Barium Ferrite by Coprecipitation — The Precipitate Is the Precursor, and the Target Phase Forms on Calcination

Back to: Mochizuki Research Group Education — Inorganic Materials Science > Virtual Experiments, Inorganic Materials Science resources

This page is the explanation that pairs with coprecip-simulator-en.html. It explains what you saw in the simulator, namely “how the pH is determined,” “which components stay in the precipitate and which go out into the supernatant,” and “in what order the phases appear on calcination,” using three tools: charge balance, solubility products, and the rate of diffusion-limited reactions. The schematic of pressing, sintering, magnetization, and the measurement of the apparent remanent magnetization (⑤), and the lab-manual data analysis that estimates the demagnetization curve from the balance readings, are also collected in §9.2 and §9.3. The sources of the numbers, the parts that remain provisional values or schematics, and the items to be confirmed with Prof. Mochizuki are also collected here. How to read the screen of each tab of the simulator is explained in Appendix A.

The backbone of this page The key point of coprecipitation is that forming the precipitate (②) and forming the target crystal phase (④) are separate steps. The reddish-brown precipitate is a precursor, a mixture of a hydrous iron oxide (often written Fe(OH)3) and BaCO3, and it does not yet contain any BaFe12O19 crystals. If Fe and Ba are finely mixed in the liquid, the distance atoms have to move during calcination becomes short, and the target phase forms at a lower temperature than in the solid-state method (§8). In return, if components are lost or left behind in the precipitation and washing steps, this shows up as a composition shift in the result of calcination (§4, §6, §8.4).
① Solutions A: Na⁺, OH⁻, CO₃²⁻ B: Fe³⁺, Ba²⁺, Cl⁻ ② Precipitate(precursor) Hydrous iron oxide + BaCO₃ Supernatant: Na⁺, Cl⁻, OH⁻ ③ Dried powder(precursor) Hydrous iron oxide + BaCO₃ + bound water +remaining salts ④ Calcined powder BaFe₁₂O₁₉ α-Fe₂O₃ and BaFe₂O₄ may also remain ⑤ Sintered body(magnet) Grains grow and densify Magnetized into a magnet Pour Wash, filter, dry Calcine Press, sinter, magnetize

Fig. 1. Process steps and the state of the sample (a schematic drawn for this page). BaFe12O19 first appears in the calcination of ④; when it is sintered into a solid and magnetized in ⑤, it becomes a magnet.

1. What Is Coprecipitation? — Precursor and Target Phase

M-type barium ferrite BaFe12O19 is an oxide with strong uniaxial magnetic anisotropy in its hexagonal crystal, and it is used in permanent magnets and magnetic recording. In the solid-state method, powders of BaCO3 and α-Fe2O3 are mixed and fired at about 1100–1200 °C, whereas in the coprecipitation method, a solution containing Fe3+ and Ba2+ is poured into an alkaline solution, and both components are precipitated at once. The precipitated grains are fine, from a few nm to a few hundred nm, and Fe and Ba are close to each other, so the calcination temperature can be lowered to about 800–900 °C (Refs. 1–4).

The precipitate itself, however, is not BaFe12O19. Fe precipitates as a hydrous oxide (often written Fe(OH)3; in the literature it is two-line ferrihydrite, with a composition of roughly FeOOH·nH2O), and Ba precipitates as the carbonate BaCO3 (Refs. 3, 5). When this precursor is heated, it changes in the order dehydration → formation of α-Fe2O3 → the intermediate phase BaFe2O4 → BaFe12O19 (§8). There are examples in which almost no intermediate phase appears, depending on how the precursor is made (Ref. 2), and the temperature boundaries shift by more than 100 °C depending on the route (Ref. 6). The simulator adopts the route (the VEL design) in which the hydrous oxide on the Fe side and BaCO3 precipitate separately.

2. Practical Procedure and Conditions, and the Provisional Values in This Simulator

This page shows separately the procedures and numbers confirmed in the 2025 edition of the lab manual for Materials Science Experiments (Inorganic Materials) of the former Tokyo Institute of Technology (§8.1–8.5; hereafter “the lab manual”) and in the practical-session videos (A1–A3, B1–B4; this page mainly uses B1–B4), the values fixed by Prof. Mochizuki’s instructions (2026-09-25), and the numbers written in none of them (provisional values). The reagent masses, water volumes, and calcination conditions also agreed with the constants in the source code of the virtual laboratory (VEL) now under separate development.

