charged value. All CD-SEMs offer the possibility to include offsets. During normal operation, the offsets obtained in the matching procedure are used. During initial matching for each CD-SEM, the applied offsets were set to zero in the measurement parameter files used in the CD-SEM matching recipes. This makes it possible to compare the absolute offset between the various CD-SEMs which are caused by hardware and/or software differences. Additionally, the current offsets can be input into the analysis software to determine the current performance. After measuring the same FEMs on the different CD-SEMs, the data is subtracted from each other. This results in a FEM containing only differences between systems. From this, the average differences or offsets are calculated over a focus range of Best Focus ± 0.3 µm (i.e. a focus range of 0.6 µm). For the normal FEMs that are typical for Stepper and Step & Scan system evaluation, this focus range contains 63 data points. When calculating the stability of the CD-SEM, the CD measurements are corrected for charging and contamination effects. Stability, long term repeatability, has been determined in a similar way as the dynamic repeatability determination. A typical value for the stability is 1.7 nm. Figure 2b shows the HV-differences, which are controlled with the stigmation settings and the beam alignment. The relative change in HV-difference is important and it is apparent that the sample has an inherent HV-difference. Because of this HV-difference the horizontal lines are approximately 6 nm wider compared to the vertical ones. 4 292 288 after acid bath 284 HCL VCL horizontal vertical 280 2 6 10 14 18 26 30 2 215 7 8 9 10 time [hours] Figure 3: 14 18 22 26 30 The magnification matching results of all CD-SEMs are very similar, they are actually plotted on top of each other. The fitted line in Figure 4 shows the linear regression results of the average of all CD-SEMs. As expected the slope equals one and the intercept equals zero. Additionally, in both cases the fitted line has a coefficient of correlation, R2, of 0.99999. . 6 10 Before the actual CD matching, the magnification matching on all CD-SEMs was done. The pitches, measured on an etched polysilicon wafer, ranged from 0.5 to 2.4 µm. We assumed that the real physical feature size of the measured pitches equals the targeted feature size on the reticle divided by four. Therefore we expect the slope of the fitted line to be one and the intercept zero. The results for vertical and horizontal magnification matching are shown in Figure 4. In these graphs the measured pitches are plotted as a function of the assumed physical pitches 218 5 6 Step 2: Magnification matching or pitch calibration. 220 4 CL Figure 2b: HV-differences as a function of time (measured daily). I-12744.ILL Critical Dimensions [nm] Hitachi S-7800H 223 3 -8 number of measurement [days] 225 2 -6 H-V 22 Figure 3 shows stability results for 0.22 µm features as a function of time, corrected for charging and contamination effects, for both the Hitachi S-7800H and Hitachi S-8C40. The same feature has been regularly measured over a time frame of approximately 8 hours. Comparison of the results indicate that the Hitachi S-7800H is not suitable for automated SEMing within ASML (measuring time is typically 3-4 hours). This is caused by the fact that this microscope is equipped with a cold field-emission electron gun that has to be flashed once a day [7]. Therefore, the Hitachi S-7800H is mainly used for tilted profile images and excluded from the matching procedure. 1 -4 stigmator not optimal Figure 2a: CD of horizontal and vertical features as a function of time (measured daily). 0 -2 -10 number of measurement [days] Hitachi S-8C40 I-12482.ILL CD (H-V) [nm] 296 I-12481.ILL Critical Dimension [nm] 0 300 Measured CD as a function of time for the Hitachi S-7800H and the Hitachi S-8C40. 