13 of 13 ideas
E.1 Structure of the atom: 1 idea
Rydberg constant from hydrogen lines using a diffraction grating
Research question. What value of the Rydberg constant, in m⁻¹, do the visible Balmer lines of hydrogen give when measured with a diffraction grating of 600 lines per mm?
- E.1 Structure of the atom
- SL and HL
- Needs care
- Rarely listed
My take. Well suited to a school lab and it gives a numerical answer that can be judged against a known value. Reasonable choice, though the RQ is best written as a value, not a vague 'consistency'.
Method, physics and where marks are lost+
- Independent variable
- Balmer line (transition from n = 3, 4, 5 and 6 to n = 2), so 3 to 4 lines, each measured at both sides of the centre and in first and second order where visible.
- What you measure
- Diffraction angle read from a spectrometer or measured with a metre rule and a laser-calibrated set-up; wavelength calculated from d sin θ = nλ.
- Controlled variables
- Grating: same one, with line spacing calibrated using a laser of known wavelength. Slit width: fixed. Distance from grating to screen: measured and held constant. Lamp: same hydrogen tube, run at steady current.
- Physics and graph
- 1/λ = R(1/2² − 1/n²). Plot 1/λ (y) against (1/4 − 1/n²) (x): a straight line through the origin with gradient R.
- SL and HL
- SL students can extract R from a gradient and compare with 1.097×10⁷ m⁻¹. Top band work calibrates the grating, uses both sides of the centre to reduce zero error, and examines the fit residuals. HL depth can come from using energy level ideas and the discussion of the Bohr model.
- Where marks are lost
- Research design: relying on a nominal grating spacing. Data analysis: fitting through the origin without justification. Evaluation: the dim red and violet lines are hard to see, which is rarely discussed.
- Data
- Needs a hydrogen discharge tube with power supply, a grating and a spectrometer or a ruler set-up; the main uncertainty is locating the centre of dim lines.
E.2 Quantum physics: 5 ideas
Planck's constant from LED current-voltage curves
Research question. What value of h results from the threshold voltage of six LEDs with peak wavelengths between 450 nm and 950 nm, and how much does the threshold definition change it?
- E.2 Quantum physics
- HL topic
- Easy data
- Common: on 2 sites
My take. A good HL choice that is fairly common, so the comparison of threshold methods makes it yours. Measure the wavelength yourself and give the systematic difference.
Method, physics and where marks are lost+
- Independent variable
- LED colour, six to eight LEDs from infrared to blue, characterised by peak wavelength measured with a diffraction grating or a spectrometer.
- What you measure
- Threshold voltage from a slowly varied power supply with a voltmeter and a milliammeter, found by extrapolating the linear part of the I–V curve to zero current and, for comparison, at a fixed current such as 1 mA. Calculate eV and 1/λ.
- Controlled variables
- Series resistor of 100 Ω kept in every circuit. Same voltmeter range and meter. Dark room for judging emission if used. LEDs at room temperature by letting them cool between readings. Same current step size for each LED.
- Physics and graph
- Photon energy eV = hc/λ so V is proportional to 1/λ. Plot V against 1/λ. Gradient is hc/e so h = gradient × e/c.
- SL and HL
- This is HL content since it uses photon energy and quantum ideas. Stronger work compares two threshold methods, gives an intercept discussion because the graph should pass through the origin, and explains why h comes out too low or high.
- Where marks are lost
- Data analysis: reading the voltage where light is first seen by eye, which depends on the observer. Evaluation: not linking the systematic error to the band gap versus emitted photon energy, and ignoring the spread of LED wavelengths. Research design: no measurement of the wavelength, using the value quoted on the packet.
- Data
- Needs LEDs, a variable supply, two meters and a diffraction grating; the main uncertainty is defining the threshold and the LED's spectral width.
Graphite atomic spacing from electron diffraction rings
Research question. What is the spacing between carbon lattice planes in graphite, found from the ring diameters of an electron diffraction tube at accelerating voltages of 2.5 to 5.0 kV?
- E.2 Quantum physics
- HL topic
- Needs care
- Rarely listed
My take. An excellent HL choice if your school has the tube. Otherwise use a simulation with clear honesty about that.
