Method for characterizing semiconductor doping using the light neutralization time constant of the corona surface charge

Abstract

Methods for characterizing semiconductor doping in wide-bandgap semiconductor samples include: measuring an initial value V0 of the surface voltage at a region on the surface of the semiconductor sample in darkness; charging the region to deep depletion in darkness by depositing a specified corona charge at the region; measuring the surface voltage value at the region in darkness after charging; and using a specific photon flux f. eff The light irradiates the charging region, and a specific photon flux f eff A photon energy above the semiconductor bandgap is used, sufficient to generate free minority carriers in the semiconductor sample, thereby causing photoneutralization of the corona charge. The relationship between the photoneutralization-induced corona charge decay and the irradiation time t at the region is monitored using non-contact time-resolved measurements of the surface voltage V(t). The monitored time-resolved surface voltage decay data V(t) is analyzed to determine the photoneutralization time constant τ. ph ; and using a specific photon flux f eff Light neutralization time constant τ ph It serves as a semiconductor doping index, and its value characterizes the semiconductor doping concentration in that region.

Description

TECHNICAL FIELD

[0001] The present application relates generally to the characterization of semiconductor doping in wide bandgap semiconductors, and more particularly to a method for non-contact measurement of doping in epitaxial layers and a system for performing the measurement. BACKGROUND

[0002] Semiconductors doped with donor or acceptor impurities change the electrical properties of the semiconductor and are a key technology element in semiconductor device manufacturing. Doping is typically accomplished by introducing impurity atoms that replace host atoms in the semiconductor lattice. Doping concentration is the concentration of doped atoms per volume of semiconductor, and its measurement is commonly used for quality control in semiconductor device production.

[0003] For example, devices based on wide bandgap semiconductors, including SiC, GaN, AlGaN, and AlGaN / GaN, are widely used in power electronics and high frequency applications. Such devices typically include doped epitaxial layers, where strict doping control of the epitaxial wafer is necessary. To meet the demands of rapidly developing wide bandgap semiconductor technology and large-scale epi-wafer manufacturing, corresponding large-scale doping testing is required. For this purpose, non-contact measurement techniques can be preferred. Non-contact metrology can eliminate the costs and time involved in manufacturing test devices and can avoid post-measurement wafer cleaning required in the case of many conventional doping measurement methods, such as the Mercury Probe method.

[0004] An example of a preparation-free non-destructive doping measurement method for wide bandgap semiconductors described in US 10,969,370 B2 can be performed using a commercially available corona non-contact capacitance voltage (CnCV) tool manufactured by Semilab SDI. CnCV uses corona charging to electrically bias the semiconductor surface to deep depletion, similar to the voltage bias of a metal Schottky barrier, however, achieved without a metal contact to the semiconductor surface. In the CnCV method, the surface voltage response ΔV is monitored with a Kelvin probe for a charge bias dose ΔQ C . The differential capacitance C = ΔQ C / ΔV in sequential charge measurements gives the non-contact C-V characteristic. The doping concentration N D is determined from the slope of 1 / C 2 vs V, which is a standard procedure in C-V methods. For a typical 12-point pattern, the CnCV measurement throughput using a commercially available tool is currently limited to about 5 wafers per hour. SUMMARY

[0005] Conventional CnCV techniques can be modified to employ a new doping sensitivity that can be exploited to increase test throughput. Conventional CnCV uses corona charge biasing to depletion, which is achieved by multiple incremental corona charge deposition steps. In response to the deposited charge, the surface voltage V changes, and is measured with a non-contact vibrating Kelvin probe. Similar to mercury probes and Schottky junction devices, CnCV employs the slope of the 1 / C-V characteristic in depletion to determine the doping concentration. CnCV is a quasi-static technique, which has many relatively slow charge biasing steps involved in acquiring the C-V characteristic. This typically limits the measurement speed and throughput in multiple wafer monitoring, where each wafer typically has 9 or 12 test sites, and has a wafer map of 49 sites. 2

[0006] The deposited corona charge can be rapidly neutralized and removed from the bare surface of wide bandgap semiconductors by short wavelength illumination. The corresponding time constant of corona charge photo-neutralization can be short, enabling rapid depletion barrier sweeping. In measurements on epitaxial SiC of different doping, a direct relationship between the corona charge photo-neutralization time constant and the doping concentration was determined. This relationship is verified for other wide bandgap semiconductors, and is valid over a wide range of doping concentrations. The enhanced throughput doping measurement method in this disclosure is based on this newly identified relationship.

[0007] Thus, in the present invention, a charge biasing method with novel doping measurement principles is described that can increase the measurement speed and throughput, for example, up to ten times compared to the conventional CnCV technique described above.

