A multi-beam mask writer MBM-2000 is developed for the 3 nm technology node. It is designed to expose EUV blanks with beamlets of total current 1.6 uA at high throughput. It also supports writing leading-edge photomasks by equipping a correction function for glass thermal expansion and high-speed data path. Fast writing modes are provided for middle-grade photomask writing. Inline function of pixel level dose correction (PLDC) is implemented to reduce mask turnaround time by replacing offline corrections with PLDC, with additional benefit of fidelity improvement by dose enhancement. In this paper, writing results of MBM-2000 are reported and discussed.
Sub-nanometer accuracy attainable with electron micrograph SEM images is the only way to “see” well enough for the mask analysis needed in EUV mask production. Because SEM images are pixel dose maps, deep learning (DL) offers an attractive alternative to the tedious and error-prone mask analysis performed by the operators and expert field application engineers in today’s mask shops. However, production demands preclude collecting a large enough variety and number of real SEM images to effectively train deep learning models. We have found that digital twins that can mimic the SEM images derived from CAD data provide an exceptional way to synthesize ample data to train effective DL models. Previous studies [1, 2, 3, 4] have shown how deep learning can be used to create digital twins. However, it was unclear if SEM images generated with digital twins would have sufficient quality to train a deep learning network to classify real SEM images. This paper shows how we built three DL tools for SEM-based mask analysis. The first tool automatically filters good quality SEM images, particularly for test chips, using a DL-based binary classifier. A second tool uses another DL model to align CAD and SEM images for applications where it is important that features on both the images are properly aligned. A third tool uses a DL multi-class classifier to categorize various types of VSB mask writer defects. In developing the three tools, we trained state-of-the-art deep neural networks on SEM images generated using digital twins to achieve accurate results on real SEM images. Furthermore, we validated the results of trained deep learning models through model visualization and accuracy-metric evaluation.
This paper introduces a simple physical model to quantitatively explain resist surface charge effect observed in EBM- 9500PLUS, our latest VSB mask writer designed for 7 nm+ generation. The model takes into account secondary electrons drawn to resist surface by an already-existing surface charge, and vertical diffusion of positive charge from resist surface to inner resist. In order to verify the model, we experimentally evaluated the surface charge densities after beam exposure on resists of different thickness (from 80 nm to 300 nm) and different dose sensitivities (from 7 μC/cm2 to 100 μC/cm2). The introduced model successfully reproduced the exposure-dose-dependent and time-dependent behaviors of those surface charge densities experimentally obtained. The model enables us to predict the amount of surface charge, and serves as one of the barometers to select the preferable resist thickness and its dose sensitivity under the pattern density and the required IP accuracy for the given product layouts. Furthermore, although the mechanism of charging had been unclear for a decade or more, the model finally provides a quantitative physical validity of our charge effect correction (CEC) system.
Deep learning has an increasing impact on our personal and professional lives. Deep learning has the potential to transform mask, semiconductor and electronics manufacturing. This paper reviews key results from the Center for Deep Learning in Electronics Manufacturing’s (CDLe’s) first year of operation. We consider results from adapting five common types of deep learning recipes to solve key challenges in the manufacture of photomasks, printed circuit boards (PCBs), and flat panel displays (FPDs). These deep learning applications include 1) grouping similar items to automatically categorize mask rule errors; 2) using U-Net architecture to construct fast mask designs; 3) using vision-based object classification to find and classify pick-and-place (PnP) errors on PCB assembly lines; 4) using anomaly detection to improve the quality of FPDs; and 5) using digital twins to create SEM images and optimize Inverse Lithography Technology (ILT). While we compare the relative benefits of these techniques, all show the importance of data to improve the success of deep learning networks and of electronics manufacturing. These applications rely on varying neural network architectures such as autoencoders, segmentation networks, deep convolutional networks, anomaly detection, and generative adversarial networks (GANs).
Over the last two decades, eBeam mask writers have added inline correction features. Particularly when minimum feature sizes on mask went below 100nm a decade ago, the need for more precision within a reasonable write time increased the demand for more corrections. Inline correction is better for turnaround time and throughput, but inline correction is computationally limited because it is unacceptable for computation to limit the machine write time.