Method, physics and where marks are lost+
- Independent variable
- Accelerating voltage, 2.5 to 5.0 kV in 0.5 kV steps, giving six values, with inner and outer ring diameters measured each time, 5 readings per ring.
- What you measure
- Ring diameter on the screen, measured with vernier calipers or a transparent ruler in mm; the electron wavelength calculated from voltage; plane spacing d found from the ring angle.
- Controlled variables
- Same tube and screen distance; filament heater voltage; room darkness for a clear ring; measurement always taken at the same ring edge.
- Physics and graph
- de Broglie: λ = h/√(2meV). Diffraction: 2d sinθ = nλ, and for small angles the ring radius r ≈ (L/d)λ times a constant. Plot ring diameter against 1/√V. The gradient gives the spacing d, with two values for two rings.
- SL and HL
- Wave properties of matter are HL only. An HL student can go beyond simple calculation by comparing d with the accepted values of 0.123 nm and 0.213 nm and treating the tube geometry critically.
- Where marks are lost
- Research design: the geometry constant is assumed without checking. Data analysis: ring edges are fuzzy, so uncertainty is underestimated. Conclusion: no comparison with literature values.
- Data
- Needs an electron diffraction tube and EHT supply, usually only available in some schools; the main uncertainty is blurred ring edges, and simulation is a fallback.
Light colour and solar cell electrical output
Research question. How does the peak wavelength of LED illumination (from about 450 nm to 650 nm, using 6 colours) affect the power output of a small silicon solar cell at fixed photon flux?
- E.2 Quantum physics
- HL topic
- Needs care
- Rarely listed
My take. Very good if you fix the photon flux or normalise by incident power, and weak if you just swap LEDs. The normalisation is the whole point.
Method, physics and where marks are lost+
- Independent variable
- LED colour or wavelength from about 450 nm to 650 nm (blue, green, yellow, orange, red and one more, checking the datasheet peak wavelength); each repeated 3 times, or use coloured filters over a white lamp.
- What you measure
- Short-circuit current and open-circuit voltage from a multimeter, or power across a fixed load. Compare with the photon rate, calculated from the light power (from a calibrated sensor) as P/(hc/λ).
- Controlled variables
- Distance from LED to cell fixed with a rail; LED drive current kept constant with a series resistor and checked; ambient light excluded with a dark tube; cell temperature and angle fixed.
- Physics and graph
- Photon energy E = hf = hc/λ. If each photon produces one electron, current is proportional to photon rate, not to frequency, so at equal power the current should rise with λ until the band gap cutoff near 1100 nm. Plot current per unit incident power against λ.
- SL and HL
- An SL student would struggle to link the response to photon energy. An HL student can test the one-photon-one-electron model and analyse the spectral response, discussing the band gap and thermalisation losses.
- Where marks are lost
- Research design: LEDs have different intensities, so frequency is confounded with power. Data analysis: no calibration of incident power. Conclusion: claiming that higher frequency gives more output without a physical model.
- Data
- Needs LEDs, a light sensor or a calibrated reference cell; the main uncertainty is the different intensity of each LED, so measure it.
Planck's constant from LED turn-on and photoelectric stopping voltage
Research question. How does the frequency of light, from about 4.3×10¹⁴ Hz to 7.5×10¹⁴ Hz, affect the stopping potential of a photocell, in volts, and what value of Planck's constant follows?
- E.2 Quantum physics
- HL topic
- Hard data
- Rarely listed
My take. Fine physics and it gives a very clean gradient if the equipment works. If your school has no photocell, do not fake it; use the LED turn-on voltage method as a clearly labelled alternative.
Method, physics and where marks are lost+
- Independent variable
- Frequency of light, 5 to 7 values, set using narrowband filters or LEDs of known peak wavelength, each measurement repeated 3 times.
- What you measure
- Stopping potential, measured with a digital voltmeter across a photocell or a commercial photoelectric apparatus. Calculated: the gradient of Vs against f, then h = gradient × e.
- Controlled variables
- Light intensity: same lamp and distance, monitored with a lux meter. Photocathode: same cell throughout. Ambient light: apparatus in a darkened box. Warm-up time: fixed before each reading.
