Non-uniformity Research Articles

Examination of nonideal film growth in batch atomic layer deposition for plasma-resistant coatings

Atomic layer deposition (ALD) can be used to fabricate protective coatings including moisture barrier layers for organic light emitting diodes, anticorrosion layers for photoelectrodes, and plasma-resistant coating for semiconductor manufacturing equipment, which necessitates the deposition of large and thick ALD films via batch ALD. However, batch ALD for the fabrication of large-area and thick coatings exhibits nonideal film growth, a phenomenon that cannot solely be explained by transient concentration distribution within the deposition chamber. This paper describes the application of precursor “exposure” (in the unit of Langmuir, or Pa s), defined as the integral of concentration over time, as a metric to assess the growth per cycle (GPC) distribution under nonideal ALD conditions, demonstrating that the local GPC correlates well with the cumulative precursor exposure at that site. Consequently, this measure can effectively predict the nonuniformity (NU) distribution of film thickness and facilitate the determination of optimal operating conditions that ensure maximal uniformity of exposure. Under this condition, the intrafilm NU of ALD-grown Al2O3 film (nominal thickness 300 nm) was reduced to 1.2%, and the interfilm NU is diminished to as low as 3.3%. These values represent reductions of 40% and 45%, respectively, compared to the NU levels observed under nonideal conditions (insufficient trimethylaluminum, TMA exposure downstream). The plasma etch rate of ALD-deposited films is merely 4.3 nm/min, representing a reduction of one-half compared to films deposited under nonideal conditions (9.8 nm/min) with overload TMA exposure downstream leading to chemical vapor deposition-like reactions.

Compensated DOE in a VHG-based waveguide display to improve uniformity

Augmented reality head-mounted displays (AR-HMDs) utilizing diffractive waveguides have emerged as a popular research focus. However, the illuminance uniformity over the fields of view (FOV) is often unsatisfactory in volume holographic grating (VHG) based waveguide displays. This paper proposes a high uniformity AR waveguide display system. Firstly, the angular uniformity of the VHG-based waveguide displays is analyzed. Subsequently, diffractive optical elements (DOEs) are seamlessly integrated onto the outer coupling surface of the waveguide substrate to improve the angular uniformity through phase compensation. To design the DOE phase, the multi-objective stochastic gradient descent (MO-SGD) algorithm is proposed. A single DOE is used to compensating various images form the image source. A hybrid loss, which includes the learned perceptual image patch similarity (LPIPS) metric, is applied to enhance the algorithm performance. Simulation results show that the proposed method effectively suppresses illumination degradation at the edge FOV in exit pupil images of the waveguide display system. In the results, the peak signal-to-noise ratio (PSNR) is improved by 5.54 dB. Optical experiments validate the effectiveness of the proposed method. The measured nonuniformity (NU) against FOVs is improved by 53.05% from 0.3749 to 0.1760.

A CT deep learning reconstruction algorithm: Image quality evaluation for brain protocol at decreasing dose indexes in comparison with FBP and statistical iterative reconstruction algorithms

PurposeTo characterise the impact of Precise Image (PI) deep learning reconstruction algorithm on image quality, compared to filtered back-projection (FBP) and iDose4 iterative reconstruction for brain computed tomography (CT) phantom images. MethodsCatphan-600 phantom was acquired with an Incisive CT scanner using a dedicated brain protocol, at six different dose levels (volume computed tomography dose index (CTDIvol): 7/14/29/49/56/67 mGy). Images were reconstructed using FBP, levels 2/5 of iDose4, and PI algorithm (Sharper/Sharp/Standard/Smooth/Smoother). Image quality was assessed by evaluating CT numbers, image histograms, noise, image non-uniformity (NU), noise power spectrum, target transfer function, and detectability index. ResultsThe five PI levels did not significantly affect the mean CT number. For a given CTDIvol using Sharper-to-Smoother levels, the spatial resolution for all the investigated materials and the detectability index increased while the noise magnitude decreased, slightly affecting noise texture. For a fixed PI level increasing the CTDIvol the detectability index increased, the noise magnitude decreased. From 29 mGy, NU values converged within 1 Hounsfield Unit from each other without a substantial improvement at higher CTDIvol values. ConclusionsThe improved performances of intermediate PI levels in brain protocols compared to conventional algorithms seem to suggest a potential reduction of CTDIvol.

