WO2005057194A1 - Evaluation differentielle de regions adjacentes pour determiner le changement de la reflectivite - Google Patents

Evaluation differentielle de regions adjacentes pour determiner le changement de la reflectivite Download PDF

Info

Publication number
WO2005057194A1
WO2005057194A1 PCT/US2004/040969 US2004040969W WO2005057194A1 WO 2005057194 A1 WO2005057194 A1 WO 2005057194A1 US 2004040969 W US2004040969 W US 2004040969W WO 2005057194 A1 WO2005057194 A1 WO 2005057194A1
Authority
WO
WIPO (PCT)
Prior art keywords
wafer
region
regions
photodetector
polarized
Prior art date
Application number
PCT/US2004/040969
Other languages
English (en)
Inventor
Peter G. Borden
Edward G. Budiarto
Original Assignee
Applied Materials, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US10/732,436 external-priority patent/US7136163B2/en
Priority claimed from US10/731,991 external-priority patent/US7190458B2/en
Application filed by Applied Materials, Inc. filed Critical Applied Materials, Inc.
Publication of WO2005057194A1 publication Critical patent/WO2005057194A1/fr

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/956—Inspecting patterns on the surface of objects
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
    • G01R31/302—Contactless testing
    • G01R31/308—Contactless testing using non-ionising electromagnetic radiation, e.g. optical radiation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/9501—Semiconductor wafers
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
    • G01R31/282—Testing of electronic circuits specially adapted for particular applications not provided for elsewhere
    • G01R31/2831—Testing of materials or semi-finished products, e.g. semiconductor wafers or substrates

