Spring 1999 Yield Management Solutions28

Inspection

Wafer Inspection Technology Challengesfor ULSI Manufacturing–Part I

by Stan Stokowski, Ph.D., Chief Scientist; Mehdi Vaez-Iravani, Ph.D., Principal Research Scientist

Evolution of the semiconductor manufacturing industry is placing ever greater demands on yield management and in par-ticular, on metrology and inspection systems. As critical dimensions shrink to 0.13 µm and 0.10 µm in the near futureand wafer sizes increase from 200 mm to 300 mm, economics will drive the industry to decrease the time for achievinghigh-yield, high-value production. Continued pressure to increase the return-on-investment for the semiconductor fabrica-tor has made it critical for inspection systems to evolve from stand-alone “tools” that just find defects to being a part of amore complete solution where detecting defects, classifying them, analyzing these results and recommending corrective actionare their functions.(Part one of two)

To understand how inspection systems willmeet the requirements of manufacturingintegrated circuits with smaller structureson larger wafers in the future, we need toconsider some of the basic physics, engineer-ing, and economic constraints imposed onthese systems.

Physics: To inspect an object we look at it viasome interrogating means, which are usuallyphotons or electrons scattered by the object.The detected scattered photons or electronsas a function of position (an image) hopeful-ly contain the information needed to deter-mine whether a defect is present. An imageprocessing system then decides if there is adefect. Thus, defect detection naturally con-sists of three main steps: first, obtaining theimage, second, processing the image, andthird, applying criteria to this processedimage to detect defects.

It is interesting to compare inspection tech-nology with that of lithography, in particu-lar, the exposure process. Lithography isalmost exclusively optical [I-line (365 nmwavelength), deep ultraviolet (DUV, 248 nm),193 nm, and eventually extreme ultraviolet(EUV, 13 nm)]. The print rate of opticallithography is now about 1010 resolution elements per second and is increasing to 1011

over the next few years. This high print rate is a conse-quence of the massive parallelism of optical techniques.On the other hand, the highest inspection rate currentlyis 6 x 108 pixels per sec. However, optical lithography has an easier task in that it does not have to process and analyze an image.

The challenge for inspection tools is then to detectsmall defects with a system resolution spot size muchlarger than the defect size. Fortunately, one does nothave to “resolve” a defect in order to detect it.Resolution, appropriately, does impact defect classifica-tion and identification. However, even for performingthese functions, we can sometimes obtain sufficientinformation without necessarily “resolving” the defect.

Even the first step of obtaining the image, by its nature,includes an optical processing step. How we illuminateand collect the resultant scattered light determines thecontrast between a defect and the background in whichit resides (surface or pattern scattering). Ideally onewants to maximize this contrast by carefully choosingthe optical arrangement. Fortunately, there are tools andtechniques for accomplishing this choice, some of whichare described here. In addition, for periodic array cellsoptical spatial filtering is effective. Finally, light polar-ization plays an extremely important role in enhancingsensitivity.

This article focuses on optical techniques for waferinspection because they are most commonly used.

F E A T U R E S

Brightfield and darkfield systemsAll optical inspection systems depend on photon scat-tering from the inspected object. Brightfield systemscollect both the scattered and reflected light throughthe same aperture to obtain an image. In addition, theobject is illuminated through the objective aperture.Basically, these systems are a high-speed microscope.Darkfield systems, on the other hand, only collect thescattered light; no part of the reflected light falls with-in the collection angle. They can have a multiplicity ofconfigurations, depending on the angle and type ofillumination, collection angles, and detector type.

Both these systems have their advantages and disadvan-tages for detecting different defect types. In general,darkfield systems are particularly useful when thedefect has some high-spatial-frequency topography,whereas brightfield is good at finding planar defects. Inmost cases darkfield systems find defects much smallerthan the system resolution or spot size; whereas, inbrightfield systems the detected defects are about thesame size as the system resolution. This fact has impor-tant implications for system throughput. Of particularimportance for system sensitivity, however, is the factthat no one optical arrangement is optimal for detect-ing all possible defect types.