5 2.5 2.0 1.5 AMAT #1 AMAT #2 AMAT #3 AMAT #4 S-8C40 0.5 0 0 0.5 1.0 1.5 2.0 2.5 2.0 1.5 AMAT #1 AMAT #2 AMAT #3 AMAT #4 S-8C40 1.0 0.5 0 0 3.0 Y = 1.005X - 0.002 R2 = 0.99999 0.5 1.0 1.5 Figure 4a: Vertical magnification matching. Measured pitches as a function of the assumed physical pitches. AMAT #1 AMAT #2 AMAT #1 AMAT #2 AMAT #3 frequency frequency AMAT #4 S-8C40 #1 10 0.35 0.36 3.0 30 AMAT #3 0 0.34 2.5 Figure 4b: Horizontal magnification matching. Measured pitches as a function of the assumed physical pitches. I-12540.ILL 30 20 2.0 pitch on reticle /4 [µm] pitch on reticle /4 [µm] 0.37 I-12539.ILL 1.0 I-12477.ILL Y = 1.002X - 0.001 R2 = 0.99999 measured pitch [µm] measured pitch [µm] 2.5 3.0 I-12478.ILL 3.0 20 S-8C40 #1 10 0 0.34 0.38 AMAT #4 0.35 0.36 0.37 0.38 critical pitch dimension [ m] critical pitch dimension [ m] Figure 5a: Vertical pitch measurements for 0.18 µm on 0.5 µm thick APEX-E represented in histograms. Figure 5b: Horizontal pitch measurements for 0.18 µm on 0.5 µm thick APEX-E represented in histograms. Special attention is given to HV-difference because of its importance when evaluating Step & Scan systems. Figure 6 shows that the maximum HV-difference is 1 nm when the magnification factor is set up correctly. Magnification matching can be verified by measuring these pitches. Therefore pitch measurements for both vertical and horizontal features are measured on a FEM. Pitches are independent of focus and energy, therefore results can be presented in a histogram. Figure 5 shows the histograms of the vertical and horizontal measurements for 0.18 µm features measured on 0.5 µm thick APEX-E. It can be concluded that after magnification matching the setup for the individual CD-SEMs independent of vendor is good with respect to pitch measurements. This includes HV-differences, which are smaller than 1% of the target CD-value. After magnification matching, the pitch differences for both orientations are less than 2 nm for all CD-SEMs, independent of the vendor. Thus, when the magnification factors are setup correctly, the pitches match within the dynamic repeatability of our CD-SEMs. The minor differences can be caused by small hardware differences (e.g. detector efficiency, tip condition and electron optics). Furthermore, the magnification factors of all CD-SEMs included in the matching procedure are correct. In combination with the correct long term repeatability, stability, the operating condition of the CD-SEMs is good and the last step of the matching procedure, matching of critical dimension through focus, can be done on these tools. 6 20 10 0.35 0.36 0.37 critical pitch dimension [ m] Figure 6: 0.38 vertical horizontal 20 10 0 0.34 0.35 0.36 0.37 S-8C40 vertical horizontal 20 10 0 0.34 0.38 critical pitch dimension [ m] 0.35 0.36 0.37 0.38 critical pitch dimension [ m] Vertical and horizontal pitch measurements for three CD-SEMs included in the matching procedure. Critical Dimension [nm] Step 3: Matching of Critical Dimension through focus. The last and most important part of the matching procedure is the actual matching of CD measurements. In this step the CD differences or offsets will be quantified in order to determine the offsets which have to be applied on the several CD-SEMs. This paragraph will be divided into two parts; first the quantification of repeatability and charging budgets will be discussed followed by the results of CD matching. 