Characterization, Testing and Materials Analytics

Introduction

Electrical engineering materials must be characterized before they can be selected, qualified, manufactured or diagnosed after failure. A conductivity value, dielectric spectrum or micrograph is not inherently trustworthy: its usefulness depends on specimen preparation, calibration, environmental control, sampling, uncertainty evaluation and appropriate statistical interpretation.

Characterization measures a property or reveals structure. Testing determines behavior under specified conditions, often against a requirement. Materials analytics integrates measurements from several techniques to support engineering decisions. For example, premature insulation failure may require dielectric spectroscopy to detect moisture, Fourier-transform infrared spectroscopy to identify oxidation, and microscopy to locate cracks or contaminants.

A sound investigation therefore connects five elements:

1. the engineering question;
2. a representative specimen and controlled test method;
3. a traceable measurement result with uncertainty;
4. statistical evidence appropriate to the population;
5. a defensible decision about suitability, failure or process control.

Learning Objectives

After studying this chapter, students should be able to:

• select electrical, dielectric, magnetic, thermal, mechanical, spectroscopic and microscopic methods;
• calculate material properties from measured quantities under stated assumptions;
• distinguish accuracy, precision, repeatability, reproducibility and traceability;
• evaluate Type A and Type B uncertainty components;
• report combined and expanded measurement uncertainty correctly;
• apply descriptive statistics, confidence intervals, capability indices and control charts;
• design a material qualification or failure-analysis plan;
• interpret complementary measurements without overstating causation;
• recognize ethical, regulatory and data-governance responsibilities.

1. Designing a Characterization Program

A characterization program should begin with a measurable question rather than an available instrument. “Why did the cable fail?” is too broad. Better questions include: Did moisture raise dielectric loss? Did thermal aging embrittle the polymer? Is conductor resistivity outside specification?

1.1 Measurand and test conditions

The measurand is the quantity intended to be measured. It must be defined sufficiently, including relevant conditions. Copper resistivity at 20 °C and the resistance of an unspecified copper specimen are not equivalent measurands.

A test plan should specify:

• material grade, batch and processing history;
• sampling location and orientation;
• specimen geometry and surface preparation;
• conditioning time, temperature and humidity;
• instrument, calibration status and measurement range;
• number of specimens and replicate readings;
• acceptance criteria and decision rule;
• methods for uncertainty and statistical analysis.

Replicate instrument readings estimate short-term measurement variation but do not replace independent specimens. Ten readings at one location cannot establish the uniformity of an entire production lot.

1.2 Destructive and nondestructive methods

Nondestructive testing preserves the component and includes visual examination, X-ray imaging, ultrasonic inspection and some electromagnetic methods. Destructive tests include tensile testing, sectioning and many breakdown tests. A failure investigation should generally progress from nondestructive observations to destructive analysis so that early actions do not erase evidence.

1.3 A multimethod workflow

Figure: Traceable measurements and statistical analysis convert an engineering question into an evidence-based materials decision.

2. Electrical Characterization

2.1 Resistance, resistivity and conductivity

For a homogeneous specimen of constant cross-section carrying uniform direct current, resistivity is

where ρ is resistivity in Ω·m, R is resistance in Ω, A is cross-sectional area in m², and L is the voltage-sensing length in m. Conductivity is σ = 1/ρ in S·m⁻¹.

The relation assumes negligible contact and spreading resistance, uniform temperature and ohmic behavior. A current–voltage sweep should be inspected for linearity before a single resistance is reported.

A two-terminal measurement includes lead and contact resistance. A four-terminal, or Kelvin, configuration separates current-carrying leads from high-impedance voltage-sensing leads and is preferred for low-resistance conductors.

For a thin, laterally large film measured by an equally spaced collinear four-point probe, the ideal sheet resistance is

where R_s is in Ω per square, V is the measured inner-probe voltage and I is the outer-probe current. Finite specimen size, probe spacing, edge proximity and film thickness require correction factors. If film thickness t is known and current is uniformly distributed, ρ = R_s t.