The effects of distance between the imaging isocenter and brain center on the image quality of cone-beam computed tomography for brain stereotactic irradiation.

In linear accelerator-based stereotactic irradiation (STI) for brain metastasis, cone-beam computed tomography (CBCT) image quality is essential for ensuring precise patient setup and tumor localization. However, CBCT images may be degraded by the deviation of the CBCT isocenter from the brain center. This study aims to investigate the effects of the distance from the brain center to the CBCT isocenter (DBI) on the image quality in STI. An anthropomorphic phantom was scanned with varying DBI in right, anterior, superior, and inferior directions. Thirty patients undergoing STI were prospectively recruited. Objective metrics, utilizing regions of interest included contrast-to-noise ratio (CNR) at the centrum semiovale, lateral ventricle, and basal ganglia levels, gray and white matter noise at the basal ganglia level, artifact index (AI), and nonuniformity (NU). Two radiation oncologists assessed subjective metrics. In this phantom study, objective measures indicated a degradation in image quality for non-zero DBI. In this patient study, there were significant correlations between the CNR at the centrum semiovale and lateral ventricle levels (rs = - 0.79 and - 0.77, respectively), gray matter noise (rs = 0.52), AI (rs = 0.72), and NU (rs = 0.91) and DBI. However, no significant correlations were observed between the CNR at the basal ganglia level, white matter noise, and subjective metrics and DBI (rs < ± 0.3). Our results demonstrate the effects of DBI on contrast, noise, artifacts in the posterior fossa, and uniformity of CBCT images in STI. Aligning the CBCT isocenter with the brain center can aid in improving image quality.

Wind tunnel investigation of blade pitch effect for performance improvement of an asymmetric bladed H-Darrieus VAWT under low wind speed condition

In this paper experimental investigation is conducted to analyse the blade pitch effect on the performance improvement (self-starting and dynamic performance) of a NACA 63415 asymmetric bladed H-Darrieus VAWT at low wind speed conditions (5.0 and 6.0 m/s). It has been found that the performance of the considered VAWT has improved with a change of blade pitch angle (β) from negative to positive value. A maximum power coefficient (Cp) of 0.086 is obtained at a tip speed ratio (TSR) of 1.73 at a wind speed of 6.0 m/s for β = + 5°. Further turbulence intensity (TI) and flow non-uniformity (NU) are also calculated. In this case, TI of 2.6% and 2.4% is calculated from experimentation along with 4% and 3.33% non-uniformity (NU) for wind speeds 5.0 and 6.0 m/s, respectively, which are responsible for Cp improvement.

Image quality and lesion detectability of deep learning-accelerated T2-weighted Dixon imaging of the cervical spine.

To validate the subjective image quality and lesion detectability of deep learning-accelerated Dixon (DL-Dixon) imaging of the cervical spine compared with routine Dixon imaging. A total of 50 patients underwent sagittal routine Dixon and DL-Dixon imaging of the cervical spine. Acquisition parameters were compared and non-uniformity (NU) values were calculated. Two radiologists independently assessed the two imaging methods for subjective image quality and lesion detectability. Interreader and intermethod agreements were estimated with weighted kappa values. Compared with the routine Dixon imaging, the DL-Dixon imaging reduced the acquisition time by 23.76%. The NU value is slightly higher in DL-Dixon imaging (p value: 0.015). DL-Dixon imaging showed superior visibility of all four anatomical structures (spinal cord, disc margin, dorsal root ganglion, and facet joint) for both readers (p value: < 0.001 ~ 0.002). The motion artifact scores were slightly higher in the DL-Dixon images than in routine Dixon images (p value = 0.785). Intermethod agreements were almost perfect for disc herniation, facet osteoarthritis, uncovertebral arthritis, central canal stenosis (κ range: 0.830 ~ 0.980, all p values < 0.001) and substantial to almost perfect for foraminal stenosis (κ = 0.955, 0.705 for each reader). There was an improvement in the interreader agreement of foraminal stenosis by DL-Dixon images, from moderate to substantial agreement. The DLR sequence can substantially decrease the acquisition time of the Dixon sequence with subjective image quality at least as good as the conventional sequence. And no significant differences in lesion detectability were observed between the two sequence types.