200 Y = 0.7X + 184 190 180 0.4 µm 0.3 µm 0.2 µm 170 1 Figure 7: Repeatability and Charging The CD at best energy has been measured ten times through focus. An example of an experiment done on the AMAT #4 for 0.18 µm features on 0.5 µm thick APEX-E is shown in Figure 7, the CD is plotted as a function of the run number for BF ± 0.4 µm. According to the method described in the previous section the repeatability and charging budget is determined. The charging budget, slope of the fitted line, per measurement for AMAT #4 equals 0.7 nm. For the other CD-SEMs similar contributions have been found. After correction for charging and contamination effects the dynamic repeatability has been determined over a focus range of BF ± 0.4 µm on two randomly chosen Applied Materials systems (AMAT #1, AMAT #4) and on the Hitachi S-8C40. The values determined for the dynamic repeatability are 2.3 nm, 2.5 nm and 2.6 nm respectively. The good repeatability of the CD-SEMs, in combination with the good stability results, allow us to match our CD-SEMs. 7 I-12541.ILL 0 0.34 30 AMAT #3 frequency vertical horizontal frequency frequency AMAT #1 I-12536.ILL 30 30 2 3 0.1 µm BF -0.1 µm 4 5 6 run number -0.2 µm -0.3 µm -0.4 µm 7 8 9 10 Averaged CD over a focus range of Best Focus ± 0.4 µm for 0.18 µm features on 0.5 µm thick APEX-E as a function of the run number. The slope of the line fitted equals the charging contribution. Critical dimension matching for charging and contamination effects (0.7 nm per measurement). For all feature sizes, CD-SEMs of the same vendor (Fig. 9a) show similar results. All offsets are smaller than 5 nm and independent of orientation (horizontal or vertical structures) and density (isolated or nested structures). The differences between CD measurements can be explained by minor differences between CD-SEMs of the same type due to e.g. different detector efficiency, different tip condition, beam current or operator dependent setup (stigmation, first aperture). CD-SEMs of different vendors show larger offsets (Figure 9b). The DUV features, using APEX-E photoresist, all show similar offsets independent of the orientation and density, namely averaged 20 nm. These differences arise from the usage of different software (e.g. measurement algorithms) and hardware (e.g. columns, electron optics) by the compared vendors. These differences are corrected by applying a constant correction offset to the CD-SEM recipes of the different CD-SEMs. The i-line features, however, show different offsets. This illustrates the importance of matching all feature sizes, especially when different photoresists are used. I-12542.ILL The final step is the matching of the critical dimension measurements. The FEM results of all CD-SEMs are compared and eventually corrected towards the reference CD-SEM, AMAT #1. The CD matching is done according to the method as described in section 2. Figure 8 shows the Bossung curves for nine different energies (for vertical dense lines) as measured on AMAT#1 and AMAT #2, left and right corner respectively. These two FEMs are subtracted from each other in the center graph. This figure clearly shows that the difference between both systems is constant through focus. Due to the optimized measurement algorithms, this is typical for all CD-SEM comparisons independent of vendor. This justifies that a constant offset through focus can be applied.The critical dimension matching procedure includes different feature sizes, 0.35 µm (365 nm I-line) and 0.25 µm, 0.22 µm and 0.18 µm (248 nm DUV). Figure 9 shows the absolute offsets of AMAT #1 and AMAT #3 (Figure 9a) and AMAT #1 and Hitachi S-8C40 (Figure 9b) for all these feature sizes after correction Figure 8: Comparison of a FEM measured on AMAT#1 (reference) and AMAT#2. The small corner graphs show Bossung curves for nine energies, while these FEMs are subtracted from each other in the center graph. 8 I-12534.ILL 0.18 m 0.22 m 0.25 m 0.35 m vertical dense L3 vertical dense L2 vertical dense L1 vertical isolated horizontal isolated I-12760.ILL 15 10 5 0 AMAT #1 - S-8C40 -10 AMAT #1 - AMAT #4 -5 vertical dense AMAT #1 - S-8C40 AMAT #1 - AMAT #3 AMAT #1 - AMAT #2 vertical dense L3 vertical dense L1 vertical dense L2 vertical isolated horizontal isolated horizontal dense AMAT #1 - AMAT #4 -5 20 AMAT #1 - AMAT #3 0 25 AMAT #1 - AMAT #2 5 Absolute differences between CD-SEMs of different vendors, AMAT #1 and Hitachi S-8C40 for different features (0.35 µm (365nm I-line) and 0.25 µm, 0.22 µm and 0.18 µm (248nm DUV)), orientations (vertical and horizontal structures), densities (isolated and nested structures). vertical dense L3 10 vertical dense -10 vertical dense L2 15 Figure 10a: -5 vertical dense L1 absolute CD offset [nm] 20 -10 0 Figure 9b: I-12535.ILL Absolute differences between CD-SEMs of the same vendor, AMAT #1 and AMAT #3 for different features (0.35 µm (365nm I-line) and 0.25 µm, 0.22 µm and 0.18 µm (248nm DUV)), orientations (vertical and horizontal structures), densities (isolated and nested structures) and 5-bar measurements. 