On this page · 5 sections
An experimental-design question gives you a chemical claim and makes you build the procedure that would test it. Or it hands you somebody else's procedure and makes you find the flaw. Either way you are not being asked what happens. You are being asked how you would know.
These questions live in both sections. In Section I they arrive as short lab stems: which glassware, which measurement, which quantity should be held constant, which graph turns curved data into a straight line. Section II always carries laboratory work somewhere. Which question it lands in is not published in advance, so expect it in a long free-response question or a short one. Section II is handwritten in a paper booklet, so your procedure has to be readable and ordered.
Scoring is analytic, not holistic. A reader works down a rubric of separate, checkable statements and awards each one independently. A typical design rubric pays for identifying the correct measurable quantity, specifying equipment whose precision fits how the number is used, describing the calculation or plot, and connecting a possible result to a conclusion. Because the points are independent, an arithmetic slip late in the problem does not erase the setup points earlier. Nothing is awarded for safety goggles, for labelling beakers, or for the phrase "repeat for accuracy" on its own. Write the chemistry, not the classroom ritual.
The steps
- 1
Read backwards from the equation you will finish with
Before you write step one, write the final equation. It names every quantity you must measure, and it does that faster than any brainstorm. Molar mass of a volatile liquid comes from M = mRT/PV, so you need mass, pressure, volume, and temperature, and nothing else. Enthalpy of reaction comes from q = mcΔT, so you need the mass of solution, the temperature change, and the moles of limiting reactant. A solubility product comes from the measured concentration of one ion plus the stoichiometry of the dissolution equation. Percent composition of a hydrate comes from mass before heating and mass after heating to constant mass. Once the equation is on the page, every variable in it becomes a numbered step, and any quantity not in it is padding you can cut. This also protects you from the most common design failure, which is a procedure that produces data you cannot actually compute with.
- 2
Name the variable you change, the one you measure, and the ones you pin down
State them explicitly, in that order, in one sentence each. If you are testing how concentration affects rate, the concentration of one reactant changes, the time to a fixed visible endpoint is measured, and temperature, total volume, catalyst, and the concentration of every other reactant stay fixed. The controlled list is where points are won, because it is where designs quietly break. Rate constants roughly double or triple for a 10 °C rise, so an uncontrolled water bath ruins a kinetics run. In a solution kinetics design you must add water so that every trial has the same total volume. Otherwise changing one reactant's volume changes both reactant concentrations at once, and no order can be isolated. Say the constant out loud in your procedure: "each trial contains 50.0 mL total, made up with distilled water."
- 3
Choose glassware by how the number is used, not by habit
Precision costs time, so spend it only where a digit reaches your final answer. A 50 mL buret is read to the hundredth of a millilitre, a volumetric pipet delivers 25.00 mL, and a volumetric flask makes a solution to a known volume at one mark. Use those whenever the volume enters a mole calculation. A graduated cylinder's resolution depends on its size: a 100 mL cylinder is read to roughly 0.1 mL, while a small 10 mL cylinder is finer. To deliver about 20 mL you would reach for the large cylinder, so treat 0.1 mL as its resolution here. A beaker's markings are decorative, so beakers are for rinse water, waste, and solvent whose exact amount does not matter. The same logic runs on balances, but state readability rather than trusting the instrument's name. Any mass you convert to moles needs a balance that reads to at least 0.001 g, and an analytical balance reads to 0.0001 g. A rougher top-loading balance is fine for masses you only need approximately. When a question asks you to justify a choice, answer with digits and consequence. A buret resolves 0.01 mL, so a titre near 20 mL carries four significant figures, while the 100 mL graduated cylinder would cap you at three and widen the uncertainty in the calculated concentration.
- 4
Design the separation before you weigh anything
Unit 3 is 20% of this exam, and its lab work is a standing source of design prompts. Gravimetric analysis is the workhorse. You precipitate the target ion, filter it, wash the solid on the filter with small portions of cold distilled water, dry it, then heat and reweigh until two successive masses agree. That last routine is heating to constant mass, and it is the only proof that all the water or all the decomposition product has left. Skipping the wash is the classic loss of a point, because unwashed precipitate carries soluble spectator ions that add mass and push the calculated percentage too high. Say which liquid you filter through, and say you dry the paper and solid together after recording the paper's own mass. Distillation separates liquids by boiling point, so collect the fraction while the thermometer at the still head holds steady, and discard the first runnings. Paper or thin-layer chromatography separates by how strongly each component sticks to the stationary phase versus the moving solvent. Spot the sample on a pencil line, keep that line above the solvent level so the spot does not dissolve into the reservoir, and compute Rf as the distance the spot travelled divided by the distance the solvent front travelled. Rf is fixed for a substance in a given solvent, so you identify an unknown by running a known beside it on the same plate.