2.2 Temperature dependence

Over a limited temperature interval, metallic resistance is often approximated by

where α_R is the temperature coefficient at reference temperature T₀, in K⁻¹. This linear model is not generally valid over large intervals or through phase changes. Semiconductors often require activated or more detailed transport models.

Hall-effect measurements provide carrier type, concentration and mobility for suitable specimens. Interpretation becomes model-dependent when multiple carrier types, nonuniform conduction or anomalous Hall effects are significant.

3. Dielectric and Insulation Testing

3.1 Complex permittivity

Under sinusoidal steady state with the e^jωt convention, a passive dielectric may be represented by

where ε' describes energy storage and ε'' describes dielectric loss. Relative permittivity is ε_r^* = ε^*/ε₀, where ε₀ is vacuum permittivity. The loss tangent is

For a parallel-plate specimen with negligible fringing,

where C is capacitance, A is electrode area and d is specimen thickness. Air gaps, electrode polarization, surface leakage and uncertain thickness can dominate the result.

If dc conductivity contributes to loss, its relative imaginary-permittivity contribution is

where ω = 2πf. This explains why low-frequency loss may rise strongly in moist or ionically contaminated insulation.

3.2 Dielectric spectroscopy and breakdown

Impedance or dielectric spectroscopy measures frequency-dependent response. Relaxation peaks and dispersion can reveal dipolar motion, interfacial polarization, charge transport and aging. Equivalent circuits can summarize behavior but are not unique physical explanations; different circuits may fit the same finite-band data.

Breakdown strength is commonly reported as voltage divided by thickness, but electric-field nonuniformity, electrode shape, ramp rate, environment and defects influence the result. Breakdown is stochastic and often modeled by a two-parameter Weibull distribution:

Here F(E) is the probability of breakdown by field E, η is the characteristic field for which F = 1-e^-1 ≈ 0.632, and β is the shape parameter. The model should be checked against data; it is not automatically valid for every insulation system.

Partial-discharge testing detects localized discharges that do not completely bridge insulation. In high-voltage qualification, inception voltage, apparent charge, phase-resolved patterns and noise discrimination are important.

4. Magnetic Characterization

Magnetic measurements relate magnetic flux density B, field strength H and magnetization M through

where B is in tesla, H and M are in A·m⁻¹, and μ₀ is vacuum permeability. For a linear, isotropic material over a specified field range, M = χH and B = μ₀(1+χ)H. Ferromagnets are nonlinear and hysteretic, so a single constant permeability is inadequate over a full cycle.

A hysteresis loop provides:

• saturation magnetization;
• remanent flux density or magnetization;
• coercive field;
• differential or amplitude permeability;
• energy dissipated per cycle, represented by the appropriate loop area.

A vibrating-sample magnetometer or superconducting quantum interference device measures magnetic moment, whereas a B–H loop tracer characterizes cyclic response. Demagnetizing fields depend on specimen geometry; corrections are essential when comparing differently shaped samples.

For ac magnetic components, total core loss depends on frequency, peak flux density, waveform, temperature, material and geometry. Empirical hysteresis–eddy-current decompositions apply only over the conditions for which their parameters were obtained.

5. Thermal Characterization

Thermal behavior affects conductor resistance, insulation aging, magnetic losses and mechanical reliability.

Thermal diffusivity a has units m²·s⁻¹. Under the ideal Parker laser-flash model for an opaque, homogeneous slab with instantaneous uniform pulse and negligible heat loss,

where L is specimen thickness and t_1/2 is the time for the rear-face temperature rise to reach half its maximum. Real analyses may require pulse-width, radiation-loss and nonuniform-heating corrections.

Thermal conductivity is then

where k is in W·m⁻¹·K⁻¹, ρ_m is mass density in kg·m⁻³ and c_p is specific heat capacity in J·kg⁻¹·K⁻¹.

Worked Example: Thermal conductivity

A ceramic has L = 1.00 mm and t_1/2 = 0.200 s. Assuming the ideal model,

With ρ_m = 3200 kg·m^-3 and c_p = 750 J·kg^-1·K^-1,

The result inherits uncertainty from diffusivity, density, heat capacity and model corrections.

6. Mechanical Characterization

Electrical materials experience assembly loads, thermal expansion, vibration and electromagnetic forces. Mechanical properties therefore influence electrical reliability.

Engineering stress and strain in a uniaxial tensile test are

where F is force, A₀ is original area, ΔL is extension and L₀ is gauge length. In the linear-elastic region, Young’s modulus is E_Y = Δσ_e/Δε_e. The slope must be obtained from an appropriate strain interval after correcting for machine compliance where necessary.

Yield strength, ultimate tensile strength and elongation characterize metals, while polymers may additionally require creep, stress-relaxation and dynamic mechanical analysis. Hardness tests measure resistance to localized indentation but should not be converted to tensile strength unless a validated material-specific correlation exists.

For linear-elastic fracture mechanics under small-scale yielding,

where K_I is mode-I stress intensity, Y is a dimensionless geometry factor, σ is applied stress and a is crack-size parameter defined by the selected geometry. Failure is expected when K_I reaches a valid fracture toughness K_IC. The criterion is inappropriate when extensive plasticity invalidates linear-elastic assumptions.

7. Spectroscopic and Microscopic Methods

7.1 Spectroscopy

Spectroscopy identifies composition and bonding through interaction with electromagnetic radiation or emitted particles.

• FTIR spectroscopy detects molecular vibrations and is valuable for polymer identification, oxidation and contamination.
• Raman spectroscopy provides complementary vibrational information and can examine carbon structure, stress and crystallinity.
• X-ray diffraction (XRD) identifies crystalline phases and lattice spacing. Amorphous material produces broad features rather than sharp Bragg peaks.
• X-ray photoelectron spectroscopy (XPS) measures surface elemental composition and chemical states over a shallow analysis depth.
• Energy-dispersive X-ray spectroscopy (EDS) in an electron microscope provides localized elemental analysis but has interaction-volume, overlap and detection-limit constraints.

For first-order diffraction, Bragg’s law is

where n is diffraction order, λ is X-ray wavelength, d is lattice-plane spacing and θ is the Bragg angle. Diffractometers commonly plot intensity against 2θ, so confusing θ with the plotted angle produces an incorrect spacing.

7.2 Microscopy

Optical microscopy reveals features down to approximately the micrometre scale, depending on numerical aperture and wavelength. Scanning electron microscopy (SEM) offers greater depth of field and surface-detail resolution. Transmission electron microscopy (TEM) can examine nanoscale structure but requires thin, carefully prepared specimens. Atomic force microscopy (AFM) maps surface topography and selected local properties using a probe.

Interpretation must distinguish real structure from preparation or imaging artifacts. Common artifacts include polishing scratches, ion-beam damage, charging, contamination, edge effects, drift and inappropriate contrast enhancement. Images require scale bars and documented acquisition conditions. A visually striking micrograph is not proof that the observed feature is representative.

8. Measurement Quality and Uncertainty

8.1 Essential metrological concepts

Accuracy describes closeness of agreement to a true quantity value, although the true value is generally unknown. Trueness concerns systematic effects; precision concerns agreement among repeated results. Repeatability refers to results under the same conditions, while reproducibility involves changed conditions such as laboratories or operators. Calibration establishes a relation between instrument indications and reference values. Metrological traceability is an unbroken, documented chain of calibrations, each contributing uncertainty.

A measurement model expresses the output as

Type A uncertainty is evaluated statistically from repeated observations. Type B uncertainty uses other information, such as calibration certificates, resolution, specifications or reference data. “Type B” does not mean subjective or less valid.

For approximately linear propagation, the combined standard uncertainty is

where c_i = ∂f/∂x_i is a sensitivity coefficient, u(x_i) is a standard uncertainty and u(x_i,x_j) is covariance. Covariance terms vanish only when inputs are uncorrelated. Monte Carlo propagation is useful for nonlinear models, non-Gaussian inputs or asymmetric output distributions.

Expanded uncertainty is

where k is a coverage factor. A value near k = 2 often gives approximately 95% coverage when the output distribution is approximately normal and effective degrees of freedom are sufficient; it is not universally an exact 95% interval.

Worked Example: Resistivity uncertainty

A bar has R = 0.200 Ω, A = 2.00 mm² = 2.00×10^-6 m² and L = 0.100 m. Thus,

Suppose the relative standard uncertainties of R, A and L are 1.0%, 0.5% and 0.2%, respectively, and are uncorrelated. For ρ = RA/L, first-order relative propagation gives

Hence u_c(ρ) = 4.54×10^-8 Ω·m. With k = 2,

The coverage basis and assumptions should accompany the result.

Standard uncertainties are not tolerance limits. Adding 1.0%, 0.5% and 0.2% does not create a worst-case tolerance bound. A worst-case analysis requires actual bounded input limits and a model showing how admissible extremes combine. If those percentages were instead independently specified maximum relative error limits and the model were monotonic over those ranges, a conservative first-order bound would be 1.7%; that conclusion cannot be drawn when they are merely standard uncertainties.

8.2 Decision rules and guard bands

When uncertainty is not negligible relative to a specification, simple comparison with the limit can create false acceptance or false rejection. A documented decision rule may use a guard band, accepting only when the result remains within an adjusted acceptance limit. The rule should state the tolerated consumer’s or producer’s risk and conform to the governing contract or standard.

9. Statistical Materials Analytics

For observations x₁,...,x_n, the sample mean and standard deviation are

The denominator n-1 makes s² unbiased for population variance under independent, identically distributed sampling with finite variance. If observations are approximately normal and independent, a two-sided confidence interval for the population mean is

Normality, independence and random sampling should be assessed rather than assumed. Material data can be skewed, censored, spatially correlated or multimodal because batches and mechanisms have been mixed.

9.1 Process monitoring and capability

A control chart distinguishes common-cause variation from evidence of special causes. Control limits are statistical monitoring limits, not engineering specification limits. For individual approximately normal observations from a stable process, limits are often constructed around the process mean at approximately three estimated standard deviations; chart choice and estimation method depend on subgrouping and data type.

For a stable, approximately normal process with upper and lower specification limits USL and LSL,

C_p measures potential capability if centered, while C_pk incorporates mean offset. Sample estimates replace μ and σ in practice. Capability indices are misleading for unstable processes or unsuitable distributional assumptions.

9.2 Measurement-system analysis

Gauge repeatability and reproducibility studies separate equipment variation, operator variation and part-to-part variation. A study requires representative parts spanning the operating range, randomized measurement order and suitable replication. A measurement system that consumes a large fraction of process variation cannot reliably support process improvement, even if its calibration certificate is current.

9.3 Correlation, regression and causation

Regression can relate properties to composition, temperature or process settings. Residual plots, uncertainty in predictors, interactions and extrapolation limits must be examined. Correlation does not establish a mechanism. A model linking moisture content to loss tangent may be predictive within one polymer system but fail after formulation or electrode changes.

10. Applied Case: Qualification and Failure Diagnosis of Insulation Film

This is an illustrative case.

Background

A manufacturer introduces a polymer film for inverter-fed motor windings. Requirements include low dielectric loss, adequate breakdown strength, thermal endurance and mechanical flexibility.

Problem

Initial qualification appears satisfactory, but one aged production lot shows elevated leakage current and occasional partial discharge. The team must determine whether the cause is random variation, measurement error or a material/process change.

Approach and analysis

1. Evidence preservation: Samples are mapped by roll position and conditioned in sealed containers. Unaged controls from the same and earlier lots are retained.
2. Measurement-system verification: Open/short compensation is performed for impedance measurements, electrode dimensions are verified, and a reference capacitor is checked. Repeat measurements show the observed lot shift is larger than repeatability variation.
3. Electrical and dielectric tests: Volume resistivity decreases after humid aging. Dielectric spectroscopy reveals increased low-frequency loss, consistent with greater mobile-charge contribution, but this alone does not identify the chemical source.
4. Breakdown testing: Ten preliminary breakdown results in MV·m⁻¹ are 298, 305, 301, 294, 309, 303, 297, 306, 300 and 302. Their mean is 301.5 MV·m⁻¹ and sample standard deviation is 4.50 MV·m⁻¹. All exceed a 285 MV·m⁻¹ minimum, but the small sample cannot by itself establish a low field-failure probability. Weibull analysis with a larger, properly censored dataset is scheduled.
5. Chemical analysis: FTIR shows increased carbonyl absorption after aging, indicating oxidation relative to controls. TGA shows a lower onset of major mass loss under the same defined atmosphere and heating program.
6. Microscopy: Cross-sectional SEM reveals microvoids concentrated near particles. EDS identifies elevated sodium near some particles. Because EDS is not highly sensitive to all light elements and has a finite interaction volume, the result is treated as localized supporting evidence rather than exact bulk composition.
7. Process investigation: Manufacturing records identify a change in filler supplier and an extended high-humidity storage interval. Controlled experiments separately vary filler lot and moisture conditioning.

Solution and outcome

The controlled study finds that the new filler has higher ionic contamination and that humid storage amplifies low-frequency loss. Incoming-ion limits, sealed storage and drying controls are introduced. A guard-banded dielectric-loss criterion is added to incoming inspection, and periodic spectroscopy is used to detect drift. Follow-up lots return to the historical distribution.

Learning points

• No single technique established the root cause.
• Repeatability checks prevented process variation from being confused with instrument noise.
• Spectroscopy and microscopy generated mechanistic evidence, while controlled experiments tested causation.
• Passing ten preliminary breakdown tests did not prove long-term population reliability.
• Corrective action addressed both material supply and environmental control.

11. Limitations, Governance and Contemporary Practice

11.1 Technical limitations

Measurements can be altered by specimen preparation, environmental history, spatial heterogeneity and instrument models. High-resolution methods often examine extremely small volumes, while bulk electrical tests average over much larger regions. Reconciling these scales is a central challenge.

Interlaboratory comparisons are valuable because local repeatability does not demonstrate reproducibility. Reference materials, proficiency testing and standardized methods improve comparability, but a standard method must still be shown suitable for the specific material and decision.

11.2 Ethics and data governance

Engineers must not discard inconvenient results without a documented technical reason, select only favorable micrographs, conceal failed specimens or report excessive precision. Raw data, metadata, calibration records, processing scripts and image transformations should be retained under controlled versioning. Personally or commercially sensitive data require appropriate access controls.

Regulated or safety-critical sectors may require laboratory competence systems such as ISO/IEC 17025, sector-specific test standards and documented chains of custody. Accreditation does not guarantee that every interpretation is correct; it supports competence within a stated scope.

11.3 Automation and artificial intelligence

Automated image segmentation, spectral classification and anomaly detection can accelerate materials analytics. However, models may learn instrument, laboratory or preparation artifacts rather than material properties. Training and validation sets should represent relevant batches, instruments and operating conditions. Human review, uncertainty estimates, data lineage and performance monitoring remain necessary. AI-generated classifications should not replace physical validation when safety decisions are involved.

12. Professional Best Practices

A defensible materials investigation should:

• define the measurand and decision before collecting data;
• use representative, randomized sampling;
• document conditioning and specimen preparation;
• verify calibration, range, resolution and measurement-system capability;
• use blanks, controls and certified reference materials where appropriate;
• retain raw data and complete metadata;
• propagate uncertainty using an explicit measurement model;
• select statistics consistent with data structure and distribution;
• distinguish control limits from specification limits;
• confirm important conclusions with complementary methods;
• report limitations, contradictory evidence and decision risks;
• connect corrective action to verified physical mechanisms.

Chapter Summary

Materials characterization converts electrical, dielectric, magnetic, thermal, mechanical, spectroscopic and microscopic observations into engineering evidence. Each method has physical assumptions, sampling scales and artifacts that determine the validity of its result. Electrical measurements require control of geometry, contact resistance and temperature; dielectric and magnetic properties are frequency-, field- and history-dependent; thermal and mechanical methods depend strongly on specimen and model conditions. Spectroscopy identifies composition and bonding, while microscopy localizes structure and damage.

Measurement uncertainty combines statistically evaluated and other traceable input information through a measurement model. Standard uncertainty, expanded uncertainty and bounded tolerance limits are distinct concepts. Statistical tools such as confidence intervals, control charts, capability indices, Weibull models and measurement-system studies support qualification and quality control only when their assumptions are appropriate. Reliable failure diagnosis integrates complementary evidence, controlled experiments and transparent data governance rather than relying on a single result.

Key Terms

Measurand: The quantity specifically intended to be measured.

Characterization: Measurement of a material’s composition, structure or properties.

Metrological traceability: Connection of a result to a reference through a documented calibration chain.

Repeatability: Precision under the same measurement conditions.

Reproducibility: Precision under changed measurement conditions.

Resistivity: Intrinsic opposition of a material to electric-current flow, expressed in Ω·m.

Sheet resistance: Resistance parameter of a thin conducting film, expressed in Ω per square.

Complex permittivity: Frequency-dependent representation of dielectric storage and loss.

Loss tangent: Ratio of dielectric loss component to storage component.

Coercivity: Reverse magnetic field required to reduce magnetization or flux density to a defined zero condition after magnetization.

Thermal diffusivity: Rate at which a temperature disturbance spreads through a material.

Fracture toughness: Resistance to crack extension under specified fracture-mechanics conditions.

Type A uncertainty: Uncertainty evaluated through statistical analysis of observations.

Type B uncertainty: Uncertainty evaluated using information other than repeated-observation statistics.

Combined standard uncertainty: Standard uncertainty obtained by propagating input uncertainties through a measurement model.

Expanded uncertainty: Combined standard uncertainty multiplied by a coverage factor.

Guard band: Adjustment of an acceptance limit to manage decision risk arising from uncertainty.

Control chart: Time-ordered statistical tool for detecting process changes.

Process capability: Ability of a stable process distribution to satisfy specification limits.

Weibull distribution: Statistical distribution frequently used to model breakdown strength and reliability data.

Review Questions

Short Answer Questions

1. Distinguish characterization from qualification testing.
2. Why is a four-terminal method preferred for low-resistance measurements?
3. What physical information is represented by the real and imaginary parts of complex permittivity?
4. Distinguish a specification limit from a control limit.
5. Why must microscopy images include scale and acquisition metadata?

Long Answer Questions

1. Explain how specimen geometry, temperature, contact resistance and instrument configuration affect resistivity measurement.
2. Compare dielectric spectroscopy, breakdown testing and partial-discharge testing as methods for insulation assessment.
3. Describe how thermal, mechanical and magnetic characterization contribute to the design of an electrical machine.
4. Develop an uncertainty budget conceptually for a parallel-plate permittivity measurement.
5. Explain how spectroscopy and microscopy can be combined in an electrical-material failure investigation, including their limitations.

Analytical / Application Questions

1. A thin film gives V = 25.0 mV at I = 10.0 mA in an ideal four-point-probe geometry. Calculate its sheet resistance and state why a geometric correction may be required.
2. Design a sampling and test plan to compare the dielectric aging resistance of two polymer formulations without confusing specimen variation with measurement repeatability.
3. A process has LSL = 90, USL = 110, estimated μ = 103 and σ = 2. Calculate C_p and C_pk, then interpret the effect of process centering.

Multiple Choice Questions with Answers

Question 1

Which statement correctly distinguishes a standard uncertainty from a tolerance limit?

A. A standard uncertainty is always the maximum possible measurement error.

B. A tolerance limit is obtained by adding all standard uncertainties.

C. A standard uncertainty quantifies dispersion, whereas a tolerance limit specifies an allowable or bounded range.

D. The two terms are interchangeable when coverage factor k=2.

Correct Answer: C. A standard uncertainty quantifies dispersion, whereas a tolerance limit specifies an allowable or bounded range.

Question 2

A homogeneous bar has R=0.50 Ω, A=1.0×10^-6 m² and L=0.20 m. What is its resistivity?

A. 2.5×10^-6 Ω·m

B. 1.0×10^-7 Ω·m

C. 1.0×10⁵ Ω·m

D. 2.5×10⁶ Ω·m

Correct Answer: A. 2.5×10^-6 Ω·m

Question 3

Why is a four-terminal connection preferred when measuring a low-resistance conductor?

A. It increases the specimen resistivity.

B. It largely excludes lead and contact voltage drops from the sensed voltage.

C. It eliminates the need to measure current.

D. It makes temperature control unnecessary.

Correct Answer: B. It largely excludes lead and contact voltage drops from the sensed voltage.

Question 4

Using the e^jωt convention, dielectric loss is represented by which quantity in ε^*=ε'-jε''?

A. ε' only

B. ε''

C. ε₀ only

D. Electrode area

Correct Answer: B. ε''

Question 5

In the ideal Parker laser-flash model, a specimen has L=2.0 mm and t_1/2=0.40 s. What is its thermal diffusivity?

A. 1.39×10^-6 m²·s^-1

B. 6.94×10^-4 m²·s^-1

C. 3.47×10^-7 m²·s^-1

D. 1.39×10^-9 m²·s^-1

Correct Answer: A. 1.39×10^-6 m²·s^-1

Question 6

Which method is most directly suited to identifying polymer functional groups and oxidation products?

A. FTIR spectroscopy

B. Four-point-probe measurement

C. Tensile testing

D. Laser-flash analysis

Correct Answer: A. FTIR spectroscopy

Question 7

For independent input quantities with relative standard uncertainties of 3% and 4% in a product model, what is the first-order combined relative standard uncertainty?

A. 1%

B. 5%

C. 7%

D. 12%

Correct Answer: B. 5%

Question 8

What does the characteristic field η represent in the two-parameter Weibull breakdown model?

A. The field at exactly 50% cumulative breakdown probability

B. The minimum possible breakdown field

C. The field at cumulative breakdown probability 1-e^-1

D. The arithmetic mean field for every dataset

Correct Answer: C. The field at cumulative breakdown probability 1-e^-1

Question 9

Which condition is required before conventional C_p and C_pk indices can be interpreted reliably?

A. The process must be statistically stable and the distributional model suitable.

B. Every observation must equal the target.

C. The measurement uncertainty must be exactly zero.

D. The upper and lower specification limits must be control limits.

Correct Answer: A. The process must be statistically stable and the distributional model suitable.

Question 10

Which practice provides the strongest evidence that elevated dielectric loss is caused by a changed filler rather than merely correlated with it?

A. Selecting the most visually striking micrograph

B. Repeating one measurement on the same specimen

C. Running controlled experiments that vary filler lot while controlling other factors

D. Discarding results that differ from the historical mean

Correct Answer: C. Running controlled experiments that vary filler lot while controlling other factors