Table 1. Procedure and conditions (fixed by the lab manual and videos) and the numbers used in this simulator“Lab manual §” gives a section of the lab manual, and “video” the practical-session videos B1–B4. “Prof. Mochizuki’s instructions” marks values fixed by his instructions of 2026-09-25. “Provisional” marks values written in none of them.
StepWhat was confirmed in the lab manual and videosNumbers used in this simulator
Solution A100 mL of distilled water in a 300 mL beaker. Dissolve NaOH (deliquescent; weighed in a plastic beaker) a little at a time, and after it has dissolved completely, add Na2CO3 (weighed on weighing paper) a little at a time (lab manual §8.1, video B1)NaOH 12.5 g, Na2CO3 3.15 g (lab manual §8.1)
Solution B50 mL of distilled water in a 100 mL beaker. Dissolve FeCl3·6H2O (large balance tray) first, and then BaCl2·2H2O (small balance tray)FeCl3·6H2O 10.14 g, BaCl2·2H2O 0.835 g (lab manual §8.1)
PrecipitationAdd solution B to solution A a little at a time while mixing well. In the video it is poured down a glass rod in several portions, with mixing by the glass rod in between (no stir bar or stirrer is used). A reddish-brown precipitate formsPouring rate 5 mL/min, 25 °C (provisional). The stirring strength of 400 rpm equivalent is a scale for comparing conditions (§5)
StandingCover with a watch glass and let stand so that the precipitate settlesStanding time 10 min (provisional; not specified in the lab manual)
WashingRemove the supernatant with a 5 mL Komagome pipette (into the waste bottle) until about 125 mL remains. Pour in distilled water heated to about 60 °C in a 500 mL beaker on a hot plate, up to about 300 mL, with a 20 mL Komagome pipette, mix, and let stand. Check the pH of the supernatant with pH test paper, and repeat until it is about 7Volume after removal 125 mL, 60 °C, 300 mL (lab manual §8.1)
Filtration and dryingAspirator and suction flask, Büchner funnel, filter paper moistened with distilled water → switch on and check the suction before pouring in the precipitate → rinse the beaker with distilled water → finish when the filter cake cracks (close the stopcock before switching off). Put it on a watch glass together with the filter paper and place it in the drying oven. When dry, the filter cake breaks into small dark-brown pieces and the filter paper is stained orange (video B2)Filter paper 5B, 105 °C for about 20 h (lab manual §8.1)
Weighing, grinding, and calcinationWeigh the dried powder in two ways, on weighing paper and in the crucible, and grind it in an alumina mortar. Calcine in a lidded alumina crucible (the staff operate the furnace). The group at the lab bench with the most dried powder sets aside about 0.5 g of dried powder for XRDHeat at about 3 °C/min, 925 °C for 2 h, furnace cooling (lab manual §8.2). The atmosphere is air (Prof. Mochizuki’s instructions). Calculated without setting any aside
Pressing and sinteringGrind the calcined powder, add 3 drops of a 5 mass% PVA aqueous solution to about 3 g, and granulate. Press in a die of φ 1.000 cm at 98.1 MPa, holding for 1 min (two pieces of different thickness; if a green body is 6 mm or thicker, press again). Heat at about 3 °C/min to 1100 °C, hold for 2 h, and furnace-cool. The group with the second-most dried powder sets aside about 0.5 g of calcined powder for XRDThe conditions are as in lab manual §8.3. Powder used for each piece: 0.6 g and 1.2 g; relative densities of the green body and the sintered body: 55 % and 80 % (provisional)
Polishing and bulk densityRemove burrs with #400; if the thickness is 5 mm or more, polish it to less than 5 mm; finish to a mirror surface with #800; and measure W, D, and h. The relative density is taken against a true density of 5.374 g/cm3Thickness removed 0.10 mm (provisional)
Magnetization and measurementMagnetize with a direct current of 10 A (about 32 V; 752 000 A/m between the pole pieces) in an electromagnet for magnetization (pole gap 5 mm) for 1 min (operated by the staff). Place the sample on the sample stage of the balance and read W; then set the reference magnet (1 × 10−6 Wb·m) on the support stand in the orientation that repels the sample and read the apparent mass Wrepul. Two pieces stuck together are also measured (lab manual §8.4–8.5, video B4)Thickness of the reference magnet 14.020 mm; distance from the lower face of the reference magnet to the upper face of the sample stage 37.349 mm (Prof. Mochizuki’s instructions; values written on the balance); Wrepul is a schematic guide (§9.2). The data analysis follows lab manual §8.5 (§9.3)

The default starting Fe/Ba is 10.97 (BaO:Fe2O3 = 1:5.487), which differs from the stoichiometric 12 (1:6). Fe is the limiting component, and the theoretical yield is 3.4747 g (calculated with the atomic weights of lab manual §8.1; the simulator uses 35.45 for the atomic weight of Cl, so it gives 3.4748 g, and the screen shows 3.475 g). The simulator shows the “starting ratio (the ratio of the amounts put in),” the “ratio remaining in the precipitate (Fe/Ba of the solid),” and the “phases obtained” as separate quantities. Under the default conditions of this model the loss of Ba in ② and ③ is about 1 %, so with a starting ratio of 10.97 Ba is in excess and a little BaFe2O4 remains (§8.4).

The heat-treatment atmosphere is air in the student experiment. In research, heat treatment is sometimes done in a tube furnace with flowing N2 or in a glove box filled with Ar (Prof. Mochizuki’s instructions). This simulator treats air only.

