Using Scanning Microwave Impedance Microscopy for Nanoscale Mapping of Permittivity and Conductivity

Scanning microwave impedance microscopy, also known as sMIM, is an atomic force microscopy (AFM)-based method used for the characterization of materials and devices. The reflected microwave signal arising from the interface between tip and sample holds data about the electrodynamic characteristics of the sample surface beneath the tip apex.

Real-time detection and processing of the reflectance enables the sMIM technique to directly access the materials’ conductivity and permittivity. When the sample surface is being scanned with an AFM-type sMIM probe, sMIM can image the differences in capacitive (sMIM-C) and resistive (sMIM-R) characteristics. In this detection method, there is no need to make electrical contact between the substrate and the sample, as the sMIM is based on the capacitive coupling between the sample and the tip.

In device characterization and materials research, a material’s physical characteristics are usually exposed by determining the response from an electromagnetic excitation. The most optimum spatial resolution for electrical characteristics that can be obtained in the far-field radiation region is on the order of a half wavelength called the Abbe's diffraction barrier. However, when subwavelength apertureless or aperture waveguide methods are used, this far-field limit collapses. This forms the core of near-field measurement, offering spatially resolving power that is characterized by the spatial level of the evanescent fields, instead of the free-space wavelength.

The free-space wavelength is over a millimeter in the microwave regime, and this makes it unfeasible to define materials on the nanoscale - below one millionth of the microwave wavelength. It was in the 1950s that near-field microscopy was initially described, approximately three decades following the original suggestion by Synge.

Near-field microscopy has now been extended to other spectral regions, such as infrared, far infrared, and visible regimes. In this method, the probe has to be extremely close to the surface of the sample because the near-field contribution of the electromagnetic field or evanescent decays quickly with distance.

The use of metal-coated or metallic probes in AFM allows near-field measurements as they are suitable for AFM imaging as well as near-field detection. This enables the acquisition of serial pixel-to-pixel data where the spatial resolution is mainly measured by the size of the tip apex. Such methods have now become commercially available; some instances include scattering SNOM in the infrared domain like Bruker's Inspire, and tip-enhanced Raman spectroscopy like Bruker's Innova-TERS. A significant example for microwave frequencies is sMIM, which is the main focus of this article.

In the sMIM method, the tip removes the microwave transmission line. The tip impedance is a function of the tip-to-sample impedance, the tip stray capacitance, and the tip-to-sample coupling capacitance. When the distance between the tip and the sample is relatively smaller than the typical tip size and proper shielding is applied, the sample "near-field" impedance dominates the tip-sample impedance, which holds data about the sample’s electrical properties. As an "open" end of the transmission line, the tip is an impedance discontinuity which leads to microwave reflection. As the reflection coefficient is a function of the tip impedance, it is therefore the sample "near-field" impedance. By imaging the reflection of the microwave, it would be easy to access conductivity, permittivity, and other local electrical properties of the surfaces of the materials below the tip.

sMIM offers a number of benefits in contrast to other AFM-based electrical modes. Firstly, it can directly image the difference of conductivity and permittivity as the tip scans across the surface of the sample. This helps to distinguish materials with a broad dynamic range, spanning from semiconductors and metals to dielectrics and insulators. Secondly, the measurement is based on the capacitive coupling or microwave reflection between the tip and the sample. So in sMIM, electrical contact is not required between the substrate and sample to map electrical characteristics, unlike the surface potential/Kelvin Probe force microscopy (KPFM) or conductive AFM. However, this poses a major concern, particularly on samples where an electrical contact cannot be introduced or where electrical contact changes the properties of the material.

Two standard examples are MoS₂ and graphene, but in effect, a large number of nanoscale 2D materials are suitable for this method. This capability can also be leveraged by measuring islands, particles, or films in an insulating material or on a non-conductive substrate. sMIM is shown to be a truly non-invasive technique for nanoelectrical measurements. Thirdly, when an AC bias is applied to modulate the sample, it is possible to capture the matching AC response of sMIM signals. In the case of non-linear materials, like semiconductor devices and materials, the AC-sMIM signal differs with doping density.