Definitions

  • FIG. 1 A shows a cross-sectional view of a MOS field effect transistor (FET) well known in the prior, art.
  • MOS transistor typically includes source region 101s and drain region 101d, source extension region 102s and drain extension region 102d, channel 103, gate insulator 104, gate 105, and well 106, all of which are formed in semiconductor substrate 107. Additional doped layers may be present, but are not shown for simplicity.
  • the source and drain regions 101s and 101d are heavily doped, typically with arsenic for n-type doping or boron for p-type doping. Doping levels are on the order of 10 20 dopant atoms per cubic centimeter.
  • the layers for regions 101s and 101d are typically 500-700 angstroms deep.
  • the extension regions 102s and 102d are also heavily doped, with the same type of dopant atoms as the source and drain regions 101s and 101d, but the extension regions are shallower - typically 300 to 500 angstroms deep.
  • FIG. 1B shows the doping profiles in the vertical directions (along the arrow A in FIG. 1A).
  • Extension regions 102s and 102d provide contacts to the channel region 103.
  • the transistor operates by applying a bias to the gate 105 while grounding the well 106.
  • the regions 101s, 102s, 102d and 101d are n- type, so that the majority carriers are electrons. If a positive voltage is placed on gate 105 with respect to the channel 103, no current will flow between the gate 105 and channel 103 because of the presence of thin gate insulator 104. However, the positive voltage will attract electrons to the channel region 103, - creating a thin layer of electrons (called an inversion layer) that connects source extension 102s to drain extension 102d, allowing current to flow between the source and drain.
  • an inversion layer a thin layer of electrons
  • the doping profiles for the various source and drain layers 101s, 101d, 102s and 102d are not perfectly abrupt (box-like). They are usually formed by diffusion processes that may involve several thermal cycles, causing the profiles to be somewhat rounded.
  • FIG. 1B shows two profiles 112A and 112B for the source extension 112s, following the arrow A in FIG. 1A. Line 112A shows a relatively abrupt profile and line 112B shows a less abrupt profile.
  • Such variation in abruptness may be encountered in different semiconductor wafers under fabrication because each step in the fabrication process has a certain tolerance. Variation in individual process steps or the cumulative variation of a series of process steps can cause a loss of abruptness in the profile (e.g. may go from line 112A to line 112B).
  • junction depth and peak doping concentration may vary depending on process properties, such as, for example, variation in annealing temperature. For example, profile 112B forms a deeper diffused profile than profile 112A, but at a smaller peak doping concentration.
  • the performance of the transistor is affected by the final doping profile after annealing.
  • Profile depth, peak concentration, and profile abruptness are carefully controlled because they directly contribute to short-channel effects and speed of the transistor.
  • a deeper and less abrupt profile in the vertical direction creates a higher off-state leakage current that leads to increased power consumption.
  • a smaller peak doping concentration increases the resistance component between the source/drain region and the transistor channel, leading to a greater voltage drop between the source 101s and drain 101d (FIG. 1A). This voltage drop reduces the ability of the transistor to drive the next stage, reducing the speed of the circuit.
  • Jiang et al. describe (in U.S. Patent 6,556,306 which is incorporated by reference herein in its entirety as background) a method for determining the index of refraction of a thin film at a desired angular frequency.
  • Jiang et al. disclose (in the abstract) generating an input desired-frequency pulse and an output detectable probe pulse. According to Jiang et al. a thin film is moved in and out of the path of the input pulse, creating an output pulse that alternates between a transmitted signal created when the film intercepts the input pulse path, and a reference signal, created when the sample is outside the input pulse path. The output pulse modulates the probe pulse, which is then detected with a photodetector and the difference between the transmitted signal and the reference signal is calculated.
  • the above steps are repeated over a plurality of delay times between the input pulse and the probe pulse until a complete field waveform of the differential signal is characterized.
  • the index of refraction is calculated by Jiang et al. as a function of a ratio between the differential signal for the thin film and the reference signal.
  • Heinrich et al. describe (in U.S. Patent 4,758,092 which is incorporated by reference herein in its entirety as background) a method and means for optical detection of charge density modulation in a semiconductor.
  • Heinrich et al. describe passing a polarized coherent light beam onto an interferometer which establishes two polarized beams. The two polarized beams are focused on a silicon device under test with one beam focused on or near an active device and one beam providing a reference. After passing through the device under test the two beams are reflected off a metal layer and back through the device under test where they are recombined by the beam splitter.
  • the charge carriers affect the index of refraction of silicon, and by modulating the electrical charge in the active device a small modulation occurs in the index of refraction.
  • the modulation affects the phase delay of the one beam near the active device in relation to the reference beam, and hence when the beams are recombined at the beam splitter they interfere and convert the relative phase modulation into an amplitude modulation which can be detected with a photodiode.
  • Heinrich et al. also state that the position of a single optical beam can be spatially modulated over a silicon wafer surface to detect stationary charge densities in one area relative to a reference area.
  • the reference area may contain no charge density variation thereby giving an absolute reference value.
  • a semiconductor wafer having at least two regions of different reflectivities on a front surface thereof is evaluated in accordance with the invention by generating a beam of substantially monochromatic electromagnetic radiation of a predetermined wavelength ⁇ , illuminating each of the two regions on the front surface with the beam either simultaneously or sequentially, while continuously measuring intensity of a portion of the beam reflected by the illuminated region during uninterrupted rhythmic movement, by use of a photodetector sensitive to the wavelength ⁇ .
  • the two regions are illuminated simultaneously by splitting the beam (which is linearly polarized) into two beams of S and P polarization each of which illuminates one of the regions.
  • the beam is scanned between the two regions, to alternately illuminate each region.
  • the signal sensed by the photodetector varies synchronously at the predetermined frequency f at which the beam is being oscillated between the two regions.
  • the intensity of the S and P polarized beams is modulated over time at a predetermined frequency f, such that the two regions are alternately illuminated.
  • the signal detected by the photodetector also varies synchronously at the predetermined frequency / and in a phase-sensitive manner.
  • the amplitude of fluctuation i.e.
  • the time varying component) of an analog electrical signal being sensed by the photodetector is thereafter used as a measure of change in reflectivity between the two regions, e.g. in process control.
  • the measurement is performed on a test structure that is fabricated at the same time as (and using the same processes as) active devices in the wafer.
  • two beams that are coaxial are split, and the S and P polarization components of each beam are incident on a test structure and form concentric spots thereon. While both beams are incident on and reflected by each region in the test structure, one of the beams is not modulated and reflection of this beam is filtered out by an optical element present in the path of the reflected beams, and the other beam is incident on and detected by the photodetector in the above-described manner.
  • the use of two beams enhances the reflectivity of, for example, a doped region due to creation of carriers by the additional beam.
  • the difference in reflectivity measured by i.e. the amplitude of fluctuation of the electrical signal generated by
  • the photodetector is larger, and any deviation therein is easier to detect.
  • the use of two beams also enhances the sensitivity of the measurement to other properties of the doped layer such as oxide charge and defect densities, due to creation of excess carriers by the additional beam.
  • two beams are used in the above- described manner, except that only one beam is split and this beam is unmodulated, whereas the other beam is not spit and has an oscillating intensity.
  • FIG. 1A shows, in a cross-sectional diagram through a prior art semiconductor wafer, the structure of a MOS field effect transistor.
  • FIG. 1B shows, in graphs, the doping profile of the prior art semiconductor wafer, along arrow A of FIG. 1 through extension layer 102s.
  • FIG. 2A illustrates, in a flow chart, acts performed in several embodiments of the invention.
  • FIG. 2B illustrates, in a cross-sectional view, the relative dimensions of absorption length Al, wafer thickness z s , well depth z w and depth z j of an interface between doped region 211 and well region 216.
  • FIG. 2C illustrates, in a plan view, multiple positions of the incident beam (one position shown solid and four other positions shown dotted) during oscillating movement along a straight line L between the doped region 211 and the undoped region 213 of a test structure 210; also shown in FIG. 2C is an integrated circuit 220 formed simultaneous with test structure 210.
  • FIGs. 2D-2H illustrate, in a cross-sectional side view along line L of FIG. 2C, multiple positions of the incident beam of FIG. 2C.
  • FIG. 21 illustrates, in a graph, a plot of an analog electrical signal along the y-axis as a function of time t along the x-axis, wherein signal on the y-axis is generated by a photodetector sensitive to a wavelength of the beam of FIG. 2C in response to oscillating movement thereof, as illustrated in FIGs. 2D-2H.
  • FIG. 2J illustrates, in a perspective three dimensional view, the relative positions of various optical components in an embodiment that uses three lasers for three measurements, uses an acoustb-optic beam deflector for causing relative motion between the beam and the wafer, and a lock-in amplifier that uses a frequency of oscillation of the beam deflector to measure amplitude of a signal generated by the photodetector.
  • FIGs. 3A and 3B illustrate various dimensions of the test structure relative to the beam incident thereon for use by a method of the type described herein.
  • FIG. 3C illustrates, in a graph, the continuous nature of the signal sensed by the photodetector, including the durations tmax and tmin for which the signal remains steady and the durations tfall and trise during which the signal falls and rises respectively.
  • FIG. 3D illustrates, in a graph, an increase in amplitude of the analog fluctuating electrical signal of FIG. 3C which can happen in case of higher dopant concentration and/or deeper profile depth (e.g. Z j ⁇ 700 micrometers), and/or longer wavelength of probe beam as compared to a structure which generates the signal of FIG. 3C.
  • dopant concentration and/or deeper profile depth e.g. Z j ⁇ 700 micrometers
  • FIG. 3E illustrates, in a graph, a reduction in duration tmax during which time signal remains at its maximum value and an increase in duration tmin during which time signal remains at its minimum value, which happens when beam positions 309a and 309b of FIG. 3B are not equidistant from a boundary 302 between the doped region 301 and undoped region 303 (specifically illustrated herein is the situation when the beam spends more time in the doped region than in the undoped region).
  • FIG. 3F illustrates, in a graph, the increase in durations trise and tfall during which time the signal respectively rises from its minimum value to maximum value and vice versa, which happens in a situation wherein a boundary 302 between the doped region and the undoped region is less abrupt than in a structure which generates the signal of FIG. 3B.
  • FIG. 3G illustrates, in a graph, a reduction in amplitude of signal wherein the beam spends equal time in the doped and undoped regions but the beam scan (distance Dscan between positions 309a and 309b of FIG. 3A) is less than the beam diameter (distance Bdiameter).
  • FIG. 3H illustrates, in a side view, oscillation of a beam along a line across alternating fingers of doped and undoped regions.
  • FIG. 31 illustrates, in a graph, the fluctuating electrical signal generated by a photodetector when oscillating the beam as illustrated in FIG. 3H.
  • FIGs. 3J-3N illustrate application of an oscillating beam measurement of the type described above to different structures in various embodiments of the invention.
  • FIG. 4A illustrates, in a flow chart, acts for use of a measurement of the type described above to decide if a process for fabricating a semiconductor wafer is in control (i.e. generates wafers within predefined tolerances of material and dimension properties).
  • FIG. 4B illustrates, a block diagram, a wafer fabrication system including a measurement apparatus used in a process flow for wafer fabrication.
  • FIG. 5A illustrates, in a cross-sectional view, a test structure for characterizing the profile of doped regions of a transistor, in one embodiment of the invention.
  • FIG. 5B illustrates, in a cross-sectional view, the test structure of FIG. 5A after removal of oxide mask 511 and annealing.
  • FIGs. 6A-6E illustrate, in a series of cross-sectional diagrams, a process of creating a test structure of the type illustrated in FIGs. 5A and 5B.
  • FIG. 7A and 7B illustrate the incidence and reflection of a beam of electromagnetic radiation in a direction perpendicular to an exposed surface of an undoped material (FIG. 7A) and a doped-material (FIG. 7B).
  • FIG. 8A shows a simplified graph of doping concentration (N) as a function of depth (z) from a front surface of a wafer in a test structure, which resembles actual doping profile in real samples in the same wafer.
  • FIG. 8B shows the derivative of the graph in FIG. 8A.
  • FIG. 9A shows, in theory, the reflected signal's intensity (along the y axis) as a function of depth of a doping profile for a silicon substrate material (along the x axis), when using a laser of wavelength of 980nm for the measurement, and for peak doping concentration and profile abruptness combinations of 8e19/cm and 100A (for graph 900C), 5e19/cm 3 and 100A (for graph 900A), and 8e19/cm 3 and 200A (for graph 900B).
  • FIG. 9B illustrates use of three signals Si, S 2 , and S 3 measured from an unknown sample having a profile depth z j , using three different wavelength laser beams.
  • Curves 901 , 902, and 903 are linear portions of a cosine response curve of the type shown in FIG. 9A, in the order of decreasing wavelength.
  • FIG. 9C illustrates use of a single signal to determine a profile depth Z j , from a linear portion of a response curve of the type shown in FIG. 9B.
  • FIG. 9D illustrates, in a three dimensional graph, signal ratio (Samplitude/Smax) plotted on the Z-axis, as a function of the peak doping concentration and the profile depth on the X and Y-axes, respectively, wherein measurements using two different wavelengths of the measuring beam are shown as two surfaces 921 and 922.
  • FIG. 9E illustrates, in a flow chart, acts performed in some embodiments of the invention.
  • FIG. 10A illustrates a screenshot of a commercially-available software program Mathcad 2001 i Professional which is programmed in accordance with certain embodiments of the invention to solve three simultaneous nonlinear equations for the determination of the following three properties: peak (N rf ) of the dopant profile, depth (zj) of the dopant profile, and abruptness (a) of the dopant profile using three amplitude measurements at three wavelengths.
  • FIGs. 10B and 10C illustrate, in a block diagram, two alternative embodiments of a three laser apparatus for performing the measurements used in the program of FIG. 10A.
  • FIG. 11A illustrates in an alternate embodiment probing of doped region 1101 and undoped region 1103 of a test structure by a carrier generation beam 1110 collinear with a measuring beam 1109.
  • FIG. 11B illustrates generation of excess carrier distributions 1113 and 1114 by beam 1110 in the doped layer 1101 and well region 1106, respectively.
  • FIG. 11C illustrates, in a plan view, multiple positions of two concentric incident beams (one position shown solid and two other positions shown dotted) during oscillation along a straight line L between the doped region and the undoped region of a test structure.
  • FIG. 12 shows the hardware configuration of some embodiments for evaluation of a test structure using either an oscillating stage (1218) or a beam deflector (1215) as the means for oscillating a spot along a straight line between two adjacent regions.
  • FIG. 13A illustrates, in a flow chart, acts performed in several embodiments of the invention wherein a beam is split into two portions that are respectively incident on two regions of a wafer to be evaluated.
  • FIG. 13B illustrates, in a plan view, at most two positions of the incident beam formed by splitting a single beam, thereby to illuminate doped region 1301 and the undoped region 1303 of a test structure 1310; also shown in FIG. 13B is an integrated circuit 1320 formed simultaneous with test structure 1310.
  • FIG. 13C illustrates, in a side view, the simultaneous illumination of doped region 1301 and undoped region 1303 of a test structure by respective beam portions 1309a and 1309b which are generated by a beam splitter (such as Wollaston prism) 1307 from a single beam 1308 incident thereon.
  • a beam splitter such as Wollaston prism
  • FIG. 13D illustrates, in a close up perspective view, the use of Wollaston prism in some embodiments to generate two beams of mutually perpendicular polarizations from a single beam of rotating linear polarization that in turn is generated by a half-wave plate from a beam of linear polarization incident thereon.
  • FIG. 13E illustrates, in a side view in the direction 13E in FIG. 13D, the optic axes of two wedges of the Wollaston prism of FIG. 13D.
  • FIG. 13F illustrates, in a graph, the continuous nature of the signal sensed by the photodetector, including the times at which the signal reaches its maximum and minimum and the durations tfall and trise during which the signal falls and rises respectively; also illustrated in FIG. 13F are the relative directions of the input beam's polarization and the half-wave plate crystal axis.
  • FIGs. 13G-13M illustrate, in a plan view, spots of illumination formed on a test structure by use of the embodiment of FIG. 13C.
  • FIG. 13N illustrates, in a three dimensional view, the relative positions of various optical components in an embodiment that uses a Wollaston prism to illuminate two regions of a wafer, and a lock-in amplifier that uses a frequency of rotation of a half-wave plate (through which a polarized beam passes) to measure amplitude of a signal generated by the photodetector.
  • FIG. 130 illustrates a hardware configuration for an embodiment of the type illustrated in FIG. 13A using a rotating half-wave plate (1204) and a Wollaston prism (1214) to illuminate the doped and undoped regions (without movement along a line of the type illustrated in FIG. 3A).
  • FIG. 14A illustrates, in a side view, an alternative embodiment wherein two beams that are coaxial with one another are both split by a Wollaston prism, for evaluation of a test structure of the type described herein.
  • FIG. 14B illustrates a hardware configuration of the type illustrated in FIGs. 13A- 131 for the alternative embodiment of FIG. 14A.
  • FIGs. 14C-14E illustrate, in a plan view, exactly two positions of one incident beam and at most two positions of another incident beam due to splitting of both beams as illustrated in FIGs. 14A and 14B.
  • FIG. 15A illustrates, in a side view, an alternative embodiment wherein one beam is split by a Wollaston prism and another beam is modulated at a predetermined frequency, used in evaluation of a test structure of the type described herein.
  • FIG. 15B illustrates a hardware configuration of the type illustrated in FIGs. 14B for the alternative embodiment of FIG. 15A.
  • FIG. 16 illustrates, in a two dimensional graph, relative angles of various beams and their polarizations for the embodiment of FIGs. 15A and 15B.
  • FIG. 17A illustrates, in a graph, the concentration of dopants as a function of depth, for an ideal case of a perfectly abrupt doping profile.
  • FIG. 17B illustrates, in a graph the sinusoidal nature of the measured signal as a function of the depth of the doping profile.
  • a test structure 210 (FIG. 2B) is also simultaneously fabricated (as per act 201 in FIG. 2A), using the same processes that are used to form integrated circuits (e.g. IC portion 220) in the wafer.
  • the only difference between regions of the test structure 210 and regions of the integrated circuit portion 220 is in dimensions.
  • the test structure 210 is designed to have dimensions sufficiently large to accommodate the presence of a beam 219 of electromagnetic radiation as described next.
  • a doped region 211 in the test structure may be a square with a side of the same size as the spot diameter. Making the doped region 211 of the test structure larger than the spot diameter accommodates error in alignment of the beam relative to the region.
  • a doped region 211 in such a test structure may made smaller than the beam spot in some embodiments (e.g. the doped region's diagonal may be same as the spot diameter), although the signal measured in such embodiments is not as sensitive to a change in properties between regions.
  • the specific active device 220 that is being fabricated in the wafer is not a critical aspect of the invention, as long as a structure 210 to be tested is formed at the same time as and in the same manner as the active device 220 (or portions thereof).
  • the above-described wafer is evaluated during semiconductor wafer fabrication by use of a laser (or other such coherent light source) to generate a beam 219 (FIG.
  • beam 219 could be polarized or unpolarized depending on the embodiment.
  • Illumination of the front surface F in all embodiments described herein allows the incident electromagnetic radiation to immediately reach (and be reflected from) regions of interest, such as doped regions, undoped regions, metallic regions or other such regions that are being fabricated at the front surface F of the wafer, without having to pass through most of the wafer's thickness.
  • regions of interest such as doped regions, undoped regions, metallic regions or other such regions that are being fabricated at the front surface F of the wafer, without having to pass through most of the wafer's thickness.
  • regions of interest such as doped regions, undoped regions, metallic regions or other such regions that are being fabricated at the front surface F of the wafer, without having to pass through most of the wafer's thickness.
  • regions of interest such as doped regions, undoped regions, metallic regions or other such regions that are being fabricated at the front surface F of the wafer, without having to pass through most of the wafer's thickness.
  • Illumination of the bottom surface as suggested by Heinrich et al. requires the incident
  • any defect in the substrate affects the measurement (especially since substrates are manufactured with a larger tolerance for defects), and (b) to ensure that the reflected energy can be meaningfully measured it is necessary that a probe beam of a large wavelength (e.g. 1.3 ⁇ m) be used, as mentioned by Heinrich et al. at column 3, lines 61-62).
  • a probe beam of a large wavelength e.g. 1.3 ⁇ m
  • the predetermined wavelength ⁇ of beam 219 (FIG. 2B) in accordance with the invention is selected to be sufficiently short to ensure that an absorption length Al of the beam in the wafer is less than the thickness z s of the wafer.