3. pH of the Solution — Determined by Charge Balance

The simulator does not take pH as an input. From the amounts of the reagents put in, it determines the pH from the condition that the charge of the whole solution balances (charge balance). Given the dissolved Na+, Cl−, Ba2+, and Fe3+ in a solution of volume V and the total amount of carbonic acid CT, the charge balance reads

[Na+]+2[Ba2+]+qFe[Fe3+]+[H+] = [Cl−]+[OH−]+α1CT+2α2CT (1)

where [OH−] = Kw/[H+], and the fractions of the carbonic acid species are

α1=Ka1hh2+Ka1h+Ka1Ka2, α2=Ka1Ka2h2+Ka1h+Ka1Ka2 (h=[H+]) (2)

Fe3+ reacts with water, Fe3+ + H2O ⇄ FeOH2+ + H+, and releases H+, so the effective charge was taken to be qFe = 3 − Kh/(Kh + h) (Kh = 6.3 × 10−3, pK ≈ 2.2). Because of this, solution B, the iron chloride solution, is strongly acidic, in the pH 1 range (pH 1.18 for the default starting amounts). The residual obtained by subtracting the right side of Eq. (1) from the left side is monotonic in pH, so it can be solved by bisection. Solution A (NaOH 12.5 g + Na2CO3 3.15 g / 100 mL) has a pH of 14.49. In concentrated solutions the activity coefficients depart greatly from 1, so pH values in the 14 range are guides calculated from the concentrations (no activity correction is made). All the constants are values at 25 °C, and changes with temperature are not treated.

Dissolving NaOH raises the temperature of the liquid through the heat of dissolution. The simulator uses a heat of dissolution of 44.51 kJ/mol and a specific heat of 4.179 J/(g·K) (both are the given conditions of the problem in lab manual §8.1), assumes that without heat loss the temperature rises by ΔT = nΔH/(cm), and shows this on the thermometer and in the readout of ①. When the default 12.5 g of NaOH is dissolved at once in 100 mL of water, the liquid is 112.5 g, ΔT ≈ 29.6 K, and the liquid temperature goes from 25 °C to about 54.6 °C (58.3 °C if only the 100 g of water is counted). If the NaOH is added in p portions and 35 % of the temperature difference is lost (an estimate) before the next portion is added, the maximum temperature difference becomes (ΔT/p)(1 − 0.65p)/0.35, which is lower the more the NaOH is divided. This is the reason for the statement in the lab manual and the video that “dissolving it all at once may make the solution boil (bump) because of the heat of dissolution.” However, this estimate is the average temperature when the whole liquid is warmed uniformly. Around the dissolving grains the liquid can locally reach the boiling point and bump, so it is dangerous even if the average does not reach 100 °C.

4. Precipitation — Amount of Base and Solubility Products

4.1 Amount of base

To turn 1 mol of Fe3+ into a hydrous oxide, 3 equivalents of base are needed (Fe3+ + 3 OH− → FeOOH + H2O). If NaOH is insufficient, CO32− accepts two H+ (HCO3−, H2CO3), and part of it escapes as CO2. In the simulator, of the amount that became H2CO3 (3nFe − nNaOH − nCO₃, from 0 up to nCO₃), the fraction given in §5 is assumed to escape as CO2. If H2CO3 and HCO3− remain without escaping, the liquid stays weakly acidic at around pH 6, and because there is little CO32−, part of the Ba does not precipitate (with 2.5 g of NaOH in ①, even though the total base is sufficient, the pH is 5.98, 52 % of the carbonic acid has escaped, and 89 % of the Ba becomes BaCO3). In this case the carbonate needed to precipitate Ba is eaten up by Fe, so Ba precipitates less easily. If the base is still insufficient, some Fe stays dissolved. With the default starting amounts, the base put in (NaOH + 2 Na2CO3) is 3.3 times the amount needed to precipitate the Fe (the NaOH needed to turn the Fe into Fe(OH)3 is 4.501 g, and the 12.5 g of NaOH alone is 2.78 times that), and the excess OH− makes the liquid strongly basic. The carbonic acid that escaped as CO2 has also accepted two H+, so it acts as a base (the Na+ stays in the liquid). The simulator does not count this “amount of base” separately but treats it automatically within the charge balance, as in 4.2 below.

4.2 Solubility products

The Fe3+ and Ba2+ remaining in the liquid are determined by the solubility products (the amount exceeding them precipitates).

[Fe3+]=Ksp,Fe(OH)3[OH−]3, [Ba2+]=Ksp,BaCO3α2CT (3)

If a pH is provisionally chosen, Eq. (3) gives the Fe remaining in the liquid (with the amount put in as the upper limit), and the smaller root of (nBa − x)(CT − x)α2 = KspV2 gives the amount x of BaCO3. The residual obtained by putting these into Eq. (1) decreases monotonically with pH, so if pH is taken as the only unknown and solved by bisection, exactly one state that satisfies the charge balance and the two solubility products at the same time is found. The equilibrate function of VEL first fixed the upper limit of the amount of precipitate from a “base budget” and then solved by fixed-point iteration, but the budget did not include the base contributed by the carbonic acid that escaped as CO2, so during pouring in the reverse direction a discrepancy appeared: “the pH is 12, yet more than 10 % of the Fe is dissolved.” This simulator uses the solution method above instead. Ba is treated as precipitating only as BaCO3 (Ba(OH)2 is very soluble, so it does not precipitate as a hydroxide). The “Fe3+ remaining in solution” and “Ba2+ remaining in solution” in the readout of ② are exactly the composition of the liquid in this solved state.