Simultaneously, the same need for linearity correction, printability enhancement, and resilience to manufacturing variation has caused much innovation in offline mask data preparation and mask process correction. Typically, the writer performs inline correction for backscatter, fogging, loading, charging and thermal effects, but leaves <10μm effects to offline correction.
With multi-beam writers, the write time is independent of shape count. Any set of input shapes is rasterized to a set of arrays of equal sized pixels that are each independently dosed to write the desired shapes. Multi-beam writers also have a certain minimum write time that is required for writing even a very small number of simple shapes. This gives rise to the possibility of providing linearity correction features, even for the short-range effects as inline correction in the writer. Such inline correction has zero impact on throughput and turnaround time of mask making.
This paper introduces the GPU-accelerated inline linearity correction capability of the NuFlare MBM-1000 for the first time.
Mask writers need to be able to write sub-50nm features accurately. Nano-imprint lithography (NIL) masters need to create sub-20nm line and space (L:S) patterns reliably. Increasingly slower resists are deployed, but mask write times need to remain reasonable. The leading edge EBM-9500 offers 1200A/cm2 current density to shoot variable shaped beam (VSB) to write the masks.
Last year, thermal effect correction (TEC) was introduced by NuFlare in the EBM-95001. It is a GPU-accelerated inline correction for the effect that the temperature of the resist has on CD. For example, a 100nm CD may print at 102nm where that area was at a comparably high temperature at the time of the shot. Since thermal effect is a temporal effect, the simulated temperature of the surface of the mask is dynamically updated for the effect of each shot in order to accurately predict the cumulative effect that is the temperature at the location of the shot at the time of the shot and therefore its impact on CD. The shot dose is changed to reverse the effects of the temperature change.
This paper for the first time reveals an enhancement to this thermal model and a simulator for it. It turns out that the temperature at the time each location receives backscatter from other shots also make a difference to the CD. The effect is secondary, but still measurable for some resists and substrates. Results of a test-chip study will be presented.
The computation required for the backscatter effect is substantial. It has been demonstrated that this calculation can be performed fast enough to be inline with the EBM-9500 with a reasonable-sized computing platform. Run-time results and the computing architecture will be presented.
We propose a new concept of tuning a point-spread function (a “kernel” function) in the modeling of electron beam
lithography using the machine learning scheme. Normally in the work of artificial intelligence, the researchers focus on the
output results from a neural network, such as success ratio in image recognition or improved production yield, etc. In this
work, we put more focus on the weights connecting the nodes in a convolutional neural network, which are naturally the
fractions of a point-spread function, and take out those weighted fractions after learning to be utilized as a tuned kernel.
Proof-of-concept of the kernel tuning has been demonstrated using the examples of proximity effect correction with 2-layer
network, and charging effect correction with 3-layer network. This type of new tuning method can be beneficial to give
researchers more insights to come up with a better model, yet it might be too early to be deployed to production to give better
critical dimension (CD) and positional accuracy almost instantly.
Semiconductor scaling is slowing down because of difficulties of device manufacturing below logic 7nm
node generation. Various lithography candidates which include ArF immersion with resolution enhancement
technology (like Inversed Lithography technology), Extreme Ultra Violet lithography and Nano Imprint
lithography are being developed to address the situation. In such advanced lithography, shot counts of mask
patterns are estimated to increase explosively in critical layers, and then it is hoped that multi beam mask
writer (MBMW) is released to handle them within realistic write time. However, ArF immersion technology
with multiple patterning will continue to be a mainstream lithography solution for most of the layers. Then,
the shot counts in less critical layers are estimated to be stable because of the limitation of resolution in ArF
immersion technology. Therefore, single beam mask writer (SBMW) can play an important role for mask
production still, relative to MBMW. Also the demand of SBMW seems actually strong for the logic 7nm
node. To realize this, we have developed a new SBMW, EBM-9500 for mask fabrication in this generation. A
newly introduced electron beam source enables higher current density of 1200A/cm2. Heating effect
correction function has also been newly introduced to satisfy the requirements for both pattern accuracy and
throughput. In this paper, we will report the configuration and performance of EBM-9500.
