Unit 2: Sampling in Chemical Analysis, Representative Population & Sample Preparation

Analytical chemistry frequently confronts a staggering scale mismatch: an analyst may be tasked with determining the gold content of a $100,000\text{-ton}$ mineral ore deposit, the pesticide residue in a $20\text{-ton}$ grain silo, or trace heavy metals across a $50\text{-km}^2$ freshwater lake. Yet modern analytical instruments consume an analytical test portion typically weighing between $0.1\text{ g}$ and $1.0\text{ g}$.

The Sampling Paradox

If the sub-gram test portion is not chemically and mineralogically representative of the bulk parent population, the most sophisticated spectrometer or ultra-high-precision balance will produce completely erroneous analytical conclusions. Experimental studies demonstrate that sampling error is frequently 10 to 100 times larger than the combined instrumental measurement error.

``` The Analytical Sampling Hierarchy ┌───────────────────────────────────────────────────────────────┐ │ Gross Target Population (e.g., 100,000 kg Mineral Ore Lot) │ └───────────────────────────────┬───────────────────────────────┘ │ Sampling Plan (Augers, Increments) ▼ ┌───────────────────────────────────────────────────────────────┐ │ Gross Sample (e.g., 50 kg Combined Increments) │ └───────────────────────────────┬───────────────────────────────┘ │ Primary Crushing & Sieve Milling ▼ ┌───────────────────────────────────────────────────────────────┐ │ Subsample / Laboratory Sample (e.g., 500 g, -100 mesh) │ └───────────────────────────────┬───────────────────────────────┘ │ Pulverization & Homogenization ▼ ┌───────────────────────────────────────────────────────────────┐ │ Analytical Test Portion (e.g., 0.2500 g Digested Aliquot) │ └───────────────────────────────────────────────────────────────┘ ```

Definitions in the Sampling Chain

1. Target Population: The entire collection of matter whose chemical composition is to be characterized.

2. Sampling Unit: Distinct portions of the population that can be separately sampled (e.g., individual bags of fertilizer, discrete soil core intervals, or hourly wastewater effluent discharges).

3. Gross Sample: The composite material formed by gathering multiple individual primary increments extracted from the target population.

4. Laboratory Sample: A reduced, pulverized, and homogenized fraction of the gross sample sent to the analytical laboratory (typically $100\text{--}500\text{ g}$).

5. Analytical Test Portion: The precise aliquot ($0.1\text{--}1.0\text{ g}$) weighed on an analytical balance and subjected to chemical dissolution and measurement.

Visman Two-Constant Sampling Model

While the Ingamells sampling constant $K_s$ models fundamental random segregation in well-mixed particulate matrices, real industrial raw materials (e.g., coals, crushed mineral ores, soil deposits) exhibit both random particulate variance and long-range spatial segregation variance. In 1969, J. Visman derived the two-constant general sampling equation:

where:

• $s^2$ is the total variance of the gross sample composed of $N_s$ increments each having mass $m_{\text{incr}}$ (total gross mass $M = N_s \times m_{\text{incr}}$).
• $A_v$ is the random variance constant (governed by particle size and mineral liberation degree, equivalent to Ingamells $K_s$ when segregation is absent).
• $B_v$ is the segregation variance constant, reflecting macroscopic spatial gradients, stratifications, or batch inhomogeneity across the bulk population.

By conducting a two-tier sampling experiment collecting small and large increment sets, an analytical laboratory can simultaneously evaluate $A_v$ and $B_v$:

Solving this linear system determines the minimum gross sample mass $M_{\text{min}}$ and minimum number of increments $N_{\text{min}}$ required to attain any pre-defined confidence tolerance.

To place sampling on a rigorous mathematical foundation, analytical chemists employ probabilistic models describing the distribution of analyte-bearing particles within segregated or randomized particulate matrices.

Ingamells Sampling Constant ($K_s$)

In 1974, C. O. Ingamells established that for a heterogeneous particulate material possessing random, independent segregation of two mineral phases (e.g., rich ore grains dispersed in barren gangue), the relative standard deviation of sampling ($\% \text{RSD}_s$) is inversely proportional to the square root of the analytical sample mass $w$:

where:

• $w$ is the mass of the analytical test portion (typically in grams).
• $\% \text{RSD}_s$ is the percentage relative standard deviation attributable strictly to sampling heterogeneity.
• $K_s$ is the Ingamells Sampling Constant, representing the theoretical minimum sample mass (in grams) required to restrict sampling uncertainty to $\pm 1.0\%$ at the $68\%$ confidence level.

Practical Application: If an ore requires $w_1 = 1.0\text{ g}$ to achieve a sampling uncertainty of $\pm 5\%$, the Ingamells constant is:

To reduce the sampling uncertainty to $\pm 1.0\%$, the required test portion mass is:

Visman Sampling Equation

J. Visman generalized sampling theory to account for both microscopic random composition variance and macroscopic spatial segregation variance:

where:

• $s_s^2$ is the total sampling variance.
• $A$ is the homogeneity constant, describing random compositional variability at the particle level (decreases with increasing subsample mass $w$).
• $B$ is the segregation constant, describing macroscopic spatial gradients or stratification across the bulk lot (decreases with increasing number of primary sample increments $N$).

To minimize overall sampling variance economically, an analytical protocol must balance increasing the mass of individual test portions ($w$) to diminish $A/w$ against increasing the total number of collected primary increments ($N$) to diminish $B/N$.

The spatial and temporal architecture of collecting increments determines whether a sample is statistically unbiased and representative. Four primary probability sampling designs are recognized in environmental and analytical protocols:

``` ┌─────────────────────────────────┐ ┌─────────────────────────────────┐ │ Simple Random Sampling │ │ Systematic Sampling │ │ • Random grid coordinates │ │ • Fixed temporal/spatial steps│ │ • Every unit equal chance │ │ • Risk: periodic bias │ └─────────────────────────────────┘ └─────────────────────────────────┘ ┌─────────────────────────────────┐ ┌─────────────────────────────────┐ │ Stratified Sampling │ │ Cluster Sampling │ │ • Divided into homogeneous │ │ • Naturally clustered units │ │ strata (e.g. soil horizons) │ │ (e.g. barrels, railcars) │ └─────────────────────────────────┘ └─────────────────────────────────┘ ```

1. Simple Random Sampling

Every individual unit or increment in the target population possesses an identical, non-zero probability of selection. Points are chosen using pseudo-random coordinate generation. Simple random sampling provides mathematically unbiased parameter estimates but can yield uneven spatial coverage across extensive geographic regions.

2. Systematic Sampling

Sample increments are gathered at uniformly spaced spatial or temporal intervals (e.g., collecting $100\text{ mL}$ of river water every 2 hours, or sampling every 50th bag of animal feed off a conveyor belt). Critical Caution: If the target population contains a hidden periodic frequency (e.g., cyclic batch discharges from an upstream factory every 4 hours), systematic sampling synchronized with the periodicity will produce massive systematic bias.

3. Stratified Sampling

The heterogeneous target population is segmented into distinct, internally homogeneous sub-populations termed strata (e.g., soil stratified by depth horizons A, B, and C; or a lake stratified into epilimnion, thermocline, and hypolimnion). Random increments are then drawn from each stratum in proportion to its total volume or mass:

Stratified sampling drastically reduces overall sampling variance compared to simple random sampling of the unsegmented bulk.

4. Cluster Sampling

Employed when the population is naturally divided into discrete clusters or containers (e.g., 500 drums of chemical waste, or 20 railcars of bauxite). A subset of clusters is randomly selected, and increments are drawn either exhaustively or randomly from within the chosen clusters.

Once the gross sample arrives at the preparation facility, it must undergo systematic particle size reduction and subsampling to generate a fine, homogeneous laboratory sample without introducing cross-contamination or chemical alteration.

Particle Size Reduction Kinetics

According to sampling theory, the number of particles $n$ contained within a test portion of mass $w$ is inversely proportional to the cube of the average particle diameter $d$:

Because sampling variance $s_s^2 \propto 1/n$, reducing particle diameter $d$ by a factor of 10 increases particle count $n$ by a factor of 1000, slashing sampling variance by three orders of magnitude!