Table 2. Comparison of the values of the solubility products (25 °C)The left column was used on the screen. The right column gives the values of the given conditions of the problem in lab manual §8.1 (VEL uses the same values).
SubstanceThis simulator (CRC Handbook table)Given conditions of the problem in lab manual §8.1Effect
Fe(OH)32.79 × 10−394 × 10−36At high pH almost all of the Fe precipitates with either value. The 4 × 10−38 of a common table differs by about 1 order of magnitude from the value in this simulator and by 2 orders of magnitude from the lab-manual value (the values vary widely among tables)
BaCO32.58 × 10−98.1 × 10−9Affects the amount of BaCO3 that dissolves in washing (√Ksp) by a factor of about 1.8. Depending on the table, 2.6 × 10−9 to 5.1 × 10−9

4.3 Supersaturation

As a guide to how much precipitate can form at once, the screen shows the supersaturation at the instant of mixing (the whole amount before precipitation), S = (ion activity product)/Ksp (an approximation that takes the activity coefficients to be 1 and uses the equilibrium value of the pH). When solution B is poured into solution A, log10 S is around 38 for Fe(OH)3 and around 6 for BaCO3. For S ≫ 1 the barrier to nucleation almost disappears, and countless nuclei form at once, giving a fine precipitate. This is why coprecipitated grains are fine; in the internal diagram the newly formed grains are drawn smaller (a schematic that does not calculate the speed of growth).

4.4 When there is little Na2CO3 (a provisional model)

In the VEL model, without carbonate Ba does not precipitate at all. However, Zahn et al. (Ref. 3) obtained single-phase BaFe12O19 without using Na2CO3, adjusting the pH to 12–13 with NaOH alone, and they cite literature saying that BaCO3 forms by absorbing CO2 from the air. We therefore do not assert that “Ba never precipitates without carbonate,” and have added a provisional term in which part of the Ba2+ that did not become BaCO3 is incorporated into the Fe precipitate at high pH: fraction incorporated = 0.5 × clamp((pH − 10)/3, 0, 1) × (precipitated fraction of Fe). Of the Ba that has been incorporated, 5 % is released at each wash. The incorporation reduces the Ba2+ in the liquid and so changes the pH, so this term is also included in the bisection of 4.2, so that the incorporation and the pH do not contradict each other. Both coefficients are provisional values with no basis; they serve only to show the tendency that “with little carbonate, less Ba precipitates, more is lost in washing, and Fe becomes excessive.” The absorption of CO2 from the air itself is not treated.

5. Pouring Direction, Rate, and Stirring (Schematic)

In the practical session no stir bar or stirrer is used; solution B is poured down a glass rod in several portions, and the mixture is stirred with the glass rod in between (video B1). When solution B is poured into solution A, each falling drop always meets a large excess of base, and Fe and Ba precipitate together on the spot. Conversely, when solution A is poured into solution B, the carbonate entering the initially acidic Fe solution escapes as CO2, and when the base put in (NaOH + 2 Na2CO3) reaches 3nFe (about 30 % of solution A for the default starting amounts), the pH rises abruptly. The Fe has precipitated by then and the Ba precipitates after the abrupt rise, so the mixing is poor. You can compare this with the default direction in the graph of ② in the simulator. Janasi et al. (Ref. 4) took the order of addition as a variable.

In the simulator, the uniformity h and the aggregate size d are obtained from VEL’s estimation formulas. The stirring strength is a scale for comparing, as a condition, how strongly the mixture is stirred with the glass rod (rpm equivalent; the default of 400 is provisional). With the stirring score s = log10(1 + rpm/60)/log10 11 and the rate score r = 1 − log10(rate/3)/1.6 (0.15–1), h = 0.985(0.45 + 0.35s + 0.30r)/1.10 and d = 0.30 µm/(0.55 + 0.45sr); for the reverse direction h is multiplied by 0.62 and d by 2.6. The carbonic acid that escapes as CO2 is taken to be only that which accepted two H+ and became H2CO3 (for example, carbonic acid added while the mixture is still acidic), and its fraction is taken as (0.70 + 0.012 × rate) from VEL’s estimation formula (up to 0.97). In the reverse direction, this corresponds to the carbonic acid added until the poured fraction f reaches feq = 3nFe/(nNaOH + 2nCO₃), which under the default conditions (5 mL/min) is about 23 % of the carbonic acid added by the end of pouring (the 76 % obtained by applying VEL’s formula at the end of pouring also counted the carbonic acid added after the pH had risen). Even in the standard direction given in the procedure, the carbonic acid that became H2CO3 because NaOH alone falls short of 3nFe (§4.1) is assumed to escape at the same fraction (the whole amount of carbonic acid when the total base is also insufficient and the mixture becomes acidic). The carbonic acid that escaped has also accepted H+, so it does not affect the pH calculation (charge balance). If the mixture is aged above 55 °C for longer than 20 min, part of it changes into crystalline goethite α-FeOOH (fraction (T − 55)/60 × aging time/180 min, with an upper limit of 0.75; an estimate). All of these are schematic and are not values fitted to experiments.