  • absorption length is used to indicate a length (also called “penetration depth") at which the original intensity E 0 Z of beam 219 falls to (Me)E 0 z . Therefore, more than a majority (about 63%) of the energy E 0 2 incident on a front surface F of the wafer does not even reach the bottom surface B of the wafer, when absorption length is less than wafer thickness. The same is true in the reverse direction, i.e.
  • absorption length less than wafer thickness (e.g. if absorption length is 690 ⁇ m and wafer thickness is 700 ⁇ m)
  • absorption length less than wafer thickness (e.g. if absorption length is 690 ⁇ m and wafer thickness is 700 ⁇ m)
  • approximately 26% of the total energy incident on front surface F is transmitted to and becomes incident on bottom surface B, and therefore approximately 7.8% (of the total incident energy on front surface F) is reflected by bottom surface B, and therefore about 2.8% (of the total incident energy on front surface F) reaches front surface F and only about 2% (of the total incident energy on front surface F) emerges outside of the front surface F.
  • beam 219 having absorption length Al less than wafer thickness z s ensures that most of the energy reflected by the wafer, other than from the front surface F, is reflected by regions and interfaces that are physically located between the front and bottom surfaces, e.g. regions 211 and 216 (FIG. 2B) that are formed in substrate 230 during semiconductor wafer fabrication. Therefore, use of such a short wavelength beam ensures that any portion of the incident beam 219 that may be reflected by the bottom surface B is limited to a negligible percentage (e.g. less than 10%) of the total energy reflected from the wafer. For example, if the total reflected energy is 30% of the incident energy, the bottom surface contribution is limited to less than 3%.
  • a negligible percentage e.g. less than 10%
  • the absorption length Al is selected to be no smaller than depth z ⁇ - of a junction (or other profile) of a region formed during semiconductor wafer fabrication. Hence, any value for the absorption length in the range (junction depth ⁇ absorption length ⁇ wafer thickness) may be selected. Therefore, the maximum absorption length is Almax ⁇ z s and the minimum absorption length is Almin > Z j .
  • junction depth Z j is in a range of 500-800 angstroms and wafer thickness is 700 ⁇ m.
  • the contribution of the bottom surface to the energy being reflected from the wafer is a negligible percentage (about 2.4%) of the total energy incident on the front surface.
  • beam 219 may be selected to be of an absorption length Al which is less than or equal to one-half the wafer thickness, i.e. less than or equal to 350 ⁇ m. In such a case, the energy being reflected by the wafer is primarily reflected from the front surface F and from regions of interest in the vicinity of front surface F.
  • the only energy emerging from the front surface is energy that is reflected by the front surface itself and energy that is reflected from the interfaces and regions that occur close to the front surface F, such as the interface 231 between well 216 and substrate 230, and interface 232 between region 211 and well 216, as well as regions therebetween such as well 216 and region 211.
  • beam 219 is selected to have a wavelength ⁇ of anywhere in the range 400 nm to 1.04 ⁇ m, wherein the lower limit is set by the absorption length being greater than junction depth (or profile depth) permitted by current technologies for Ultra Shallow (US) junctions, and the upper limit is set by the absorption length being less than the wafer thickness.
  • a beam of wavelength ⁇ of 1.1 ⁇ m or greater cannot be used as described herein because its absorption length of 1500 ⁇ m is more than twice the wafer thickness of 700 ⁇ m (noted above).
  • beam 219 can have any power depending on the embodiment, e.g. between 5 mW and 100 mW.
  • beam 219 can be generated by a conventional laser diode, such as a Fiber Bragg Grating (FBG) stabilized AIGaAs laser with a wavelength of 980 nm and a power of 100 mW (part # 26-8052-100) available from JDS Uniphase, San Jose, California.
  • FBG Fiber Bragg Grating
  • Beam 219 (FIG. 2B) of the type described above forms a spot on a front surface F of a wafer being evaluated, and such a spot (shown at position 219A in FIG. 2C) is moved relative to the wafer (see act 204), in a physical oscillation at a predetermined frequency / between two regions of a test structure, e.g. doped region 211 and undoped region 213 (FIG. 2C).
  • region 213 is formed of the same material as and is coextensive with a well region 216, and for this reason, region 213 is shown by a dashed line in FIGs. 2B and 2C.
  • Beam 219 may be initially located at position 219A over a doped region 211 (FIG. 2C) and during act 204 the beam is moved continuously from position 219A to a symmetric position 219E over undoped region 213, transitioning through positions 219B, 219C and 219D therebetween along a straight line L. As shown in FIG. 2C, regions 211 and 213 are adjacent to one another and share a common boundary 212.
  • Positions 219A and 219E are extreme end positions at which movement of the beam 219 is reversed. Therefore, after position 219E is reached, movement is reversed and the beam returns back to position 219A after transitioning through positions
  • FIG. 2C Side views of the five positions 219A-219E (FIG. 2C) are illustrated in the respective FIGs. 2D-2H.
  • intensity of a portion of the beam reflected by the illuminated region is continuously measured (as per act 203) by a photodetector sensitive to the wavelength ⁇ of the incident beam.
  • the intensity of the radiation being reflected changes during the relative motion, because the difference in reflectivity at the two end positions is the same as the difference in reflectivity between the two regions.
  • the beam is exactly 50% in each of the two regions and the reflectivity should be an average of the two reflectivities. In practice, reflectivity at the central location may significantly deviate from the average due to presence of discontinuities that may be inherent in the boundary between the two regions.
  • the reflectivity is a weighted average of the area of the two regions illuminated by the incident beam.
  • a photodetector 215 that is sensitive to the portion of the beam reflected from the wafer generates a continuous analog electrical signal.
  • the generated electrical signal has a magnitude that varies with time, at the predetermined frequency / of relative movement between the beam and the wafer.
  • an amplitude of fluctuation (i.e. the time varying component) of the electrical signal is synchronously detected at frequency / of the relative movement, e.g. by use of a lock-in amplifier.
  • a beam that is incident on the wafer is generated continuously, and the beam is used to continuously illuminate the test structure.
  • the continuous illumination of the test structure distinguishes this invention from the disclosure of Jiang et al. (in U.S. Patent 6,556,306) which appears to require the use of pulses.
  • Jiang appears to use the term "desired frequency" to mean a specific range of frequencies, e.g. GHz - THz (column 1 lines 8-27 and column 3, line 66), instead of a monochromatic beam.
  • Jiang uses a pulse, and a pulse normally has a frequency spread that is a significant multiple (e.g. 3 terahertz) of the center frequency (e.g. 1 terahertz).
  • Jiang's pulse is not monochromatic (in the same sense that this term is used herein to describe a laser-generated beam which is used in some embodiments of Applicants' invention). .
  • the amplitude of fluctuation (i.e. the time varying component) of the electrical signal generated by the photodetector is also referred to herein as a reflectivity change measure.
  • the reflectivity change measure indicates a difference in reflectance between the two regions, and is thereafter used (1) to obtain a measure of a change in one or more properties between the two regions (based on modeling or previously collected data); or (2) to perform process control (by comparison with predetermined limits and changing a process parameter if outside a range).
  • the just-described measurements are performed by use of at least one beam of electromagnetic radiation that is incident normal to a surface F of the wafer (FIG. 2D), and a reflected portion of this beam is detected.
  • FIGs. 2D-2H depict a beam splitter 218 that is shown to be in-line with the incident beam 219, to indicate that specular reflection from surface F in a direction opposite to the direction of incidence is being measured (wherein both directions are along a line 214 normal to the surface F).
  • Beam splitter 218 passes a laser beam that is generated by laser 217 through to wafer 230. Beam splitter 218 splits off a portion (e.g.
  • beam splitter 218 is also referred to elsewhere herein as "detection system beam splitter". Detection system beam splitter 218 is non-polarizing in many embodiments of the type described herein. Detector 215 is offset from the line of incidence and reflection 214 in the usual manner.
  • the above-described measurements are performed by the detector 215 continuously and sequentially (i.e. one after another), during relative motion between the beam 219 and the wafer 230, to generate an analog electrical signal of the type illustrated in FIG. 21.
  • An amplitude of fluctuation of the signal Samplitude is obtained as an average over a number of cycles N, by use of a lock-in device.
  • the amplitude of fluctuation Samplitude (FIG. 21) indicates a change in reflectivity between the two regions.
  • Detecting the amplitude of fluctuation eliminates the need to determine a constant present in the individual reflectivity measurement (as a baseline Smin in the analog electrical signal).
  • the electrical signal varies between Smin and Smax as illustrated in FIG. 21.
  • signal that is being measured by detector 215 is at its minimum Smin between time t1 and t2 (FIG. 21) when the beam 219 is incident wholly within the doped region 211 (FIG. 2D) and signal reaches its maximum Smax between time t3 and t4 (FIG. 21) when the beam 219 is incident wholly within the undoped region 213 (FIG. 2H).
  • An amplitude Samplitude of fluctuation of the periodic signal is detected using a phase- sensitive lock-in technique to isolate the ac (alternating current) component as compared to a dc (direct current) component.
  • the difference Smax- Smin Samplitude is measured by the lock-in device, and eliminates the need to calibrate the effect of Smin which is present as a baseline in the electrical signal S.
  • the Samplitude in each of a number of cycles is the same, i.e.
  • the number N can be one hundred or more (e.g. oscillate 100-200 times between the two regions) although in some embodiments that have a very high SNR the cycles may be as few in number as a handful (say 5-10 times).
  • Presence of spikes can be detected by tuning a lock-in amplifier to twice the frequency / of relative movement, and such spikes are used in some embodiments to indicate to the user a property of boundary 212.
  • the spikes are filtered out of signal which is otherwise fluctuating at the frequency / of relative motion.
  • the spikes are treated as anomalies in signal from the photodetector, and are eliminated by applying a filter, in the form of a series of electrical gates. The gates are synchronized to twice the frequency / of relative movement, and the filtered signal is thereafter provided to the lock-in amplifier (for synchronous detection at the frequency of relative motion).
  • Relative motion between beam 219 and wafer 230 of the type illustrated in FIGs. 2C-2H can be accomplished in any manner, e.g. in one embodiment an optical bench (carrying the laser and the photodetector) is moved relative to a stage (carrying the wafer) which is kept stationary, while in another embodiment the stage is moved while the bench is kept stationary, while in still another embodiment the stage and the bench are both moved.
  • the stage and/or bench can be reciprocated using a piezoelectric actuator of the type well known in the art. See stage 1218 attached to a piezoelectric actuator 1219 of FIG. 12 (described below) which is coupled to receive an electrical signal fluctuating at the frequency / at which the lock-in detector is operated to determine Samplitude.
  • neither a stage nor a bench is moved (relative to ground), and instead a beam that passes therebetween is scanned relative to the wafer, e.g. between positions 275A and 275B in FIG. 2J.
  • the beam can be scanned using a beam deflector 274 in a path of the beam.
  • a beam deflector is a device that takes an incoming beam traveling in a certain direction, and causes that beam to change its direction in a periodic manner about its original un-deviated direction.
  • Beam deflector 274 can be, for example a scanning galvanometer mirror or an acousto-optic deflector, although any other optical device that can oscillate a beam in space can be used in other embodiments.
  • FIG. 2J illustrates a beam deflector 274 that receives a beam 275 along the y axis from a dichroic mirror 272 (after passing through a beam splitter), and mirror 272 in turn receives the beam from one of lasers 271A-271C along the x axis.
  • Beam deflector 274 oscillates beam 275 between positions 275A and 275B that are both within the x-y plane.
  • beam deflector 274 is an acousto-optic beam deflector.
  • optical bench 279 has a planar surface 279S that is parallel to the just-described x-y plane, and the above-described optical elements illustrated in FIG. 2J are mounted on surface 279S. Note that in this embodiment, lasers 271A-271C are mounted on surface 279S, adjacent to one another, along three positions on a line L2 that is parallel to the y axis.
  • mirror 272 is translatable between three locations that are opposite to the three laser positions.
  • the three locations of mirror 272 are along another line L1 that is parallel to the yaxis and therefore parallel to line L2.
  • beam deflector 274 is located in a path of beam 275, along line L1.
  • a beam splitter 276 is also located along line L1 , between mirror 272 and beam deflector 274.
  • One method uses the wafer stage to move the wafer back- and-forth under the measuring beam while the beam and detector remain stationary.
  • the stage is oscillated over a distance sufficiently large to cover both the doped and undoped regions; for example a scan distance of 10 ⁇ m is sufficient in some embodiments.
  • the frequency / of oscillation of the stage can be, for example, 100 hertz (one hundred oscillations per second).
  • the specific frequency / that is used in any given embodiment depends on a number of factors, such as the type of mechanism used to implement oscillation, the delay (also called “latency") in electronics (e.g. lock-in amplifier) used to detect an electrical signal fluctuating at frequency /, the duration available for each measurement and the number of cycles to be used therein. Note that any frequency in the range of 1 Hz to 20,000 Hz can be used. For example, if the modulation frequency is 1000 Hz, and if at least 10 cycles are required by the lock-in amplifier to generate a reflectance measurement, then 10 milliseconds are required to perform each reflectance measurement.
  • the throughput is 30 wafers per hour, or 120 seconds per wafer, with each wafer having a measurement taken in at least ten regions.
  • an optical bench carrying all the optical elements can itself be oscillated, although the mass of the optical bench is at least an order of magnitude larger than the mass of a stage.
  • Another method uses an acousto-optic deflector to oscillate the laser beam slightly about the propagation axis.
  • the photodetector stays stationary since it is sufficiently large to capture the reflected radiation from the end points of oscillation.
  • the photodetector may also be oscillated, along with the laser beam, e.g. if the photodetector is smaller than the amplitude of oscillation (i.e. the distance between the two endpoints thereof).
  • a sound-wave traveling inside an optical crystal sets-up a diffraction grating for a light beam traveling in the perpendicular direction to the sound wave.
  • the diffraction grating deflects the light beam at an angle proportional to the frequency of the sound-wave. Scanning of the light beam is achieved either by changing the sound-wave frequency at the predetermined frequency / or by modulating the amplitude of the sound-wave at the predetermined frequency /.
  • both the stage and the acousto-optic deflector are operated simultaneously at different frequencies, and the signals at the sum and/or difference frequency is detected by the lock-in amplifier.
  • stage 281 can be used to move wafer 230 relative to the oscillating beam in the x, y and z directions.
  • stage 281 can be used to move wafer 230 in the vertical direction along the z axis to adjust focus, and in a horizontal plane to adjust the position of a test structure relative to the incident beam from mirror 282 that in turn receives the oscillating beam from acousto-optic deflector 274 and sends it to objective lens 283.
  • beam positions 275A and 275B are shown exaggerated in FIG. 21, and it is to be understood that they are sufficiently close so that the reflected beam passes back through the same optical elements (such as objective lens 283 and beam deflector 274) and is incident on a beam splitter 276 that deflects a portion of the reflected light to a photodetector 277.
  • a signal indicative of the frequency / of oscillation of deflector 274 is supplied to/from (depending on the embodiment) a lock-in amplifier to which photodetector 277 is coupled, as illustrated by the dashed arrow 278.
  • a beam that is incident on the wafer is same as the beam that is generated by a laser.
  • there is no intermediate secondary source for the incident beam which is opposite to the teaching of Jiang et al. because Jiang et al. require an emitter 14 to be excited by a pulse P1 (from an optical source 12.
  • the portion of the reflected beam from wafer 230 that is detected by photodetector 277 is directly converted to electrical signal indicative of the properties of wafer 230.
  • there is no intermediate optical element such as an electro-optic sensor that is utilized in conjunction with a second beam to facilitate the detection of the probe beam.
  • This detection method is in contrast to the teaching of Jiang et al. because Jiang et al. require the probing pulse to be detected via its interaction with a second pulse P2 in an electro-optic sensor 32 made of a 2 mm thick ⁇ 110> ZnTe crystal.
  • a specific laser of FIG. 2J that is to be used in a measurement of the type described herein is selected in some embodiments by appropriately positioning dichroic mirror 272 along the y axis, at a predetermined location opposite to the selected laser.
  • mirror 272 is illustrated as being located opposite to laser 271 C, although mirror 272 may be moved (along a rail that is not shown) to another location opposite to. laser 271 B or to yet another location opposite to laser 271 A, thereby to select the wavelength to be used in illuminating the front surface of the wafer.
  • each mirror is used to reflect a beam only when its respective laser is turned on, by using dichroic mirrors with different transition or cut-off wavelengths.
  • dichroic mirrors with different transition or cut-off wavelengths.
  • lasers 271 B when laser 271 B is in use, lasers 271 A and 271 C are turned off, and a mirror 272B (not shown) located opposite to laser 271 B reflects the beam from laser 271 B along the y axis, while a corresponding mirror 272C (FIG. 2J) in the path of such a beam simply passes this beam therethrough.
  • lasers 271 A, 271 B, and 271 C must be arranged in the order of decreasing wavelength, i.e. laser 271 A has the longest wavelength ( ⁇ 1), laser 271 C has the shortest wavelength ( ⁇ 3), and laser 271 B has a wavelength in between ( ⁇ 2). And then mirror 272B is designed to reflect wavelength ⁇ 2 but transmit wavelength ⁇ 1, whereas mirror 272C is designed to reflect wavelength ⁇ 3 and transmit both wavelengths ⁇ 1 and ⁇ 2. Mirror 272A, being the first one in the series, could be chosen to reflect all wavelength.
  • a beam that is oscillated forms spots at positions 309a and 309b (FIG. 3A) on the wafer at two different times which occur at the two ends of beam oscillation along a straight line L.
  • a scanning distance Dscan between the spot positions 309a and 309b is selected to be sufficiently large so that the spots do not overlap one another (at the extreme ends of the oscillation). Therefore, the scanning distance Dscan is selected to be (in these embodiments) greater than a diameter of the two spots, namely Bdiameter.
  • line L is selected to be perpendicular to a boundary 302 between doped region 301 and undoped region 303. Note that since the undoped region 303 is co-extensive with body 306, and for this reason region 303 is shown not hatched in FIGs 3A and 3B.
  • the width and length of doped region 301 of a test structure as shown in a plan view in FIG. 3A are selected to be sufficiently large for region 301 to completely enclose the spot at position 309a (which is formed at an extreme end of oscillation, as described in the previous paragraph). Specifically, Rwidth and Rlength are selected to be greater than the Bdiameter.
  • Such a test structure is used in some embodiments wherein dimensions of regions of transistors of integrated circuits are much finer than Bdiameter.
  • doped region 301 is identical to other doped regions that are normally formed as portions of integrated circuits that are to be diced from the wafer for use in the normal manner. - ⁇
  • depth of the doped region 301 is same as a corresponding depth of such integrated circuit doped regions (such as junction depth of source/drain regions or channel regions of a transistor).
  • junction depth is used to indicate the distance from the front surface to which dopants have been implanted in case of a junction (i.e. distance of interface 304 (FIG. 3B) from front surface F)
  • profile depth is used in a more generic situation to indicate a corresponding depth for any implant or other region that may or may not form a pn junction.
  • test structures may be formed and evaluated to obtain a measure of corresponding properties of the regions to be evaluated (e.g. one test structure may be formed for evaluating source regions whereas another test structure may be formed for evaluating well regions).
  • test structures in relation to the number of integrated circuits (e.g. there may be a couple of test structures for each integrated circuit to evaluate the source and well regions therein, or a pair of test structures may be used to evaluate the source and well regions of a group of integrated circuits).
  • an analog electrical signal is generated by a photodetector, and this signal has a number of characteristics that depend on the various dimensions of the test structure and on the frequency of oscillation f.
  • the duration tmax for which the signal remains at its maximum value S1 is determined by the duration for which a spot at position 309b remains completely within undoped region 303 (at any time between t1 and t2 in FIG. 3C).
  • signal starts dropping (at time t2 in FIG. 3C).
  • the drop in signal continues while the beam straddles the boundary 302, until the beam has completely crossed the boundary 302 (at time t3). Thereafter, signal remains steady at its minimum value S2 while the beam reaches its extreme end of oscillation at spot position 309a, reverses the direction of oscillation to approach boundary 302 (along line L in the direction from right to left). At this time, signal starts to rise, and continues to rise while the beam straddles the boundary 302, until the beam has completely crossed the boundary 302 (at time t5). Then signal remains steady at the maximum value S1 while a spot at position 309b formed by the beam is completely enclosed in region 303.
  • the just-described reflectivity change measure i.e. Samplitude
  • certain preset maximum and minimum limits on such a reflectivity change measure which may be predetermined from a reference wafer or from a model
  • an increase in Samplitude (as shown in FIG. 3C) from a structure known to be good (as shown in FIG. 3B) can indicate any of (a) higher doping concentration, (b) deeper profile depth, and (c) less abrupt doping profile.
  • a total of three values of the reflectivity change measure (in the form of Samplitude) are obtained using each of three laser beams of different wavelengths.
  • the three values of the reflectivity change measure are used to obtain measures of peak (Nj) of the dopant profile, depth (zj) of the dopant profile, and abruptness (a) of the dopant profile.
  • Nj peak
  • zj depth
  • a abruptness