Particle scatteringParticles or their effects are the source of a majority ofdefects in ICs. Thus, understanding particle scatteringhelps to design sensitive inspection tools. KLA-Tencor’sproprietary application software that calculates thepolarized scattering from a sphere into the 2π hemi-sphere above the substrate was used to calculate thescattering patterns described in this article, unless otherwise stated.

Figure 1 shows our definition of the spherical coordi-nates used in discussing scattering from particles, surfaces and defects. The polar angle is defined fromthe surface normal and the azimuthal angle counter-clockwise from the reflected beam projected onto thesurface plane. The illumination polarization definitions

Spring 1999 Yield Management Solutions 29

F E A T U R E S

are s (E perpendicular to the incidence plane) and p(E parallel to the incidence plane). The scattered fieldpolarization is s or p relative to the plane containingthe surface normal and the scattered light direction. Tounderstand some of the basic scattering rules, we startwith a polystyrene latex sphere (PSL) on silicon.Although PSL spheres are not found in IC fabs, theyare convenient for calibrating inspection systemsbecause they are spheres of known diameter.

Figure 2 shows scattering from PSL spheres as a func-tion of diameter, polarization, and angle of incidence.Of particular interest is the advantage of using p-polarized light for detecting small particles. Thetotal scattering cross section for a 60 nm PSL spherewith p-polarized illumination at 70º incident angle is86 times that with s-polarized light and 42 times thatwith normal incidence. Also note that most of the scattered light under oblique p-polarized illuminationis in the polar angular range of 20º to 70º. Thus, anoptimum system for detecting small particles usesobliquely incident p-polarized light and collects thescattered light over a large solid angle, 20º to 70º inpolar angle and almost 360º in azimuth.

The advantage of p-polarization for small particledetection is a consequence of the standing E-M wave

Surface Normal

ScatteredLight (s)

ScatteringPlane

ReflectedLight

IncidentLight (i)

IncidencePlane

Ep

Es

Ep(s)

Es(s)s0i0

sο

Figure 1. Diagram for coordinate and polarization definitions.

Figure 2. Scattered intensity patterns of PSL spheres on silicon as a

function of diameter, polarization, and incidence angle. The 488 nm

plane wave comes in from the left at 70° incidence and view is about

–90° in azimuth from the incidence plane. The last column of images

is for normal incidence, circular polarization. The numbers correspond

to the peak dif ferential cross sections and the total integrated cross

sections in µm2 divided by the cosine of the incidence angle.

where d is the spherediameter, λ is the illumi-nating wavelength, n isthe refractive index of thesphere and E is the elec-tric field at the sphere. Thus, higher refractive indexmaterials, such as semiconductors and metals, scattermore light. Figure 5 compares the total integratedscattering (TIS) for PSL, silicon, and aluminum sphereson silicon. As a consequence, if a system can detect 60nm PSL spheres on silicon, it can detect 40 nm alu-minum spheres on silicon.

Particle sizing is always of interest. Typically theindustry uses PSL spheres as a calibration standard. Ifan inspection system uses the total scattered lightintensity as an indication of particle size, figure 4reveals a problem: the intensity is not a monotonicfunction of the sphere diameter (oblique incidence hasless of a problem than normal incidence). Furthermore,the scattered intensity from spheres of other materialsobviously does not relate in a simple fashion to the PSLsphere response unless one compares the curves forsphere diameters less than 100 nm. To obtain bettersizing one needs to use more than one configuration ormode as suggested, for example, by the responsesshown in figure 4.

Surface scatteringFor unpatterned wafers the background noise comesfrom surface scattering. We will only describe here thekey parameters that determine surface scattering.

For surfaces that are rough, but with height variationsmuch less than the light wavelength, the scatteredpower per unit solid angle as a function of the polarangle and azimuth is:

where Pi is the input power, dΩ is the differential solidangle, θi, θs, φs are defined in figure 1, Qij(θi, θs, φs ) isthe polarization factor and PSD(fx, fy) is the powerspectral density of the surface height variation as a

fields above a surface. We can best describe this effectby realizing that for a small enough particle, the parti-cle acts as a probe of the near field because the far-fieldscattering depends on the E-M field present at the particle. If the particle is small enough, it does notsubstantially perturb the field that would be present inthe absence of the particle.

