Engineer Chemical Analysis common mistakes
Engineer Chemical Analysis · Understand the traps and how to avoid them.
When computing a calibration or standard-addition result, the candidate divides the blank-corrected signal by the slope but forgets to subtract the intercept, or converts a mass concentration to a molar concentration without dividing by molar mass, so the reported value is off by a constant or a factor.
Candidates memorize 'y = mx + b' as a phrase but practice only proportional calibrations through the origin; under exam pressure they treat the slope as a pure conversion factor and skip the algebra that the intercept represents a real background signal.
What to do instead
Before any numeric substitution, write the regression equation explicitly and rearrange it symbolically for the unknown, then plug in units with every term. Do a dimensional check: slope units must be signal per concentration, so dividing signal by slope yields concentration directly. Verify with one calibration point: back-calculate a standard's concentration from its signal and confirm you recover the prepared value.
The candidate states a detection or quantification limit as simply 3x or 10x the standard deviation of the blank, without stating which standard deviation (blank replicates, residual standard deviation of the calibration, or low-level spike replicates) and without the slope term.
Textbook shorthand ('LOD = 3s') is remembered without its denominator; candidates do not realize that s must be expressed in concentration units, which requires dividing the signal standard deviation by the calibration slope.
What to do instead
Always write the full expression: LOD = 3·s/slope and LOQ = 10·s/slope, and state the source of s. For the blank-based approach, s is the standard deviation of ≥10 blank measurements converted to concentration via the slope; for the calibration approach, use the residual standard deviation of the regression. Note that the blank-based and calibration-based values are not interchangeable, and report which one the method specifies.
When asked to select or justify a sample container, preservative, or holding time, the candidate answers from general laboratory habit (e.g., 'use a plastic bottle', 'acidify everything') rather than matching the analyte class to the specified container material, preservation, and maximum holding time.
Holding times and preservation requirements are numerous and easy to conflate; candidates prepare by memorizing a few headline rules instead of the logic that preservation controls adsorption, volatilization, microbial action, and speciation change.
What to do instead
Build a table organized by analyte behaviour: dissolved metals (acidified, glass or acid-washed plastic, adsorption control), mercury (oxidized and acidified, glass, short holding time), anions and nutrients (cooled, unpreserved or minimally preserved, filtration timing matters), volatile organics (zero headspace, no acid, cold), and hexavalent chromium (alkaline preservation to prevent reduction). For each, state the failure the preservation prevents, so the rule is reconstructed rather than recalled.
In a titration calculation the candidate uses the wrong stoichiometric ratio or forgets to account for the dilution of an aliquot, reporting a result that is exactly a factor of the aliquot fraction or the reaction coefficient away from correct.
Candidates focus on the arithmetic of moles = concentration × volume and treat the balanced equation as decoration; back-titrations and aliquot subsampling add two extra bookkeeping steps that are easy to drop.
What to do instead
Adopt a fixed written template: (1) write and balance the reaction, (2) compute moles of titrant actually consumed, (3) apply the mole ratio from the balanced equation, (4) scale from aliquot to total sample mass or volume, (5) divide by sample size and convert to the requested reporting unit. For back-titrations, compute moles consumed as moles added minus moles of excess titrant found. Sanity-check the magnitude against the expected result before finalizing.
When interpreting an infrared spectrum, the candidate assigns every strong band to a functional group by matching a memorized wavenumber, ignoring band shape, intensity, hydrogen bonding, and the fingerprint region, and so misidentifies or over-assigns peaks.
Wavenumber tables are memorized as isolated numbers; candidates do not learn that the same group appears at different positions depending on conjugation, hydrogen bonding, and phase, and that shape (broad vs sharp) carries diagnostic information.
What to do instead
Interpret in a fixed order: identify the carbonyl region first (roughly 1650–1750 cm⁻¹) and use its exact position and shape to distinguish ester, ketone, aldehyde, and amide; then O–H and N–H stretching (broad vs sharp); then C–H stretching to separate aromatic, alkene, and alkane; then the fingerprint region for substitution patterns. State the uncertainty of each assignment and cross-check with a second technique rather than committing to one band.