Stages of Mechanical Comminution

1. Primary Crushing: Jaw crushers reduce coarse rocks ($10\text{--}50\text{ cm}$) down to pebble size ($< 1\text{ cm}$).

2. Secondary Grinding: Disc pulverizers and ball mills reduce granules down to coarse sand ($< 1\text{ mm}$).

3. Fine Pulverization: Planetary ball mills, agate mortar and pestle, or shatterboxes grind the material to pass standard laboratory sieves (typically $-100\text{ mesh}$ or $-200\text{ mesh}$, corresponding to particle diameters $< 74\,\mu\text{m}$).

Contamination Hazards During Grinding

Grinding equipment abrades during comminution, introducing foreign elements into the sample:

• Steel/Cast Iron Mills: Introduce severe iron, chromium, nickel, and manganese contamination.
• Tungsten Carbide Mills: Introduce tungsten and cobalt binder contamination.
• Agate Mortars: Composed of pure silica ($\text{SiO}_2$); safe for metal determinations but unsuitable for silicate analysis.

Subsampling: Coning and Quartering

To split a multi-kilogram coarse sample into an unbiased subsample without segregation bias:

1. The pulverized sample is poured into a symmetrical conical pile.
2. The apex of the cone is flattened into a uniform disc.
3. The disc is divided into four equal quadrants across perpendicular diameters.
4. Two diagonally opposite quarters are retained and blended; the remaining two quarters are discarded.
5. The procedure is repeated recursively until the desired laboratory sample mass is achieved.

Water content is an insidious source of error in quantitative solid-state analysis. An uncalibrated moisture content inflates sample mass, causing systematic underestimation of analyte concentrations.

Classification of Water in Minerals and Solids

Analytical chemistry classifies water into two major categories:

1. Non-Essential Water

Water not required for the stoichiometric characterization or crystal lattice identity of the solid:

• Adsorbed Moisture: Water molecules bound weakly to the outer particle surfaces via dipole interactions; fluctuates with ambient relative humidity.
• Occluded Water: Microscopic liquid droplets trapped mechanically within microscopic cavities during crystal growth.
• Inverted / Sorbed Water: Held in porous materials (e.g., zeolites, silica gels).

Removal: Readily expelled by oven heating at $105^\circ\text{C}\text{--}110^\circ\text{C}$ for 2 hours to constant mass.

2. Essential Water

Water present in stoichiometric proportions as an integral part of the crystal structure:

• Water of Crystallization (Hydrates): Stoichiometrically bound in the lattice, such as $\text{CuSO}_4 \cdot 5\text{H}_2\text{O}$ or $\text{BaCl}_2 \cdot 2\text{H}_2\text{O}$. Expelled at elevated temperatures ($120^\circ\text{C}\text{--}250^\circ\text{C}$).
• Constitution Water (Hydroxyl Groups): Present as structural $\text{OH}^-$ ions (e.g., in clays, micas, $\text{Ca(OH)}_2$). Dehydroxylates only upon severe thermal calcination ($500^\circ\text{C}\text{--}1000^\circ\text{C}$):

Loss on Ignition (LOI)

Loss on Ignition measures the total percentage mass lost when a dried sample is ignited in a muffle furnace at $950^\circ\text{C}\text{--}1050^\circ\text{C}$ to constant mass:

LOI accounts for the simultaneous release of constitutional water, decomposition of carbonates ($\text{CO}_3^{2-} \to \text{CO}_2\uparrow$), and combustion of organic matter.

Most modern instrumental techniques (AAS, ICP-OES, UV-Vis, titrimetry) require samples in homogeneous aqueous solution. Converting refractory solid matrices into clear solutions demands deep understanding of mineral acid oxidation-reduction potentials, complexation chemistry, and thermochemistry.