For 70º incidence figure 3 shows the electric field as afunction of distance above a silicon surface for s- and p-polarization. Note that s-polarized light has a low field

at the surface (“dark fringe”), whereas, p-polarizedlight is at a maximum. It follows then that small particle scattering is greatest for p-polarization.

Experimental results confirm the utility of the scatter-ing model. For example, figure 4 shows the agreementbetween measurements and modeling results as a func-tion of incidence angle and polarization.

The sensitivity of an inspection system for small parti-cle detection depends on the particle material. In theRayleigh limit the total integrated scattering of asphere in a medium depends on

Spring 1999 Yield Management Solutions30

F E A T U R E S

Figure 3. Magnitude of the electric field as a function of distance

above a silicon surface for s-polarization (solid line) and p-polarization

(dashed line) for 70° incidence and an input field amplitude of 1.

Figure 4. Scattering model calculations agree with measurements

for PSL spheres on silicon with 70° incidence angle, s-polarization

(squares) and p-polarization (triangles), and normal incidence

(diamonds). The collector covers the polar angles from about 25° to

72° and nearly 360° in azimuth.

Figure 5. Total integrated

scattering cross sections for

PSL (dotted line), silicon (solid

line), and aluminum (dashed

line) spheres on silicon.

d6 (n2 - 1) 2

λ4 (n2+ 2)• • E 2

dP 16π2

dΩ λ4= Pi • • [cos (θi) • cos2 (θs) • Qp, q (θi,θs,φs)] • PSD(fx, fy)

(1)

(2)

able from particles in a single channel detection sys-tem. The wafer manufacturers need to classify pits andparticles on silicon wafers. Pits are octahedral voids inCzochralski-grown silicon that have been exposed atthe surface by the polishing process. They are alsoknown as crystal-originated particles (COP), obviouslya misnomer. They sometimes are a single pit and, in alarge number of cases, partially overlapping doublepits.

Pits and scratches are “surface-breaking” defects; i.e.,they are into the surface. The scattering characteristics,therefore, of pits and particles are different and as aconsequence, we can classify detected defects as pits orparticles if we have information from multiple channelsor modes.

The first difference between pits and particles comesfrom their responses to normal and oblique illumina-tion. Figure 7 is a simple illustration of this difference.Part (a) shows the normal illumination with a sphereon a surface intercepting a cross section of the beam.Part (b) is the condition for an oblique beam where theilluminated area on the surface is the same as in part(a). Note that in this plane the same-sized sphere inter-cepts a larger fraction of the incident beam cross sec-tion. Thus, a sphere will scatter significantly more withoblique incidence (see figure 2). Part (c), however,shows that with oblique incidence a pit is at a signifi-cant disadvantage relative to a sphere on the surface forscattering light. Thus, comparing the scattered light innormal and oblique incidence can help classify pits andparticles.

A more important difference between pits and particlesis the angular pattern of the scattering. Both theoreti-cal calculations and experimental results show that particles scatter light principally into the polar anglerange from 20º to 70º when illuminated with p-polar-ized light. In contrast, pits scatter primarily toward the

function of the x and y components of the surface spatial frequency (1). The frequency components, fx and fy, are, in turn, related to the scattering anglesthrough the diffraction equations.

Once the surface PSD characteristic is known, we cancalculate, to a good approximation, the angular distri-bution of the light scattered by the surface.

Obviously, one tries to minimize surface scattering toobtain good defect sensitivity on rough surfaces. Ofparticular importance is the polarization factor of equa-tion 2. Figure 6 shows the variation of this factor forsilicon over the full scattering hemisphere for the fourcombinations of input and scattered polarizations and70º incidence. To minimize surface scattering, usingthe ss polarization combination and collecting light inthe vicinity of 90º and 270º azimuth is very effective.In addition, depending on the underlying material, thepp polarization combination and collecting scatteredlight in the forward direction is useful.

In equation 2 we can also see that the cos(θi) andcos2(θs) terms also imply that greater sensitivity todetecting particles is obtained in the double darkfieldconfiguration, where both angles are >45º.