6. Decantation Washing — Diluting, Dissolving, and Coming Apart

6.1 Diluting

If the supernatant is removed down to Vd (the volume after removal) and water is added up to Vr = 300 mL, the dissolved salts are multiplied in one cycle by (volume after removal) ÷ (volume before removal). The volume before removal is V0 = 150 mL, that of the slurry of ② (100 mL of solution A + 50 mL of solution B), only for the first cycle, and Vr from the second cycle on, so

nk=n0VdV0(VdVr)k−1 (4)

For Vd = 125 mL, the factor is 125/150 = 0.833 for the first cycle and 125/300 = 0.417 from the second cycle on. Only 22–40 % of the Na+ adsorbed on the surface of the gel comes out in each cycle (more when warmer; an estimate), so it decreases a little more slowly than Eq. (4) predicts. This adsorption is assumed to occur only at a pH at which the surface of the hydrous iron oxide is negatively charged (above the isoelectric point ≈ 8) (≡FeO−Na+; an estimate). When the adsorbed Na+ leaves, the ≡FeO− on the surface accepts H+ from water, so the same amount of OH− as Na+ is added to the liquid (treated automatically in the charge balance). A slurry that has become acidic because of insufficient base (for example 2 g of NaOH in ①) has no adsorbed Na+, and on washing, the acidic supernatant is diluted and the pH approaches 7 from below. The Fe3+ remaining in the liquid at that time precipitates as hydrous iron oxide, by the amount that exceeds the solubility product as the pH rises, and releases H+, so the pH rises slowly. The electrical conductivity is a guide obtained by multiplying the sum of the limiting molar conductivities of the ions by a Kohlrausch-type correction 1/(1 + 0.6√I).

6.2 Dissolving

BaCO3 dissolves slightly. When the supernatant already contains [Ba]0 and [CO3]0, the amount that dissolves, x, is the positive root of ([Ba]0 + x)([CO3]0 + x) = Ksp, and over the standing time t the system is assumed to approach equilibrium by the fraction 1 − exp(−t/12 min) (an estimate). The more the carbonate is washed away, the weaker the common-ion effect becomes, and x approaches √Ksp ≈ 5 × 10−5 mol/L. Washing too much removes Ba little by little and raises the Fe/Ba of the solid (10.97 → 11.07 after 13 washes under the default conditions).

This model is a closed system that does not exchange CO2 with the air, so even after the salts are almost removed, the dissolved CO32− reacts with water and the pH of the supernatant stops in the 9 range. This plateau value is not the saturation value; it is the value at which the fraction of approach to equilibrium in one standing (10 min), 1 − exp(−t/12 min), balances the dilution by a factor of 125/300 in each cycle. Solving for saturated water of BaCO3 in a closed system with the same constants gives pH 9.87, but this model counts the amount that dissolves using only free CO32− and does not include the CO32− that changes into HCO3− (a simplification), so the pH stays at about 9.4 even if the standing time is made 30 min. Even if a saturated aqueous solution of BaCO3 in equilibrium with CO2 in the air (about 400 ppm, Henry’s constant 10−1.47 mol L−1 atm−1) is calculated with the same constants, from 2[Ba2+] ≈ [HCO3−] the pH is ≈ 8.2, and, like water in equilibrium with calcium carbonate (around pH 8.3), it does not fall as far as pH 7. It is therefore as expected that the pH does not reach 7 in the model. The practical-session instruction “until the pH of the supernatant is about 7” (lab manual §8.1) is a guide read with pH test paper, and it is thought that the reading can come out near 7 because of the uncertainty in reading the test paper (roughly ±0.5–1), differences in the amount removed and the standing time, and removing the supernatant before the BaCO3 has finished dissolving (not checked; §12). As guides for ending the washing, one can also use the electrical conductivity having fallen as far as it will go, and a supernatant that does not turn white and cloudy when an AgNO3 aqueous solution is added (no AgCl forms = the Cl− is almost gone).

6.3 Coming apart (peptization)

The surface of a precipitate grain carries charge. While the ionic strength I of the liquid is high, the electric double layer is thin, and the grains aggregate and settle quickly. When washing reduces the ions, the double layer thickens and the grains repel each other, so the aggregation comes apart (peptization) and fine grains stay suspended in the supernatant. In VEL’s estimation formula, peptization is taken to occur when the peptization risk = clamp((0.05 − I)/0.05, 0, 1) × (1 − 0.9 × clamp((water temperature − 20)/40, 0, 1)) exceeds 0.45, and the solid lost in the outflow is the suspended fraction (time constant of settling 1.8 + 45 × risk min for the 4 % fine particles, 1.2 min for the lumps) × the fraction of the volume removed. The tendency for warm water to peptize less follows the design of VEL, and no quantitative basis has been checked.

7. Filtration and Drying — Remaining Water and Remaining Salt

In suction filtration, the filter paper is moistened with distilled water, the aspirator is switched on and the suction is checked before the precipitate is poured in, the beaker is rinsed with distilled water, and suction is continued until the filter cake cracks (lab manual §8.1, video B1). In the model, the size of the aggregates is taken to follow a lognormal distribution (geometric standard deviation 3), and 35 % of the part finer than the retention size of the filter paper (Type 2 5 µm, 5A 1 µm, 5B 0.4 µm, 5C 0.25 µm; estimates) is assumed to pass through. If the filter paper is placed without being moistened, a further 6.5 % or so goes around the edge (an estimate). The gel-like filter cake holds more than twice its solid mass of water (water/solid = 4.4 − 2.6 × degree of squeezing; an estimate), and the Na+ and Cl− dissolved in that water remain after drying. If the washing is insufficient, they become NaCl crystals when dry.