The specifications for critical dimension (CD) accuracy and line edge roughness are getting tighter to promote every photomask manufacturer to choose electron beam resists of lower sensitivity. When the resist is exposed by too many electrons, it is excessively heated up to have higher sensitivity at a higher temperature, which results in degraded CD uniformity. This effect is called “resist heating effect” and is now the most critical error source in CD control on a variable shaped beam (VSB) mask writer. We have developed an on-tool, real-time correction system for the resist heating effect. The system is composed of correction software based on a simple thermal diffusion model and computational hardware equipped with more than 100 graphical processing unit chips. We have demonstrated that the designed correction accuracy was obtained and the runtime of correction was sufficiently shorter than the writing time. The system is ready to be deployed for our VSB mask writers to retain the writing time as short as possible for lower sensitivity resists by removing the need for increased pass count.
Resist heating effect which is caused in electron beam lithography by rise in substrate temperature of a few tens or hundreds of degrees changes resist sensitivity and leads to degradation of local critical dimension uniformity (LCDU). Increasing writing pass count and reducing dose per pass is one way to avoid the resist heating effect, but it worsens writing throughput. As an alternative way, NuFlare Technology is developing a heating effect correction system which corrects CD deviation induced by resist heating effect and mitigates LCDU degradation even in high dose per pass conditions. Our developing correction model is based on a dose modulation method. Therefore, a kind of conversion equation to modify the dose corresponding to CD change by temperature rise is necessary. For this purpose, a CD variation model depending on local pattern density was introduced and its validity was confirmed by experiments and temperature simulations. And then the dose modulation rate which is a parameter to be used in the heating effect correction system was defined as ideally irrelevant to the local pattern density, and the actual values were also determined with the experimental results for several resist types. The accuracy of the heating effect correction was also discussed. Even when deviations depending on the pattern density slightly remains in the dose modulation rates (i.e., not ideal in actual), the estimated residual errors in the correction are sufficiently small and acceptable for practical 2 pass writing with the constant dose modulation rates. In these results, it is demonstrated that the CD variation model is effective for the heating effect correction system.
In the half pitch (hp) 16nm generation, the shot count on a mask is expected to become bipolar. The multi-patterning
technology in lithography seems to maintain the shot count around 300G shots instead of increase in the number of
masks needed for one layer. However, as a result of mask multiplication, the better positional accuracy would be
required especially in Mask-to-Mask overlay. On the other hand, in complex OPC, the shot count on a mask is expected
to exceed 1T shots.
In addition, regardless of the shot count forecast, the resist sensitivity needs to be lower to reduce the shot noise effect so
as to get better LER. In other words, slow resist would appear on main stream, in near future. Hence, such trend would
result in longer write time than that of the previous generations. At the same time, most mask makers request masks to
be written within 24 hours. Thus, a faster mask writer with better writing accuracy than those of previous generations is
With this background, a new electron beam mask writing system, EBM- 9000, has been developed to satisfy such
requirements of the hp 16nm generation. The development of EBM-9000 has focused on improving throughput for
larger shot counts and improving the writing accuracy.
We quantitatively evaluate Nuflare’s latest resist charging effect correction (CEC) model for advanced photomask
production using e-beam lithography. Functionality of this CEC model includes the simulation of static and timedependent
charging effects together with an improved calibration method. CEC model calibration is performed by
polynomial fitting of image placement distortions induced by various beam scattering effects on a special test design
with writing density variations. CEC model parameters can be fine tuned for different photomask blank materials
facilitating resist charging compensation maps for different product layers. Application of this CEC model into
production yields a significant reduction in photomask image placement (IP), as well as improving photomask overlay
between critical neighbouring layers. The correlations between IP improvement facilitated by this CEC model and single
mask parameters are presented and discussed. The layer design specifics, resist and blank materials, coupled with their
required exposure parameters are observed to be the major influences on CEC model performance.
In our previous work, we reported the static portion of the surface charging on EBM-8000 and compared it with that on EBM-6000. The scope of this paper is to report the analysis of charging decay component on EBM-8000 and compare it with EBM-6000. We confirmed that our fundamental modeling scheme of the charging decay worked well on EBM-8000 as well as on EBM-6000. However, we found totally different charging decay behaviors between EBM-8000 and EBM-6000. To explain the results, we propose a conceptual model of the charging decay phenomena both on EBM-8000 and EBM-6000.