- Physics and graph
- eVs = hf − φ. Plot Vs (y) against f (x): gradient = h/e, y-intercept = −φ/e, x-intercept gives the threshold frequency.
- SL and HL
- Photoelectric effect is HL content, so this suits HL. Top band work compares h and φ with accepted values, quantifies uncertainty via maximum and minimum gradients, and discusses why the voltage reading drifts as the cell charges.
- Where marks are lost
- Research design: using LED wavelengths from a datasheet without checking the spread. Data analysis: reading the stopping voltage at an ill-defined point on the curve. Evaluation: not addressing the contact potential and the leakage current.
- Data
- Needs a photoelectric apparatus or vacuum photocell with a high-impedance voltmeter; the main uncertainty is defining the stopping voltage and the wide spectrum of filtered light.
Work function of a metal from LED stopping voltages
Research question. What is the work function of a metal, in eV, found from stopping potentials measured with light frequencies from 5.0 × 10¹⁴ to 8.0 × 10¹⁴ Hz?
- E.2 Quantum physics
- HL topic
- Needs care
- Rarely listed
My take. Good, classic and worth doing. It is a physics measurement, so do it carefully and compare h with the known value.
Method, physics and where marks are lost+
- Independent variable
- Light frequency, from 5 to 6 filters or coloured LEDs (red to violet) covering 5.0 to 8.0 × 10¹⁴ Hz, using known wavelengths from a spectrometer; 5 readings each.
- What you measure
- Stopping potential in V from a high-impedance voltmeter across a photocell; the work function found from the intercept of the graph.
- Controlled variables
- Same photocell and distance from source; same light intensity, checked with a lux meter; dark room; same warm-up time for LEDs.
- Physics and graph
- Einstein: eVs = hf − φ. Plot stopping potential against frequency: the gradient is h/e and the intercept on the vertical axis is −φ/e. The gradient checks against h/e = 4.14 × 10⁻¹⁵ V s.
- SL and HL
- Photoelectric effect is HL only. Extra depth: use the gradient as an independent test of h, and consider why LED bandwidth widens the uncertainty on frequency.
- Where marks are lost
- Research design: LED wavelength taken from the packaging rather than measured. Data analysis: stopping voltage read while still drifting. Evaluation: the contact potential is not discussed.
- Data
- Needs a photocell or photoelectric kit and voltmeter; the main uncertainty is the wavelength spread of LEDs and slow voltage drift.
E.3 Radioactive decay: 3 ideas
Absorption of β and γ radiation by aluminium and lead
Research question. How does the thickness of aluminium sheet (0 to 4 mm, 8 values) affect the corrected count rate of a Sr-90 β source at a fixed 5 cm from a Geiger tube?
- E.3 Radioactive decay
- SL and HL
- Needs care
- Overdone: on 4 sites
My take. Very overdone, so it needs a twist. Try comparing several materials with one thickness in g cm⁻² to show mass thickness matters, or use a simulation or open dataset if no source is available.
Method, physics and where marks are lost+
- Independent variable
- Thickness of aluminium absorber, 0 to 4 mm, at least eight values from stacked sheets measured with a micrometer. Alternatively lead for γ. Repeat each count three times.
- What you measure
- Counts in 120 s with a Geiger-Müller tube and counter, divided by time. Subtract the background count rate measured before and after (about 20 minutes total). Calculate the uncertainty as √N.
- Controlled variables
- Source to detector distance, fixed with a clamp. The same source and tube with the same voltage. Absorber positioned at the same place. Count time long enough for over 400 counts at the lowest rate.
- Physics and graph
- Exponential attenuation, I = I₀e−μx. Plot ln(corrected rate) against x. The gradient is −μ, and the half-value thickness is ln2/μ. For β the curve is only approximately exponential, so discuss range.
- SL and HL
- SL students can plot ln R against thickness and find μ. Top marks require proper Poisson uncertainty, error bars on the ln plot and a discussion of why β is not truly exponential. HL adds nothing needed but the link to nuclear physics is nice.
- Where marks are lost
- Research design: short counting times giving large √N errors. Data analysis: background not subtracted, or uncertainties in ln values ignored. Conclusion: claiming pure exponential for β. Evaluation: source distance and absorber gaps not discussed. Also this is a very common topic (four sources) so examiners expect more. Safety and school licence rules for sources need to be handled by the teacher.
- Data
- Needs a sealed source and GM tube from school, so you may only have access under the teacher; the main uncertainty is count statistics.
Effect of counting time on decay constant precision
Research question. How does the counting interval, from 5 s to 60 s, affect the percentage uncertainty in the decay constant of a Ba-137m source fitted over 10 minutes?
- E.3 Radioactive decay
- SL and HL
- Needs care
- Rarely listed
My take. Good if you have access to a short-half-life source, and the statistics gives a distinct angle. If the school only has long-lived sources, do it as a simulation and say so.
Method, physics and where marks are lost+
- Independent variable
- Counting interval, 5 values (5, 10, 20, 40 and 60 s), with the same source data set binned in different ways or with separate runs.
- What you measure
- Counts recorded by a Geiger-Müller tube and scaler, with background subtracted; the decay constant λ calculated from the gradient of ln(A) against time, and the uncertainty in it taken from the fit.
- Controlled variables
- Source and tube geometry: fixed with a clamp at set distance. Background: measured for 10 minutes before and after. Voltage on the tube: kept at the operating value. Total elapsed time: same for every interval.
- Physics and graph
- N = N₀e−λt, so ln(count rate) against t is linear with gradient −λ. Counting error is √N, so the uncertainty on ln(rate) is about 1/√N. Compare λ with ln 2 / 2.55 min.
- SL and HL
- SL students can fit the decay and compare with the accepted half-life. Top band work propagates the Poisson uncertainties into weighted fits and shows how the precision scales with the interval. HL depth is not syllabus-specific, but the statistics adds sophistication.
- Where marks are lost
- Research design: forgetting the background and not repeating the run. Data analysis: taking logs of low counts without treating the error bars. Evaluation: ignoring dead time and the source running out during the experiment.
- Data
- Needs a protactinium or Ba-137m generator kit, a GM tube and a counter; the main uncertainty is the low count rate at long times.
Testing inverse square fall off with a γ source
Research question. Does the corrected count rate from a sealed γ source follow an inverse square law for distances between 3 cm and 30 cm from a GM tube?
- E.3 Radioactive decay
- SL and HL
- Needs care
- Rarely listed
My take. A good option if your school holds a source. The offset method is what lifts it from a routine check to real analysis.
Method, physics and where marks are lost+
- Independent variable
- Distance from source to tube window: 3, 5, 8, 12, 16, 20, 25, 30 cm (8 values), each measured from the centre of the source to the sensitive element, with count times of 2 to 5 minutes.
- What you measure
- Counts recorded over a timed interval with a GM tube and scaler, converted to count rate in counts per second. Background is measured for 10 minutes before and after, and subtracted. The uncertainty is taken as √N.
- Controlled variables
- Same source and same tube, fixed on a rail. Background measured at the same location. Source and tube axes aligned. Counting time chosen so that every point has at least 1000 counts where possible.
- Physics and graph
- For a point source, corrected rate ∝ 1/(x + x₀)², where x₀ is the unknown offset to the effective centre of the tube and source. Plot 1/√(corrected rate) against x; it should be a straight line, and the negative x intercept gives the offset.
- SL and HL
- SL: plot rate against 1/x² and comment on linearity. Top band: use the 1/√R against x method to find the hidden offset and quantify it, and discuss the tube's dead time. HL and SL alike gain from a proper treatment of Poisson uncertainty.
- Where marks are lost
- Research design: safety and source handling not considered, or too short counting times. Data analysis: forgetting to subtract background or using distance from the case, not the source. Conclusion: claiming an exact power of 2 with no uncertainty on the gradient. Evaluation: ignoring absorption in air and the finite size of the source and tube.
- Data
- Needs a school GM tube, scaler and a sealed source from the physics department, handled by the technician; the main uncertainty is the source to tube offset and low counts at large distance.