In drying, the free water leaves at a first-order rate k = 182.9 exp(−45 kJ mol−1/RT) s−1 (time constant ≈ 2.5 h at 105 °C; an estimate). The hydrous iron oxide keeps its gel water even at 105 °C, and 1.25 mol of H2O per mol of Fe is assumed to remain (an estimate). This bound water leaves during the temperature rise of calcination. In the practical session the sample is put on a watch glass together with the filter paper and dried at 105 °C for about 20 h (lab manual §8.1). When dry, the filter cake breaks into small dark-brown pieces and the filter paper beneath is stained orange (video B2). The dried powder is a precursor, not BaFe12O19. In the practical session, the group at the lab bench with the most dried powder sets aside about 0.5 g of dried powder, and the group with the second most sets aside about 0.5 g of calcined powder, for XRD (this simulator calculates without setting any aside).

8. Calcination — Order of Phase Appearance and Diffusion-Limited Reaction

8.1 Reactions

①Loss of free water and bound water (from around 100 °C) ②2FeOOH→α-Fe2O3+H2O ③BaCO3+Fe2O3→BaFe2O4+CO2 ④BaFe2O4+5Fe2O3→BaFe12O19(rate-limiting) (5)

If the intermediate phase is insufficient in reaction ④, BaFe12O19 is made directly from BaCO3 + 6 Fe2O3. The Ba that was incorporated (§4.4) is counted as equivalent to Ba(OH)2, becomes BaO at the same rate as reaction ②, and reacts like BaCO3 in reactions ③ and ④ (no CO2 is released). The NaCl left after incomplete washing melts at 801 °C and becomes a flux, speeding up the reaction by up to a factor of 10 and grain growth by up to a factor of 46 (saturating when NaCl is 3 % or more of the solid), and then volatilizes slowly (time constant about 30 min at 900 °C; all estimates). The mass of volatilized NaCl is shown in the “Lost” line of the TG-like plot, alongside H2O and CO2. The vertical axis of the TG-like plot is normally 65–101 %, and it is extended downward for a powder that is not yet fully dry (with a lot of free water). Of the Na left after washing, the part that is not paired with Cl− (NaOH, Na2CO3, and the Na on the surface of the precipitate) becomes Na2O above 400 °C (NaOH releases H2O, Na2CO3 releases CO2, and the oxygen stays in the solid), and by Na2O + Fe2O3 → 2 NaFeO2 it is assumed to become a Na–Fe oxide impurity (time constant 10 min at 700 °C; an estimate). Zahn et al. (Ref. 3) found 0.5–0.7 wt% of Na-based impurities such as FeNaO2 in insufficiently washed samples. NaFeO2 takes up Fe that could have been used for BaFe12O19, so Ba is left over and BaFe2O4 remains, and the fraction of BaFe12O19 and the saturation magnetization decrease.

8.2 Equation for diffusion-limited reaction

Reactions ③ and ④ are considered to be determined by the speed at which atoms diffuse through the product layer, and the Ginstling–Brounshtein equation for spherical particles (Ref. 8) is used.

1−23α−(1−α)2/3=kt, k=2Dr2, D=D0exp(−EaRT) (6)

α is the fraction of the Ba-containing phases that has become product, and r is the effective radius of the reaction, taken as aggregate size × (2 − uniformity) (the poorer the mixing, the longer the diffusion distance). Because the temperature changes with time, kΔt is added to g(α) in steps of 120 s or less and α is solved again. The constants are those of VEL: for reaction ④, D0 = 1.56 × 10−3 cm2/s and Ea = 250 kJ/mol (an estimate chosen so that an aggregate size of 0.3 µm gives α ≈ 0.9 at 900 °C for 2 h), for reaction ③, 8.46 × 10−4 cm2/s and 215 kJ/mol (an estimate), and reaction ② is a first-order reaction with A = 1.66 × 106 s−1 and 110 kJ/mol (time constant about 17 min at 350 °C; an estimate). The growth of the crystallites follows G3 − G03 = k0 exp(−Q/RT) t (k0 = 2.5 × 1021 nm3/s, Q = 450 kJ/mol; an estimate).

8.3 Comparison with the literature

Table 3. Calculated values when the default dried powder (13 washes) is heated at 3 °C/min and held for 2 hThe heating rate is the value of lab manual §8.2. The phases are mass fractions of the solid (excluding water and residual salts). The magnetic property is the guide given in §9.
Holding temperatureHydrous iron oxideBaCO3α-Fe2O3BaFe2O4BaFe12O19Mass lossCrystallite sizeHc
300 °C27 %18 %56 %0 %0 %17.1 %−−
400 °C0 %18 %82 %0 %0 %24.4 %−−
600 °C0 %16 %80 %3 %1 %24.7 %12 nm−
700 °C0 %10 %71 %11 %8 %25.8 %12 nm181 kA/m
800 °C0 %0 %46 %21 %33 %27.5 %16 nm270 kA/m
900 °C0 %0 %7 %6 %87 %27.5 %56 nm412 kA/m
925 °C0 %0 %0 %3 %97 %27.5 %78 nm422 kA/m
1000 °C0 %0 %0 %3 %97 %27.5 %189 nm434 kA/m
1100 °C0 %0 %0 %3 %97 %27.5 %531 nm247 kA/m