  • the signal generated by the photodetector contains additional information other than the reflectivity change measure Samplitude.
  • the maximum signal value Smax, and the minimum signal value Smin are respectively indicative of the reflectance of the two regions where the measurements are made.
  • the duration tmax for which the signal stays at its maximum value, and the corresponding duration tmin for which the signal stays at its minimum value can also be compared with each other. As shown in FIG. 3E, if tmin is greater than tmax, the beam is spending more time in the undoped region than in the doped region (i.e. unequal beam scan).
  • the duration trise (FIG. 3F) during which the signal rises to its maximum value from its minimum values can be indicative of a property of the boundary.
  • trise of a production wafer is greater than trise of a reference wafer, then this is an indication that the boundary between the doped region and the undoped region of the production wafer (FIG. 3F) is more diffuse than the boundary of the reference wafer (FIG. 3C).
  • the beam scan distance Dscan (FIG. 3A) across the doped and undoped regions is preferably greater than the diameter Bdiameter (which is around 2 microns in one example) of a spot formed on the wafer, in order to maximize the accuracy in detection of signal Samplitude. Therefore, in one example, Dscan is 4 microns.
  • the scan distance is reduced in any region, the beam spends less time with its spot fully enclosed in that region, thereby to reduce the duration for which the signal remains at its maximum value or minimum value.
  • the value of Samplitude is reduced (i.e. smaller than S1-S2).
  • the fluctuating signal has a reduced amplitude S5-S4 which in turn results in reduced sensitivity (due to truncated Samplitude) to the change in reflectivity between the regions.
  • some embodiments generate such a truncated oscillating signal (which oscillates between S4 and S5 as shown in FIG. 3G although the true maximum and minimum are at S2 and S1 respectively), most embodiments are designed to allow the oscillating signal to reach its full maximum value and full minimum value (by ensuring that the beam's spot is at least momentarily enclosed fully within each region by making the regions sufficiently large).
  • the lower limit on the scan distance Dscan depends on the smallest value of Samplitude that can still be detected and correlated to the change in reflectivity between the regions, which in turn is determined by the noise of the system.
  • the minimum detectable value of Samplitude is set at a signal-to-noise ratio of 1.
  • a beam is scanned between two adjacent regions
  • the beam may be scanned between two regions that are separated from one another, e.g. by one or more other regions.
  • FIG. 3H illustrates the scanning of a beam between extreme positions 388a and 388d that are respectively located over an undoped region 383L and a doped region 381 R, and two additional regions 38 L and 383R are present therebetween.
  • the beam is swept through positions 388b (at which the beam is shown currently located) and 388c that are respectively located over doped region 381 L and undoped region 383R.
  • a fluctuating electrical signal generated by a photodetector in the just-described embodiment has a period that is one-third of the period of oscillation of the beam, as illustrated in FIG. 31. Therefore, although in certain embodiments the frequency of oscillation of the beam (also called “beam-scan frequency”) is same as the frequency of fluctuation of the electrical signal (also called “signal frequency”), in other embodiments the two frequencies can have a non-identical relationship (i.e. need not be equal).
  • a signal frequency that is an integral multiple of the beam-scan frequency (in FIG. 31, one doped region 381 L and one undoped region 383R are together considered a single pair that intervene between regions 383L and 381 R).
  • the test structure is designed with the width of the lines larger than or equal to Bdiameter.
  • a fluctuating component in signal may be identified by digitally sampling the photodetector's output signal and a computer analyzing the samples to identify Samplitude at the appropriate frequency (which is based on the frequency of mechanical oscillation of the beam relative to the test structure and also based on a pattern of the test structure across which the beam is swept).
  • a pair of lines instead of a pair of lines, a single line of a different material (than the two regions) may intervene, in which case the electrical signal has three steps (horizontal flat regions) instead of two steps.
  • regions 301 and 303 have been described above as being doped and undoped regions, reflectivity change across regions of other materials and/or dimensions can also be determined by acts 202-206 described above.
  • FIG. 3J illustrates measurement (by acts 202-206) of reflectivity change between a lightly doped region 303a and doped region 301.
  • FIG. 3K illustrates measurement (by acts 202-206) of reflectivity change between a shallow doped region 303b and doped region 301.
  • a change in the Samplitude measurement between successive wafers being fabricated is used for process control.
  • the Samplitude measurement may be directly compared against predetermined limits, and if not within the range then a change is made in a process control parameter (also called "process parameter") used in fabrication of regions 303b and 301.
  • a process control parameter also called "process parameter”
  • the unknown properties in regions 303a and 303b may be looked up using Samplitude data obtained from reference samples with known properties.
  • FIG. 3L illustrates measurement (by acts 202-206) of reflectivity change between undoped region 303 and a region 301a that is formed by a layer 391 of dielectric or conductive material (e.g. gate oxide or metal).
  • a layer 391 of dielectric or conductive material e.g. gate oxide or metal.
  • the front surface F of the wafer is not planar, because the just-described layer 391 is formed over a planar surface 304a of the semiconductor material that forms the substrate (substrate 307 is explicitly shown in FIG. 3N). Therefore, front surface F of the wafer is planar in region 303 of the substrate, discontinuous at boundary 302a and planar again over region 301 a of the dielectric/conductive layer 391.
  • the vertical distance between the two planar regions 301a and 303 is thickness z ⁇ of layer 391.
  • Surface 304a forms an interface between substrate 306 and layer 391.
  • a change in thickness z / . of layer 391 from one wafer to another wafer during semiconductor wafer fabrication results in a corresponding change in the Samplitude measurement which is used for process control in some embodiments.
  • the thickness of layer 391 is much less than the depth of focus of the incident beam and hence the magnitude of a change in thickness z L of layer 391 has negligible effect on beam focusing.
  • the incident beam and the reflected portion are both normal (i.e. perpendicular) to the front surface of the wafer (see line 214 normal to surface F in FIGs. 2D-2H), in other embodiments, there may be an angle langle (FIG. 3M) which is other than 90°, between an axis 314 of the incident beam at position 309b and surface F of wafer. In such a case, a photodetector 308b is oriented at the same angle to receive the reflected portion, as will be apparent to the skilled artisan in view of this disclosure.
  • the regions being evaluated are of a test structure, in other embodiments the regions of an integrated circuit are directly evaluated as illustrated in FIG. 3N, in view of this disclosure.
  • regions of an integrated circuit can be directly evaluated depends on a number of factors such as the relative dimensions of the regions and a spot formed by the incident beam.
  • the spot formed on the wafer in turn depends on, for example, the wavelength of the beam, and the focal length of an objective lens that focuses the beam on the wafer.
  • the contribution of an extension region 302s (which is located between a doped region 301s and an undoped region 303) to the measurement is ignored, by filtering out any components in the analog electrical signal that are not at the frequency of relative movement (between regions 301s and 303), e.g.
  • test structure reflectance is used, for example, to estimate the depth of a doped region in a transistor, e.g. by forming a test structure of doped regions of the same dosage as in the transistor.
  • the estimates can be of a dimensional property (such as junction depth which is one example of profile depth) and/or a material property (such as peak dopant concentration) and/or a mixed dimensional/material property (such as the gradient of dopant concentration which is also called profile abruptness).
  • a dimensional property such as junction depth which is one example of profile depth
  • a material property such as peak dopant concentration
  • a mixed dimensional/material property such as the gradient of dopant concentration which is also called profile abruptness
  • Estimates of the type described herein can be made by comparing an attribute of a signal measured on the test structure of a wafer under fabrication with a corresponding structure on a set reference wafers having a set of known properties forming a calibration or look-up table.
  • use of a test structure in a production wafer as described herein provides a convenient method for determining various such properties because one or more dimensions of the test structure need not be limited to dimensions of the transistor or other devices to be used in an integrated circuit (being formed).
  • Several embodiments use the measurement to control ion implantation and/or annealing processes for forming such regions in the next production wafer, e.g. by changing one or more process parameters such as implanted dose, annealing temperature, and thermal exposure time.
  • Embodiments of the type described herein have numerous advantages over the prior art methods. From the physics point-of-view, the method/apparatus of certain embodiments of the invention measure the active doping profile, since the measurement is sensitive to the majority carrier distribution, which closely follows the active doping profile. Some embodiments extract the following three properties of a semiconductor wafer resulting from doping: profile depth, profile abruptness, and peak doping concentration. From the hardware set-up and stability point-of-view, such embodiments of the invention offer easier optical beam alignment and long-term stability over the prior art. FIGs. 4A and 4B illustrate the use of certain embodiments in process control, to control the wafer fabrication process in a feedback loop. Specifically, one or more test structures are formed in a production wafer, e.g.
  • test patterns are preferably identical in doping characteristics to a transistor layer that is of interest. For example, if source and drain formation processes used to form regions 101 and 102 are to be controlled, then the same processes are used to form the test structure(s).
  • test structures may be open areas several microns on a side situated in open areas such as the streets between integrated circuits, and may be formed simultaneously with the formation of source/drain 101 and 102.
  • junction depth of doped regions of a transistor is to be estimated, and the test structure formed in act 401 requires implantation of dopant atoms and annealing, and depending on the embodiment such implantation and annealing may be performed simultaneously with implantation and annealing for formation of transistors of the wafer.
  • a wafer containing the test structures is aligned to a measurement system (see act 402 in FIG. 4A), followed by measurement of a signal indicative of a property of the doped region, using a non-contact probe (see act 403 in FIG. 4A).
  • a measurement system see act 402 in FIG. 4A
  • a non-contact probe see act 403 in FIG. 4A.
  • acts 402 and 403 may be repeatedly performed, e.g. for multiple test structures as illustrated by act 408, and may be interleaved with or performed simultaneously with other kinds of measurements as would be apparent to the skilled artisan.
  • Act 403 may be performed by any method described herein, e.g. by performing acts 202-206 (FIG. 2A).
  • the signals from the measurements performed in act 403 are optionally processed to convert the signals into a more meaningful quantity related to one or more properties of the doped region, such as the peak, the depth and the abruptness of a profile of the implanted dopants (see act 404 in FIG. 4A). As noted elsewhere herein, such conversion is not necessary in some embodiments wherein the raw measurements are directly used in act 405 for process control (described next).
  • the results of the analysis in act 404 are compared with predetermined control limits (see act 405 in FIG. 4A), and if the measurements fall within the limits, fabrication of the wafer is continued (see act 407 in FIG.
  • a measurement tool 413 (FIG. 4B) is co-located with other wafer fabrication tools, such as an annealer 414, an ion implanter 411 , a patterning tool 410 and an oxide mask removal tool 412.
  • a wafer 430 (FIG. 4B) may enter a patterning tool 410, wherein patterns associated with the source and drain extensions of to-be formed transistors, and also doped regions of one or more test structures are formed on wafer 430.
  • wafer 430 is inserted into an ion implanter 41 1 wherein dopant atoms are implanted to form, for example, doped region 21 1 (FIGs. 2B and 2C) of a test structure 210 as well as one or more doped regions 221 of one or more integrated circuits 220 in the wafer.
  • the implant mask is removed by tool 412 (FIG. 4B), and the test structures in the wafer are evaluated by measurement tool 413 as described above in reference to act 403. Therefore, measurement tool 413 contains one or more items illustrated in FIG. 2J.
  • measurement tool 413 contains all of the items in FIG. 2J, oriented relative to one another in the manner illustrated in FIG. 2J, and in addition also contains a stage that is located adjacent to and to the right side of optical bench 279 (FIG. 2J). Measurement tool 413 of such embodiments also contains a mirror to deflect the beam by 90° (regardless of its position being at 275A or at 275B or anywhere therebetween), and an objective lens that is located between the mirror and the wafer 230 (FIG. 2J). After act 403 (FIG. 4A) is completed in measurement 413 (FIG. 4B), and if the wafer is found to have the implant properties within predetermined limits, the wafer is further processed.
  • the wafer is annealed in a rapid thermal annealer 414, and afterwards the test structures in the wafer are evaluated again by measurement tool 415 which is similar or identical to the above-described measurement tool 413. Since measurement tool 415 is used only after the wafer has been annealed, it is also referred to as anneal measurement tool whereas tool 413 is also referred to as an implant measurement tool.
  • measurement signals generated by units 413 and 415 are supplied on a bus 420 that is connected via connection 422 to ion implanter 411 , and via connection 421 to annealer 414, thereby to provide feedback signals to these tools 411 and 414.
  • the measurement signals on bus 420 may be provided via connection 423 to a computer 416 that is mounted on a factory floor adjacent to tools 411 , 412 and 414 as well as units 413 and 415 (that are also mounted to the same floor). Due to being located on the factory floor, computer 415 is also referred to herein as a factory computer.
  • Factory computer 416 may archive the measurement signals from units* 413 and 415 for later correlation to electrical performance of electronic devices being fabricated on wafer 430. If the , measurements are outside the control limits, factory computer 416 can provide corrections to ion implanter 411 and annealer 414 to return the process back to the control limit. As noted above, the feedback to ion implanter 411 and annealer 414 may also be provided directly from either or both of measurements units 413 and 415.
  • FIGs. 5A and 5B show in cross-sectional views one example of such a test structure (which forms only a portion of the wafer and it is understood that the remainder of the wafer contains at least one or more integrated circuits and may contain additional test structures).
  • the test structure includes an ion-implanted region 513 (one region is shown by way of example, although more may be used depending on other considerations, such as the ability of the measurement system to align to a pattern, or space limitations within the integrated circuit die).
  • Region 513 of the test structure typically has a width Rwidth of >20 ⁇ m, while its length Rlength in the direction normal to the cross section (see FIG. 3A) may also be on the order of 20 ⁇ m, but may in some cases be as large as 500 ⁇ m.
  • an ion implanter used to form region 513 has exactly the same energy and dose as the ion implanter used to form source/drain extension regions of a transistor that is present (as a part of the normal circuitry) in the semiconductor wafer.
  • the region 513 is formed in the same ion implantation step used to form the source/drain structures on the transistors, ensuring that the ion implant in the transistors and test structures are identical.
  • the energy and dosage for a test structure may be selected to be the same as the MOS transistor for two reasons: first, it best represents the real transistor doping and second it requires no additional process steps.
  • the dimensions of the test structure(s) are sufficiently larger than the dimensions of the regions of transistor(s), so as to enable a signal sensed from the test structure to be evaluated in the manner described herein.
  • dimensions of the test structures are an order of magnitude or more larger than transistor regions because the transistor regions are normally printed much finer than the diameter of a spot formed at the surface of the wafer by a beam of illumination incident thereon during evaluation of the type described herein.
  • the dimensions of the test structure regions are selected to be of the same order of magnitude as the beam diameter (and depending on the embodiment, dimensions of test regions may be slightly smaller or slightly larger than the beam diameter).
  • the dopant atoms may be implanted to a very shallow depth, typically ⁇ 50 ⁇ A.
  • the implant parameters are the implant specie (B, As, P, Sb, etc), the energy (0.2 to 2 keV typically), and the dose (1x10 14 to 3x10 15 atoms/cm 2 typically).
  • the anneal parameters are typically the temperature (on order of 1000°C), time (instantaneous to 10 sec), ramp-up rate (50 to 200°C/sec) to temperature and ramp-down time (same as ramp-up rate). After annealing, the size of the implanted region increases due to diffusion, as shown in FIG. 5B.
  • region 513 is shown with dashed lines as region 514a in FIG. 5B, and after annealing it becomes larger and is labeled as region 514b. Also, in FIG. 5B the ion implantation mask 511 has been removed, which was done prior to annealing.
  • FIGs. 6A-6E A process for making a test structure of the type described above in reference to FIGs. 5A and 5B is illustrated in FIGs. 6A-6E, for one embodiment.
  • a photoresist layer 611 is applied to the surface of substrate 606 (FIG. 6A).
  • the photoresist layer 611 is patterned by exposing and developing the resist, creating a hole 612, in the photoresist layer 611 (FIG. 6B) for the test structure.
  • the just-described acts are also used simultaneously for the creation of one or more portions of a transistor in the silicon wafer. For example, the source and drain regions, and extensions thereof may be formed simultaneously with formation of region 613, depending on the embodiment.
  • layer 611 has holes (such as hole 602) at the locations of the to-be-formed regions of the transistors, in addition to the hole 612 required for forming the test structure. Hole 602 typically has smaller dimensions than hole 612, depending on the embodiment.
  • all of the regions of the various transistors in a wafer may be formed by acts separate and different from the just-described acts for formation of a test structure, again depending on the embodiment.
  • Ion implantation is applied, to form region 613 beneath the hole 612 (and beneath any additional holes 602 that may be present for the formation of transistors as noted above).
  • Photoresist layer 611 blocks the ion implantation elsewhere (FIG. 6C).
  • Photoresist 611 is then removed, leaving implanted region 613 in substrate 606 (FIG. 6D), and any additional regions (such as region 603) for the transistors.
  • the wafer is annealed, causing diffusion of the implanted region 613 resulting in an expanded doped region 614 (FIG. 6E). Any additional region 603 if present in the wafer also experiences diffusion during annealing, resulting in a correspondingly expanded doped region 604.
  • a portion of the adjacent doped region 614 of the test structure in a semiconductor wafer is illuminated with a laser beam. Initially, when the beam is incident on region 614, it reflects off the surface of the wafer because of index of refraction difference between the air and the semiconductor material. During oscillation of the beam into an adjacent undoped area, the optical reflection occurs only at the surface of the wafer as described next.
  • FIG. 7A illustrates a beam 704 incident on a front surface F of a wafer in an undoped or low doped region 701 , a portion 705 of the beam 704 being reflected from surface F, and another portion 710 being transmitted into and absorbed in region 701.
  • the intensity of reflected beam portion 705 is given by:
  • the optical reflection consists of an interference between two reflection components; one component 706 (FIG. 7B) comes from the front surface F of the wafer.
  • the other component can be either (1 ) component 708 which comes from interface 703 (between doped layer 702 and the underlying region 701 ) or (2) component 709 which comes from every location throughout doped layer 702 having doping concentration gradient.
  • FIG. 7B illustrates the reflection process in detail. For the moment, reflection component 709 is ignored; it will be incorporated into the analysis later on.
  • reflection component 709 is ignored; it will be incorporated into the analysis later on.
  • incident beam 704 encounters the surface F of the sample, part of it is reflected as beam component 706, and part of it is transmitted into doped layer 702 as beam 707.
  • Beam 707 is reflected from interface 703 and passes through front surface F to emerge as reflected beam component 708.
  • ⁇ 7 _ 2 T o ⁇ 9 >- N and z,- are the doping concentration and thickness (junction depth) of the doped layer, respectively, e is the electron charge, m is the carrier effective mass in the semiconductor material, ⁇ 0 is the permittivity of free space, c is the speed of light in vacuum, and ⁇ 0 is the wavelength of the measuring beam.
  • the amplitude of reflection from the junction (component 708) is ⁇ .
  • Both the front surface reflection component 706 and the interface reflection component 708 from the doped region contain information about the properties of the doped layer, since the free carriers contributed by the dopant atoms after annealing modify the refractive index of the semiconductor material.
  • the surface reflection component 706 contains information about the peak doping concentration, while the interface reflection component 708 contains additional information about profile depth and abruptness of the profile. Therefore, in several embodiments, by comparing the reflectivity of the doped and undoped region, the aforementioned properties of the doped region are measured.
  • FIG. 8A depicts a simplified profile 801 of doping concentration as a function of depth (z).
  • the graph in FIG. 8A models adequately real doping profiles in production wafers, following ion-implantation and thermal-annealing process.
  • the doping concentration is constant and maximum near the surface (N d ), and then it gradually drops to the background doping concentration N s .
  • processes used to form background doping concentration N s are well controlled using any prior art method, and therefore, the value of N s is known.
  • the slope of the doping concentration profile 801 is commonly referred to as the abruptness of the profile, and it is an important property of a semiconductor wafer, in addition to the peak concentration N d and the profile depth z j .
  • the profile depth Z j is defined as the distance from the front surface F to a location in the interface 703 (FIG. 7B) where the concentration has dropped to a level halfway between N d and N s (FIG. 8A); whereas the profile abruptness is defined as the slope (N d - N s )/2a (see the slanted portion of profile 801 in FIG. 8A).