[0008] The method described here can replace the sequence of cumulative charge steps with a single large corona charge to deep depletion. Thereafter, the deposited charge on the semiconductor surface is photo-neutralized using near UV illumination that generates excess carriers. The photo- neutralization time constant is determined from time-resolved surface voltage measurements, and this time constant is the doping measurement parameter.

[0009] During photo-neutralization, the photo-generated minority carriers within the charge-induced surface depletion layer are mobile. They are guided to the surface by the depletion electric field, and are captured by the corona ions with opposite polarity, thereby neutralizing the corona charge and reducing the depletion width. This technique includes measuring and analyzing the corresponding change in surface voltage versus illumination time.

[0010] ​In wide bandgap semiconductors, photo-neutralization of corona ions is irreversible. The corona charge does not recover after photo-neutralization, and the surface voltage change induced by photo-neutralization does not reverse in the dark. This enables monitoring of the process by measuring the surface voltage while the charged area is illuminated or after a continuous illumination pulse. The present disclosure encompasses both measurement configurations. Specifically, the measurement while the charged area is illuminated employs a system configuration in which the illumination is underneath the surface voltage probe. The measurement after an illumination pulse is used in a system configuration in which the illumination is at a location separate from the surface voltage probe.

[0011] The measurement technique takes advantage of the irreversible nature of the surface voltage change induced by illumination to separate the charge photo-neutralization from other surface photovoltage effects. The corresponding test involves monitoring the surface voltage in the dark after stopping the illumination. For good measurement conditions, the dark recovery should be negligible, e.g., in the range of 0.1%.

[0012] In certain embodiments, the measurement technique employs the depletion layer Schottky barrier electrostatic charge-voltage relationship to extract the charge photo-neutralization rate and determine the photo-neutralization time constant τ ph from the logarithmic dependence of the surface voltage on the illumination time. ph The doping concentration N D is determined based on the calibrated dependence of the time constant τ ph vs N D . The time constant vs surface voltage characteristic τ ph -V provides a means for assessing the doping depth profile. A uniform doping is identified by a constant τ ph value. In the logarithmic photo-neutralization surface voltage characteristic, a multilayer film of different doping will show different τ ph values over a certain time period. The corresponding surface voltage reflects different depths underneath the surface.

[0013] In general, the measurement apparatus includes three system components with the following capabilities:

[0014] (a) - corona charging to bias the semiconductor surface charge to a depth depletion.

[0015] (b) - non-contact surface voltage measurement.

[0016] (c) - short wavelength illumination with photon energy higher than the semiconductor bandgap.

[0017] The semiconductor sample wafer can be placed on a conductive chuck with fast movement capability for positioning the test sample area (i.e., test sites on the wafer) under the system components (a), (b), and (c) according to the measurement cycle.

[0018] A housing for the measurement system can be provided to prevent stray light (e.g., short wavelength stray light) from reaching the wafer. Otherwise, such stray light would cause uncontrolled corona charge neutralization on the wafer surface.

[0019] time constant τ ph may be used directly as a relative doping concentration index. For accurate quantitative measurement of the doping concentration, the method can include a calibration function of the time constant, which is determined by measurements of a reference sample wafer with known doping concentration. Alternatively or additionally, wafer specific calibration can be performed with corona C-V doping measurements performed on the same sample area (same wafer site) as the measurement of the photo- neutralization time constant τ ph .

[0020] In conventional corona charging methods for semiconductor and dielectric characterization, the response to deposited corona charge density variations is typically monitored by measuring the corresponding variation of the surface voltage. In the present technology, measurements of the surface voltage are also employed, however, the variation of the charge density of the corona deposition is achieved by photo-neutralization of the charges rather than by charge deposition. The measurement of the charge photo-neutralization and the surface voltage can be performed simultaneously or sequentially, depending on the configuration of the light source and the surface voltage probe. Illumination under the probe enables the measurement to be made during the actual corona photo-neutralization. This allows the illumination and measurement of the sample site without transferring the wafer between the illumination and the measurement probe, which can benefit the speed and throughput of the measurement. The configuration with a separate light source can facilitate higher illumination uniformity than the illumination under the surface voltage probe. This can benefit the accuracy and tool-to-tool matching.

[0021] The disclosed technology can use multiple light sources with different configurations and wavelengths, which can be selected and optimized according to the requirements of the specific semiconductor wafer fabrication that uses the technology.