In the literature, BaFe12O19 from a coprecipitated precursor appears in XRD from about 800 °C (Ref. 4), there is an example of a single phase at 800 °C for 1 h (Ref. 1, with mechanical activation), an example in which the coercivity exceeds 370 kA/m at 800–1000 °C for 3 h and falls to 45–70 kA/m at 1200 °C (Ref. 3), and an example in which, for a sol–gel precursor, α-Fe2O3 appears as an intermediate phase at 550–900 °C and BaFe2O4 at 650–750 °C (cited in Ref. 7). In Table 3, BaFe12O19 appears from around 700 °C and is almost complete at about 900 °C for 2–3 h (87 % at 900 °C for 2 h, 97 % for 3 h, and 93 % at 910 °C for 2 h), and above about 1050 °C (when the crystallite size of BaFe12O19 exceeds the critical single-domain size of 0.3 µm (provisional)) grain growth lowers the coercivity (for a 2 h hold, 434, 405, and 247 kA/m at 1000, 1050, and 1100 °C). The order in which the phases appear and the direction of the decrease of coercivity at high temperature agree with the literature. However, the temperature at which it is complete corresponds to the upper end of the “800–900 °C” of the literature, and near 800 °C the model is slower (39 % at 800 °C for 3 h, coercivity 290 kA/m; Ref. 3 reports a single phase and more than 370 kA/m at 800 °C for 3 h). This is because VEL’s D0 was matched to α ≈ 0.9 for an aggregate size of 0.3 µm held at 900 °C for 2 h, whereas this simulator has an effective radius of about 0.375 µm and also includes the time of the temperature rise. The temperature boundaries move by 100 °C or more depending on how the precursor is made, so the revision of D0 awaits the results of the practical session (§12). The numbers themselves are not predictions.

The mass loss is the sum of the adsorbed water, bound water, and H2O of dehydroxylation, and the CO2 of decarbonation. The simulator includes the bound water (1.25 mol per mol of Fe; an estimate), so the mass loss is larger than the stoichiometric value that assumes Fe(OH)3 or FeOOH (24.89 % for BaCO3 1 mol + Fe(OH)3 12 mol → BaFe12O19, shown as a comparison in the readout of ④; the calculation for the default dried powder is the 27.5 % of Table 3). We have confirmed that the balance of amounts of substance (the total amounts of Fe and Ba, and initial mass = remaining mass + H2O lost + CO2 + volatilized NaCl) is preserved at every stage of the calculation (for samples washed only a few times, the volatilized NaCl is added; 0.078 g for 2 washes).

8.4 What remains when the composition is off

When all the Ba goes into BaFe12O19 and the Fe/Ba of the solid is R, for R > 12 there is a surplus of (R − 12)/2 mol of α-Fe2O3 per formula unit.

wα-Fe2O3=MFe2O3(R−12)/2MBaFe12O19+MFe2O3(R−12)/2 (7)

For R < 12, the surplus Ba remains as BaFe2O4 (in the simulator’s model it does not disappear even at high temperature). For a starting ratio of 13, the solid after washing is 13.1, and α-Fe2O3 from Eq. (7) is 7.5 %. At the end of 925 °C for 2 h (3 °C/min), 11.2 % of α-Fe2O3 and 1.4 % of BaFe2O4 remain; of this, the 7.5 % of Eq. (7) is due to the composition shift, and the remaining 3.7 % or so is the part still in the middle of reacting (at 925 °C for 10 h or at 1000 °C for 2 h it falls to 7.5 % and no further). With a starting ratio of 8, about 16 % of BaFe2O4 remains (the same at 925 °C and at 1100 °C). Even for the default starting ratio of 10.97, in this model almost all (99.99 %) of the Ba precipitates in ②, the Ba lost in excess of Fe in 13 washes is about 0.9 % (Fe/Ba of the solid 10.97 → 11.07), so Ba is left over and 2.8 % of BaFe2O4 remains (the same at 925 °C and at 1200 °C). Even with a starting ratio of exactly 12, Ba is lost in washing and the solid becomes 12.1, and α-Fe2O3 from Eq. (7) is 0.8 % (4.7 % at the end of 925 °C for 2 h, including the unreacted part). The short comment at the end of ④ in the simulator shows the unreacted part and the composition-shift part separately, from the difference between the present phases and this “phases when the reaction has gone to completion.” The literature also summarizes that in the solid-state method α-Fe2O3 coexists at Fe:Ba ≥ 12, and BaFe2O4 coexists below 1000 °C at Fe:Ba < 11 (Ref. 7).

9. Guides to Magnetism, and Pressing, Sintering, Magnetization, and Apparent Remanent Magnetization (Schematic)

9.1 Guides to the magnetism of the calcined powder (④)

The saturation magnetization is obtained by multiplying the room-temperature value of BaFe12O19, 72 A·m2/kg (cited from Ref. 3), by the mass fraction of BaFe12O19, and reducing it by the magnetically dead layer at the surface of fine crystals (15 % at 20 nm, 0 at 200 nm and above; VEL’s estimate). α-Fe2O3 and BaFe2O4 are assumed to contribute almost nothing to the room-temperature magnetization.

Hc=Hmax·min(1,DsdG)·[1−(DspG)3/2] (8)

G is the crystallite size, Hmax = 5500 Oe ≈ 438 kA/m (VEL’s value; about the same as the measured 440 kA/m (Ref. 4) and 436.7 kA/m (Ref. 1) of coprecipitated powders), the critical single-domain size is Dsd = 0.3 µm (provisional; in the literature it ranges from 100–300 nm to about 1 µm; VEL uses 0.7 µm), and the critical superparamagnetic size Dsp is 8.4 nm from KuV = 25kBT (Ku = 3.3 × 105 J/m3, 300 K). This is a schematic that shows only the trend that the coercivity decreases when a crystallite exceeds the single-domain size and domain walls form; the grain size and shape anisotropy of the sintered body are not treated.