Properties and Applications of Digestion Mineral Acids

1. Hydrochloric Acid ($\text{HCl}$, $12\text{ M}$, $37\text{ wt}\%$)

A non-oxidizing acid ($E^\circ_{\text{H}^+/\text{H}_2} = 0.00\text{ V}$) acting through hydronium attack and strong chloro-complexation:

Dissolves metal carbonates, phosphates, oxides, and sulfides. Forms soluble chloride complexes with $\text{Fe}^{3+}, \text{Zn}^{2+}, \text{Sn}^{4+}$. Incompatible with $\text{Ag}^+, \text{Pb}^{2+}, \text{Hg}_2^{2+}$ due to insoluble chloride precipitation.

2. Nitric Acid ($\text{HNO}_3$, $16\text{ M}$, $70\text{ wt}\%$)

A powerful oxidizing mineral acid ($E^\circ = +0.96\text{ V}$ in acid):

Oxidizes base metals and decomposes biological/organic matrices into $\text{CO}_2$ and $\text{H}_2\text{O}$. Nearly all metal nitrates are universally soluble in water.

3. Aqua Regia ($3:1 \ \text{v/v} \ \text{HCl} : \text{HNO}_3$)

Combines the ferocious oxidation power of nitric acid with the intense complexing capability of chloride ions. In situ reaction generates nitrosyl chloride and nascent chlorine:

Readily dissolves noble metals (gold, platinum) by lowering the formal redox potential via chloro-complexation:

4. Hydrofluoric Acid ($\text{HF}$, $29\text{ M}$, $48\text{ wt}\%$)

The unique mineral acid capable of dissolving refractory silicate matrices by cleaving extremely strong $\text{Si}-\text{O}$ bonds ($\text{BDE} \approx 460\text{ kJ}\cdot\text{mol}^{-1}$) to form volatile silicon tetrafluoride gas:

Crucial Safety & Operational Rules: HF dissolves laboratory borosilicate glassware; must be handled exclusively in Teflon (PTFE) or platinum vessels. Severe contact poison: penetrates skin and precipitates bone calcium ($\text{CaF}_2$), causing systemic cardiac arrest.

5. Perchloric Acid ($\text{HClO}_4$, $70\text{ wt}\%$)

When cold and dilute, perchloric acid acts as a harmless strong acid. However, when concentrated ($70\%$) and heated to its boiling point ($203^\circ\text{C}$), it becomes one of the most ferocious oxidizing agents known ($E^\circ \approx +1.39\text{ V}$). Explosion Hazard: Extremely hazardous in the presence of easily oxidizable organic matter, rubber, or wooden fume hoods. Must be manipulated exclusively in dedicated wash-down fume hoods with water scrubbing.

Refractory ores (zircon, chromite, corundum, rutile) and volatile trace analytes (arsenic, mercury, selenium) present insurmountable hurdles for open-beaker hot-plate digestion. Two advanced techniques resolve these challenges: closed-vessel microwave digestion and high-temperature alkali fusion.