25 5 vertical dense 0.18 m 0.25 m 0.35 m vertical dense L3 vertical dense L1 vertical dense L2 vertical isolated horizontal isolated horizontal dense -10 0.22 m -5 10 vertical isolated 0 15 horizontal isolated 5 20 horizontal dense 10 25 horizontal dense 15 Figure 9a: absolute CD offset [nm] absolute CD offset [nm] I-12533.ILL 20 vertical dense absolute CD offset [nm] 25 Figure 10b: Differences after applying one constant correction offset between all CD-SEMs with the reference CD-SEM, AMAT #1, for 0.18 µm features for different orientations (vertical and horizontal structures), densities (isolated and nested structures) and 5-bar measurements. Absolute differences between all CD-SEMs with the reference CD-SEM, AMAT #1, for 0.18 µm features for different orientations (vertical and horizontal structures), densities (isolated and nested structures) and 5-bar measurements. 9 Critical dimension matching becomes more important when going to smaller feature sizes. Figure 10 shows the 0.18 µm matching results for all CD-SEMs compared to the reference system. All Applied Materials CD-SEMs show offsets less then 5 nm towards the reference system, since all these systems are equal with respect to hard- and software. As discussed previously, the Hitachi S-8C40 shows an averaged offset of 20 nm. It should be noted that this is not a problem because they match through focus and, thus, can be corrected using only one constant offset. CD-SEMs of one vendor show only small differences due to minor hardware differences (e.g. different detector efficiency, different tip condition, beam current or operator dependent setup). By comparing CD measurements of CD-SEMs of different vendors (Applied Materials and Hitachi) larger difference are observed, namely 20 nm. These offsets, caused by major hardware and software differences, can easily be corrected by applying one constant offset through focus. 6. FUTURE WORK Additionally, figure 9 and 10 show that the same correction offsets can be applied for all measured features (e.g. horizontal and vertical, isolated and nested). The differences in CD offsets for all these features is smaller than 4 nm. Therefore, after applying an averaged correction offset, the CD-SEMs are matched within specification and smaller than the dynamic repeatability of the used CD-SEMs, namely generally significant smaller than 2 nm. The main concern is maintaining the tight matching specifications over time. It is unfeasible to perform the matching procedure on regular basis, since a lot of expensive CD-SEM time will be consumed. Therefore, once CD-SEMs are matched, we’re looking into the possibility to study the resolution targets using Fourier transformation [8]. By this, the performance of all CD-SEMs can be monitored and adjusted if necessary. Next to that, all CD-SEMs will be linked to a central recipe server in order to ensure usage of common releases of CD-SEM recipes. Finally, accuracy will continue to become more important with decreasing feature sizes. Therefore, a lot of effort has to be put into absolute calibration and correlation with using cross-section SEM and AFM. Finally, in agreement with the pitch measurements, no significant HV-differences have been observed. HV-difference is very near to the dynamic repeatability of the CD-SEMs since the magnification matching for both horizontal and vertical features is performed