EBM-9000 equipped with new features such as new electron optics, high current density (800A/cm2) and high speed deflection control has been developed for the 11nm technology node(tn) (half pitch (hp) 16nm). Also in parallel of aggressive introduction of new technologies, EBM-9000 inherits the 50kV variable shaped electron beam / vector scan architecture, continuous stage motion and VSB-12 data format handling from the preceding EBM series to maintain high reliability accepted by many customers. This paper will report our technical challenges and results obtained through the development.
This paper gives a review and summary of the key messages from the 2013 PMJ panel discussion topic: “Future
mask patterning technologies in the next decade: searching for the best mix solution." The results of the special
survey conducted for the panel are reviewed along with a brief summary of each panelist’s point of view.
In this paper, we report our modeling results of the resist surface charging effect on our newer e-beam mask writer
EBM-8000. We show that our fundamental modeling scheme we have developed for EBM-6000 can be adapted on
EBM-8000 platform without major modifications. We also discuss the significant differences in the charging effect between
EBM-6000 and EBM-8000 in terms of its amplitude, its spatial distribution, and its dependency on the pattern density.
To enhance global CDU attained by our EB mask writer EBM-8000, we examined extending the loading effect correction (LEC) function to treat plural of loading effects, for instance, develop and etch loading. Here, we propose a LEC dose composition method, assuming uniquely-defined relation between amount of dose modulation and resultant CD change. Sets of LEC dose maps (pairs of base dose maps and proximity backscattering ratio maps) are converted to sets of CD change maps which are summarized to create a set of dose maps used for writing. This paper describes the correction procedure and possible applications of the method.
Improvement of pattern placement accuracy is essential to solve upcoming challenges in mask making. Placement
errors are driven by multiple effects with electron mediated resist surface charging being a major error source. Modeling
this systematic effect thus allows the determination of the placement errors before plate processing. This opens the door
to an effective charging compensation.
In this paper we study the simulated benefit of two distinct charging compensation models in the context of full-scale
mask production layouts. The potential pattern placement improvements are evaluated using actual placement results
obtained without charging effect corrections. An in depth comparison of the two models is presented, demonstrating the
differences in placement error prediction between using a static or a dynamic charging model. We find that substantial
improvements can be achieved using the dynamic charging model. Productive implementation of this functionality is
the natural next step.
We report our development of fogging effect correction method aimed for EBM-8000, our newest series of EB mask
writers for mask production of 22nm half-pitch generation and for mask development of 16nm half-pitch generation. We
refined the method of fogging effect correction by taking account of dose modulation for proximity effects correction
and loading effect correction into fogging effect correction, greatly reducing theoretical error. Writing experiment has
shown that our method based on the threshold dose model is effective, though deviation from the model is observed.
Many lithography candidates, such as ArF immersion lithography with double-patterning/double-exposure techniques,
EUV lithography and nano-imprint lithography, show promising capability for 22-nm half-pitch generation lithography.
ArF immersion lithography with double-patterning/double-exposure techniques remains the leading choice as other
techniques still lack the conclusive evidence as the practical solution for actual production. Each of the prospective
lithography techniques at 22-nm half-pitch generation requires masks with improved accuracy and increased complexity.
We have developed a new electron beam mask writer, EBM-8000, as the tool for mask production of 22-nm half-pitch
generation and for mask development of 16nm half-pitch generation, which is necessary for the practical application of
these promising lithography technologies.
The development of EBM-8000 was focused on increasing throughput and improving beam positioning accuracy. Three
new major features of the tool are: new electron gun with higher brightness to achieve current density of 400 A/cm2,
high speed DAC amplifier to accurately position the beam with shorter settling time, and additional temperature control
to reduce the beam drift.
The improved image placement accuracy and repeatability, and higher throughput of EBM-8000 have been confirmed
by actual writing tests with our in-house tool.
A new method to describe the resist surface charging effect more accurately is proposed. In our previous work, we handled
only the static portion of the surface charging and it was applicable only to a limited situation. The scope of this paper is to
add a new model to handle the dynamic, discharging behavior on top of the existing static model to make the whole charging
model closer to what is really happening on the plate during the exposure. With the new model, the correction accuracy has
been improved not only for the equilibrium state but also for the state when the tool is dynamically writing the main pattern.