This AC output, analogous to conventional scanning capacitance microscopy (SCM), possesses amplitude and phase signals that mirror the varied types of dopants as well as the difference in the doping density. In comparison to SCM, sMIM can also offer capacitive and resistive channels, in addition to the AC response of the resistive information. For failure analysis, critical data is obtained through resistance-voltage and local and site-specific capacitance- voltage spectroscopy. In this manner, sMIM can directly access the conductivity and permittivity, which remain relatively complicated for other techniques, including SCM.

sMIM is also based on the interaction between the matter and microwave, where the electromagnetic field can enter a specific distance within the sample - exponential decay within a medium. This enables electrical imaging of subsurface structures. The depth sensitivity relies on the near field’s spatial distribution within the medium. These benefits enable users to apply the sMIM technique to a variety of materials, such as polymers, biological samples, ferroelectrics, low-dimensional structures, semiconductor devices, composites, and so on. This article describes some examples that exploit the special capabilities of Bruker's AFM systems.

Performing sMIM with Bruker AFMs

Bruker AFM is configured with the sMIM setup, which includes a probe holder that not only houses standard AFM probes, but also accommodates customized, shielded AFM probes. During sMIM operation, the probe becomes a part of the transmission line which transmits the microwave to the interface between the tip and sample. Hence, a conductive probe has to be used for electrical measurement. In order to justify the minimal topographical convolution and the noise level, probes with a coaxial shielding structure can be helpful. This commercially-available coaxially shielded probe is batch-fabricated and is known to be a major factor that measures very high sensitivity, a factor of 10 higher than conventional methods. The probe features a core metal trace embedded in the insulating dielectric layers (Figure 1). It transmits the microwave signal between the metal tip and electrode. A coaxially shielded structure is formed by the metal layers on either side of the cantilever.

Figure 1. (A) Illustration of the coaxially shielded structure of the sMIM probe with a tip radius <50 nm. The metal layers on both sides of the cantilever form a coaxial shield to eliminate stray and parasitic capacitance, and reduce noise from the surrounding environment. The central metal line transmits the microwave to the tip apex. (B) and (C) are SEM images of a typical sMIM probe.

This shielding considerably prevents parasitic and stray capacitance, and also minimizes other noises from the adjacent environment. It even makes sure that interaction occurs only between the tip apex and the sample during the scanning process. The coaxially shielded structure, along with the <50 nm radius, offers high spatial resolution and electric sensitivity. Further, when a conductive probe is used on Bruker's flexible AFM platforms, it helps to combine other electrical measurements with sMIM to deliver concurrent multimodal data like surface potential and piezoelectricity.

In a standard sMIM system, a corresponding network facilitates the transmission of a low-power signal of 3GHz to the tip apex. By default, this electronics output is set at -20 dBm, but can be differed between -10 and -40 dBm. Usually, the optimal frequency tends to have a small deviation from the 3 GHz signal, because of the difference of separate probe properties, and is chosen to reduce the reflectance (r) from the probe interface assembly and the probe. If the tip is scanning across the surface of the sample, sMIM has the ability to locate the reflected microwave signals as a result of a change in the system impedance from Z₀:

The Ztip, which represents the probe-sample impedance, can be defined as a lumped element model (Figure 2). The reflected microwave signal arising from the tip-sample interface is sent back inside the circuit. The out-of-phase (imaginary) and the in-phase (real) signals are resolved by an RF mixer, and these signals are matched with the permittivity and conductivity of the sample, respectively. Both output signals are marked as sMIM-C and sMIM-R. After making an alteration to the demodulation phase on a reference sample, the capacitive (sMIM-C) as well as the resistive (sMIM-R) channels are separated, imaged spatially, and shown in parallel as "maps." or images.

Figure 2. Schematic diagram of sMIM. The tip-sample impedance is described as a lumped element circuit model. The reflected microwave is resolved by an RF mixer (Mixer 1) into sMIM-C (capacitive) and sMIM-R (resistive) channels, which are related to permittivity and conductivity of the sample, respectively. When the sample is AC biased, the AC component of the sMIM signals is further resolved by lock-in amplifiers to provide dC/dV information for carrier profiling and dR/dV channels for other analysis.

If the sample is AC biased, lock-in amplifiers further resolve the AC component of the sMIM signals to afford the out-of-phase and in-phase signals. For the sMIM-C channel, this is relative to the phase signals (akin to SCM) and dC/dV amplitude. In this manner, sMIM can perform carrier profiling (as SCM) and at the same time affords the low-frequency (3 GHz) dielectric contrast of the surface of the sample. This aspect significantly extends the utility of sMIM in semiconductor failure analysis as opposed to SCM. Further, the sMIM method offers the specialized channel dR/dV, providing yet additional information for the characterization of devices and materials. It can also be operated when the cantilever is oscillating, or in contact mode.

Considering these sMIM capabilities, Bruker's unique Peakforce Tapping mode has a major significance and provides several benefits. Firstly, Peakforce Tapping allows concurrent mapping of electrical characteristics with mechanical characteristics such as adhesion, deformation modulus, and dissipation via PeakForce QNM. Secondly, very low imaging forces (typically <100 pN) can be used, thanks to accurate, linear force control in PeakForce Tapping, allowing the analysis of soft and delicate materials, or materials that are loosely fixed to the substrate. For the same basis, the lifetime of the tip, which is a major concern in electrical AFM modes, is considerably increased.

The more the interleave mode is used, the more data, such as piezoelectricity or surface potential, can be simultaneously obtained.True multi-dimensional nanoscale characterization can be realized by machining this data with surface morphology. Therefore, sMIM serves as a powerful tool for device failure analysis and materials characterization. This article includes a range of application examples for a variety of applications.

sMIM Case Studies

Static Random Access Memory (SRAM)

SRAM samples that are polished on the front side possess a range of topographical features with respect to the implant regions. They are utilized as a test and reference sample for SCM measurements to validate the fundamental capabilities to image doped samples and distinguishing the types of carriers and the enormous doping levels on a device. When users scan in SCM mode, a small capacitor is formed by the sample and tip. It is discovered that the capacitance differs with the accumulation or depletion depth adjusted by an AC sample bias applied to the sample.

SCM is capable of determining AC-modulated capacitance with variation in applied AC sample bias that is dC/dV. Both the amplitude and dC/dV phase are used to distinguish doping concentrations and dopant types, respectively. While sMIM is characterized by analogous functionality, it also introduces the capability of mapping dielectric characteristics.

An instance of sMIM measurements on SRAM samples is shown in Figure 3, and the topography in 3D rendering is illustrated in Figure 3A. The imaged area measures 15 x 7.5 µm with contact mode displaying a flat substrate, a rough region measuring about 100 nm high, and features measuring about 150 nm high.

Figure 3. sMIM images of a SRAM samples. (A) Topography channel, 15 µm x 7.5 µm, and ~150 nm tall features; (B) sMIM-C channel as a skin covers on topography; (C) sMIM dC/dV phase channel on topography; (D) sMIM dC/dV amplitude channel on topography; (E) a topographically featureless (roughness < 0.1nm) region with two clearly different doping regions; and (F) cross-sectional analysis of the phase transition as indicated by the red line on the image in (E).

In Figure 3B, sMIM-C channel reveals a difference in permittivity across the surface of the sample. On each 150 nm tall plateau, the edges exhort permittivity variation as opposed to the plateau itself. In Figure 3C, the dC/dV phase image overcomes the entire details of this "O" structure, namely n-type low-doped drain, epitaxial p-type substrate, and n-type and p-type gate channels. Also, the image of relative doping density reveals the depletion layer when shifting from p- to n-type channels, as denoted by the dark lines present on the flat plateau (Figure 3D). Figure 3E shows that it is possible to obtain high-resolution electrical imaging on surfaces that are topographically featureless. In Figure 3F, the cross-sectional examination of the phase image is plotted, revealing a clear transition of less than 10 nm from p- to n-type doped regions.

Inverted Vertical Insulated Gate Bipolar Transistors (IGBTs)

One of the key types of discrete power semiconductor devices is IGBTs. These types of devices were designed to integrate rapid switching and high efficiency. Nanoscale characterization of the architecture and a better understanding of the carrier type and doping level at varied locations in the device are important to ensure optimum design and performance of IGBT devices.

Figure 4. SEM and sMIM studies of an IGBT cross-section-polished sample from Chipworks. (A) Comparison of SEM and sMIM-C results on resolving different materials; (B) sMIM dC/dV phase image; (C) sMIM dC/dV amplitude channel; and (D) An image constructed from (B) and (C), similar to a traditional SCM image. Scale bar is1 µm. SEM image courtesy of Chipworks.

An example of microscopic analyses on an IGBT cross-sectioned sample is shown in Figure 4. As shown in Figure 4A, SEM is extensively utilized to expose the diverse device components, like poly-Si trench gate, gate oxide, metal contacts, and SC-Si (single crystal silicon) source and body. When the local dielectric difference is being imaged by sMIM-C, SC-Si domains with varied doping densities are also differentiated. This is clearly shown in Figure 4A through the distinct emitter domains in the sMIM-C channel. The emitters directly link to the tungsten contact strip. The rough emitter/source metal contacts as well as the ~125 nm tall tungsten contacts may present difficulties for tip-lifetime and imaging, and may also lead to topographical convolution within the electrical data.

This issue can be effectively prevented by combining PeakForce Tapping with sMIM, which is discussed in the subsequent sections. As shown by the IGBT sample, sMIM possesses analogous capabilities like SCM for carrier profiling. The dC/dV amplitude and phase In Figure 4C and Figure 4D provide critical data regarding the carrier concentration and carrier type, in that order. Based on the dC/dV amplitude and phase, it is possible to design an image, showing the dopant types and doping density, analogous to a conventional SCM image (Figure 4D).

For failure analysis, SCM, SEM and other conventional imaging techniques show restrictions in studying the various aspects of specific devices, like the IGBT illustrated here. In the case of SEM, unique, difficult-to-implement etching methods are needed to preferably etch doping regions for adequate contrast. As a result, SCM is generally applied to overcome the varied doping regions. SCM, however, is only susceptible to doped semiconductor regions or nonlinear regions, and does not reveal any contrast in dielectrics, metals or regions with uneven oxide thickness. sMIM, on the contrary, requires no special sample preparation to overcome the doped regions, and can even distinguish metals, oxides, and semiconductors.

Complementary Metal-Oxide Semiconductor (CMOS) Image Sensors

A front-side illuminated global-shutter CMOS image sensor was employed to demonstrate sMIM’s ultra-high sensitivity to various doping concentrations, and the contrast from those linear regions as shallow trench separation, dielectrics, and polysilicon (Figure 5).

Figure 5. sMIM images of a front-side illuminated global-shutter CMOS image sensor from Chipworks.14 The sample features 3µm pitch pixels. (A) Topography; (B) sMIM-C; (C) sMIM-R; (D) dC/dV phase, solid yellow, 90° and solid blue, -90°; (E) dC/dV amplitude; and (F) site-specific capacitance-voltage curves at locations labelled in (B). The numbering in (B) features: (1) n-well storage diffusion; (2) n-well photocathode diffusion; (3) shallow trench isolation; (4) contact; and (5) p-type substrate surrounding cathode. The sMIM-C channel illustrates success in differentiating materials with a large dynamic range: metallic, semiconducting, and insulating. Scan size is 5µm x 5µm. For color code from (B) to (E), the yellower indicates the lower value. (B) and (C) have the sample scale bar in color contrast (-30mV to 30mV). These images were captured with a Dimension Edge AFM at Chipworks.

3 µm pitch pixels are included in the sample. In Figure 5 B, the sMIM-C channel resolves all of the important features, including n-well photocathode diffusion; n-well storage diffusion; shallow trench isolation; p-type substrate surrounding cathode; and contact. At the same time, the sMIM-R signal is captured as well (Figure 5C). When compared to the sMIM-C, the sMIM-R response is weaker which can be probably attributed to the presence of the surface oxide, and also due to the fact that domains are highly insulating or conductive. At intermediate conductivity, the sMIM-R signal is related to resistive peaks and loss.

The dC/dV phase emerges as predicted and analogous to traditional SCM imaging (Figure 5D). Differentiation is done to p-type regions (solid blue, -90°) and N-type regions (solid yellow, 90°), while all of the other remaining areas emerge with a phase signal about 0°. In Figure 5E, it can be seen that the dC/dV amplitude has varied doping densities with respect to non-linear material, for example Si, in this sample. The dopant sensitivity spans between 10¹⁴ cm^-3 and 10²⁰ cm^{- 3} for sMIM. The dC/dV amplitude channel appears to be analogous to the sMIM-C; however, the sMIM-C channel showed a wider dynamic range. For instance, when locations #4 (highly doped Si) and #3 (oxide) in the sMIM-C channel are compared with the dC/dV amplitude channel, the SMIM-C signal in #3 is found to be different from #4, while dC/dV amplitude displays no contrast between #3 and #4. While the sMIM-C signal remains monotonic with dopant concentration, the variation in the signal can be attributed to dC/dV amplitude being peaked at intermediate dopant values.

sMIM is also capable of conducting capacitance-voltage (C-V) spectroscopy utilizing the sMIM-C signal. For semiconductor devices, site-specific C-V spectroscopy can be utilized as a failure analysis tool. Plot #1 and #2 substantiate n-type regions (n-well photocathode diffusion and n-well storage diffusion) plot #3 is in parallel to the oxide of shallow trench isolation, and plot #5 shows the p-type substrate enclosing the cathode (Figure 5F). Figure 6 demonstrates an example of another type of CMOS image sensor device, further showing the ability of sMIM in acquiring a nuanced view of the structure of the device beyond the carrier type and doping data,

Figure 6. sMIM images of a Samsung S5K2P2XX CMOS image sensor from Chipworks. This sensor has 1.1µm pixel features. SEM imaging is challenging to differentiate different implant regions. Only under certain etching conditions will the SEM image display some n- and p-type contrast.

1.1 µm pixel features are integrated into the sensor and the same can be fully characterized by sMIM. The doping gradient of the photosensitive p-doped region is very important to the device. sMIM-C and dC/dV amplitude images afford high sensitivity to small dopant differences. Moreover, sMIM dC/dV phase distinctly reveals the types of doping carriers with a high degree of contrast. The pinning layer is also resolved with excellent detail, making it easy to directly measure the spacing (0.8 µm) and the thickness of the layer (~140 nm).

Semiconductor Metal Oxide Films

In different phases, iron oxide nanoparticles are extensively utilized in a wide range of applications, such as photovoltaics, bio sensors, storage devices, and solar-fuels generators. The electrical properties of these iron oxide nanoparticles significantly affect the performance of a device. Here, sMIM again is a powerful tool for the characterization of nanoparticle materials.

A γ-Fe₂O₃ nanoparticle film utilized in recording media is shown in Figure 7A. Spindle-like particles, measuring 100 x 500 nm in size, were casted onto a continuous film. Conductivity inhomogeneity is spread across the surface of this sample, which was utilized as a reference in conductive AFM modes. The sample’s local impedance differs across the surface of the sample and also reveals contrast on the sMIM-C as well as on the sMIM-R channels (Figure 7).

Figure 7. sMIM images and local capacitance- / resistance-voltage spectroscopy of a γ-Fe₂O₃ nanoparticle sample. (A) Topography shows 100 nm x 500 nm spindle particles; (B) sMIM-C reveals the location permittivity variation; (C) dC/dV amplitude; (D) dC/dV phase indicates different dopant polarities; (E) sMIM-R demonstrates different local conductivities; (F) dR/dV amplitude; (D) dR/dV phase; (H) highresolution dC/dV phase image; (I) site-specific capacitance-voltage spectroscopy. Locations are indicated in (H); (J) high-resolution dR/dV phase image at the same location as (H); (K) site-specific resistance-voltage spectroscopy.

This sample clearly shows that sMIM can be applied to both semiconductor devices and other types of materials systems. In the case of the y- Fe₂O₃ nanoparticle film, iron oxides containing rich oxygen vacancies are assumed to be n-type semiconductor materials. Numerous bulk measurements have demonstrated this factor. Conversely, in Figure 7D, the sMIM dC/dV phase signals display phase contact with a difference of 180°, suggesting different types of dopants for different nanoparticles spread across the film at the nanoscale. In Figure 7F and Figure 7G, sMIM dR/dV amplitude and phase signals clearly reveal the domains and amplitude contrast with phase differences of 180°. This aspect further substantiates the fact that this iron oxide is a non-linear material. When dR/dV was compared with dC/dV images, clear differences were seen. The contrast responses in the phase images clearly show the various aspects of the grains, within a certain grain (Figure 7H and Figure 7J).

As depicted in Figure 7I, C-V curves were acquired from two clear phase domains, shown in Figure 7H, which anticipates reverse signs of the slopes for both these curves. Also, the slopes represent higher doping concentration on location #1 than on location #2. This provides a better understanding about the material’s physics. In a similar way, voltage-dependent sMIM-R sweeps were also carried out at locations illustrated in Figure 7J, and the ensuing R-V curves are displayed in Figure 7K that further validates the uneven non-linear conductivity across the surface of the sample. This would present an issue during device design, because when an external voltage is applied, the conductivity of the samples changes at the nanoscale.

Buried Structures

The sMIM method is built on the electromagnetic interaction between matter and microwave. The electromagnetic field, concentrated, can enter into dielectric materials while decomposing exponentially, which is a facet of near field or evanescent interaction. When imaging buried structures, this long-range microwave-matter interaction has been shown to be quite useful. Figure 8 shows the sMIM’s subsurface imaging ability on a flat sample with dielectric differences resulting from the buried structures.

A patterned SiO₂ structure in the sample is buried 133 nm within a Si3N4 film (Figure 8A), and to prevent residual topographic features, the surface was fully polished. A 0.4 nm surface roughness is obtained, as illustrated in Figure 8B. Due to the permittivity variation between the nitride (ε = 7.5) and the oxide (ε = 3.9), the sMIM-C channel (Figure 8C) reveals a response from the oxide patterns hidden within the nitride. The sMIM determines a lower capacitance where there is an oxide, contrary to where there is no feature. As predicted, the nitride and oxide are both insulating, but no conductivity difference was found over the sample, and no contrast was seen in the sMIM-R channel in Figure 8D.

Figure 8. (A) A buried structure reference sample was fabricated on a Si wafer. 90nm thermal SiO₂ was grown on the wafer and patterned. 1um LPCVD Si₃N₄ was deposited to fully cover the SiO₂. Chemical mechanical polishing was used to polish the sample until the Si₃N₄ was about 223 nm; (B) The topography of the polished reference surface with a roughness of 0.44 nm in the imaged area. (C) and (D) are sMIM-C and sMIM-R channels respectively. As expected, contrast only appears in the dielectric channel (sMIM-C) and vanishes in the conductivity channel (sMIM-R).

PeakForce Imaging and sMIM

In the case of PeakForce Tapping, the sample and the probe are allowed to come into contact intermittently as the tip scans over the surface of the sample. Compared to the cantilever's resonance of more than 10 kHz, the 1~ 2 kHz tapping frequency is considerably lower. Through this intermittent contact mechanism, lateral forces are prevented during imaging, and this prevents damages to the sample and the tip from the shear force. The peak force or maximum force on the tip is controlled by the feedback loop for each tapping cycle. The control algorithm directly responds to the force interaction between the tip and sample through the sinusoidal, low-frequency, off-resonance mechanical modulation.

In PeakForce Tapping, the sinusoidal ramping enables the tip to approach the surface of the sample with a near-zero speed, unlike the standard linear ramping with a triangular wave, as utilized in Force Volume. This allows stable and direct force control so that the peak force remains below 100 pN or below 50 pN. This gentle and well-controlled force together with the 1 ~2 kHz tapping frequency, PeakForce Tapping protects the sample and the tip against damage and also provides high-resolution imaging while keeping usual imaging rates. For each tapping cycle by tip movement towards the Z direction, PeakForce Tapping obtains a force curve, allowing concurrent imaging of quantitative mechanical characteristics that are directly associated with the topographic data.

sMIM can also be combined with PeakForce Tapping: PeakForce sMIM, just like Bruker’s other popular PeakForce-enabled electrical modes such as PeakForce SSRM™, PeakForce TUNA™, and PeakForce KPFM™. However, it should be noted that the oscillation frequency (1 to 2 kHz) of PeakForce Tapping leads to a contact time of several tens to hundreds of microseconds, but this contact time is long enough to acquire the sMIM signals. sMIM can be directly carried out to capture the cycle-averaged sMIM signals, or alternatively with a sensor circuit akin to PeakForce TUNA to collect signals with a better signal-to-noise ratio. Interleave mode can also be used, just like the PeakForce KPFM mode. In this mode, dark lift scanning through the AFM laser off in the lift scan can help characterize light-sensitive materials.

PeakForce sMIM for Carbon Nanotubes

Figure 9 shows a typical instance of PeakForce sMIM imaging of aligned carbon nanotubes (CNTs) fixed loosely to the insulating substrate. This sample can be easily imaged in PeakForce Tapping mode, while the same would be too complex for contact mode.

Figure 9. PeakForce sMIM images. (A) Topography; (B) adhesion; (C) DMTModulus; and (D) sMIM-R maps of carbon nanotubes (CNTs) aligned flat on an insulating substrate. Mechanical channels have higher sensitivity on visualizing the CNTs compared to topography. The sMIM-R channel confirms these CNTs have different conductivities as indicated by the square boxes on adhesion and sMIM-R channels. Note that no electrical contact is required as sMIM is performed based on the tip-sample capacitive coupling. Sample courtesy Greg Michael Pitner and Professor H.-S. Philip Wong, Stanford University.

The topography channel is shown in Figure 9A, displaying the aligned CNTs on the insulating substrate. The mechanical channels - modulus and adhesion - shown in Figure 9B and Figure 9C, can resolve the CNTs in a better way, as they are quite sensitive to local mechanical characteristics. As shown by the sMIM-R channel, different CNTs can possess different conductivities. Comparison between the white box in Figure 9D and the black box in Figure 9B shows that these CNTs exhibit a similar adhesion yet varied conductivity. All of the debris particles on the surface disappear in the sMIM-R channels because they exhibit an analogous sMIM-R response as the substrate. Most importantly, no electrical contact was made during the measurement.

Since sMIM imaging is based on capturing the microwave reflectance from the impedance discontinuity in the transmission line, electrical contact was not needed unless modulation of the sample with an AC bias is required for carrier profiling. This provides immense benefits for examining nano- and micro-size delicate materials where it is difficult to make electrical contact, or where electrical contact can change the properties of a sample.

PeakForce sMIM for IGBTs

For failure analysis, SEM is often used to characterize IGBT devices for material structure. Just like the contact mode sMIM (Figure 10), PeakForce sMIM’s sMIM-C channel reveals those features as the metallic gate contact, the dielectric gate oxide layer, and the semiconducting emitter region. The emitter region is challenging to SEM and is not displayed on the SEM image. When PeakForce Tapping is used, it becomes easy to scan on the metal contact region, which is significant for contact mode. The tip will wear off when this region is imaged in contact mode, reducing the image resolution. These problems can be resolved by utilizing PeakForce sMIM. As illustrated in Figure 10, distinct electrical and topographic features with more nuanced details can be observed in the semiconductor as well as in rough metal regions. The same probe was utilized for over 10 images without affecting the performance.

Figure 10. PeakForce sMIM imaging of an IGBT device. Compared to contact mode sMIM in Figure 4, clear topographic and electrical features with more nuanced details can be seen in both the rough metal and semiconductor regions. Additionally, the tip lifetime is greatly improved.

PeakForce sMIM for SRAM

Peakforce sMIM also has the ability to perform carrier profiling that integrates the benefits of sMIM and PeakForce Tapping in a single method. An example of an SRAM sample is shown in Figure 11, where varied electrical channels are utilized as color skins on the 3D topographic views.

Figure 11. PeakForce sMIM images of an SRAM sample. Scan size is 12 pm x 4 pm. (A) sMIM-C channel as a skin covers on topography; (B) sMIM dC/dV phase channel on topography; (C) sMIM dC/dV amplitude channel on topography.

In Figure 11A, the sMIM-C skin is obtained from raw data without applying any image post processing. This image reveals only reduced effects from the stray capacitance, confirming the excellent performance of PeakForce sMIM.

Figure 11B and Figure 11C are the dC/dV phase and dC/dV amplitude, respectively. These images reveal varied electronic areas of the device, as marked in Figure 11A. Unlike standard sMIM, these images display better overall sharpness and lateral resolution. Through sMIM-C and dC/dV amplitude images, prospective mask defects in the epitaxial p-type region can be seen, where the doping concentration has a local difference close to the two n-channels, as shown by the black dotted line in Figure 11C.

Conclusion

sMIM has been shown to be a viable method for the characterization of conductivity, permittivity, and other material electrical properties. Combined with AFM’s accurate topographic imaging capabilities, the devices and materials’ function can be studied with less than 20nm resolution, introducing novel dimensions of data that correlates directly with the topographic channel. Most prominently, this approach of measuring the electrodynamic properties of materials eliminates the need for making electrical contact with the substrate. This means, engineers and researchers no longer have to perform the dreary work of soldering and wiring that could change the electrical properties of samples and which would not be possible for nanoscale materials.

By using Bruker's flexible AFM platforms, it would be easy to image both the resistive and capacitive properties. However, the AC components ensuing from sample bias modulation are scanned and subsequently overcome for dC/dV-type semiconductor carrier profiling. Unique data from the dR/ dV analysis can also be added.

Further, site-specific resistance-voltage and capacitance-voltage spectroscopy introduces novel tools for basic studies of new materials and for device failure analysis. sMIM, when integrated with Bruker's PeakForce Tapping, can be extended to make complex measurements on delicate samples, while offering a concurrent mapping of correlated mechanical characteristics. The local difference of a sample can be directly imaged at the tens of nanometer length scales of sMIM, paving the way for new avenues of research and applications. Subsurface patterns, nanoparticle oxide films, CNTs, and a range of semiconductor devices are just some of the examples elucidated in this article. The combination of sMIM and Bruker's versatile AFM platforms will enable device engineers and material researchers to investigate the fundamental principles underlying functionality, and carry out more sophisticated and in-depth device failure analysis and materials characterization.

This information has been sourced, reviewed and adapted from materials provided by Bruker Nano Surfaces.

For more information on this source, please visit Bruker Nano Surfaces.