prior to CD matching. Thus, it can be concluded that CD-SEMs of different vendors can be matched through focus within 5 nm for 0.18 µm lines/spaces, using only one correction offset and identical recipe database on CD-SEMs of one vendor. When going to smaller feature sizes the demands on the CD-SEMs become more demanding. According to the 1997 SIA Roadmap [1], for 0.13 µm technology CD-SEM precision must decrease to 2 nm. To reach this goal, a lot of effort has to be put in increasing the stability of the CD-SEMs and decreasing both the dynamic repeatability and the matching budget. By matching smaller features additional problems will arise such as severe charging and contamination, photoresist profile deviation due to e-beam sensitivity. An example of charging and photoresist deviation effects on 0.13 µm features using 193nm photoresist is shown in Figure 11. 5. CONCLUSIONS Critical Dimension Scanning Electron Microscopes can be matched through focus within 5 nm for 0.18 µm lines/spaces, using only one correction offset which is constant through focus. Thus, the offset is independent of orientation (vertical and horizontal structures) and density (isolated and nested structures). The three step matching procedure as followed within ASML shows good matching results. Firstly, the stability of the CD-SEMs included in the matching procedure is continuously monitored. Since the CD change over a month is less than 2 nm, it allows us to match our systems within tight specifications. Secondly, the magnification factors, vertical as well as horizontal, are checked and corrected on every CD-SEM. As a result, the pitch measurements are matched and the HV-differences are minimized. Thirdly, the actual CD measurements are matched. a b Figure 11: A 0.13 µm photoresist line measured for the first time (CD = 130 nm) (a) and after 4 measurements (CD = 147 nm) (b). 10 7. ACKNOWLEDGEMENTS 8. REFERENCES The authors wish to thank Erik van Brederode, Yin Fong Choi , Frank Duray, Mariëtte Hoogendijk, Ted der Kinderen, Bart Rijpers and Jenny Swinkels in the Veldhoven office and Shawn Cassel of the Demo Lab in Tempe for their support. Furthermore, they wish to thank Guy Davies, Jo Finders and Paul Luehrmann for their useful discussions with the authors and inputs during the course of this work. [1] Semiconductor Industry Association, ‘National Technology Roadmap for Semiconductors’ (1997). [2] Bowley R., Beecher J., Cogley R., Dupuis S., Farrington D., ‘Matching Analysis on Seven Manufacturing CDSEMs’, Proceedings of SPIE, Metrology, Inspection, and Process Control for Microlithography XII, Vol. 3332 (1998), p94-99. [3] Erikson D., Sullivan N., Elliott R., ‘Statistical verification of multiple CD SEM matching’, Proceedings of SPIE, Metrology, Inspection, and Process Control for Microlithography XII, Vol. 3050 (1997), p93-100. [4] Allgair J., Archie C., Banke B., Bogardus H., Griffith J., Marchman H., Postek M., Saraf L., Schlesinger J., Singh B., Sullivan N., Trimble l., Vladar A., Yanof A., ‘Towards a Unified Advanced CD-SEM Specification for Sub-0.18 mm Technology’, Proceedings of SPIE, Metrology, Inspection, and Process Control for Microlithography XII, Vol. 3332 (1998) , p138-150. [5] Semiconductor Equipment and Materials International Draft Doc. #2929, ‘Guide for gauge measurement capability analysis’. [6] Schoot, J. van, Suddendorf M., Spek, J. van der, Bornebroek F., Stoeten J., Hunter A., Rümmer, P., ‘ 0.7 NA DUV Step & Scan System for 150nm Imaging’, Presented at SPIE 1999. [7] Reimer L., ‘Image Formation in Low Voltage Scanning Electron Microscopy’, SPIE Optical Engineering Press, Volume TT 12.0 [8] Postek M.T., Vladar A.E., Davidson M.P., ‘Fourier transform feedback tool for scanning electron microscopes used in semiconductor metrology’, Proceedings of SPIE, Metrology, Inspection and Process control for Microlithography XII, Volume 3050 (1997), p68-79 11 4022 502 94097
© Copyright 2025 Paperzz