  • FIG. 8B shows the derivative of this simplified doping profile 801. As seen in FIG. 8B, the slope is zero everywhere except in the range 2a centered around profile depth Z j .
  • the signal of interest is the change in amplitude between reflected signals from the doped and the undoped regions.
  • Equation (13) indicates that the measurement of Samplitude is dependent on three properties of the doping profile, namely the profile depth Z j , the maximum or peak doping concentration N d , and a, which is a property associated with profile abruptness. Note that Equation (13) includes the interference effects of reflected components 706, 708 and 709 (FIG. 7B).
  • FIG. 9A shows the measurement of Samplitude (normalized by dividing it with Smax, wherein Smax is the maximum signal value as shown in FIG. 21) plotted along the y-axis as a function of profile depth for a silicon substrate material plotted along the x-axis, when probed using a beam of wavelength 980nm.
  • Smax is the maximum signal value as shown in FIG. 21
  • three graphs 900A-900C are plotted in FIG. 9A for three combinations of the peak doping concentration N d and the abruptness property a (see legend at the top of FIG. 9A).
  • the abruptness property a is defined in FIG. 8A as being one half of the distance around profile depth Z j wherein the doping concentration N d changes in substantially linear manner as a function of depth.
  • the peak doping concentration primarily modifies the amplitude of the cosine curve, while the abruptness property modifies both the amplitude and the offset.
  • the effect of the profile abruptness is quite small (as compared to the effect of a change in dopant concentration).
  • a correlation curve A relationship between signal measurement (i.e. Samplitude) and profile depth, such as shown in FIG. 9A, is called a correlation curve, since it can be used to convert measured signals from unknown samples to their respective profile depths, provided that the other two properties N d and a are known.
  • a measured signal of 0.01 corresponds to a profile depth of 393A for the unknown sample.
  • the smaller depth is the correct answer because in most embodiments the implant process does not generate implants of the large depth 100 ⁇ A, because the wafer is undergoing formation of Ultra Shallow (US) junctions. Normally, in process control embodiments this ambiguity is not an issue because initially the wafer fabrication process starts with fabrication of wafers having the smaller implant depth, and any deviations therefrom are measured (and corrected as appropriate).
  • the correlation curve can be constructed using a set of samples, called calibration or reference samples, having known doping profiles. The doping profiles are characterized using an independent method such as SIMS or spreading resistance measurement.
  • Three wavelengths require three A coefficients, and they are obtained from a calibration procedure using a single calibration sample.
  • One patterned wafer is prepared with known active doping concentration (N d ), profile depth (zj) and abruptness property (a). These properties are collected from a combination of SIMS analysis and sheet resistivity measurement.
  • the calibration wafer is measured successively using each of the three selected wavelengths of the probing beam. For each wavelength, the coefficients is calculated using equation (13).
  • the undoped area with known background doping concentration N s are used to calibrate the A coefficients.
  • the signal from the undoped area is a DC signal given by:
  • the coefficient A DC may be different from the A coefficient in equation (13) due to the difference in amplifier gain between the AC and the DC signals.
  • A is calculated from Anc, since the optical transmission loss and the photodetector conversion efficiency factors are the same.
  • the calibration proceeds in the same way as with a calibration sample, the undoped area is measured successively using each of the three selected wavelengths, and for each wavelength, the coefficient A DC is calculated using equation (14). Once all the coefficients in equation (13) are determined, the measurement of the doping properties of an unknown wafer proceeds by sequential probing of a test structure containing the doped region at three pre-selected wavelengths.
  • linear portions 901 A, 901 B, and 901 C of the cosine response curve corresponding to the three wavelengths indicate the values Si, S 2 , and S 3 that are measured for a given (unknown) set of profile depth (zj), peak doping concentration (N d ), and profile abruptness (a).
  • zj profile depth
  • N d peak doping concentration
  • a profile abruptness
  • data on the behavior of Samplitude as a function of profile depth are collected ahead of time prior to evaluation of a production wafer.
  • the above- described method of oscillating a beam source and detector is performed on doped and undoped regions on a number of reference wafers to obtain the values (Samplitude/Smax) for a corresponding number of different profile depths Z
  • a plot of the graph from such measurements is illustrated in FIG. 9C. Note that the measurements that are plotted in FIG. 9C are for a given set of all other properties and measurement conditions, including a given wavelength of the probe beam.
  • an Samplitude measurement on a production wafer (generated by oscillating the beam source and detector) is used to look up the profile depth, and if the resulting profile depth falls outside a predetermined range, then an appropriate change is made in a fabrication process.
  • a pre-anneal measurement is made to determine the dosage, and only wafers that are within predetermined tolerance limits are further processed.
  • the profile depth and profile abruptness are both changed, although the impact on profile depth is larger than the impact on profile abruptness. Therefore, assuming that profile abruptness has not changed, any change in the measured Samplitude is treated as indicating a corresponding change in profile depth.
  • the look-up table approach is utilized to monitor variations in any of the three properties Z j , N d , and a, without knowing beforehand which property is changing.
  • a set of reference samples with known variations in all three properties Z j , N d , and a, clustered around a center (nominal) process condition, are measured using three laser wavelengths to construct the look-up table.
  • the data from the look-up table can be plotted, and if all three properties are used, the resulting plot is a 4-dimensional surface for the signal for each wavelength.
  • FIG. 9D shows portions of such 3D surfaces for two different wavelengths of the measuring beam (since two properties only requires two wavelength measurements).
  • the signal ratio (Samplitude/Smax) is plotted on the Z-axis, as a function of the peak doping concentration and the profile depth on the X and Y-axes, respectively.
  • Surfaces 921 and 922 are constructed from measuring a given set of reference samples using illumination of two respective wavelengths ⁇ 1 and ⁇ 2, where ⁇ 1 is larger than ⁇ 2.
  • the one reference sample with the nominal process condition is represented as cross 923 on the XY-plane in FIG. 9D, with concentration N d c and profile depth ZJC-
  • This nominal reference sample produces signals S1 and S2 on surfaces 921 and 922 respectively.
  • the process condition moves from cross 923 to cross 924 on the XY-plane.
  • the signals shift from S1 and S2 to a new pair of signals S1a and S2b on surfaces 921 and 922, as indicated by arrows 926 and 928 respectively.
  • the signal difference at the two wavelengths are (S1 a-S1 ) and (S2a-S2) for the doping concentration shift, and (S1 b-S1 ) and (S2b-S2) for the profile depth shift.
  • the ratio of the difference signal at the two wavelengths, (S1a-S1)/ (S2a-S2) or (S1b-S1)/(S2b-S2), identifies the process parameter and/or property of the wafer that is causing the shift, e.g. as discussed below.
  • the look-up table constructed by the measurements on a set of reference samples, contains the information on the size of the signal difference ratio that can be expected to enable the identification process. FIG.
  • 9E further illustrates a step-by-step procedure that is used in some embodiments to implement process monitoring described in the previous paragraphs.
  • a set of reference wafers are fabricated with known variations in zj and Nd, clustered around a center (nominal) process condition, as illustrated in act 931 of FIG. 9E.
  • the reference wafers are then measured at multiple wavelengths, in this case 2 wavelengths (see act 932 in FIG. 9E).
  • a set of look-up tables is then generated, as indicated by act 933 of FIG. 9E. Examples of such look-up tables are shown in Tables 1 and 2 below, for measurements at 670nm and 980nm wavelengths.
  • Table 1 Look-up table generated by a set of reference samples with variations in profile depth Z j .
  • Table 2 Look-up table generated by a set of reference samples with variations in peak doping concentration N d -
  • the "Change in Signal” columns refer to the signal difference as a result of a property shift from the nominal condition.
  • the "Ratio of Signal Change” in the look-up table refers to the difference signal at 670nm divided by the difference signal at 980nm.
  • Table 3 Measurement results of five production wafers with identification of the property that has changed based on the ratio of signal change.
  • such a look-up table approach is also capable of estimating the amount of shift in the out of spec property by performing a data interpolation using the look-up table values (see act 938 in FIG. 9E). Using such a shift, an appropriate adjustment is made in the fabrication process (see act 939 in FIG. 9E).
  • a feed- back loop may be implemented by incrementally changing a process control parameter during fabrication of production wafers, and finding an incremental improvement in a corresponding property.
  • act 938 is not performed in some embodiments, i.e. the precise amount of change in a property of the wafer need not be determined in embodiments wherein act 939 is performed directly after act 937. For example, if the calculated ratios fall within a predetermined range of ratio values then a table of predetermined data (which is determined experimentally) may be used to identify the change to be made in one or more process control parameter(s).
  • wafer #3 has signal values of 166.4 and 217.1 at
  • the method as described above can be extended by using more than two wavelengths for the measurement, and considering all the possible combinations of the signal difference ratio (or Ratio of Signal Change). Specifically, if three different process parameters may change, then three measurements are made on each reference wafer (one with each of three probe beams of different wavelengths), and three different signal changes are computed, and then three different ratios are computed. Thereafter, during wafer fabrication, each production wafer is the subject of three measurements at the three wavelengths, and once again three different ratios are computed, and they are used to identify the process parameters that have changed (again by identifying whichever ratio falls within the range of ratios in a table).
  • the signal difference ratio or Ratio of Signal Change
  • the measurements on each reference wafer may be made at sufficiently close intervals in the change of each process parameter, to generate a sufficiently large "cluster" of measurements centered around nominal values (of the process parameters) so as to ensure that a single process parameter change during fabrication of production wafers is easily identified by such tables.
  • FIG. 10A shows the screenshot of a commercial software, Mathcad 2001 i Professional, which is programmed to solve the three simultaneous nonlinear equations.
  • Block 1001 defines the equation that needs solving, i.e. equation (13), and its associated components. Some definitions for the constants are not shown in FIG. 10A, but they are defined elsewhere within the program.
  • Block 1002 specifies certain initial guess values for the unknown variables Na, z, and a.
  • Block 1003 is the "solve block", consisting of a set of constraints (block 1004), the solving function "Find” and the solutions (block 005).
  • Block 1004 forms the set of three simultaneous equations to be solved, with the values 5.019x10 "3 , 4.404x10 "3 , and 3.836x10 "3 corresponding to the signals Si, S 2 , and S 3 , respectively.
  • Block 1005 displays the solution to the unknown variables N , z, and a as 8.028x10 19 /cm 3 , 399A, and 102.7A, respectively, for this hypothetical unknown sample.
  • Full text of an example of the MathCAD program shown in FIG. 10A is provided as an illustration on the following page.
  • Measurement lasers 1201a, 1201b, and 1201c are semiconductor diode lasers with wavelengths that can range anywhere from 400nm to 110Onm. Typical values of the laser wavelength that may be used are 405nm, 635nm, 730nm, 830nm, and 980nm. The maximum output power of the measurement lasers is around 10OmW.
  • the output beams of the lasers 1201 a, 1201b, and 1201c are collimated with collimating lenses 1202a, 1202b, and 1202c, providing collimated beam 1203. Only one of the three lasers is operated at any given time. Not shown is a mechanical shutter in front of each of the lasers to block the beams when they are not in use.
  • a dichroic mirror 1210 (such as a partially transmissive mirror, e.g. part number 1918-b available from Dominar of Santa Clara, Calif.) is used with one of the lasers, to select a beam 1211 that is to be incident on a surface of the wafer.
  • dichroic mirror 1210 is mounted on a small movable platform, and moved to an appropriate location to select the wavelength of beam 1211.
  • Beam 1211 then passes through 50:50 non-polarizing beam splitter 1212 (e.g. part number 10BC17MB.2 from Newport Corp., Irvine, CA) which is also referred to as a "detection system beam splitter” and a 90:10 beam splitter 1213 which is also referred to as a "vision system beam splitter”.
  • vision system beam splitter 1213 is available from Precision Applied Products of Fullerton, Calif., by specifying 93.3% transmission at 0.83 microns wavelength and 90% transmission at 1.48 microns wavelength. Note that the specifications for the vision system beam splitter 1213 are selected based on the wavelength of the probe beam to ensure that a majority of the power is transmitted therethrough.
  • a 90:10 beam splitter is used in some embodiments as being used as a vision system beam splitter, other embodiments may use beam splitters that split the beam in different ratios, such as 95:5 or 80:20.
  • a 50:50 beam splitter is used in some embodiments as a detection system beam splitter, other embodiments may use beam splitters that split the beam in different ratios, such as 60:40.
  • the power transmitted through beam splitter 1213 thereafter reaches a beam deflector 1215 which may be, for example, an acousto-optic beam deflector or a galvanometer mirror as described elsewhere herein.
  • An oscillating beam from deflector 1215 then passes through an objective lens 1216 (such as a 100X, 0.8 NA lens made by Olympus of Tokyo Japan), and is thereafter incident on wafer 1217.
  • Deflector 1215 receives a modulation signal from lock-in amplifier 1231 via connection 1235, which causes beam 1211 to be deflected slightly in a periodic manner about the propagation axis.
  • the periodic deflection of beam 1211 by deflector 1215 causes beam 1211 to scan along a line on wafer 1217.
  • Lens 1216 focuses beam 1211 onto the surface of wafer 1217, which is mounted on a stage 1218. With proper alignment, beam 1211 scans across doped and undoped regions of a test structure of interest.
  • the portion of beam 1211 which is specularly'reflected (in the direction of incidence which is normal) from wafer 1217 is recollimated by lens 1216.
  • Beam splitter 1213 diverts 10% of the reflected beam to lens 1226 and camera 1227, which provide a system to align the beam spot on a surface of wafer 1217 to the test structure of interest.
  • an autofocus system that consists of a pinhole and a detector, which also uses the portion of the reflected beam diverted by beam splitter 1213.
  • lens 1226 is tube lens 81845 available from-Nikon of Tokyo, Japan
  • an example of camera 1227 is a CCD camera, e.g. model 85400 available from FJW Industries of Palatine, Illinois.
  • the signal provided by camera 1227 is fed (in one embodiment) into a vision system such as model ASP-60CR-11-S available from Cognex Corporation, Boston, Mass.
  • detector 1223 is a photocell (such as a silicon photodiode, e.g. PIN-44DP from UDT Sensors, Inc. of Hawthorne, CA, USA) that converts the incident electromagnetic radiation into a current.
  • the current from detector 1223 is converted to a voltage using transimpedance amplifier 1230, the output of which goes to lock-in amplifier 1231.
  • the signal from detector 1223 to lock-in amplifier 1231 is an analog continuous electrical signal.
  • lock-in amplifier 1231 goes to a digital computer 1232, which receives the signal and presents it (or a value derived therefrom as described herein) to the user or to other data collection systems.
  • Lock-in amplifier 1231 includes a frequency reference that is used to modulate deflection of the beam by deflector 1215 via connection 1235. As noted above, the lock-in amplifier 1231 supplies to deflector 1215 a signal oscillating at a predetermined frequency that is generated by a reference oscillator included in the lock-in amplifier 1231.
  • embodiments of the type described herein can be extended to use more than three wavelengths to provide increased accuracy and/or to measure other properties. Additionally, for process monitoring purposes, measurement(s) of the type described above can be performed without any calibration to determine the actual value of a property.
  • a reference signal value is initially obtained from a reference wafer having known good properties (e.g. formed by known process parameters). Monitoring of subsequently processed wafers in a production line utilizes a Statistical Process Control (SPC) chart, in a manner well known in the art, to track deviations from the reference signal. The signal deviations are directly related to actual process deviations, and one or more process limits are used in process control. Use of previously-identified process limit(s) results in manufacture of wafers within preset manufacturing tolerances.
  • SPC Statistical Process Control
  • the apparatus is identical to the above-described apparatus illustrated in FIG. 10B except for the following difference. Specifically, deflector 1215 of FIG. 10B is not used in the apparatus of FIG. 10C. Instead, in the apparatus of FIG. 10C, stage 1218 is movable and is used to move wafer 1217 back and forth in a single direction at a fixed frequency. The signal to oscillate stage 1218 comes from lock-in amplifier 1231 via a connection 1237, as shown in FIG. 10C.
  • FIGs. 11 A and 11 B illustrate additional embodiment of this invention, that use two laser beams: a first laser beam 1109 is of the type described above (also called “probe beam”), and a second laser beam 1110 (also called “pump beam”) has photon energy greater than the bandgap of the semiconductor material.
  • the pump beam 1110 (FIG. 11B) is combined with the probe beam 1109 using a dichroic mirror (not shown in FIG. 11B) to form dual collinear beams. These two beams are thereafter moved together between doped region 1101 and undoped region 1103 of a test structure.
  • the two beams are illustrated in FIG. 11A at positions 1109a and 1109b respectively at the end points of oscillation of the probe beam, and positions 1110a and 1110b respectively for the end points of oscillation of the pump beam.
  • the pump beam has a slightly different (in this example slightly larger) spot size of diameter Bpump than the probe beam's diameter Bprobe.
  • FIG. 11B illustrates in more detail the effect of using the pump beam 1109 in addition to the probe beam 1110.
  • Portions of the incident probe beam 1109 and pump beam 1110 are transmitted into the doped layer 1101 (with thickness zj) and the underlying well region 1106 as beams 1111 and 1112, respectively.
  • the transmitted pump beam 1112 generates excess carrier distributions 1113 and 1114 inside the doped layer 1101 and the well region 1106, respectively. Due to the short carrier diffusion length inside the doped layer, the excess carrier distribution 1113 decays rapidly outside the diameter of the pump beam.
  • the carrier diffusion length is much larger, resulting in a nearly flat carrier distribution 1114.
  • the excess carrier distribution is strongly influenced by the underlying doping profile.
  • the excess carrier distribution is constant with depth.
  • the difference in reflectivity of the probe beam between two regions that are illuminated by the pump beam is now a measure of the difference in excess carrier distribution between the doped and undoped regions. Therefore, an analog electrical signal that is measured by a photodetector (based on reflectivity of the probe beam) is still related to the doping parameters such as profile depth and ,peak doping concentration. However, the signal may also be sensitive to other parameters such as oxide charge and defect densities.
  • a carrier-generation beam i.e. the pump beam
  • a reflectivity measurement of the probe beam is made more sensitive to other properties of the doped layer.
  • phase-sensitive detection of reflectivity of the probe beam enhances a small difference in reflectivity between the doped and undoped regions.
  • the two beams together form a combined beam which is oscillated between the doped and undoped regions in a periodic manner at a fixed frequency.
  • a reflection of the combined beam is filtered to pass only the probe beam reflected portion to a photodetector.
  • An analog continuous electrical signal at the output of the photodetector is periodic, and is detected by a lock-in amplifier at the frequency of and in phase with oscillation of the combined beam. The amplitude of this electrical signal is proportional to the difference in reflectivity between the doped and undoped areas, and contains information on one or more properties of the doped layer.
  • N d and N s in equation (13) are replaced by Ncs and Nc ⁇ which refer to the excess carrier concentrations at the undoped and doped regions, respectively.
  • the apparatus of FIG. 12 includes a fourth laser 1205.
  • the fourth laser 1205 is used as a carrier generation laser, an example of which is a semiconductor diode laser operating at 830nm wavelength. In this example, its maximum output power is 200mW, to generate sufficient carrier density within the samples.
  • Laser 1205 can also be used as a measurement laser. Not shown is a mechanical shutter in front of laser 1205 to block the beam when not in use.
  • the output beam of laser 1205 is collimated using collimating lens 1206, providing collimated beam 1207.
  • the above-described beam 1203 from one of the measurement lasers 1201a- 1201c is combined with beam 1207 from the carrier generation laser using dichroic mirror 1210 to create combined and collinear beam 1211.
  • dichroic mirror 1210 is mounted on a small movable stage/platform to allow wavelength selection for the measuring beam.
  • Combined beam 1211 of these embodiments then passes through 50:50 beam splitter 1212, 90:10 beam splitter 1213, deflector 1215, and objective lens 1216 (lOOX from Olympus).
  • deflector 1215 receives a modulation signal from lock-in amplifier 1231 via connection 1235, which causes the combined beam 1211 to be deflected slightly in a periodic manner about the propagation axis.
  • Lens 1216 focuses the combined beam 1211 onto the surface of wafer 1217, which is mounted on a stage 1218.
  • stage 1218 is used to move the wafer 1217 back and forth in one direction at a fixed frequency as described above in reference to FIG. 10C. Both types of embodiments are illustrated in FIG. 12, although it is to be understood that only one of the two mechanisms for providing relative motion need be used in any given embodiment.
  • the reflected beam (which includes components from the probe beam and the pump beam) is recollimated by lens 1216.
  • Beam splitter 1213 diverts 10%) of the combined reflected beam to lens 1226 and camera 1227.
  • the combined reflected beam then enters beam splitter 1212, which deflects it to an optical filter 1221 (FIG. 12).
  • Filter 1221 passes the light from measurement laser 1201, but blocks light from generation laser 1205. Note that such a filter is not required in the embodiments of FIGs. 10B and 10C because of their use of a single beam.
