Evaluation method of fine pattern feature, its equipment, and method of semiconductor device fabrication

An evaluation method of fine pattern feature, comprising the steps of:

• scanning a device under test with an electron beam irradiating thereon, detecting secondary electrons emitted from the device under test or reflected electrons, and converting a two-dimensional distribution of the intensity into a gray-scaled image;
• measuring the evaluation target pattern within the image along a reference straight line in a perpendicular direction to the reference straight line at constant intervals to generate roughness data of pattern dimensions;
• Fourier transforming the roughness data;
• computing a total value by adding all the absolute values of Fourier coefficients squared for a frequency in a specific frequency region according to inputs for specifying a specific frequency region or a specific frequency region being set up in advance;
• and
• outputting at least one or more values selected from among the total value, a square root of the total value, two times the square root of the total value, three times the square root of the total value, and six times the square root of the total value as an index representing characteristics of the evaluation target pattern feature,
• wherein an upper limit or lower limit of the specific frequency region that is specified by inputs or set up in advance is an inverse of the pattern length of an article that is made.

An evaluation method of fine pattern feature, comprising the steps of:

• measuring a dimension of fine pattern along a reference straight line at constant intervals and generating roughness data of pattern dimensions;
• computing the standard deviation σ0 of a distribution of the roughness data;
• dividing the roughness data into N groups each consisting of consecutive M data;
• computing the average and the standard deviation of M data in each group;
• obtaining the average σ1 of the N standard deviations obtained in the previous step and computing the standard deviation σ2 of the N averages obtained in the previous step;
• finding a reference curve defined in advance and fitting to values of the σ0, σ1, and σ2;
• computing both a predicted value of variation in pattern dimensions within an area of a length L and a predicted value of variation among typical dimension values that are obtained for a plurality of areas of a length L from the obtained fitting curve and the set-up dimension L;
• calculating a value of an index of a short-period roughness and an index of a long-period roughness;
• and
• outputting at least one value onto a screen.