Simulation and Experimental Investigation of the Radial Groove Effect on Slurry Flow in Oxide Chemical Mechanical Polishing

Slurry flow on the pad surface and its effects on oxide chemical mechanical polishing (CMP) performance were investigated in simulations and experiments. A concentric groove pad and the same pad with radial grooves were used to quantitatively compare the slurry saturation time (SST), material removal rate (MRR), and non-uniformity (NU) in polishing. The monitored coefficient of friction (COF) and its slope were analyzed and used to determine SSTs of 25.52 s for the concentric groove pad and 16.06 s for a certain radial groove pad. These values were well correlated with the simulation prediction, with around 5% error. Both the laminar flow and turbulent flow were included in the sliding mesh model. The back mixing effect, which delays fresh slurry supply, was found in the pressure distribution of the wafer–pad interface.

Detection and quantification of pore, solid and gravel spaces in CT images of a 3D soil sample

In this study, we scanned the core of a cylindrical soil sample (60mm diameter and 100mm height) by X-ray Computed Tomography (CT) producing 300 consecutive 2D digital images with 16-bit gray level depth and a resolution of 32 microns (image size 676 × 676 pixels). The aim of this work was to determine the geometry and spatial distribution of the elements in a sample, related in this case to pore, solid and gravel, inside each 2D image for the latter reconstruction of the corresponding 3D approximation of the elements using the total set of 300 soil images. Therefore, it was possible to determine the relative percentage of each element present in each 2D image and, correspondingly, the structure and total percentage in the 3D reconstruction. The identification of elements in the 2D image slices was very well accomplished using three standard segmentation algorithms: k-Means, Fuzzy c-Means and Otsu multilevel. In order to compare and evaluate the quality of results, a non-uniformity (NU) measure was applied such that low values were indicative of homogeneous regions. Due to the depth of the greyscale of the images, the results were very similar with comparable statistics and homogeneity (NU values) among the detected materials of the three algorithms. That suggests that the pore, solid and gravel spaces were very well identified, and this is reflected through their connectivity in the 3D reconstruction. Additionally, the gray level depth was reduced to 8 bits and the same study was undertaken. In this case, the quality of results was comparable to the previous ones, as the number of elements and NU values were very close. However, this also depends largely on the high resolution of the images. Thereby, the soil sample of this work was very well characterized using the simplest and most common algorithms for image segmentation thanks to the high contrast and resolution, and regardless the depth of the grey-level.

Performance Assessment of Low-Cost Thermal Cameras for Medical Applications.

Thermal imaging is a promising technology in the medical field. Recent developments in low-cost infrared (IR) sensors, compatible with smartphones, provide competitive advantages for home-monitoring applications. However, these sensors present reduced capabilities compared to more expensive high-end devices. In this work, the characterization of thermal cameras is described and carried out. This characterization includes non-uniformity (NU) effects and correction as well as the thermal cameras’ dependence on room temperature, noise-equivalent temperature difference (NETD), and response curve stability with temperature. Results show that low-cost thermal cameras offer good performance, especially when used in temperature-controlled environments, providing evidence of the suitability of such sensors for medical applications, particularly in the assessment of diabetic foot ulcers on which we focused this study.

Selection Criteria for Determination of Optimal Reconstruction Method for Cu-64 Trastuzumab Dosimetry on Siemens Inveon PET Scanner.

The goal of this study was to suggest criteria for the determination of the optimal image reconstruction algorithm for image-based dosimetry of Cu-64 trastuzumab PET in a mouse model. Image qualities, such as recovery coefficient (RC), spill-over ratio (SOR), and non-uniformity (NU), were measured according to National Electrical Manufacturers Association (NEMA) NU4-2008. Mice bearing a subcutaneous tumor (, HER2 NCI N87) were injected with monoclonal antibodies (trastuzumab) with Cu-64. Preclinical mouse PET images were acquired at 4 time points after injection (2, 15, 40 and 64 h). Phantom and Cu-64 trastuzumab PET images were reconstructed using various reconstruction algorithms (filtered back projection (FBP), 3D reprojection algorithm (FBP-3DRP), 2D ordered subset expectation maximization (OSEM 2D), and OSEM 3D maximum a posteriori (OSEM3D-MAP)) and filters. The absorbed dose for the tumor and the effective dose for organs for Cu-64 trastuzumab PET were calculated using the OLINDA/EXM program with various reconstruction algorithms. Absorbed dose for the tumor ranged from 923 mGy/MBq to 1830 mGy/MBq with application of reconstruction algorithms and filters. When OSEM2D was used, the effective osteogenic dose increased from 0.0031 to 0.0245 with an increase in the iteration number (1 to 10). In the region of kidney, the effective dose increased from 0.1870 to 1.4100 when OSEM2D was used with iteration number 1 to 10. To determine the optimal reconstruction algorithms and filters, a correlation between RC and NU was plotted and selection criteria (0.9 < RC < 1.0 and < 10% of NU) were suggested. According to the selection criteria, OSEM2D (iteration 1) was chosen for the optimal reconstruction algorithm. OSEM2D (iteration 10) provided 154.7% overestimated effective dose and FBP with a Butterworth filter provided 20.9% underestimated effective dose. We suggested OSEM2D (iteration 1) for the calculation of the effective dose of Cu-64 trastuzumab on an Inveon PET scanner.