  • the reflected beam (after being filtered by filter 1221) at the measurement wavelength reaches detector 1223, which generates the above-described analog electrical signal.
  • Filter 1221 of some embodiments of the apparatus of FIG. 12 is mounted on an actuator that enables it to be moved out of the path of the reflected beams, so that laser 1205 may be used by itself in such a configuration to provide a measurement wavelength (e.g. with all the other lasers 1201 a-1201c turned off).
  • a test structure 1305 (FIG. 13B) is formed (as per act 1311 in FIG. 13A) using the same processes as and at the same time as the integrated circuit 1320 in a production wafer 1330, in the manner described above for act 201 (FIG. 2A).
  • wafer 1330 of FIG. 13B may be same as wafer 230 of FIG. 2A, i.e. the same wafer can be evaluated by each of the two methods illustrated in FIGs. 2A and 13A respectively. In the method of FIG.
  • a beam of linearly-polarized electromagnetic radiation is generated in a continuous manner, with a wavelength of the beam being selected'to ensure that the absorption length is less than a thickness of the wafer but greater that the profile depth (see act 1312 in FIG. 13A).
  • This act 1312 is similar to act 202 described above for FIG. 2A, except that a beam 1308 (see FIG. 13C) of linear polarization is generated in act 1312.
  • the linearly-polarized beam 1308 (FIG. 13C) is split (as per act 1313) into two beams of S and P polarizations (these polarization directions are orthogonal to one another).
  • the two beams of S and P polarizations are used to illuminate (as per act 1314) two regions of a test structure (or of the integrated circuits depending on the embodiment) on a front surface F of the wafer 1330 (FIG. 13C).
  • the two regions 1301 and 1303 (FIG. 13B) are illuminated simultaneously by the S and P beams forming the respective spots 1309a and 1309b.
  • doped region 1301 is illustrated as having a spot 1309a formed by the P beam, and undoped region 1303 has spot 1309b formed by the S beam, an inverse of this orientation can also be used (i.e. S beam illuminating doped region and P beam illuminating undoped region). Regardless of the relative orientation, the spots 1309a and 1309b are positioned symmetrically relative to a boundary 1304 between the two regions 1301 and 1303.
  • the intensity of the two beams incident on the two regions in the respective spots 1309a and 1309b is modulated as a function of time, as illustrated in act 1315.
  • act 1315 For example, when intensity of the S polarization is at the maximum and the intensity of the P polarization is at the minimum, then the intensity of the S polarization is gradually reduced and concurrently therewith the intensity of the P polarization is gradually increased. While the intensities of the two beams are modulated, portion of the two beams reflected by the wafer is measured (as per act 1316).
  • a linearly-polarized beam 1302 (FIG. 13C) is passed through a half-wave plate 1310 (FIG. 13D) that is being rotated, to generate a beam 1308 that has polarization direction R which changes as a function of time.
  • a beam 1308 that has polarization direction R which changes as a function of time.
  • an angle ⁇ between direction R of polarization and the z axis is a function of time t. Therefore, in such embodiments, linearly polarized beam 1308 (FIGs. 13C and 13D) is split by passing it through a beam splitter 1307, to generate the S and P beams which are separated by a small angle 0.
  • beam splitter 1307 e.g. a prism formed by cementing two wedges.
  • polarizing beam splitters can also be used as described herein, as long as the beam splitter has a sufficiently small separation angle (e.g. 0 smaller than or equal to 20°).
  • the small separation angle is required to provide a certain amount of physical distance Pd (FIG. 13C) between the beam splitter 1307 and front surface F of wafer 1330, to accommodate an objective lens (not shown) therebetween.
  • the separation angle 0 is selected to be sufficiently large (in view of the distance Pd from the wafer) to ensure that the two beams S and P that are obtained from splitting do not form spots 1309a and 1309b (FIG. 13B) that overlap. Spots 1309a and 1309b may be separated from one another by any distance, e.g. 1 micron or even 10 microns.
  • beam splitter 1307 is a Wollaston prism.
  • the intensity of reflection detected by the photodetector indicates a reflectivity of the wafer 1330 which is average (indicated in FIG. 13F by Savg at the time tA) of the reflectivities of the two regions 1301 and 1303 (FIG. 13B).
  • the input beam 1308 is linearly polarized at a 45° angle in the direction R with respect to the prism's optical axes x and z, and the beam is refracted into two equal- intensity, orthogonally-polarized beams S and P, with each polarization in the same direction as the respective optical axes.
  • the intensity of the output beams from beam splitter 1307 is alternated between one refracted direction S to the other refracted direction P, and therefore the resulting signal detected by the photodetector has a triangular waveform illustrated in FIG. 13F.
  • the refracted beams are aligned to the doped and undoped regions, and these two regions are then alternately probed in a periodic manner, generating a fluctuating signal detectable by a lock-in amplifier.
  • the signal period of the analog electrical signal generated by the photodetector (which is provided to the lock-in amplifier) is one fourth the rotation period of the half-wave plate.
  • this method of beam splitting does not produce anomalous spikes in the signal train, since the probing beam never crosses the boundary between doped and undoped regions.
  • FIG. 13N illustrates one such embodiment of the invention, where a single incoming laser beam 1308 having rotating polarization is split into two orthogonally polarized beams S and P by a beam splitter 1307, in order to illuminate the doped and undoped regions of a wafer (not shown).
  • the S and P beams each have an intensity that changes over time, depending on the direction R of polarization of the input beam 1308 relative to the prism's two optical axes which are shown in FIG. 13E as being in the directions z and x respectively, and these directions are illustrated in FIG. 13D relative to the direction of travel of the beam 1308.
  • the two beams S and P are shown in FIG. 13N as being in a plane parallel to the x-y plane which is also the plane of an optical bench 1380.
  • Optical bench 1380 supports various optical components, such as lasers 1381A- 1381 C for generating the beam 1302, half-wave plate 1310, dichroic mirror 1382, and beam splitter 1307.
  • Dichroic mirror 1382 is mounted on a rail on the optical bench to allow it to translate along the y axis, so as to select one of lasers
  • a non-polarizing beam splitter 1383 (also called “detection system beam splitter”) which is in line between dichroic mirror 1382 and beam splitter 1307, and is configured to allow the beam from mirror 1382 to pass therethrough to beam splitter 1307, but at least a portion of a reflection in the opposite direction is diverted to a photodetector 1384 that in turn is coupled to a lock-in amplifier (not shown).
  • the lock-in amplifier is electrically connected (by wire 1389) to a motor (also not shown) that turns the half-wave plate 1310.
  • FIG. 13O shows a hardware configuration for some embodiments which use split beams with changing intensities to probe the doped and undoped region of a test structure.
  • Half-wave plates 1204a, 1204b, and 1204c are placed in the path of beam 1203 and are rotated at a fixed frequency about the beam propagation axis.
  • the linear polarization of the measuring beam is also rotated, so that when the beam passes through Wollaston prism 1214, the beam is split into two orthogonally polarized beams 1219a and 1219b with alternating intensities.
  • beams 1219a and 1219b are placed within the doped and undoped regions, respectively.
  • the beams are recombined by Wollaston prism 1214, such that the signal detected by photodetector 1223 is an analog fluctuating electrical signal with amplitude proportional to the reflectivity difference between the doped and undoped regions.
  • a connection such as one of wires 1391 a-1391c
  • lock-in amplifier 1231 to a motor that can rotate the half-wave plate (e.g. one of plates 1204a-1204c) at a frequency set by lock-in amplifier 1231.
  • Other means of rotating the polarization of the measuring beam may be employed; they include but not limited to electro- optic and liquid-crystal modulators.
  • several Wollaston prisms 1214 may be used, mounted on a turret or variable position actuator. Each prism sets the spacing between the two beams, so that a selection of spacings is available to accommodate different patterns to be measured.
  • the same portion of the doped 1301 and undoped 1303 regions are illuminated with an additional laser beam (also called “pump” beam) 1404 (FIG. 14A) that is collinear with the above-described laser beam (also called “probe” beam) 1408.
  • Both beams are split to form a pair of concentric (or overlapping) spots for the S polarized light in one region and also form another pair of concentric (or overlapping) spots for the P polarized light in another region, as illustrated in FIG. 14C. Note that only the P and S components of probe beam 1404 have their intensity modulated, while the P and S components of pump beam 1408 do not have modulating intensities.
  • the S component of the pump beam continues to illuminate an undoped region even when the S component of the probe beam has zero intensity.
  • FIG. 14E the intensity of probe beam 1408 is modulated between the two regions by use of a rotating half-wave plate, in the manner described above in reference to FIGs. 13A-13N.
  • Pump beam 1404 has photon energy greater than the band-gap of the semiconductor, in order to generate excess carrier distribution, in the same manner as the method described in US patent 6,323,951 and 6,049,220 by Borden.
  • the value of Samplitude obtained from measuring the fluctuation in an analog electric signal sensed by a photodetector as the intensity of reflected radiation at the wavelength of probe beam 1408 is now a measure of the difference in excess carrier distribution between the doped and undoped regions.
  • Such an Samplitude is sensitive to other parameters of the doped layer, such as defect density and oxide charge, in addition to its sensitivity the profile depth, the doping concentration, and the profile abruptness.
  • the difference signal of interest is given by equation (13).
  • Pump beam 1404 can be unpolarized, or circularly polarized or polarized linearly (at any angle relative to the two axes of the Wollaston prism, although in one embodiment it is at 45 degrees thereby to ensure that the two regions have equal illumination of the pump beam). Note that if pump beam 1404 is linearly polarized along the direction of one axis of the Wollaston prism, then only one region (e.g. the doped region) receives the full intensity of the pump beam, and the other region receives no energy.
  • FIG. 14B shows a hardware configuration for some embodiments which use a probe beam with changing intensities of S and P components simultaneously with a pump beam with fixed intensities of the S and P components. The hardware configuration in FIG.
  • the apparatus also includes an additional laser source 1205 and a collimating lens 1206 located in a path to the dichroic mirror 1210.
  • the beam from laser source 1205 is oriented along the direction of travel of the dichroic mirror 1210.
  • probe beam forms the two components 1509a and 1509b.
  • components 1509a and 1509b can be made respectively incident on the doped and undoped regions (at a front surface F of a wafer 1500), although in other embodiments the two components may be made incident on the same type region (e.g. both only on a doped region).
  • This beam configuration can be achieved by using a Wollaston prism, a circularly polarized probe beam, and a linearly polarized pump beam with its polarization direction aligned to one of the Wollaston prism's optic axis.
  • the Wollaston prism splits the measuring beam into two orthogonally polarized beams 1509a and 1509b with equal intensities. With proper alignment, beams 1509a and 1509b can be placed within the doped region 1501 and undoped region 1503, respectively.
  • the linear polarization of the pump beam is chosen for example to be oriented along the x-direction of the Wollaston prism axis (see FIG.
  • the pump beam may have a slightly different spot size than the measuring (probe) beam.
  • the Wollaston prism recombines the reflected probe beams from the doped and undoped areas as described above.
  • an intensity of the pump beam (also called “carrier generation beam”) is modulated as a function of time, at a predetermined frequency.
  • the frequency is selected to be sufficiently small to avoid the creation of a wave of carriers in an illuminated region. Reflected beams from the doped and undoped areas are recombined by the Wollaston prism.
  • the reflected combined beam from the Wollaston prism is then split again (by a beam splitter that is polarizing) into two orthogonally polarized beams, before they are detected separately by two photodetectors.
  • the axis of the polarizing beam splitter (also called “photodiode beam splitter” due to its adjacency to two photodiodes that receive their respective signals therefrom) is oriented at 45 angle with respect to the two axes of the Wollaston prism, such that one photodetector measures the sum while the other photodetector measures the difference of reflected electromagnetic radiation, from the doped and undoped area at the wavelength of the probe beam.
  • wavelengths ⁇ 1 and ⁇ 2 of the probe beam and the pump beam respectively is a critical aspect in one embodiment and leads to unexpected results.
  • the pump beam contains photons having energy above silicon's bandgap energy and the probe beam contains photons having energy approximately the same as or less than the bandgap energy.
  • the 830 nm and 1033 nm wavelength beams provide one or more benefits described herein.
  • Wavelength 830 nm is considered particularly suitable for the pump beam because the absorption length in silicon is about 15 microns. Thus, the absorption length is much greater than the junction depth, and creation of excess charge carriers is nearly uniform over the depth of concern in the measurement. Because the photon energy is close to the bandgap energy, photon generation is more efficient, with less energy going directly into heating the semiconductor.
  • the absorption length at wavelength 1033 nm is about 373 microns, and therefore the number of excess carriers being created by such a probe beam is sufficiently low to ensure minimum perturbation to the excess carrier distribution.
  • the absorption length at wavelength 1033 is short enough that very little reflection from a back surface of the wafer is seen (wafers are typically 600- 800 microns thick), since the back surface reflection can potentially cause spurious signals.
  • Direction 1601 is the direction of the beam splitter axis, which is at a 45 angle with respect to the S polarization direction 1602, and the P polarization direction 1603 of the probe beam.
  • directions 1602 and 1603 coincide with the z- and y-axes of FIG. 13N, respectively.
  • one photodetector detects the sum of signals 1604 and 1607, while another photodetector detects the difference between signals 1605 and 1606 of the S and P polarizations.
  • the intensities of the S and P polarizations of the probe beam reflected by the wafer are slightly different due to the different properties of the doped and undoped regions; their difference is exaggerated in FIG. 16 for clarity.
  • the electrical signals from the output of each of the two photodetectors are sinusoidal waveforms that are electronically subtracted from one another and amplified. Such a difference signal is indicative of the properties of the doped region, and an amplitude of this signal is detected using a lock-in amplifier to enhance the signal to noise ratio (since the pump beam is modulated at a fixed frequency).
  • FIG. 15B A hardware configuration illustrative of embodiments of the type described above is shown in FIG. 15B.
  • a single measurement laser 1201 which is a semiconductor diode laser operating at 980nm wavelength. Its output beam is collimated by collimating lens 1202, and then passes through quarter-wave plate 1204 to produce collimated beam 1203 that is circularly polarized.
  • the carrier generation laser 1205 is a semiconductor diode laser operating at 830nm wavelength. Its output is collimated by collimating lens 1206, and then passes through half-wave plate 1208 to produce collimated beam 1207 that is linearly polarized at 45 angle with respect to the normal direction of the optical bench (z-axis direction in FIG. 13N).
  • Beams 1203 and 1207 are combined using dichroic mirror 1210 (such as a partially transmissive mirror, e.g. part number SWP45-Rp1047-Tp830 available from CVI Laser of Albuquerque, New Mexico.) to create a combined and collinear beam 1211.
  • This combined beam passes through 50:50 non-polarizing beam splitter 1212 (also called “detection system beam splitter”), e.g.
  • part number 10BC17MB.2 from Newport Corp., Irvine, CA
  • a 90:10 beam splitter 1213 available from Precision Applied Products of Fullerton, Calif., by specifying 93.3% transmission at 0.83 microns wavelength and 90% transmission at 1.48 microns wavelength
  • a Wollaston prism 1214 and an objective lens 1216 (such as a 100X, 0.8 NA lens made by Olympus of Tokyo Japan).
  • Objective lens 1216 focuses the combined beam 1211 onto wafer 1217.
  • the 90:10 beam splitter 1213 is also called "vision system beam splitter”.
  • Wollaston prism 1214 splits the probe beam into S and P polarized portions before entering the objective lens 1216, resulting in two focused beams 1219a and 1219b on the front surface of wafer 1217.
  • the spots formed by beams 1219a and 1219b are typically separated by a distance of 5-20 ⁇ m. With proper alignment, spots formed by beams 1219a and 1219b are placed within the doped and undoped regions, respectively of wafer 1217.
  • the optic axes of the Wollaston prism 1214 is oriented at 0° and 90 with respect to the normal direction of the optical bench (or along the z- and x-directions of FIG. 13N), such that the two orthogonally polarized beams1219a and 1219b have polarizations oriented at 0 and 90 with respect to the optical bench.
  • the reflected beams are re-collimated by lens 1216 and recombined by Wollaston prism 1214.
  • Beam splitter 1213 diverts 10% of the return beam to lens 1226 and camera 1227, which provide a system to align the two beam spots to regions in the test structure of interest.
  • an auto-focus system that consists of a pinhole and a detector, which also uses the portion of the return beam diverted by beam splitter 1213.
  • the remaining portion of the return beam then enters detection system beam splitter 1212, which deflects a portion of it to optical filter 1221.
  • Filter 1221 passes the light from measurement laser 1201 , but blocks light from generation laser 1205 (i.e. any reflected portion of the pump beam).
  • the portion of the probe beam from filter 1221 is thereafter split into two orthogonally polarized beams by a polarizing beam splitter 1222, (also called “photodiode beam splitter”) whose axis can be rotated.
  • a polarizing beam splitter 1222 also called “photodiode beam splitter”
  • the axis of photodiode beam splitter 1222 is oriented at 45 angle with respect to the two axes of Wollaston prism 1214, such that one detector 1223 detects the sum of the reflected probe beams coming from the doped and undoped regions, while detector 1224 detects the difference.
  • Detectors 1223 and 1224 are silicon photodiodes.
  • lock-in amplifier 1231 The photodiode currents are converted to voltages and subtracted from each other electronically using amplifier 1230, the output of which goes to lock-in amplifier 1231.
  • the output of lock-in 1231 goes to a computer 1232 which receives the signal and presents it to the user or other data collection systems.
  • Lock-in 1231 includes a frequency reference that is used to modulate laser driver 1234, which provides a modulated drive output for generation laser 1205.
  • the exponential factor in equation (17) accounts for the 90° ( ⁇ /2) phase shift between S and P polarization in the probe beam because it is circularly polarized.
  • One photodetector output is proportional to the intensity of the sum of the two reflected amplitudes Es and Ep, while the other photodetector output is proportional to the intensity of the difference of these same two amplitudes.
  • the signal of interest is the difference between these two photodetector outputs, i.e.:
  • equation (18) the difference signal is given by: where A is a wavelength dependent coefficient representing transmission loss through optical components, photodetector conversion efficiency, and amplifier gain factor. All of the other parameters have the same definitions defined previously.
  • N refers to the excess carrier concentration in the doped region generated by beam 1510.
  • the form of equation (19) allows the possibility of an arbitrary doping profile leading to an excess carrier profile as a function of depth N(z).
  • the resulting excess carrier profile resembles a step function, shown as curve 1702 in FIG. 17A.
  • the signal has a sine dependence on the depth of the doping profile (zj), as shown in FIG. 17B.
  • the above-described sine dependence has the advantage that the signal being measured provides improved resolution as profile depth (zj) gets smaller in response to improvements in wafer fabrication technology. This is because the sine curve passes through the origin (in contrast to a cosine curve that remains practically unchanged in a small neighborhood around the origin).
  • This embodiment can also be implemented on an un-patterned sample, i.e. a wafer that is doped over its entire area. In this case, no alignment on a particular pattern is needed, the two beams can be placed anywhere on the wafer. Since one of the probe beam is not accompanied by a carrier generation beam, its reflection property is very similar to the reflection from an undoped region, so the signal is the same as given by equation (19).
  • the method of some embodiments of the invention may be applied to measure the properties of pre-annealed implant wafers.
  • the method of may be applied to measure the properties of Silicon-On-lnsulator (SOI) wafers, both pre and post-annealed.
  • SOI Silicon-On-lnsulator
  • the measurement may also be used to determine the difference in dielectric coating thickness in two regions.
  • embodiments of the type described herein may also be used for process development. For example, if a development engineer wants to compare the profile abruptness that is possible with different laser anneal treatments, an Samplitude measurement of the type described herein can provide such information. For this case, the types of test structures that can be used for evaluation are expanded than some of the test structures described above, because it is no longer necessary to manufacture the test structures within a standard flow for the manufacture of integrated circuits.
  • any of the following properties can be determined by use of one or more Samplitude measurements of the type described herein on a wafer prior to annealing: the dose of implants in a doped region, profile depth and profile abruptness. Moreover, similar measurements after anneal are used in some embodiments to determine the depth of an annealed semiconductor junction, the dose of a doped region after annealing, doping concentration after annealing, thermal exposure of the layer, and a profile of doping concentration after annealing.
  • a lock-in amplifier as a synchronous detector
  • other implementations of a synchronous detector such as boxcar averager and transient digitizer are used in several other embodiments, as will be apparent to the skilled artisan.
  • a Rochon prism can be used in alternative embodiments.
  • An example of the Rochon prism that can be used in embodiments of the type described herein is available from CVI Laser as part# RCHP-10-CA-670-1064, 10 degree separation, 670-1064nm wavelength.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Pathology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Electromagnetism (AREA)
  • Toxicology (AREA)
  • Testing Or Measuring Of Semiconductors Or The Like (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