- 5
Write measurements, not verbs
"Mass the crucible" is not a step. "Record the mass of the empty crucible to 0.001 g" is. Every instruction should state what is recorded, to what precision, and at what moment. Include the readings students forget because they feel like setup rather than data: the initial buret reading, the initial temperature of the calorimeter before mixing, the barometric pressure for a gas collected over water, the mass of the container before and after. Write the steps numbered and in physical order, so a classmate could run them without asking you a single question. That readability test is the real standard, because a reader who cannot follow the sequence cannot award the point. Keep it short. Six to ten precise steps beat a page of prose, and a rubric never pays for length.
- 6
Say how the data settles the question
A procedure that stops at data collection loses the analysis points, and those are usually worth more than the procedure itself. Finish by stating what you compute or plot, and what result supports which claim. Learn the linearizations, because the exam loves them. Concentration against time is straight for a zero-order reaction with slope −k. The natural log of concentration against time is straight for first order with slope −k. The reciprocal of concentration against time is straight for second order with slope +k. Absorbance against concentration is straight through the origin, and the slope of that calibration curve lets you read an unknown concentration off the line. The integrated rate laws are printed on the AP Chemistry equations sheet, so there the graded skill is choosing the right plot and interpreting the slope. The Arrhenius equation is not on that sheet. Memorise ln k = −(Ea/R)(1/T) + ln A, and remember that plotting ln k against 1/T gives a straight line of slope −Ea/R.
- 7
Build in the comparison that makes the result mean something
A single number proves nothing on its own. Design in what it will be compared against. That might be repeated trials so you can average and spot an outlier, a control run with the catalyst omitted, a blank cuvette of pure solvent to zero the spectrophotometer, or a calibration curve of standards of known concentration. It might be a literature value, a second independent method aimed at the same quantity, or heating to constant mass so you know decomposition finished. When a prompt asks whether the data supports a claim, the comparison is the answer. Say what result would confirm the claim and what result would contradict it, in the same breath. A design that can only come out one way is not an experiment, and readers notice.
- 8
Do error analysis with a named step, a direction, and a mechanism
Error questions want three things and reject anything vaguer. Name the specific step that went wrong. State whether the final calculated value comes out too high or too low. Explain the path from one to the other using the equation. Example: the buret was rinsed with distilled water and not with titrant, so the titrant is diluted, so more volume is needed to reach the endpoint, so the calculated moles of analyte come out too high. Example: some solid was lost when transferring the precipitate, so the recovered mass is low, so the calculated percent of that ion in the sample is too low. Also know which mistakes do nothing. Adding extra distilled water to the flask holding your analyte changes the concentration but not the moles present, so the equivalence volume in a titration is unchanged. Random error shrinks with more trials. A systematic error does not, no matter how many times you repeat the run.
Worked through
Prompt: a student has a solution of an unknown weak monoprotic acid, HA. Using standardized 0.100 M NaOH and a pH meter, design a procedure to determine the Ka of HA, and describe how the data give the answer.
Step 0, the target equation. At the half-equivalence point of a weak acid titration, exactly half the HA has been converted to A⁻, so [HA] = [A⁻]. The Henderson-Hasselbalch expression pH = pKa + log([A⁻]/[HA]) has a log of 1 in it, which is zero, so pH = pKa there. That identity assumes the titration runs at one steady temperature, since Ka is temperature-dependent, and it ignores activity effects in concentrated solution. That single fact dictates the whole procedure: I need a full titration curve, and I need to read one pH off it.
Variables. I change the volume of NaOH added. I measure pH. I hold constant the volume and identity of the acid sample, the concentration of the NaOH, and the temperature.
Procedure. 1. Use a volumetric pipet to transfer 25.00 mL of the HA solution into a 250 mL beaker. Add a stir bar and start slow stirring. 2. Calibrate the pH meter with buffers that bracket the range you will actually read. This curve runs past pH 11, so use pH 7 and pH 10 buffers, or a three-point calibration at 4, 7, and 10. Rinse the electrode with distilled water, then immerse it in the acid. 3. Record the initial pH before any base is added. 4. Fill a 50 mL buret with the 0.100 M NaOH and record the initial reading to 0.01 mL. 5. Add NaOH in portions of about 1 mL. After each addition, record the buret reading to 0.01 mL and the pH once the meter is steady. Switch to portions of about 0.1 mL as soon as the pH starts to climb quickly. 6. Continue until the pH levels off above about 11, then repeat the whole titration with a fresh 25.00 mL sample.