We conclude that our Charging Effect Correction (CEC) was advanced by this new model to become completely production
Optical lithography is facing resolution limit. To overcome this issue, highly complicated patterns with high data volume
are being adopted for optical mask fabrications. With this background, new electron beam mask writing system, EBM-
7000 is developed to satisfy requirements of hp 32nm generation. Electron optical system with low aberrations is
developed to resolve finer patterns like 30nm L/S. In addition, high current density of 200 A/cm2 is realized to avoid
writing time increase. In data path, distributed processing system is newly built to handle large amounts of data
efficiently. The data processing speed of 500MB/s, fast enough to process all the necessary data within exposure time in
parallel for hp32nm generation, is achieved. And this also makes it possible to handle such large volume dense data as
2G shots/mm2 local pattern density.
In this paper, system configuration of EBM-7000 with accuracy data obtained are presented.
The impending need of double patterning/double exposure techniques is accelerating the demand for higher pattern
placement accuracy to be achieved in the upcoming lithography generations. One of the biggest error sources of pattern
placement accuracy on an EB mask writer is the resist charging effect. In this paper, we provide a model to describe the resist
charging behavior on a photomask written on our EBM-6000 system. We found this model was very effective in correcting
and reducing the beam position error induced by the charging effect.
In order to comply with the demanding technology requirements for 45 nm half pitch (HP) node (32 nm technology
node), Nuflare Technology Inc. (NFT) has developed Electron-beam mask writing equipment, EBM-6000, with
increased current density (70A/cm2), while its other primary features basically remain unchanged, namely 50 kV
acceleration voltage, Variable Shaped Beam (VSB)/vector scan, like its predecessors [1-5]. In addition, new
functionalities and capabilities such as astigmatism correction in subfield, optimized variable stage speed control,
electron gun with multiple cathodes (Turret electron gun), and optimized data handling system have been
employed to improve writing accuracy, throughput, and up-time. VSB-12 is the standard input data format for
EBM-6000, and as optional features to be selected by users, direct input function for VSB-11 and CREF-flatpoly
are offered as well.
In this paper, the new features and capabilities of EBM-6000 together with supporting technologies are reported to
solidly prove the viability of EBM-6000 for 45 nm HP node.
EBM-5000 equipped with the new feature of high current density (50A/cm2) has been developed for 45 nm technology node (half pitch (hp) 65 nm). EBM-5000 adopts 50 kV variable shaped electron beam (VSB)/vector scan architecture and continuous motion stage, following the steps of preceding EBM series. In addition to the high current density, new technologies such as high resolution electron optics, finer increment for beam position and exposure time control, and new data format "VSB-12" to handle large data volume have been introduced on EBM-5000. These new technologies address two conflicting issues: improvement of throughput and better accuracy. This paper will report the key challenging technologies, certain results of EBM-5000 operation and findings obtained through our development efforts that can be applied to future generation tools. The fundamental local CD uniformity (LCDU) limit is also discussed.
Optical lithography will be extended down to 65nm to 50 nm. However, a mask with high accurate CD uniformity and resolution enhancement technology (RET) such as optical proximity effect correction (OPC) and phase shifting mask (PSM) are required to achieve resolution by exposure wave length. The mask technology is the key of the optical
lithography extension. We developed the electron beam mask writer EBM-3000 for 180-150nm design rule 1), 2), and EBM-3500 for 150-130nm design rule 3), to achieve high accuracy CD uniformity mask and small OPC pattern writing. They were variable shaped electron beam mask writing system with continuous moving stage, at 50kV acceleration
voltage, and had the functions of multi-pass field shift writing, real-time proximity effect correction, grid matching correction, and automatic adjustment for election optical column.The LSI road map calls for such small minimum feature size as that so close to optical resolution limitation where increasingly complex optical proximity corrections (OPC) as well as extremely good mask CD uniformity are required. What is making the challenge even more difficult is that writing time is exponentially increasing as the shot number is exploding to primarily cope with the complex and voluminous OPC and extremely good CD uniformity requirements. Thus the newly developed electron beam mask lithography system EBM-4000 is designed to overcome all these difficult problems associated with 100nm as well as 70nm node masks. In order to increase throughput, triangle/rectangle beam optical column, high current density/high resolution lens, and high speed DAC amplifiers have been developed. To achieve accurate CD uniformity, foggy electron correction/loading effect correction functions are developed.