Pour évaluer une plaquette de semi-conducteurs comportant deux régions à profils de concentration de dopant, on effectue au moins deux mesures dans les deux régions puis on compare les mesures des deux régions pour obtenir une mesure du changement de la réflectivité indiquant une différence de réflectivité entre les deux régions. L'analyse de la mesure du changement de la réflectivité fait ressortir au moins une propriété d'une des régions si des propriétés correspondantes de l'autre région sont connues. Par exemple, si une des régions est dopée et que l'autre région ne l'est pas (par exemple, les régions de source/drain et de canal d'un transistor), alors un changement de la réflectivité entre les deux régions peut déterminer au moins une des propriétés suivantes dans la région dopée: (1) la concentration de dopant, (2) la profondeur de la jonction ou du profil et (3) le caractère abrupt (c'est-à-dire la pente) d'un profil de la concentration de dopant au niveau de la jonction. Une des régions est insolée avec une des polarisations S et P d'un faisceau de rayonnement cohérent et l'autre région est insolée avec l'autre polarisation S ou P. Les intensités des composantes de polarisation S et P respectives sont modulées dans le temps.
PCT/US2004/040969 2003-12-09 2004-12-08 Evaluation differentielle de regions adjacentes pour determiner le changement de la reflectivite WO2005057194A1 (fr)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
US10/731,991 2003-12-09
US10/732,436 2003-12-09
US10/732,436 US7136163B2 (en) 2003-12-09 2003-12-09 Differential evaluation of adjacent regions for change in reflectivity
US10/731,991 US7190458B2 (en) 2003-12-09 2003-12-09 Use of scanning beam for differential evaluation of adjacent regions for change in reflectivity