Analysis. Plot pH on the vertical axis against volume of NaOH on the horizontal axis. The equivalence point is the centre of the steep rise, located as the volume where ΔpH/ΔV is largest. Suppose that lands at 22.4 mL. The half-equivalence volume is half of that, 11.2 mL. Read the pH on the curve at 11.2 mL. Suppose it is 4.77. Then pKa = 4.77 and Ka = 10^−4.77 = 1.7 × 10⁻⁵.
A second calculation from the same run. This one is not an independent check, because it reuses the equivalence volume read off the same curve. At equivalence, moles of NaOH equal moles of HA: 0.0224 L × 0.100 M = 2.24 × 10⁻³ mol. In 25.00 mL that is an initial concentration of 0.0896 M. If the measured initial pH was 2.91, then [H⁺] = 10^−2.91 = 1.23 × 10⁻³ M, and Ka = (1.23 × 10⁻³)² / (0.0896 − 0.00123) = 1.7 × 10⁻⁵. The genuine comparisons here are the duplicate titration in step 6 and, if you had the time, a second method aimed at the same quantity.
Error question. Suppose the student rinsed the buret with distilled water and forgot to rinse it with NaOH. The titrant is now more dilute than 0.100 M, so more volume is required, and the measured equivalence volume is too large. Because the second calculation is built on that volume, it inflates the calculated initial concentration of HA and returns a wrong Ka. The half-equivalence method survives it. Half of the observed equivalence volume is still the point where half the acid has reacted, so the pH there still equals pKa. Knowing which method a given error corrupts, and which it leaves alone, is exactly the reasoning these questions are built to reward.
Common mistakes
Listing equipment instead of measurements. "Use a buret" earns nothing. "Record the buret volume before and after the titration to 0.01 mL" earns the point, because the reader can see what number you will hold.
Matching precision to habit instead of to use. Measuring a volume with a 100 mL graduated cylinder when that volume goes into a mole calculation costs a significant figure and the justification point. Using a volumetric flask for rinse water wastes time you do not have.
Letting two variables move at once. In a solution kinetics design, changing one reactant's volume without topping every trial up to the same total volume changes both concentrations, and the order with respect to either reactant becomes unmeasurable.
Skipping the wash and the constant-mass check in a gravimetric procedure. An unwashed precipitate carries soluble spectator ions, so the mass reads high, and a solid weighed once may still hold water that a second heating would drive off.
Stopping at data collection. A procedure that never says what is plotted or computed, and never says which result would support which claim, forfeits the analysis points, which are usually the larger half of the rubric.
Writing error analysis with no direction and no mechanism. "Human error" and "the reading could be off" score zero. Name the step, say whether the final value comes out high or low, and trace it through the equation.
Treating "repeat the trials" as a fix for everything. More trials average away random scatter. They do nothing to a systematic error such as an uncalibrated balance or a contaminated titrant, and repetition alone is not the analysis step.
Checklist
- Can you write the final equation for a design prompt before you write step one, and read your list of measurements straight off it?
- Can you name the independent variable, the dependent variable, and at least two controlled variables for any lab prompt in one sentence each?
- Can you justify choosing a buret over a 100 mL graduated cylinder using readable digits and the effect on significant figures in the final answer?
- Can you state the readability a balance needs for a mass you will convert to moles, without leaning on the word "analytical" to carry the argument?
- Can you state what to plot to identify a zero, first, and second order reaction, and say what the slope equals in each case?
- Can you write ln k = −(Ea/R)(1/T) + ln A from memory, knowing it is not printed on the equations sheet?
- Can you write a gravimetric procedure that includes washing the precipitate and heating to constant mass, and say what error each of those steps prevents?
- Can you define Rf and explain why the pencil line must sit above the solvent level?
- Can you describe a calibration-curve procedure using Beer's law, including the blank and the standards of known concentration?
- Can you explain, without doing a calculation, why the pH at half-equivalence equals the pKa of a weak acid, and name the condition that identity assumes?
- Can you take a named procedural error and state the direction of its effect on the calculated result and the reason?
- Can you write a six-to-ten step procedure a classmate could run without asking you a single clarifying question?