[0022] In general, in one aspect, the invention features a method of characterizing a semiconductor dopant in a wide bandgap semiconductor sample, the method comprising: measuring an initial value V0 of a surface voltage at a region of a surface of the semiconductor sample in the dark; charging the region to deep depletion in the dark by depositing a prescribed corona charge at the region; measuring a surface voltage value at the region after the charging in the dark; illuminating the charged region with light of a specific photon flux f eff having a photon energy higher than the semiconductor bandgap sufficient to generate free minority carriers in the semiconductor sample to cause photo-neutralization of the corona charge; monitoring the photo-neutralization induced corona charge decay vs illumination time t at the region using a non-contact time resolved measurement of the surface voltage V(t); analyzing the monitored time resolved surface voltage decay data V(t) to determine a photo- neutralization time constant τ eff .ph ; and using a specific photon flux f eff Light neutralization time constant τ ph It serves as a semiconductor doping index, and its value characterizes the semiconductor doping concentration in that region.

[0023] Implementations of this method may include one or more of the following features.

[0024] This method may include a specific corona charge light neutralization time constant τ based on surface voltage. ph vs V is used to characterize the depth distribution of doping concentration in semiconductor samples.

[0025] This method may include average τ based on the surface voltage range ph The value is used to characterize the doping of a single epitaxial layer.

[0026] In some examples, τ is determined according to the following equation. ph :

[0027]

[0028] Where t is the irradiation time, and V const This refers to surface voltage probe offset. The method may include using a probe with a known doping concentration value N. D τ of one or more reference samples ph The calibration measurement determines the calibration function τ. ph vs ND and inverse calibration function N D vsτ ph (represented as f) cal In some examples, the method includes adjusting the effective photon flux φ. eff The time constant τ of the measurement ph Normalized to calibrated photon flux φ eff ,as follows: And using the normalized time constant according to the following equation Determine the absolute doping concentration:

[0029] This method may include calibrating τ for doping concentration. ph The calibration is specific to the semiconductor wafer sample and is based on τ at multiple sites (e.g., 9 sites, 12 sites, or 49 sites) on the wafer sample. ph During characterization, calibration is performed at one site on the measured wafer sample, where all τ ph Measurements are performed under the same irradiation conditions. Calibration may include measuring the corona charge light neutralization time constant τ using a non-contact corona CV (CnCV) method. ph and independent measurement of doping concentration N Dwhere two measurements are made at the same wafer site and in a similar range of corona charge-induced depletion voltages, one after the other. The method can include using τ ph and N D to determine a wafer-specific inverse calibration function value f D that satisfies N ph = τ cal · f cal .

[0030] In some examples, the method includes determining the doping concentration at all different wafer sites based on determining τ ph at each wafer site and using the wafer-specific value of f cal determined on a single site.

[0031] The capacitive probe can be a non-contacting vibrating Kelvin probe.

[0032] The area can be illuminated while the surface voltage is monitored, or it can be illuminated separately from monitoring the surface voltage.

[0033] The semiconductor sample can include a semiconductor selected from the group consisting of SiC, GaN, and AlGaN.

[0034] The method can include identifying a defective region of the semiconductor sample based on V Dark and a dark decay rate ΔV Dark / Δt Dark measured after corona charging but before illumination.

[0035] The method can include identifying a contribution to the monitored surface voltage from surface photovoltaic effects other than corona photoneutralization based on a dark decay amplitude ΔV Dark measured after illumination.

[0036] Other features and advantages will be apparent from the accompanying drawings, description, and claims. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1A is a schematic diagram of an example system for measuring doping of a semiconductor wafer using corona charge deposition followed by illumination-induced charge photoneutralization and photoneutralization time constant determination from non-contact time-resolved surface photovoltage. In this example, the configuration includes illumination located below the surface voltage probe.

[0038] Figure 1B is a schematic diagram of an alternative configuration of the example system shown in Figure 1A where the illumination is at a location separate from the surface voltage probe.

[0039] Figure 2is a diagram of steps in an example measurement procedure for determining photo-neutralization time constant from time-resolved surface voltage decay.

[0040] Figure 3A is a plot of example results showing surface voltage measured during corona charge photo-neutralization versus irradiation time.

[0041] Figure 3B is a plot of an example of photo-neutralization time constant corresponding to a transient in Figure 3A .

[0042] Figure 3C is a plot of surface voltage specific τ ph for a uniform doping depth profile.

[0043] Figure 4 is a plot of an example showing corresponding logarithmic surface voltage decay for an n-type SiC epitaxial wafer with 3 layers of different doping.

[0044] Figure 5 is a plot of example doping calibration results showing photo-neutralization time constant τ ph versus doping concentration N D measured on n-type 4H-SiC. Measurements were made using a negative corona charging charge density of -4E12 q / cm 2 . A 325 nm wavelength irradiation with incident photon flux of 1.2 x 10 13 photon / cm 2 s was used. Doping concentration N D corresponds to C-V mercury probe results.

[0045] Figure 6 is a band diagram showing photo-neutralization of corona charge on an n-type wide bandgap semiconductor. The threshold energy hv is the bandgap E g .