Publications (1)

Publication Number Publication Date
WO2005057194A1 true WO2005057194A1 (fr) 2005-06-23

Family

ID=34681744

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2004/040969 WO2005057194A1 (fr) 2003-12-09 2004-12-08 Evaluation differentielle de regions adjacentes pour determiner le changement de la reflectivite

Country Status (1)

Country Link
WO (1) WO2005057194A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107271452A (zh) * 2017-07-25 2017-10-20 常州工程职业技术学院 一种盘形悬式瓷绝缘子视觉检测装置
CN109883986A (zh) * 2019-03-27 2019-06-14 电子科技大学 基于激光振镜的太赫兹单像素成像系统及成像方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4758092A (en) * 1986-03-04 1988-07-19 Stanford University Method and means for optical detection of charge density modulation in a semiconductor
EP0313681A1 (fr) * 1987-10-30 1989-05-03 Ibm Deutschland Gmbh Alignement interférométrique masque-plaquette semi-conductrice sensible à la phase
JPH02102404A (ja) * 1988-10-11 1990-04-16 Dainippon Printing Co Ltd 周期性パターンの検査方法
US5610718A (en) * 1991-09-27 1997-03-11 Canon Kabushiki Kaisha Apparatus and method for detecting a relative displacement between first and second diffraction gratings arranged close to each other wherein said gratings have different pitch sizes
JP2000021748A (ja) * 1998-06-30 2000-01-21 Canon Inc 露光方法および露光装置
US6072179A (en) * 1998-08-07 2000-06-06 Intel Corporation Method and apparatus using an infrared laser based optical probe for measuring voltages directly from active regions in an integrated circuit
US6411389B1 (en) * 2000-05-03 2002-06-25 The Regents Of The University Of Claifornia Optical monitor for real time thickness change measurements via lateral-translation induced phase-stepping interferometry

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4758092A (en) * 1986-03-04 1988-07-19 Stanford University Method and means for optical detection of charge density modulation in a semiconductor
EP0313681A1 (fr) * 1987-10-30 1989-05-03 Ibm Deutschland Gmbh Alignement interférométrique masque-plaquette semi-conductrice sensible à la phase
JPH02102404A (ja) * 1988-10-11 1990-04-16 Dainippon Printing Co Ltd 周期性パターンの検査方法
US5610718A (en) * 1991-09-27 1997-03-11 Canon Kabushiki Kaisha Apparatus and method for detecting a relative displacement between first and second diffraction gratings arranged close to each other wherein said gratings have different pitch sizes
JP2000021748A (ja) * 1998-06-30 2000-01-21 Canon Inc 露光方法および露光装置
US6072179A (en) * 1998-08-07 2000-06-06 Intel Corporation Method and apparatus using an infrared laser based optical probe for measuring voltages directly from active regions in an integrated circuit
US6411389B1 (en) * 2000-05-03 2002-06-25 The Regents Of The University Of Claifornia Optical monitor for real time thickness change measurements via lateral-translation induced phase-stepping interferometry

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
HEINRICH H K: "PICOSECOND BACKSIDE OPTICAL DETECTION OF INTERNAL SIGNALS IN FLIP- CHIP MOUNTED SILICON VLSI CIRCUITS", MICROELECTRONIC ENGINEERING, ELSEVIER PUBLISHERS BV., AMSTERDAM, NL, vol. 16, no. 1 / 4, 1 March 1992 (1992-03-01), pages 313 - 324, XP000245360, ISSN: 0167-9317 *
PATENT ABSTRACTS OF JAPAN vol. 014, no. 317 (P - 1073) 9 July 1990 (1990-07-09) *
PATENT ABSTRACTS OF JAPAN vol. 2000, no. 04 31 August 2000 (2000-08-31) *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107271452A (zh) * 2017-07-25 2017-10-20 常州工程职业技术学院 一种盘形悬式瓷绝缘子视觉检测装置
CN109883986A (zh) * 2019-03-27 2019-06-14 电子科技大学 基于激光振镜的太赫兹单像素成像系统及成像方法
CN109883986B (zh) * 2019-03-27 2021-08-24 电子科技大学 基于激光振镜的太赫兹单像素成像系统及成像方法

Similar Documents

Publication Publication Date Title
US7190458B2 (en) Use of scanning beam for differential evaluation of adjacent regions for change in reflectivity
US6489801B1 (en) Apparatus and method for evaluating a semiconductor wafer
US6426644B1 (en) Apparatus and method for determining the active dopant profile in a semiconductor wafer
US6532070B1 (en) Method for determining ion concentration and energy of shallow junction implants
US6008906A (en) Optical method for the characterization of the electrical properties of semiconductors and insulating films
US6812717B2 (en) Use of a coefficient of a power curve to evaluate a semiconductor wafer
JP2002540396A5 (fr)
US7136163B2 (en) Differential evaluation of adjacent regions for change in reflectivity
US6878559B2 (en) Measurement of lateral diffusion of diffused layers
US6963393B2 (en) Measurement of lateral diffusion of diffused layers
US7499168B2 (en) Combined modulated optical reflectance and electrical system for ultra-shallow junctions applications
US7751035B2 (en) Method and device to quantify active carrier profiles in ultra-shallow semiconductor structures
WO2005057194A1 (fr) Evaluation differentielle de regions adjacentes pour determiner le changement de la reflectivite
WO1999013318A1 (fr) Procede optique de caracterisation des proprietes electriques de semi-conducteurs et pellicules isolantes
US20060114478A1 (en) Evaluating effects of tilt angle in ion implantation
JASON Laser Reflectance Modulation in Silicon Integrated Circuits

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase