JPS59217124A - Analyzing device of polarization of light - Google Patents

Abstract

PURPOSE:To obtain a polarization-analyzing device of high accuracy by a construction wherein an angle is formed between the planes of incidence and emission of a phase plate, and a multiple reflection beam passing through an analyzer is intercepted by a stop means so as to detect only a main beam. CONSTITUTION:An incident light passes through a polarizer 10 and a sample 12 to be measured and enters a lambda/4 plate 14 whose planes of incidence and emission form an angle to each other, so as to be subjected to multiple reflection. A multiple reflection return beam 24 thus obtained separates from a main beam, passes through an analyzer 16 and is intercepted by a stop mechanism 18, and thus only the main beam is detected by a light-receiving unit 20. According to this constitution, polarization-analyzing device of high accuracy can be obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an ellipsometer that optically measures the thickness of a thin film.

A thin oxide film (2
At present, an ellipsometer is known for determining the film thickness such as 1111j (around 01). Therefore, I.C.
Manufacturers install this in a clean room with constant temperature and humidity, and manually operate it by experienced personnel to achieve an appearance accuracy of ±0. With an accuracy of lA, 5iCh I11',! The thickness is 1.f.

In contrast, many measuring instruments are attempting to develop automated measuring devices with such measurement accuracy, but so far they have not been successful. Therefore.

In order to solve these problems, we first developed a quarter-wave plate (hereinafter referred to as input/4
It is desired to lower the cost and increase the precision of the plate (referred to as a plate), and there is also a need to develop an optical system that supports this.

Traditionally, in polarization analyzers (ellipsometers) that optically measure the thickness of a thin strip, when high-precision measurement is required, an extinction-type measuring device like the one shown in Figure 1 has been used. It is being Here, 2 is a polarizer, 4 is a sample to be measured, 6 is a 74-plate plate, and 8 is an analyzer.

In the above-mentioned extinction measurement, the polarizer (prism) 2 and analyzer (prism) 8 are used to satisfy the crossed Nicol condition, and the rotation angle of the polarizer 2 and analyzer 8 is determined when the amount of transmitted light is minimized. The thickness reduction and refractive index are determined from

Therefore, as shown in FIG. 2, (extinction ratio)=(transmitted light amount)/(incident light amount) (hereinafter referred to as γ) The smaller the value, the higher the accuracy in measuring the extinction angle. The illustrated curve (a) shows the case where the extinction ratio is small, and the curve (b) shows the case where the extinction ratio is large.

The curve shown in FIG. 2 is in principle a part of a sine function, and can be approximated to a quadratic curve near the extinction angle, or the shape of the quadratic curve changes depending on the magnitude of the extinction ratio.

In order to reduce (improve) the extinction ratio of the entire system shown in FIG. It is necessary to improve the extinction ratio of each component such as the analyzer 8 and the input/fourth plate 6. Among these, polarizer 2 and analyzer 8 are ranked 4 (5X to 4.t XIO", 5X to
-', t x to') (this is
Measurement base with extremely good extinction ratio (measured values when forming orthogonal nicols with a prism).

This extinction ratio is a function of the number of light scattering defects in the single crystal that makes up the prism, that is, the aperture area and the optical path length. However, the extinction ratio can be substantially increased.

When actually using a laser beam to form orthogonal Nicols, the extinction ratio .gamma. can be set to about 1 to 3.times.10'.

Further, as shown in FIG. 3, if the amount of transmitted light at the extinction angle is IO, and the angular width that provides less than 2° is αe, then αe=2F (radians) holds true in a normal optical system. For example, when γ=IXlO"1, αe=2
．． 2 (minutes).

Incidentally, the extinction ratio of the input/quarter plate is determined by various factors. Mica has been used for a long time and has cleavage properties, so
Although the parallelism on both sides is excellent, the yield rate for obtaining accurate input/four plate characteristics is low, and good quality products without defects cannot be obtained, so digestibility is not very good.

On the other hand, advances in manufacturing technology for artificial quartz have made it possible to obtain defect-free artificial quartz, and improvements in processing technology have put into practical use a wide variety of configurations of insert plates. For example, as shown in (■) to (■) below.

■ λ/4 plate for ellipsometer A conventional input/4 plate using quartz crystal is cut out so that the normal line of the plate surface and the optical axis of the crystal are perpendicular to each other, and used so that the light beam is incident perpendicularly to the plate surface. was common. Therefore, the thickness of the zero-order mode heating plate is approximately 17 g for a He-Ne laser (input = b328A), for example.
, m, making processing extremely difficult.

In an actual ellipsometer, 3/4 plates (thickness 52.7
4gm, tertiary wire) is used.

However, it is difficult to achieve the required precision in thickness and parallelism on both sides, making it very expensive. Furthermore, since it is thin and needs to be used by adhering it to a glass substrate, only a soft coating is applied as surface anti-reflection 11Q treatment, which has the disadvantage that not only is the film easy to peel off, but the anti-reflection measures are not sufficient. Another disadvantage is that the glass 7' and disc adhesive layer do not have uniform optical properties and are prone to flare. Furthermore, since the tertiary mode is used, there are problems such as poor temperature and humidity characteristics compared to the zero-order mode.

■ Assignment for bonding/4 plates This involves placing two crystal plates with thicknesses d and d2 together so that their optical axes are perpendicular to each other, so that the thickness becomes (d2).
d+) to obtain a composite input/4 board equivalent to in fact,! A combination of a 3/4 board and a 12/4 board is used, but in normal use, the extinction ratio will not improve even if anti-reflection treatment is applied.

Currently, it is technically possible to mechanically grind the phase plate so that the parallelism on both sides is 30 seconds or less. However, when constructing one composite phase plate 4 using two phase plates finished with such precision, it is difficult to obtain parallelism with the same precision, and the yield is also low. Furthermore, even though anti-reflection coaching is applied (the limit is 0.2% anti-reflection), multiple reflection components will inevitably remain.

■ Bias cut 2/4 plate In a normal phase plate, the optical axis direction and the normal line of the bias cut 2/4 plate are orthogonal (that is, the optical axis direction is parallel to the plate surface), so the 0th mode type , the thickness of the phase plate becomes extremely thin. Therefore, if the angle between the optical axis and the plate surface normal is set to about 136, for example, a thickness of about 350 lumen is sufficient even in the zero-order mode using a He-Ne laser (input = 8328A), resulting in a highly accurate laser beam. Mechanical polishing becomes possible.

However, if the laser beam has a "spread" (usually about 1 milliradian), the incident angle will change at the periphery of the beam by the same amount, so in the beam cross section, there will be some side beams as shown in Figure 4. It only quenches. Further, even if the beam speed is made parallel by the lens diameter, the light beam reflected from the sample to be measured will not necessarily be parallel, and no improvement in the extinction ratio can be expected. However, the zero-order mode is good in single board! Therefore, if it becomes possible to obtain a material with a value of about 50 ALm due to advances in processing technology, it is considered that it will be fully suitable for a simple ellipsometer.

In view of the above points, an object of the present invention is to provide a highly accurate polarization analyzer using a combination of a phase plate and an optical system that supports the phase plate.

In order to achieve this objective, in the present invention, the incident and exit surfaces of the phase plate are not made parallel but have a predetermined angle, and the optical path is changed each time the multiple reflected beam passes through the 0'l phase plate. It is configured to be separated from the main beam, and the multiple reflected beams that have passed through the analyzer are blocked by a diaphragm means, so that only the main beam is detected and extinguished.

Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 5 shows the basic configuration of an optical system of an ellipsometer to which the present invention is applied. In addition, FIGS. 6(A) to 6(D) show the fifth
The polarization state of the transmitted beam in the optical system shown in the figure is shown.

In FIG. 5, 10 is a polarizer, 12 is a sample to be measured, and 14 is a polarizer.
16 is an analyzer, 18 is an aperture, 2o is a photoelectric conversion unit using a photodiode, etc., 22 is a main axis, 24
is the multiple reflected return beam. Further, FIG. 6(A) shows linearly polarized light, FIG. 6(B) shows elliptically polarized light, and FIG. 6(C) and (It) show linearly polarized light. However, in FIG. 6(C), the solid line arrow indicates the main beam (linearly polarized light), and the broken line arrow indicates the multiple reflected return beam (linearly polarized light). Furthermore, the dashed arrow in FIG. 6(D) indicates the multiple reflection return beam (
linearly polarized light).

In this embodiment, in order to construct an ellipsometer that improves the extinction ratio and thereby makes it possible to accurately measure the rotation angles of the polarizer and analyzer, the input /4 plate 1 is constructed as shown in Figure 5.
Both sides of 4 are not parallel, but have a certain angle α.

In addition, the unnecessary beam light that is reflected by both the front and back surfaces of the input/fourth plate 14 and returns is tilted with respect to the main beam optical path.
A light receiving section (light converting section) 20 is used which blocks this light with a diaphragm mechanism 18 so that only the main beam light can be detected. Although the structure of the transmission sample All standard type is shown in 5th A, the same applies to the reflection sample measurement type.

Physical constants of the sample 12 to be measured (for example, refraction (・heto 11
thickness) is the thickness of the polarizer 10 when the extinction condition is satisfied.
- It is removed from the rotation angle T and Δ between the analyzers 16 (see FIG. 6(c)). Here, the extinction state refers to a state in which the output signal level of the light receiving section (photoelectric conversion section) 20 shows the minimum value. At this time, the main beam white line polarized light is blocked by the analyzer (prism) 1B, but the linearly polarized light component of the multiple reflected return beam that is perpendicular to this passes through the analyzer 16 as it is, so in the conventional ellipsometer, the extinction angle is The accuracy could not be determined.

That is, since the minimum value of the received light output signal is given by these unnecessary beams, the extinction ratio deteriorates as shown in FIG. 3, making it impossible to accurately measure the extinction angle. In other words, the input/4 plate 14 is
This is because the incident beam in the elliptically polarized state is converted into two linearly polarized states perpendicular to the elliptically polarized state and then exits.

In this way, a part of the main beam is primarily reflected at the entrance/fourth plate exit surface, and a part of this is secondarily reflected at the entrance/fourth board entrance surface to become a return beam. The surface is not limited to a λ/4 plate surface. In particular, primary reflection also poses a problem at the entrance and exit surfaces of the analyzer 18. More generally speaking, among the primary and secondary reflection components that occur in the ellipso optical system, the component that passes through the λ/4 plate twice becomes a problem.

Therefore, in order to cause the optical axis to change each time the light passes through the input/quarter plate, a small angle α is provided between the incident surface and the output surface of the input/quarter plate. Then, every time passing through /4& (n-1)
The optical path is tilted with respect to the main beam axis by - (here, n is the refractive index of the material in which the phase plate is also 111), and the noari beam optical path is tilted with respect to the main beam optical path by 2(n-1)・α. . Thus, by providing a certain amount of span, light can be blocked by the aperture mechanism of the light receiver.

In the above-mentioned input /4 plate, if the face angle α is increased (that is, the parallelism is made bad), if the cross-sectional area of the transmitted beam is large, the retardation angle will be changed to a predetermined π/2 depending on the location where the beam is transmitted. As a result, as shown in FIG. 4, only the center part can be extinguished, and the extinction ratio becomes worse.

As a countermeasure for this, as shown in FIG.
Two parallel plane phase plates A and B are pasted together at an angle α,
Make up 4 boards.

Here, phase plate A is a (n11)/2 plate (retardation = (n+t)/π/2), phase plate B is an n/2 plate (retardation = nπ/2), and an adhesive layer is used. For C, an adhesive substance that does not cause birefringence and has a refractive index equal to that of the phase plate is used.

Since the optical axis of the phase plate A and the optical axis of the phase plate B described above are made to intersect, the total retardation angle R is as follows, and it is possible to obtain a zero-order mode to 74 plates. As a result, the optical path length through which the beam passes through the phase plate becomes equal, and there is no change in retardation at the location where the beam passes. In addition, optical rotation is a problem with phase plates using quartz crystal, but by configuring phase plates A and B using dextrorotating quartz and levorotary quartz, the 0th-order mode input/4 plate with small optical rotation can be used. 1) Can.

FIG. 8 shows a method for measuring parallelism on both sides of a phase plate.

That is, irradiate the laser beam as shown in this figure, and
(Two beam spores)
It is measured from a distance of As a result, if the accuracy of the parallelism of the bonded surfaces is set to about ±1 minute and the parallelism bias angle α (target i+ti) is set to 10 minutes, it is possible to produce a phase plate with a parallelism of 9 to 10 with a good yield. can.

Next, we will discuss the deterioration of the extinction ratio in the case of using two λ/4 plates bonded together with a parallelism of 10 minutes, and the problems related to the characteristics of the λ/4 plate. Here, the parallelism of the phase plates A and B, which are 4-degree elements, is considered to be very good and is not taken into consideration.