[0046] Figure 7 is a plot of example results showing correlation of photo-neutralization time constant with incident photon flux. Results are for n-type SiC with doping concentration N D = 6.11e15 cm -3 and N D = 1.85E16 cm -3 . N D values are based on C-V mercury probe results. DETAILED DESCRIPTION

[0047] Reference is made to Figure 1A, schematically illustrates an example apparatus 100 for corona charge-voltage optical neutralization characterization of a semiconductor wafer sample 101. The wafer sample includes a substrate 104 and an epitaxial layer 103 supported by the substrate 104. The apparatus 100 includes a corona charge source module 120 for precise charge deposition on a test site 126, a surface voltage measurement module 130 for surface voltage measurement, and a light source module 140 for illuminating the wafer surface.

[0048] The wafer 101 is supported by a wafer chuck 110, which is positioned on a moveable stage 112. The stage 112 can move the wafer chuck by translation and rotation. A Coulomb meter 114 is connected to the wafer chuck 110.

[0049] The corona charge source module includes a corona charge electrode arranged to deposit a corona charge 124 on a site 126 on the surface of the wafer 101. The surface voltage measurement module 130 includes a Kelvin probe having an electrode 132 arranged to vibrate a small distance (e.g., one millimeter or less) above the surface of the wafer 101. While the example described here uses a Kelvin probe, other capacitive probes can be used.

[0050] The light source module 140 includes a light source 142 arranged to illuminate the surface of the wafer 101 directly below the Kelvin probe electrode. During operation, the light source 142 delivers illumination having a wavelength suitable for corona charge optical neutralization at the wafer surface directly below the Kelvin probe electrode 132. The light source 142 can include, for example, a light emitting diode (LED) or a near-UV laser. For an LED, the light beam can be additionally monochromated by passing through a narrow bandpass filter.

[0051] Reference Figure 1B, showing an alternative apparatus 100’ in which the light source module 140’ comprises a light source 144 positioned to illuminate a different location on the surface of the wafer 101 than the location underneath the electrode 132. Like the light source 142, the light source 144 can comprise an LED or a laser capable of emitting light having a wavelength suitable for corona charge neutralization. In general, the wavelength of illumination provided by the light sources 142 and 144 can be selected according to the wafer being measured. In general, the photon energy should be higher than the semiconductor energy gap. For measurements of 4H-SiC having an energy gap of 3.26 eV, suitable light wavelengths include = 355 nm or = 325 nm corresponding to photon energies hv = 3.49 eV and hv = 3.82 eV, respectively. Both photon energies exceed the SiC energy gap, which is required to generate free carriers (electrons-holes) involved in the corona charge neutralization process. For measurements on other semiconductors, the light wavelength can be selected based on the values of the energy gap and the absorption coefficient. Shorter wavelength = 315 nm illumination can be suitable for measuring doping on GaN and AlGaN materials having an energy gap greater than that of SiC. Other wavelengths of light of about 380 nm or less can be used depending on the application. The light sources 142 and 144 are arranged to provide uniform illumination within the surface area of the corona charging spot, e.g., a circular spot having a diameter of about 10 mm or less (e.g., a diameter of about 6 mm).

[0052] The computer controller 150 is used to control the operation of the apparatus and to perform data analysis to determine information about the doping of the sample 101 using the measurement and analysis steps described below.

[0053] Figure 1A and Figure 1B The example apparatuses 100 and 100’ shown in FIGS. 1 and 1’ respectively show two illumination configurations: i.e., a light source 142 (LED 1) for illumination underneath the vibrating electrode of the Kelvin probe electrode 132, and a light source 144 (LED 2) for illumination separate from the Kelvin probe 132. Standard high-precision Kelvin probes can use opaque gold electrodes. With the Kelvin probe electrode 132 being opaque, low-angle illumination from the side can be used, as schematically shown in the light source 142 in FIG. 2. In some embodiments, transparent Kelvin probes can also be used with illumination through the electrode. In general, with separate illumination using the light source 144, it can be easier to achieve high uniformity of illumination, as shown in FIG. 3. Figure 1A Figure 1B

[0054] ​​In the Kelvin probe method, a capacitance electrode oscillating a fraction of a millimeter above the surface under test generates an alternating current that is zeroed by compensating for the DC bias, thereby providing a measure of the voltage difference between the electrode and the semiconductor. For the present application, a Kelvin probe that is accurate and fast responding is preferred, such as a probe with a time constant of about 5 ms and a 0.2 mV accuracy. The surface voltage measurement range is typically -100 V to +100 V for corona charges that are deeply depleted to n-type (negative bias) and p-type (positive bias) semiconductors. A 1 mm or 2 mm electrode diameter is suitable for measuring surface voltage in the uniform central region of the corona charging and illumination area.

[0055] Corona charging is achieved using a corona discharge in air generated by a high DC voltage applied to the discharge electrode 124 from the power supply 122. Negative polarity discharges in air generate ions, while positive polarity discharges generate (h2O) n H + ions. The corona discharge electrode (a needle for point charging or a wire for entire wafer charging) can be confined in an enclosure, and ion deposition is not field driven; rather, ions diffuse through the aperture opening to the wafer surface. Given a very short mean free path of 10 -5 cm in the room atmosphere, ions lose kinetic energy and become thermalized before reaching the semiconductor surface. Such a configuration enables non-invasive corona charge biasing. Precise corona charging is typically performed in a clean room environment with controlled humidity and temperature. Control of the deposited corona charge density is achieved by setting the high voltage, discharge current, and deposition time period of the corona power supply 122. The deposition time period can be about 1 second or less. The deposited charge density dose can be monitored in situ with the coulomb meter 114.

[0056] Corona charging to deep depletion extends from about 1 x 10 12 q / cm 2 to about 1 x 10 13 q / cm 2 and can be achieved in a single charging step. The charge deposited on the surface at the test site 126 is mirrored as opposite charge in the semiconductor surface space-charge region and acts as an electrical bias. The charge dose is much greater than the initial existing surface charge, typically about 1 x 10 11 q / cm 2 . Thus, depending on the doping concentration N D , the charge dose Q C can be selected to obtain the desired depletion width W D = -Q C / qN D .

[0057] Wide bandgap semiconductor devices are primarily fabricated on n-type epitaxial layers, with 4H-SiC being a commonly used material. Accordingly, exemplary results for a method for n-type 4H-SiC and negative polarity corona charging are presented in the present disclosure. To measure such wafers, the method can also include stabilizing new epitaxial 4H-SiC wafers using a UV pre-treatment chamber, which can be part of the apparatus 100. Immediately after the epitaxial growth process, 4H-SiC can exhibit a rapid dissipation of deposited corona charge. This is caused by surface diffusion of corona ions. This effect can interfere with the measurement of photo-induced charge photoneutralization if present. The pre-treatment to eliminate the surface diffusion effect can be performed simultaneously with the doping measurements on multiple sets of wafers. This capability can be added to an automated version of the apparatus 100.

[0058] A stable bare surface of depleted SiC typically exhibits good corona charge stability in the dark. The large bandgap prevents thermal generation of free minority carriers that can neutralize the corona charge. In the method, the charge stability in the dark after charging is verified by the corresponding stability of the surface voltage. For this purpose, the method can include a prescribed period of surface voltage measurement in the dark after charging but before illumination. Such a measurement is an example of a dark decay rate measurement, and produces a decay rate based on the change in the dark surface voltage V dark A large dark surface voltage decay rate and a nominal expected surface voltage amplitude that is lower than the deposited charge density can indicate defects that cause charge dissipation and thus interfere with the corona charge photoneutralization measurement. In this case, the measurement can be repeated on an adjacent wafer site that is free of defects and has a negligible dark decay rate. In epitaxial 4H-SiC used as an example in the present disclosure, the interfering defects are device-killing triangular defects, downfall defects, or carrot defects. These defects can have sub-millimeter dimensions, and shifting the measurement location by a distance larger than the Kelvin probe diameter (e.g., 3-5 mm in the case of a 2 mm diameter Kelvin probe) can be sufficient to eliminate the interference.

[0059] In some embodiments, additional verification steps can also be performed. For example, such a step can include a measurement of the surface voltage performed in the dark after turning off the photo-induced charge photoneutralization. This verification involves the irreversible nature of the photoneutralization, and is aimed at identifying other surface photovoltaic effects that are reversible in the surface depletion layer.

[0060] Typically, measurements are performed under conditions where corona charge neutralization is dominant. This can be tested in measurements of surface voltage recovery in the dark, immediately after irradiation. For optimized measurement conditions, the magnitude of this surface voltage recovery, indicating other effects, should be insignificant compared to the magnitude of corona charge neutralization, for example, in the range of about 100 mV or less for corona charge neutralization magnitudes of about 20 V or higher. In SiC, measurements are performed using approximately 10... 13 To about 10 14 photons / cm 2 This insignificant contribution can be achieved with low effective photon flux in the s range.

[0061] Characterization of semiconductor doping includes determining the time constant τ for corona charge neutralization under surface depletion conditions. ph Once τ is determined ph It is used as the doping index. τ ph The value is converted to a doping concentration of N. D =τ ph ·f cal , where f cal This is the so-called inverse calibration function. The time constant τ will be discussed further below. ph and calibration function f cal The determination.

[0062] Figure 2 The table shows the steps involved in the measurement process that can be used to acquire time-resolved surface voltage data to determine the light neutralization time constant. Figure 2 The included graph shows the surface voltage measured at each step. Surface voltage measurement is performed using equipment 100 or 100' and Figure 1A The illumination configuration shown measures the characteristics of a typical n-type epitaxial wafer of 4H-SiC. The wafer is held on a conductive chuck 110 by vacuum suction. During corona charging and measurement, the chuck 110 provides a back-side electrode coupled to the wafer capacitance. The non-contact characteristic of this method means that there is no physical contact with the front surface of the wafer (in this case, the surface of the epitaxial layer 103). All light in the system is initially turned off, and the wafer is in darkness. The wafer chuck is movable and for... Figure 2 Step 1 of the process involves positioning the wafer test site below the Kelvin probe electrode 132 to measure the initial surface voltage V0 in darkness, for example, V0 = 0.8V. After the initial surface voltage measurement, the test site 126 is moved to a position below the corona charging source (e.g., ...). Figure 1A and Figure 1B As shown), and in step 2, a charge density Q is specified. C (e.g., -1.5x10) 12 q / cm 2The negative charge is deposited on the surface of the epitaxial layer 103 at test site 126. As a result of the charge deposition, the surface is biased to be deeply depleted. For step 3, the charging site is shifted to a position below the Kelvin probe electrode 132. The voltage of the deeply depleted surface after charging is then measured, giving a value after charging, such as V. D = -59V. In step 4, the test site is illuminated with short-wavelength light having a photon energy higher than the 3.26 eV band gap of 4H-SiC, suitable for photonic neutralization of corona charge. For example, a 325 nm wavelength (hv = 3.82 eV) from a near-UV LED can be used for illumination. Figure 1A In the example shown, illumination below the Kelvin probe electrode 132 with light source 142 can be used, enabling continuous monitoring of the depletion layer surface voltage V(t) while illuminating the charging region. Figure 1B In the case of the exemplary device 100' shown, the illumination is applied to a location separate from the surface voltage probe. Therefore, monitoring cannot be performed simultaneously with illuminating the charging region. Instead, it is accomplished using a sequence of steps including a short illumination period (typically about 10 ms to 20 ms in duration) and surface voltage measurement. In this sequence, the charging region moves from a position below the surface voltage measurement probe 132 to a position below the illumination source 144, and then back to the surface voltage probe, and so on.

[0063] The final step, step 5, of the measurement cycle involves recording the surface voltage after the short-wavelength light is turned off. This value is... Figure 2 The dark voltage V is referred to as V. after Very small depletion voltage recovery was used to confirm the negligible contribution of the surface photovoltaic effect, in addition to corona charge light neutralization.

[0064] An example of the measured V(t) is in Figure 3A As shown in the figure, the initial portion of the voltage transient in the dark shows no decay before illumination. This dark portion is an indication of negligible charge dissipation due to charge neutralization associated with surface diffusion or any defects. During illumination, the rapid optical neutralization of corona charge is evident as a corresponding decay of the surface voltage. This recorded V(t) time correlation can be reliably analyzed using the Schottky barrier voltage-charge equation, thus providing the time constant τ for determining the optical neutralization time of corona charge. ph The method involves depleting the surface potential barrier and surface charge density (V). D ~Q C 2 The quadratic relationship between τ and τ can be used to derive the surface voltage decay corresponding to the corona charge decay. In practice, τ ph It can be determined by the inverse slope of the logarithmic surface voltage time decay during the light neutralization period. Where V0 is the initial precharge value, including the offset related to Kelvin probe calibration. For Figure 3A The V(t) data in, which is in Figure 3B As shown in the image.

[0065] This fitting process yields the average τ within the specified surface voltage range. ph The value, in this case, is in the range of -5V and -59V. Figure 3B A good linear fit indicates a constant τ within that range. ph This value is consistent with the uniform doping depth in the epitaxial layer.

[0066] For the doping depth distribution, a voltage-specific time constant corresponding to a given depletion surface voltage V can be used, based on the light neutralization V vs irradiation time characteristic V(t). It is calculated from the relative surface voltage decay rate as follows: Figure 3C It shows that for Figure 3A The data in the figure are voltage-specific τ as a function of the depletion surface voltage. ph It indicates a constant τ within the voltage range that corresponds to a uniform doping depth distribution in the epitaxial layer. ph value.

[0067] Figure 4 The example illustrates the logarithmic surface voltage time decay of an epitaxial wafer with three differently doped layers. Each layer is characterized by a corresponding τ that is proportional to the doping value. ph1 τ ph2 and τ ph3 Layers 1, 2, and 3 are counted from the top of the epitaxial layer structure.

[0068] Figure 5 This illustrates the application of n-type 4H-SiC with wavelength λ = 325 nm and effective incident photon flux φ. eff =1.2x10 13 photons / cm 2 Light irradiation by s, light neutralization time constant τ ph With doping concentration N D The correlation. This correlation can be used to determine the calibration function N. D vsτ ph It corresponds to a specific doping range N D and irradiation conditions λ and

[0069] Light neutralization time constant τ ph With dopant concentration N D The increase is usually for the same corona charge Q. C The smaller surface exhausts the width W D =Q Cq N D The result is a smaller photogeneration of holes in the depletion. As shown in Figure 6 , the free holes h + are active in the light and in the depletion, and they are guided to the surface by the depletion field. At the surface, the negative corona ions capture the holes, Q - + h + → Q↑ and the neutralized corona species desorb from the surface.

[0070] The smaller hole generation caused by the reduced depletion width under increasing doping can be overcome by increasing the incident photon flux Q eff . As shown in Figure 7 , τ ph is inversely proportional to the incident photon flux. Thus, by increasing the incident photon flux, a fast measurement can be achieved at higher N D . In this approach, the measurement speed is the main advantage. In this regard, the inverse proportionality of the time constant τ ph and the incident photon flux ( ph ~ 1 / φ eff ) can be used to optimize the measurement. For example, for N 16 in the range of 10 -3 cm D , a time constant τ 14 in the range of 0.1 s can be achieved by increasing the photon flux to 10 2 photons / cm ph s. In combination with an increased photon flux and a shorter wavelength of the light, e.g. λ = 315 nm or λ = 300 nm, a fast measurement of a high doping, e.g. 10 18 cm -3 , is possible.

[0071] As mentioned before, a calibration can be used to convert the measured time constant into a doping concentration. For this purpose, a family of specific calibration functions can be determined for a specific doping range and (λ, φ eff ) values. For an accurate measurement, individual specific calibration functions can be limited to a subset of the entire range of values for N D and , e.g. an accurate calibration function can cover a range of one order of magnitude for N D and φ eff . The calibration functions can be determined using measurements and accurate settings of the illumination conditions on reference wafers with known doping concentrations. For example, wafers with N D pre-measured with a mercury probe C-V method or wafers pre-measured with a corona non-contact C-V method (CnCV) can be used. The exact same illumination conditions can then be used when measuring the doping in a manufactured epitaxial layer in a production environment.

[0072] From the calibrated τ ph vs N D Data quantifies the doping concentration using the inverse function N cal vs τ D ph . The measurement is performed at the same wavelength as used for calibration, e.g. Figure 5 and Figure 7 with λ = 325 nm. The photon flux φ eff used during the measurement is typically different from the photon flux φ cal used for calibration, as it is adjusted to take into account the measurement speed requirements. The dependence of τ ph on the incident photon flux is used to normalize the measured τ ph . The normalized value is then used in the inverse function f D of N ph vs τ cal to calculate the doping concentration as In the example of Figure 5 , φ cal calibration is performed at φ 2 = 12 e 13 photons / cm cal s. However, applied to different photon fluxes as shown in Figure 7 , this calibration gives correct N D values, within 0.5%.

[0073] An alternative method uses a wafer-specific calibration performed during the actual measurement of a given wafer. The apparatus 100 and 100' shown in Figure 1A and Figure 1B respectively are suitable for this purpose. The wafer-specific calibration involves: (1) a measurement of the corona charge neutralization time constant τ ph and (2) a different non-contact C-V (CnCV) measurement of the doping concentration ND. These two independent measurements are performed one after the other on the same wafer site. The results are used to determine the inverse calibration value function f cal that satisfies N D = τ ph .f cal . On all other test locations of this wafer, only τ phk (k is the location number) is measured. The doping N phk at different k locations on the wafer is determined using the measured τ cal and the same value of the wafer-specific function f Dk = τ phk .f cal .

[0074] ​Generally, the measurement with wafer-specific calibration is done for all measured sites with the same illumination conditions. This condition can be optimized for different doping ranges. However, no precise knowledge of the incident photon flux is required, since the method uses the measured τ ph Instead of normalizing to the photon flux The above-described measurements can be performed at any number of sites on a wafer. A typical production line test involves 12 wafer site measurements or 49 site wafer mapping. The wafer-specific calibration adds a unit site CnCV doping measurement time of typically 40 seconds. The τ ph measurement time at a single site is shorter, e.g., about 4 seconds. For a 12 site test with wafer-specific calibration, the measurement time per wafer can be about 90 seconds. This measurement speed can give a throughput of about 30 wafers / hour, including wafer handling overhead time. For a 49 site wafer mapping with 40 second additional calibration time, resulting in a total measurement time of about 240 seconds and a throughput of about 13 wafers / hour with wafer handling overhead.

[0075] The measurement using wafer-specific calibration can reduce the uncertainty by reproducing the illumination conditions and still provide very fast, high throughput, epitaxial SiC doping monitoring.

[0076] The above-described photo-neutralization method employs a specific photo- neutralization time constant definition based on a logarithmic analysis of the charge decay. However, it should be noted that in general, the sensitivity to the photo- neutralization kinetics of the doping determined in the present application can use different time constant definitions for the doping measurement, e.g., the half-life of the charge decay, the stretched or compressed exponential time constant of the charge decay, etc.

[0077] Although the single application of the photo-neutralization method is described above, the technique and benefits are not limited to bare SiC doping monitoring only. More generally, the corona photo-neutralization technique can be applied to characterize other wide bandgap materials and structures, e.g., GaN and HEMTs. For example, the technique can be applied to determine HEMT characteristics, e.g., pinch-off voltage (V p ) and 2DEG sheet charge.

[0078] Other embodiments are therefore in the attached claims.

Claims

1. A method for characterizing semiconductor doping in a wide-bandgap semiconductor sample, the method comprising: The initial value V0 of the surface voltage at a region on the surface of the semiconductor sample is measured in darkness; By depositing a prescribed corona charge in the area, the region is charged to deep depletion in the dark; The surface voltage at the area was measured in the dark after charging. Using a specific photon flux φ eff The specific photon flux φ is generated when the area being charged is illuminated by light. eff It has photon energy above the band gap of the semiconductor, sufficient to generate free minority carriers in the semiconductor sample and thus cause optical neutralization of the corona charge; Non-contact time-resolved measurements of surface voltage V were used to monitor the relationship between light neutralization-induced corona charge decay and irradiation time t in the region. Analyze the monitored time-resolved surface voltage decay data V(t) to determine the light neutralization time constant τ. ph ; as well as Using a specific photon flux φ eff The light neutralization time constant τ ph It serves as a semiconductor doping index, and its value characterizes the semiconductor doping concentration in the region.

2. The method according to claim 1, further comprising a specific corona charge light neutralization time constant τ based on the surface voltage. ph vs V is used to characterize the depth distribution of doping concentration in the semiconductor sample.

3. The method of claim 1, further comprising, based on the average τ within the surface voltage range ph The value is used to characterize the doping of a single epitaxial layer.

4. The method according to claim 1, wherein, τ ph Determined according to the following equation: Where t is the irradiation time, and V const It is the surface voltage probe offset.

5. The method of claim 4, further comprising using a known doping concentration value N. D τ of one or more reference samples ph The calibration measurement determines the calibration function τ. ph vs N D And represented as f cal Inverse calibration function N D vs τ ph .

6. The method according to claim 5, further comprising: In the effective photon flux φ eff The time constant τ of the measurement ph Normalized to calibrated photon flux φ cal ,as follows as well as Use the normalized time constant according to the following equation. Determine the absolute doping concentration:

7. The method of claim 4, further comprising calibrating τ for doping concentration. ph ,in, The calibration is specific to a semiconductor wafer sample, and the calibration is based on τ at multiple sites on the wafer sample. ph The characterization process is performed at one site on the measured wafer sample, where all τ ph Measurements were performed under the same irradiation conditions.

8. The method according to claim 7, wherein, The calibration includes measuring the corona charge neutralization time constant τ using a non-contact corona CV (CnCV) method. ph and independent measurement of doping concentration N D The two measurements are performed sequentially, one after the other, at the same wafer site and within a range of depletion voltages caused by similar corona charges.

9. The method of claim 8, further comprising using τ ph and N D Determine if N is satisfied D =τ ph .f cal The chip-specific inverse calibration function value f cal .

10. The method of claim 7, further comprising determining τ at each wafer site. ph And it uses a wafer-specific inverse calibration function value f determined at a single site. cal To determine the doping concentration at all different wafer sites.

11. The method according to claim 1, wherein, The surface voltage was measured using a non-contact vibrating Kelvin probe.

12. The method according to claim 1, wherein, The area is irradiated while the surface voltage is monitored.

13. The method according to claim 1, wherein, The area is irradiated separately from the surface voltage being monitored.

14. The method according to claim 1, wherein, The semiconductor sample includes semiconductors selected from the group consisting of SiC, GaN, and AlGaN.

15. The method of claim 1, further comprising a dark surface voltage V measured after corona charging but before the irradiation. Dark and dark decay rate ΔV Dark / Δt Dark To identify defective regions in the semiconductor sample.

16. The method of claim 1, further comprising a dark attenuation magnitude ΔV measured after the irradiation. Dark To identify the contribution of surface photovoltaic effects, other than the corona light neutralization, to the monitored surface voltage.

17. The method according to claim 2, wherein, Characterizing the depth distribution of doping concentration in an epitaxial wafer with multiple layers having different doping concentrations includes determining the transition surface voltage values ​​between different doping layer concentrations.

18. The method of claim 1, wherein, The surface voltage was measured using a transparent probe.

Images