Numerical Simulation of the Relative Sliding Distance of a Wafer on Groove-Patterned Pads in Chemical Mechanical Planarization System

This paper presents the numerical calculation of the relative sliding distance between the pad and the wafer in a rotary-type chemical-mechanical planarization (CMP) system. The numerical scheme developed in this paper considered most of the possible kinematical and geometrical variables. While the pad was rotating, and the wafer was rotating and sweeping, the point trajectories on the wafer were calculated by standard coordinate transformations. The effect of groove patterns on the pad surface was investigated for the relative sliding distance of the wafer. Three types of pad surface geometries, namely flat, circular-type, and rectangular-type grooves, were considered. From the simulation results, we evaluated the nonuniformity (NU) of the relative sliding distance distribution on the polished wafer surface. It was found that the groove patterns have a significant effect on the result of the CMP, as well as the width and pitch of the groove. It was found that the groove patterns increase the NU of the sliding distance distribution. The circular-type groove caused the greater NU than the rectangular-type groove under the identical operating conditions. The optimized operating conditions were varied for different types of the groove patterns. The simulation method proposed in this paper may be further extended to analyze other process variables.

Pore detection in 3‐D CT soil samples through an improved sub‐segmentation method

SummaryX‐ray computer tomography (CT) is a non‐invasive technique for image acquisition. Recent technological advances have enabled reliable and high‐resolution images to be obtained. In soil samples, for example, this subserves the identification of pores and their structure and the analysis of their geometric characteristics. However, the lack of contrast between pores and solids in soil samples makes it difficult to identify the pores, and it poses problems for their connectivity when a three‐dimensional (3‐D) reconstruction is made from a group of consecutive 2‐D images obtained with a scanner. To solve this problem, an improved sub‐segmentation method, which had been developed and tested previously, was applied in this research to achieve a better identification of the pore space and consequently the solid space in the 2‐D slices of the image, followed by a 3‐D reconstruction of the soil sample. In this study, two soil samples were used, one real soil sample with 255 2‐D CT consecutive images and a synthetic image with 215 2‐D images. The latter sample was used only to evaluate the robustness of the improved sub‐segmentation method and the results from analysis of the pore connectivity in a known structure. The results obtained with the improved sub‐segmentation were compared with those of traditional clustering algorithms for image segmentation by k‐means, fuzzy c‐means and Otsu's methods. The results were promising, and the 3‐D reconstruction presents a realistic structure for the continuity and coincidence of the shapes of the pores in the consecutive images. In addition, the pore regions detected have a small non‐uniformity (NU) value, which indicates both good pore detection and homogeneity, which facilitates pore connectivity between the different 2‐D images.HighlightsPore structure is not disturbed by the acquisition of images with computer tomography. Pore spaces were identified properly with the improved sub‐segmentation. Pore continuity by position, size and shape was observed through the 2‐D consecutive images. Realistic 3‐D reconstruction was feasible for the pore spaces in soil samples.

Intensity non-uniformity correction of aerothermal images via ℓp-regularized minimization.

Aerothermal-induced intensity non-uniformity (NU) effects severely influence the effective performance of infrared (IR) imaging systems in high-speed flight. In this paper we propose a ℓp-regularized minimization method to remove intensity NU in IR images. Different from the existing NU correction methods, we consider and study important priors from the NU noise and the IR image, respectively. We assume spatial smoothness of the NU noise and piecewise continuity of the IR image, where the ℓp regularization term is employed in the correction model. A computationally efficient numerical algorithm based on half-quadratic regularization is adopted to solve the optimization problem. To tackle the non-convex ℓp-norm minimization sub-problem in this scheme, a generalized iterated shrinkage algorithm is used. Significant improvement on the image quality is obtained on both simulation and experimental studies. Both quantitative and qualitative comparisons to specialized state-of-the-art algorithms demonstrate its superiority.

Modeling and Compensating Temperature-Dependent Non-Uniformity Noise in IR Microbolometer Cameras.

Images rendered by uncooled microbolometer-based infrared (IR) cameras are severely degraded by the spatial non-uniformity (NU) noise. The NU noise imposes a fixed-pattern over the true images, and the intensity of the pattern changes with time due to the temperature instability of such cameras. In this paper, we present a novel model and a compensation algorithm for the spatial NU noise and its temperature-dependent variations. The model separates the NU noise into two components: a constant term, which corresponds to a set of NU parameters determining the spatial structure of the noise, and a dynamic term, which scales linearly with the fluctuations of the temperature surrounding the array of microbolometers. We use a black-body radiator and samples of the temperature surrounding the IR array to offline characterize both the constant and the temperature-dependent NU noise parameters. Next, the temperature-dependent variations are estimated online using both a spatially uniform Hammerstein-Wiener estimator and a pixelwise least mean squares (LMS) estimator. We compensate for the NU noise in IR images from two long-wave IR cameras. Results show an excellent NU correction performance and a root mean square error of less than 0.25 C, when the array’s temperature varies by approximately 15 C.

Effect of Radionuclide Activity Concentration on PET-CT Image Uniformity.

Assessment of radionuclide activity concentration on positron emission tomography-computedr tomography (PET-CT) image uniformity has been carried out quantitatively. Tomographic PET-CT images of cylindrical phantom containing F-18 fluorodeoxyglucose (FDG) activity concentration was acquired and used for the assessment. Activity concentrations were varied and PET-CT images were acquired at the constant acquisition parameters of time, matrix size, and reconstruction algorithm, respectively. Using midtransaxial image slices, quantitative index of nonuniformity (NU), and coefficient of uniformity variation were estimated for the different activity concentrations. Maximum NUs of 17.6%, 26.3%, 32.7%, 36.2%, and 38.5% were estimated for activity concentrations of 16.87 kBq/mL, 14.06 kBq/mL, 11.25 kBq/mL, 8.43 kBq/mL, and 5.62 kBq/mL, respectively. The coefficient of uniformity variation established an inverse quadratic relationship with activity concentration. Activity concentrations of 16.87 kBq/mL, 14.06 kBq/mL, 11.25 kBq/mL, 8.43 kBq/mL, and 5.62 kBq/mL produced uniformity variations of 1.47%, 2.52%, 4.23%, 5.12%, and 4.98%, respectively. Increasing activity concentration resulted in decreasing coefficient of uniformity and hence, an increase in image uniformity. The uniformity estimates compared well with the standards set internationally.

The effects of a spray slurry nozzle on copper CMP for reduction in slurry consumption

The environmental impact of semiconductor manufacturing has been a big social problem, like greenhouse gas emission. Chemical mechanical planarization (CMP), a wet process which consumes chemical slurries, seriously impacts environmental sustainability and cost-effectiveness. This paper demonstrates the superiority of a full-cone spray slurry nozzle to the conventional tube-type slurry nozzle in Cu CMP. It was observed that the spray nozzle made a weak slurry wave at the retaining ring unlike a conventional nozzle, because the slurry was supplied uniformly in broader areas. Experiments were implemented with different slurry flow rates and spray nozzle heights. Spray nozzle performance is controlled by the spray angle and spray height. The process temperature was obtained with an infrared (IR) sensor and an IR thermal imaging camera to investigate the cooling effect of the spray. The results show that the spray nozzle provides a higher Material removal rate (MRR), lower non-uniformity (NU), and lower temperature than the conventional nozzle. Computational fluid dynamics techniques show that the turbulence kinetic energy and slurry velocity of the spray nozzle are much higher than those of the conventional nozzle. Finally, it can be summarized that the spray nozzle plays a significant role in slurry efficiency by theory of Minimum quantity lubrication (MQL).

Effect of filters and reconstruction algorithms on I-124 PET in Siemens Inveon PET scanner

Purpose: To assess the effects of filtering and reconstruction on Siemens I-124 PET data.Methods: A Siemens Inveon PET was used. Spatial resolution of I-124 was measured to a transverse offset of 50 mm from the center FBP, 2D ordered subset expectation maximization (OSEM2D), 3D re-projection algorithm (3DRP), and maximum a posteriori (MAP) methods were tested. Non-uniformity (NU), recovery coefficient (RC), and spillover ratio (SOR) parameterized image quality. Mini deluxe phantom data of I-124 was also assessed. Results: Volumetric resolution was 7.3 mm3 from the transverse FOV center when FBP reconstruction algorithms with ramp filter was used. MAP yielded minimal NU with β =1.5. OSEM2D yielded maximal RC. SOR was below 4% for FBP with ramp, Hamming, Hanning, or Shepp-Logan filters. Based on the mini deluxe phantom results, an FBP with Hanning or Parzen filters, or a 3DRP with Hanning filter yielded feasible I-124 PET data.Conclusions: Reconstruction algorithms and filters were compared. FBP with Hanning or Parzen filters, or 3DRP with Hanning filter yielded feasible data for quantifying I-124 PET.

Thermal crosstalk simulation and measurement of linear terahertz detector arrays

Thermal simulation of differently structured linear terahertz detector arrays (TDAs) based on lithium tantalate was performed by finite element analysis (FEA). Simulation results revealed that a relatively simple TDA structure can have good thermal insulation, i.e., low thermal crosstalk effect (TCE), between adjacent pixels, which was thus selected for the real fabrication of TDA sample. Current responsivity (Ri) of the sample for a 2.52THz source was measured to be 6.66×10−6A/W and non-uniformity (NU) of Ri was 4.1%, showing good performance of the sample. TCE test result demonstrated that small TCE existed in the sample, which was in good agreement with the simulation results.

Improving feedback-integrated scene cancellation nonuniformity correction through optimal selection of available camera motion

Scene-based nonuniformity correction (SBNUC) techniques provide a means of identifying and correcting focal plane array nonuniformity (NU) through algorithmic analysis of the camera output. SBNUC techniques rely almost universally on camera motion to provide a means of separating the scene from the NU pattern. A simulation is developed that is used to explore the role and effect of camera motion on two representative registration-based SBNUC algorithms: interframe registration-based least mean square (IRLMS) by Zuo et al. and feedback-integrated scene-cancellation (FiSC) by Black and Tyo. The effect of camera motion velocity and direction between frames is examined. The high spatial frequency portion of NU is shown to be corrected by both IRLMS and FiSC, and this correction is relatively indifferent to nonzero camera motion parameters. The FiSC algorithm was specifically designed to incorporate the low spatial frequency component into registration-based SBNUC, but demonstrates a strong dependency on camera motion. Techniques for mitigation of camera motion parameter effects through tradespace and buffering are presented and tested. With proper mitigation and camera motion, FiSC is shown to correct most high and low spatial frequency NU with fewer than 100 framepairs repeatedly processed using techniques suitable for real-time processing implementation.

선형 Roll-CMP에서 공정변수에 관한 통계적 분석

Nowadays, most micro-patterns are manufactured during flow line production. However, a con- ventional rotary chemical mechanical polishing (CMP) system has a limited throughput for the fabrication of large and flexible electronics. To overcome this problem, we propose a novel linear roll-CMP system for the pla- narization of large-area electronics. In this paper, we present a statistical analysis on the linear roll-CMP process of copper-clad laminate (CCL) to determine the impacts of process parameters on the material removal rate (MRR) and its non-uniformity (NU). In the linear roll-CMP process, process parameters such as the slurry flow rate, roll speed, table feed rate, and down force affect the MRR and NU. To determine the polishing char- acteristics of roll-CMP, we use Taguchi's orthogonal array L16 (44) for the experimental design and F-values obtained by the analysis of variance (ANOVA). We investigate the signal-to-noise (S/N) ratio to identify the prominent control parameters. The higher is better for the MRR and lower is better for the NU were selected for obtaining optimum CMP performance characteristics. The experimental and statistical results indicate that the down force and roll speed mainly affect the MRR and the down force and table feed rate determine the NU in the linear roll-CMP process. However, over 186.3 N of down force deteriorates the NU because of the bending of substrate. Roll speed has little relationship to the NU and the table feed rate does not impact on the MRR. This study provides information on the design parameter of roll-CMP machine and process optimization. Keywordslinear roll-CMP (선형 롤-화학기계적 연마), taguchi method (다구찌법), material removal rate (연마율), non-uniformity (불균일도)