Unpatterned inspection systems measure the back-ground scattering level, which the industry refers to as“haze”. The measured haze value obviously depends onwhere in the hemisphere we collect the scattered light.Obviously haze is related to the PSD characteristics of a surface, but the relationship is not necessarily asimple one.

Pit scatteringPits are of great interest to silicon wafer manufacturers.Pits have been a problem for inspection systemsbecause they also scatter light and are indistinguish-

Spring 1999 Yield Management Solutions 31

F E A T U R E S

Figure 6. Relative magnitudes of the optical polarization factors

Qi,j.cos2(θ) for 70° incidence on silicon: ss, pp, sp, and ps polariza-

tion combinations over the scattering hemisphere. (Gray scale conver t-

ed from color: bright band contour is 0.5 of the maximum.) The 488-nm

plane wave comes in from the left; view is near –90° in azimuth from

the incidence plane.

Figure 7. Schematic

illustrating the dif fer-

ence between parti-

cles and pits relative

to the illumination

incidence angle.

Scratches also scatter primarily toward the normal,similar to pits. Furthermore, for uniformly detectingscratches of any orientation, normal incidence is preferred.

Dielectric film effectsOn patterned wafers, dielectric films are present. Theselead to a couple of complications. One is the interfer-ence effect that produces color under broad band illu-mination and contrast variation under monochromaticillumination. These effects are particularly troublesomeif the film thickness is not uniform, and one is tryingto do a die-to-die comparison. In brightfield systemsbroad-band illumination has helped. In darkfield systems circularly polarized light is extremely useful in minimizing the film effect.

As a simple example, the scattering cross section of aPSL sphere on silicon dioxide on aluminum as a func-tion of the oxide thickness is shown in figure 9. Notethe substantial variations of total scattered light withfilm thickness with both s- and p-polarization withoblique incidence. However, because the s and p scat-tering are out-of-phase with respect to each other, scat-tering with circular polarization, which has both, is much less affected by film non-uniformity. For normal incidence, s, p, and circular are all equivalentand the film effect is worse than that seen with obliqueincidence.

Digs and scratch detection will also be affected bydielectric film thickness and polarization, but theirvariation in scattering is not in phase with the particlescattered intensity.

Previous layer defectsOne may or may not want to see previous layer defects,depending on the system application. Usually, whilemonitoring equipment one does not want to see downinto the previous layers. Oblique illumination with spolarization has much less penetration of energythrough transparent dielectrics than normal illumina-tion and thus is preferred for detecting current layerdefects.

Part II of this article addresses System Considerations to Meet theDesign Shrink Challenge and Future Needs and Developments in WaferInspection Technology and will appear in the next issue. To view thewhole article, you may also request an advance copy through theBRC or visit our website at www.kla-tencor.com/corpmag.

1. Church, E.L., Jenkinson, H.A., Zavada, J.M., Opt. Eng. 18, 125-138(1979)

normal; therefore, comparing the light scattered intohigher angles with those toward the normal will alsoclassify pits and particles. Even for normal incidencethis separation works; however, oblique incidenceworks best. We show experimental results for the p-polarized oblique incidence case are shown in figure 8.

Scratch scatteringScratches are important in CMP processes and may be yield-limiting. Scratches preferentially scatter perpendicular to their long dimension. Real scratchesare not perfect linear defects; in many cases they havecross-sectional variations along the scratch, may haveparticulate debris nearby, and commonly are “chattermarks.” These “chatter marks” or “micro-scratches”actually are a series of short small scratches along aline perpendicular to the long dimension of thescratches.

Spring 1999 Yield Management Solutions32

F E A T U R E S

Figure 8. Signal levels of PSL spheres and silicon pits with about 25°

to 72° collection vs. about 6° to 20° collection, shows superior classi-

fication of pits and particles using oblique incidence.

Figure 9. Scattered intensity for a 100 nm PSL sphere and a

250 nm PSL sphere on silicon dioxide on aluminum as a function of

oxide thickness and input polarization, p-polarization (long dash),

s-polarization (shor t dash), and circular polarization (solid). Scattered

light collected from about 25° to 72° polar angle.