As shown in FIG. 7, when a beam is irradiated perpendicularly to the incident plane of the incident quarter plates bonded together at an angle α, the incident angle to the B plane is α. Therefore, if the thickness of the phase plate B is expressed as a term, then the optical path difference (∆ term = 1) 2 document α2 compared to normal incidence
). The retardation error ΔR for 13 people/4 boards is given by the following equation.

= 0.087 milliradian = 0.3 minute Therefore, even if there is a beam spread (about 1 milliradian), unless the optical rotation is particularly large, at a bias angle of this degree,
Uniform extinction in the beam cross section can be easily achieved.

On the other hand, the retardation error caused by the parallelism error of the phase plate becomes extremely severe because it becomes more effective in the first-order term. That is, the retardation error ΔR due to the parallelism error of component A or B is. Therefore, now Δθ = 20" = 0.1 milliradian, = 1.8 milliradian = 6 minutes. In this way, depending on the bonding angle α, III
It can be seen that the error is much smaller than the parallelism of the constituent elements. Furthermore, the 1 error due to optical rotation is similarly small.

FIG. 9 shows an example of separating the main beam and multiple reflected beams. Since the multiple reflections on the input/four plates are the most difficult to block,
To give a concrete example, when α is 10 minutes and the distance between the input/quarter plate and the light-receiving surface is 1 m, the distance between the optical paths at the light-receiving surface is 3
It becomes mm. Therefore, if a laser beam with a beam spread of ±1 milliradian is used, the spot on the light-receiving surface will be about 2 mm in diameter and can be sufficiently separated.

As explained above, according to the present invention, (1) multiple reflected beams within the ellipsometer can be blocked and only the main beam can be detected, so the extinction ratio can be improved and the extinction angle can be measured with high precision. Therefore, the reproducibility of oxide films of 20A or less is 1. It becomes possible to evaluate +-.

■ Many of the optical elements in the ellipsometer optical system are difficult to apply anti-reflection coating to (for example, analyzer/polarizer prisms, etc.), and although multiple reflections between these elements are unavoidable, it is possible to divert the optical axis. By doing so, the influence can be removed and a device with high measurement accuracy can be obtained.

■ If no measurement sample is placed in the ellipsometer optical system,
That is, even when calibrating the 4111 constant system, complete extinction can be obtained by telescopic observation. Therefore, by quantifying the change in the extinction state when a measurement sample is placed and measured, the refractive index calculated from the rotation angle of the polarizer/analyzer, the film thickness, and its low temperature 3 parameters (refractive index or The standard deviation value of film thickness) can be evaluated.

[Brief explanation of drawings]

Figure 1 is No. 1 of the extinction type ellipsome make (Motoichi configuration diagram, Figure 2 is a line diagram showing the extinction curve, Figure 3 is a line diagram showing the extinction river width, Figure 4 is the case where only the center is removed) FIG. 5 is a diagram showing an embodiment of the present invention, FIG. 6 (A)
~(D) is a diagram showing the polarization state of the transmitted light beam in FIG. 5, and FIG. 7 is a configuration diagram of a quarter-wave plate using two phase plates. Figure '08 is a diagram showing a method for measuring parallelism on both sides of a phase plate, and 9th UA is a diagram explaining separation of a main beam and multiple reflected beams. 2... Polarizer, 4... Sample to be measured, 6... Quarter wavelength plate (included/quarter plate), 8... Analyzer, 10... Polarizer, 12... Sample to be measured, 14... Quarter wavelength plate (in/4&), 16... Analyzer, 18... Diaphragm, 20... Charging converter, 22... Main axis, 24...・Multiple reflection return beam. Patent issued by ρn Toyo Tsushinki Co., Ltd. Applicant H Hon Broadcasting Corporation Figure 2 Figure 1

Claims (1)

[Claims] l) The entrance and exit surfaces of the phase plate are not made parallel but have a predetermined angle, and each time the multiple reflected beam passes through the phase My, the optical path is changed so that the main beam is 1. A polarization analyzer characterized in that the multiple reflected beams passing through the analyzer are blocked by a diaphragm means, and only the 11J main beam is detected and extinguished. 2) The polarization analyzer according to claim 1, wherein a quarter wavelength plate is used as the phase plate.