The candidate treats a matrix interference in atomic spectroscopy as a fixed correction factor, applying the same internal standard recovery, spike recovery, or background correction to all samples regardless of matrix, and ignores that suppression or enhancement depends on the sample composition and the plasma or flame conditions.
Candidates learn that internal standards and matrix matching 'correct for matrix effects' but not that the correction is only valid when the interferent and the analyte behave proportionally; they also confuse physical, chemical, ionization, and spectral interferences, which have different remedies.
What to do instead
Classify the interference first: physical (viscosity, nebulization efficiency), chemical (refractory compound formation in a flame), ionization (easily ionizable elements in a plasma), or spectral (overlap, molecular bands, polyatomic ions). Then apply the matching remedy: matrix matching, standard addition, internal standardization, ionization buffer, or collision/reaction cell and interference equations. Verify with a spike recovery and a dilution test, since a true matrix effect changes with dilution while a spectral overlap does not.
The candidate diagnoses a chromatographic problem by adjusting one parameter at a time without reading peak shape and retention together, so tailing, fronting, splitting, and retention drift are attributed to the wrong cause, and method adjustments are proposed that violate the validated method.
Peak-shape diagnosis requires linking a symptom to a physical cause (extra-column volume, injection solvent strength, column overload, active sites, temperature, mobile-phase pH); candidates learn symptoms in isolation and do not build the causal map.
What to do instead
Learn a symptom-to-cause matrix and read it as a whole: fronting suggests column overload (mass or volume) or a mismatched injection solvent; tailing suggests secondary interactions with active silanols, pH outside the buffered range, or a degrading column; splitting suggests a void or a badly cut column end; retention drift suggests mobile-phase evaporation, temperature instability, or gradient equilibration. For each proposed change, check whether it is within the validated method's allowed adjustments before recommending it.
The candidate reports a measurement uncertainty budget that lists only the repeatability of replicate measurements, omitting calibration, purity of the reference standard, volumetric and gravimetric operations, and sample inhomogeneity.
Repeatability is the only component candidates can compute easily, so it becomes the whole budget; the concept that uncertainty propagates from every step of the measurement chain is understood in principle but not applied.
What to do instead
Build the budget from a cause-and-effect diagram of the whole procedure: sample homogeneity, subsampling, weighing (calibration and repeatability of the balance), volumetric glassware (calibration and temperature), standard purity and preparation, calibration curve fit, and repeatability. Express each as a standard uncertainty, combine in quadrature, and multiply by the coverage factor for the stated confidence. Report the expanded uncertainty with its coverage factor and confidence level, and identify the dominant contributor so the report is actionable.
When asked about method validation, the candidate lists the validation parameters as a memorized sequence without tying each to a performance criterion, and cannot say which parameters are required for a limit test versus a quantitative assay or for a qualitative identification.
Validation parameters are learned as a vocabulary list; candidates do not connect each parameter to the question it answers and therefore cannot select the appropriate set for a given method purpose.
What to do instead
Learn validation as purpose-driven: a limit test needs detection limit, specificity, and robustness; a quantitative assay needs linearity, range, accuracy, precision, specificity, LOD/LOQ, and robustness; an identification method needs specificity only. For each parameter, state the acceptance criterion, the experimental design that produces it, and the risk if it is omitted. Distinguish validation from verification and from system suitability, which are frequently conflated.
The candidate answers data-integrity questions by describing good intentions ('follow the SOP', 'review the data') instead of naming specific controls such as audit trails, unique user accounts, raw-data retention, electronic signatures, and change control, and cannot distinguish a correction from an alteration.
Data integrity is treated as an attitude rather than a set of auditable controls; candidates do not practice mapping a scenario to the specific control that would have prevented it.
What to do instead
For any scenario, identify the control that detects or prevents the failure: attributable entries via unique logins and no shared accounts, contemporaneous recording, original records retained with the correction and its reason, audit trail enabled and reviewed, electronic signatures linked to the record, and no deletion of raw data. Practice rewriting vague answers into control-specific ones, and state which record (raw data, workbook, audit trail) each control protects.