The electron beam (EB) writing system with high acceleration voltage must be used for the mask fabrication because of its fine resolution. In this case, the resist heating effect becomes one of the serious problems in CD control. This paper discusses the controllability of the resist heating effect and shows that; (1) The CD variation caused by the effect increases with higher pattern coverage and larger shot size, which supports qualitatively results of temperature simulation based on Ralf's model. (2) The multiple exposure is effective to suppress the temperature rise in a substrate and the CD variation. The shifting-type exposure is more effective than the non-shifting-type exposure for suppression of the effect. (4) The CD variation for ZEP7000 can be suppressed to less than 5.0 [nm] (range) provided the shot size is less than or equal to 1.0 [micrometer] and the shifting-type exposure is adopted. Thus, the resist heating effect can be controlled and the CD variation by the effect can be suppressed enough for fabricating the masks to produce 0.15 micrometer devices and beyond.
Toshiba and Toshiba Machine have developed an advanced electron beam writing system EX-11 for next-generation mask fabrication. EX-11 is a 50 kV variable-shaped beam lithography system for manufacturing 4x masks for 0.15 - 0.18 micrometer technology generation. Many breakthroughs were studied and applied to EX-11 to meet future mask-fabrication requirements, such as critical dimension and positioning accuracy. We have verified the accuracy required for 0.15 - 0.18 micrometer generation.
Background exposure of a resist caused by scattered electrons (the fogging effect) degrades critical dimension accuracy when the pattern density changes over the specimen. We measured the fogging effect in an electron beam optical column. In order to reduce the fogging effect, a scattered electron absorber plate having a converging holes structure was attached to the lower surface of the objective lens. When the most severe pattern for the fogging effect was applied, we achieved the size variation caused by the fogging effect less than 8 nm. The converging holes effectively trap the scattered electrons and greatly reduce the fogging effect.
CD uniformity to be patterned by electron-beam (EB) writing system with a variable-shaped beam was evaluated. The experimental EB writing system, EX-8D, was used under conditions of current density of 20 A/cm2 and acceleration voltage of 50 keV. Quartz reticles coated with positive tone resist ZEP7000TM (Nippon Zeon Co., Ltd.) were applied. Test patterns of 1-micrometer-width design were written by shaped beam shots of 1 micrometer square with different exposure doses. Since higher measurement repeatability was confirmed, line width of test patterns without shot stitching points was measured by Nikon XY-3I with a circle-spot probe of 1 micrometer. Line width of clear patterns on resist film was measured after development, and line width of clear patterns on chrome (Cr) film of one mask was measured at same points after wet-etching. The other mask was measured at the same points after dry-etching process by conventional reactive ion etching (RIE). Certain comparisons in this study indicate the importance of evaluating CD uniformity on Cr film after dry- etching process. Expect for resist heating contribution by four-pass writing method, the uncertainty of CD error was quantified as follows: 4 nm (3(sigma) ) on resist film at the applied dose of 19 (mu) C/cm2, and 4 nm (3(sigma) ) on Cr film at the applied dose of 27 (mu) C/cm2.
The critical dimension uniformity required in the fabrication of photomasks for 1 gigabit DRAMs will be more stringent that 20 nm in terms of 3 sigma. High-voltage variable-shaped e-beam (VSB) writing is advantageous because of its high resolution, linewidth stability, and throughput performance. However, stitching errors in VSB writing have been a critical problem in the fabrication of advanced photomasks. In this paper, an improved method to calibrate the size of a VSB shot and reduce shot stitching errors is proposed. The accuracy of the calibration method depends on that of the linewidth measurement system, and shot-size calibration with an accuracy of +/- 10 nm can be achieved using existing measurement systems. The positioning accuracy of VSB shots was enhanced by a multiple pass exposure scheme. With these procedures applied to a 50 kV VSB system, the linewidth variation of a photomask in a local area such as a square region of 200 micrometers X 200 micrometers was reduced to less than 20 nm.