E.5 Fusion and stars: 4 ideas
Blackbody fits to archive spectra for star temperatures
Research question. How closely does the temperature found from Wien's law applied to archive spectra of 8 to 12 stars with published temperatures between 3500 K and 10000 K agree with the catalogue values, in percentage difference?
- E.5 Fusion and stars
- SL and HL
- Needs care
- database
- Rarely listed
My take. Worth choosing if you like data work and want astrophysics without a telescope. Make it your own by picking a themed sample, such as stars from one constellation or cluster, and being listed on only one site means examiners have not seen it often.
Method, physics and where marks are lost+
- Independent variable
- Published catalogue temperature of each star, 8 to 12 stars spanning about 3500 K to 10000 K, chosen from one archive so the spectra are reduced in the same way.
- What you measure
- Peak wavelength read from each flux calibrated spectrum with a spreadsheet or Python, giving temperature from λmax = b/T; percentage difference from the catalogue value is then calculated.
- Controlled variables
- Same archive and instrument for every spectrum, so calibration is alike. Same wavelength range used when locating the peak. Same smoothing window applied to every spectrum. Stars chosen with little reddening, so dust does not shift the peak.
- Physics and graph
- Wien's displacement law λmax = b/T and the Stefan-Boltzmann relation for stars. Plot λmax against 1/Tcatalogue, expecting a straight line through the origin with gradient b = 2.898e-3 m K. HL students may fit the full Planck curve.
- SL and HL
- SL: peak reading, the graph and a percentage comparison. Top band: fit the Planck function, compare it with the peak method, and explain why stellar spectra with absorption lines and limited wavelength coverage bias the peak. HL: propagate fit uncertainties.
- Where marks are lost
- Research design: no reason given for choosing these stars or archive. Data analysis: peak read by eye with no uncertainty. Conclusion: agreement claimed without a percentage difference. Evaluation: ignores that stars are not perfect blackbodies and that the spectrum may not cover the peak.
- Data
- Needs a public spectral archive and a spreadsheet or Python; the main uncertainty is the peak position where the spectrum is noisy or truncated.
Estimating the Hubble constant from a galaxy catalogue
Research question. What value of the Hubble constant, in km s⁻¹ Mpc⁻¹, follows from a linear fit of recession speed against distance for 20 to 30 galaxies within 300 Mpc?
- E.5 Fusion and stars
- SL and HL
- Easy data
- database
- Rarely listed
My take. Easy data, but it is a popular choice, so it needs a twist. Compare two distance indicators or two redshift ranges, and explain why the results disagree.
Method, physics and where marks are lost+
- Independent variable
- Distance to the galaxy, 20 to 30 galaxies spread from about 10 Mpc to 300 Mpc, taken from a public database.
- What you measure
- Recession speed calculated from redshift, v = cz for small z; the Hubble constant found as the gradient of v against d.
- Controlled variables
- Distance method: one indicator only, such as Type Ia supernovae. Data source: one database (for example NED). Redshift range: below 0.1, so the low-speed formula holds. Peculiar velocities: nearby galaxies excluded.
- Physics and graph
- v = H₀d. Plot v (y) against d (x): gradient H₀, and 1/H₀ gives an approximate age of the universe. Compare with about 70 km s⁻¹ Mpc⁻¹.
- SL and HL
- SL students can get a straight line and a value with an uncertainty from the gradient. Top band work compares the results for different distance indicators and explains the scatter in terms of peculiar velocity. HL depth can come from using the relativistic Doppler formula at higher redshift.
- Where marks are lost
- Research design: mixing distance methods and not stating the selection criteria. Data analysis: no uncertainty on the gradient and ignoring outliers without reasoning. Evaluation: forgetting that the local scatter does not reflect the measurement quality.
- Data
- Needs only a spreadsheet and free online catalogue data; the main uncertainty is the systematic error in the distance ladder.
Planet size from public transit light curve data
Research question. How does the fractional dip in brightness in TESS or Kepler light curves for 6 to 8 known exoplanets compare with the value predicted from published planet and star radii?
- E.5 Fusion and stars
- SL and HL
- Needs care
- database
- Rarely listed
My take. A good database study with real data. Twist: choose systems with a range of sizes, including one small planet where the noise matters, and be honest about it.
Method, physics and where marks are lost+
- Independent variable
- Exoplanet system, 6 to 8 targets chosen with transit depths from about 0.3% to 3%.
- What you measure
- Transit depth from the flux drop in downloaded light curve data, found with a spreadsheet or Python; planet radius calculated as Rp = Rstar × √depth.
- Controlled variables
- Data source: same mission and same pipeline for every target. Data quality: only targets with a clear, high signal-to-noise transit. Detrending: the same method applied throughout. Stellar radius: taken from one catalogue.
- Physics and graph
- Depth ≈ (Rp/Rstar)². Plot measured depth (y) against (published Rp/Rstar)² (x): a line through the origin with gradient 1. Duration relates to orbital speed via v = 2πa/T.
- SL and HL
- SL students can measure depths and compare radii. Top band work handles the noise, limb darkening and the propagation of uncertainty. HL depth can come from using Kepler's third law with orbital period to get the orbit radius and the star's mass.
- Where marks are lost
- Research design: choosing targets without stating criteria. Data analysis: reading the depth by eye and giving no uncertainty. Evaluation: not discussing stellar variability and limb darkening.
- Data
- Uses free archive data such as the NASA Exoplanet Archive or MAST, and a spreadsheet; the main uncertainty is the noise in the baseline flux.
Testing L = 4πR²σT⁴ with catalogued stellar data
Research question. For about 40 main-sequence and giant stars from a catalogue, does the luminosity follow L proportional to R²T⁴ with a fitted constant close to 4πσ, using surface temperatures from 3000 K to 30 000 K?
- E.5 Fusion and stars
- SL and HL
- Needs care
- database
- Rarely listed
My take. Good database study if you fix the flawed idea that luminosity simply rises with temperature. Testing the constant is what makes it personal and worth choosing.
Method, physics and where marks are lost+
- Independent variable
- Star's R²T⁴, calculated for 30 to 40 stars spanning 3000 to 30 000 K, chosen from the Hipparcos or Gaia catalogue via SIMBAD or VizieR.
- What you measure
- Luminosity in watts, found from the catalogued absolute magnitude or from apparent brightness and parallax distance. Radius is taken from the catalogue's interferometric or angular diameter values.
- Controlled variables
- Selection rule: stars with parallax error below 5 percent, one catalogue for all values. Stellar type: separate main sequence and giants when plotting. Data source: the same catalogue release. Units: converted to SI once at the start.
- Physics and graph
- The Stefan-Boltzmann law L = 4πσR²T⁴. Plot L (y) against R²T⁴ (x); gradient should equal 4πσ = 7.12×10⁻⁷ W m⁻² K⁻⁴. A log-log plot of L against T at fixed R also gives an exponent of 4.
- SL and HL
- SL students plot the graph and compare the gradient with 4πσ. Depth comes from uncertainties in distance, radius and temperature, and discussing why hotter stars are not always brighter, since radius matters. The original claim that brighter means hotter is wrong, so fix it.
- Where marks are lost
- Research design: choosing stars without a stated rule, or plotting L against T only, which does not test the law. Data analysis: log plots without uncertainties. Evaluation: not discussing that stars are not perfect black bodies.
- Data
- Needs a spreadsheet and free catalogue access; the main uncertainty is the catalogued radius and temperature.
Ideas were collected from public lists and published IA titles, merged when they are the same investigation, and rewritten from scratch. “On N sites” counts how many public lists carry the same investigation. How this list was made.
Frequently asked questions
What are good IB Physics IA ideas on nuclear and quantum physics?
+
Good starting points with easy data that few sites list are estimating the hubble constant from a galaxy catalogue. Each one gives a straight-line graph from school equipment, which is what the Data analysis and Conclusion criteria need.
Which nuclear and quantum physics IA ideas are overdone?
+
absorption of beta and gamma radiation by aluminium and lead appear on three or more public lists. They still work, but they need a twist that shows your own thinking.
Can I do a nuclear and quantum physics IA at SL?
+
8 of the 13 ideas use SL physics. The others rely on HL-only content and are marked as HL topics.