Closed-Vessel Microwave-Assisted Digestion

Samples and concentrated acid mixtures are sealed in heavy-walled Teflon (PTFE or PFA) vessels encased in high-strength composite polymer sleeves and irradiated with microwave energy ($2.45\text{ GHz}$).

``` Closed-Vessel Microwave Digestion Chamber ┌────────────────────────────────┐ │ Safety Burst Vent Rupture │ └───────────────┬────────────────┘ │ ┌───────────────┴────────────────┐ │ PFA / TFM Teflon Liner │ │ │ │ Acid Vapor (P ~ 40-80 bar) │ │ │ │ Liquid Core (T ~ 220-260°C) │ │ • Dipole rotation heating │ │ • Ionic conduction heating │ │ • Complete recovery of Hg/As │ └────────────────────────────────┘ ```

Physical Mechanisms of Microwave Heating

1. Dipole Rotation: Polar solvent molecules (water, $\text{HNO}_3$) oscillate at $2.45 \times 10^9\text{ cycles/sec}$ to align with the oscillating electric field, dissipating kinetic energy as thermal friction.

2. Ionic Conduction: Dissociated ions migrate back and forth with the alternating field, colliding with solvent molecules to generate volumetric core heating.

Advantages Over Open Systems

• Superheating: High pressure ($40\text{--}80\text{ bar}$) elevates acid boiling points from $120^\circ\text{C}$ to $> 240^\circ\text{C}$, boosting reaction kinetics by orders of magnitude via the Arrhenius equation.
• Volatile Retention: Sealed digestion vessels quantitatively retain volatile elements ($\text{Hg}, \text{As}, \text{Se}, \text{B}, \text{Sn}$) that would escape open beakers.
• Reduced Reagent Consumption: Requires only $5\text{--}10\text{ mL}$ of ultra-pure acid, slashing procedural blank contamination.

High-Temperature Alkali Flux Fusion Chemistry

When mineral lattices are completely inert to acid digestion (e.g., corundum $\alpha\text{-Al}_2\text{O}_3$, rutile $\text{TiO}_2$, zircon $\text{ZrSiO}_4$), they are blended with a 10- to 20-fold mass excess of an anhydrous inorganic salt (flux) and melted at $900^\circ\text{C}\text{--}1100^\circ\text{C}$ in platinum, graphite, or nickel crucibles.

| Flux Reagent | Melting Point | Crucible Type | Target Refractory Minerals | | :--- | :---: | :---: | :--- | | Lithium Metaborate ($\text{LiBO}_2$) | $845^\circ\text{C}$ | Graphite, Platinum | Silicates, aluminosilicates, basaltic rocks | | Lithium Tetraborate ($\text{Li}_2\text{B}_4\text{O}_7$) | $920^\circ\text{C}$ | Platinum | Basic rocks, iron ores, bauxites | | Sodium Carbonate ($\text{Na}_2\text{CO}_3$) | $851^\circ\text{C}$ | Platinum | Silicates, quartz, baryte ($\text{BaSO}_4$) | | Potassium Pyrosulfate ($\text{K}_2\text{S}_2\text{O}_7$) | $419^\circ\text{C}$ | Vycor, Platinum | Basic metal oxides ($\text{Fe}_2\text{O}_3, \text{TiO}_2, \text{ZrO}_2$) | | Sodium Peroxide ($\text{Na}_2\text{O}_2$) | $460^\circ\text{C}$ | Zirconium, Nickel | Chromite ($\text{FeCr}_2\text{O}_4$), sulfides, platinum ores |

Reaction Chemistry: Borate Fusion of Silicates

Molten lithium metaborate acts as a Lewis acid/base solvent, disrupting refractory network-covalent silicate lattices into soluble mononuclear borate-silicate complexes:

Upon cooling, the molten bead is quenched into dilute nitric or hydrochloric acid, where it dissolves rapidly and completely to yield a clear, stable analytical solution.

Before solid analytical samples (geological ores, metallurgical alloys, environmental soils, biological tissues) can be introduced into liquid chromatography or atomic spectrometers, their insoluble solid matrices must be quantitatively dissolved into a clear, single-phase aqueous solution without loss of volatile analytes or contamination.

``` Modern Sample Digestion Methodologies +----------------------------------------------------------------------+ | 1. Open-Vessel Wet Acid Digestion: | | HNO₃, HClO₄, HF, H₂SO₄ on hot plate. Atmospheric pressure. | | High reagent consumption, risk of volatile analyte loss (As, Hg). | | -------------------------------------------------------------------- | | 2. Closed-Vessel Microwave Bomb Digestion: | | TFM/PTFE vessels inside microwave cavity. P > 40-100 bar, T > 250°C| | Zero volatile loss, accelerates reaction rate by 100-1000x! | | -------------------------------------------------------------------- | | 3. High-Temperature Alkali Flux Fusion: | | LiBO₂ / Li₂B₄O₇ or Na₂O₂ in Pt/Zr crucible at 1000-1100°C. | | Dissolves refractory silicates, zircon, chromite, corundum. | +----------------------------------------------------------------------+ ```

Chemistry of Mineral Acid Digestion Mixtures

1. Nitric Acid ($\text{HNO}_3$, $68\%\text{ w/w}$, b.p. $120.5^\circ\text{C}$):

A versatile oxidizing acid that destroys organic matter by nitration, esterification, and oxidation to $\text{CO}_2$ and $\text{H}_2\text{O}$:

2. Perchloric Acid ($\text{HClO}_4$, $70\%\text{ w/w}$, b.p. $203^\circ\text{C}$):

An extraordinarily powerful oxidizing acid when hot and concentrated ($\text{HClO}_4 \to \text{Cl}_2 + \text{O}_2 + \text{H}_2\text{O}$). In cold dilute solutions, it behaves merely as a strong non-oxidizing acid. Safety Rule: Must always be preceded by nitric acid digestion to eliminate easily oxidizable organic matter, preventing violent explosive deflagration.

3. Hydrofluoric Acid ($\text{HF}$, $48\%\text{ w/w}$, b.p. $112^\circ\text{C}$):

The indispensable reagent for dissolving refractory silicate minerals ($\text{SiO}_2$, feldspars, clays), converting silica into volatile silicon tetrafluoride gas:

Microwave-Assisted Closed-Vessel Bomb Digestion

In closed polytetrafluoroethylene (PTFE or TFM) microwave vessels, acid mixtures absorb microwave radiation ($2.45\text{ GHz}$) through dipole rotation and ionic conduction. Because the vessel is hermetically sealed, autogenous pressure builds to $40\text{--}100\text{ bar}$, elevating the boiling point of the acid mixture from $120^\circ\text{C}$ to $> 260^\circ\text{C}$. According to the Arrhenius relation ($k = A e^{-E_a/RT}$), this $140^\circ\text{C}$ temperature elevation accelerates digestion kinetics by over three orders of magnitude, completing refractory digests in $15\text{--}30\text{ minutes}$ while quantitatively retaining volatile elements ($\text{Hg}, \text{As}, \text{Se}, \text{Pb}$).

Alkali Flux Fusion Thermodynamics

Refractory minerals such as chromite ($\text{FeCr}_2\text{O}_4$), corundum ($\alpha\text{-Al}_2\text{O}_3$), and zircon ($\text{ZrSiO}_4$) are completely impervious to mineral acids. They are solubilized by molten salt fusion at $1000\text{--}1100^\circ\text{C}$ in platinum or vitreous carbon crucibles:

Upon cooling, the molten bead forms a glassy cake that dissolves instantaneously in dilute nitric or hydrochloric acid.

## Advanced University Honors Research Monograph: Pierre Gy's Theory of Sampling (TOS) & The Seven Errors Pierre Gy's unified Theory of Sampling (TOS) is the only mathematically complete physical theory governing particulate sampling in science and industry. Gy proved that the total sampling variance $s_{\text{TE}}^2$ is the direct sum of seven fundamentally independent sampling errors:

1. Fundamental Sampling Error ($s_{\text{FSE}}^2$): Arises from constitution heterogeneity (inherent differences between individual mineral grains). It is the only error that cannot be eliminated by mechanical design; it can only be reduced by comminution (reducing top particle size $d$).

where $C$ is the sampling constant, $M_s$ is sample mass, and $M_L$ is lot mass.

2. Grouping and Segregation Error ($s_{\text{GSE}}^2$): Arises from distribution heterogeneity (gravitational settling, density stratification). Suppressed by collecting many small increments ($N \ge 30$) rather than few large scoops.

3. Increment Delimitation Error ($s_{\text{IDE}}^2$): Geometric error caused when the sampling cutter does not cut a strictly parallel, complete cross-section of the stream.

4. Increment Extraction Error ($s_{\text{IEE}}^2$): Physical error when particles bounce out of or are deflected away from the cutter blade edges.

5. Preparation Errors ($s_{\text{PE}}^2$): Cross-contamination, dust loss, moisture absorption, or chemical alteration during crushing and milling.

Compliance with ISO 3082 and ASTM E300 mandates that $s_{\text{IDE}}^2 = s_{\text{IEE}}^2 = 0$ by strict mechanical cutter geometry, ensuring that total error approaches the irreducible physical fundamental limit $s_{\text{FSE}}^2$.