9.2 Pressing, sintering, magnetization, and apparent remanent magnetization (⑤, schematic)

In the practical session (lab manual §8.3–8.5, videos B3 and B4), the calcined powder is ground, 3 drops of a 5 mass% PVA aqueous solution are added to about 3 g and granulated, and two green bodies of different thickness are made in a die of φ 1.000 cm at 98.1 MPa with a 1 min hold. They are sintered at 1100 °C for 2 h (about 3 °C/min, furnace cooling), polished with #400 → #800 to bring the thickness below 5 mm, and the bulk density is obtained from the mass W, diameter D, and thickness h. They are magnetized in the electromagnet for magnetization with a direct current of 10 A (752 000 A/m between the pole pieces) for 1 min; then the reference magnet is brought close, in the orientation that repels the sample, from above the sample placed on the sample stage of the balance, and the apparent mass Wrepul is read. The ⑤ tab draws this flow continuing from the calcined powder of ④. The model is the following schematic and is not a prediction of measurements.

Under the default conditions (13 washes in ③, 925 °C for 2 h in ④), the rough guide for the increase in the display, Wrepul − W, is about 0.27 g for A (0.6 g), about 0.62 g for B (1.2 g), and about 1.04 g for A+B (the balance readings Wrepul are about 0.84, 1.80, and 2.79 g, respectively). The increase is larger for thicker samples and grows faster than the ratio of the volumes (because the thicker the sample, the closer its center is to the reference magnet and the weaker the demagnetizing field). When the increase is less than 0.001 g (the last digit shown on the screen) or the BaFe12O19 in the sintered body is less than 2 %, “hardly increases” is displayed. Typical values in the practical session have not been checked, so these values are only a rough guide (§12).

9.3 The lab-manual analysis — estimating the demagnetization curve from the balance readings (readout of ⑤)

The analysis of results in lab manual §8.5 is a procedure that obtains the magnetization of the sample from the balance readings alone, corrects for the effect of the demagnetizing field, and estimates the demagnetization curve (the B–H curve in the second quadrant). When the measurement in ⑤ is finished, the simulator takes the readings on the screen as input, carries out this procedure as it is, and shows the result in the readout under “Analysis as in the lab manual.” The units are the same E–H correspondence MKSA system as in the lab manual (magnetization I and magnetic flux density B in Wb/m2, magnetic field H in A/m, magnetic moment in Wb·m), with μ0 = 4π × 10−7 Wb/(A·m), g = 9.81 m/s2, and the magnetic moment of the reference magnet μref = 1 × 10−6 Wb·m.

  1. Distance: The distance between the centers of the reference magnet and the sample is the distance L′ from the lower face of the reference magnet to the sample stage, plus half the thickness of the reference magnet, minus half the thickness of the sample.
    L=L′+href−h2(8-13)
  2. Magnetic moment: Regarding the two magnets as point dipoles lined up on the axis, the repulsive force is 3μμref/(2πμ0L4), and this equals the force obtained by multiplying the increase in the reading, (Wrepul − W) (kg), by g, so
    μ=2πμ0gL4(Wrepul−W)3μref(8-14)
  3. Magnetic charge and magnetization: Regarding the sample as a magnet of length h with magnetic charges ±md on its upper and lower faces, μ = mdh. Since the magnetization is the magnetic charge of a face divided by its area,
    md=μh,Id=4mdπD2(8-15), (8-16)
    Id is the magnetization that remains with the sample’s own demagnetizing field acting (the apparent remanent magnetization).
  4. Demagnetizing factor: The two points of Table 7-1 of the lab manual, N = 1 at dimension ratio (length/diameter) 0 and N = 0.27 at 1, are joined by a straight line, giving Ni ≈ 1 − 0.73 h/D (i = 1, 2, 3 for A, B, A+B).
  5. Points on the demagnetization curve: Inside the sample, B = μ0H + I and the demagnetizing field H = −NI/μ0 hold. Putting in I = Idi gives
    Bdi=(1−Ni)Idi,B=(1−1Ni)μ0H(8-17), (8-18)
    Here the right-hand equation is the straight line of the demagnetizing field drawn in the B–H plane (the form with I eliminated). The point on it where B = Bdi is (−Hdi, Bdi), where, as in the lab manual, Hdi = NiIdi/μ0 is the magnitude of the demagnetizing field (a positive value).
  6. Estimation: When the three points (−Hdi, Bdi) for the different thicknesses are joined by the least-squares line B = a + bH, the intercept at H = 0 is the remanent flux density Br (= Ir), and the magnitude of the field at which B = 0, a/b, is the coercivity Hc. The reason for making two samples of different thicknesses and also measuring the two attached together as one piece is to obtain these three points.
Table 4. Values obtained by analyzing the readings under the default conditions (13 washes in ③, 925 °C for 2 h in ④, 0.6 g and 1.2 g of powder)The same calculation as in the readout of ⑤ in the simulator. The diameter D is 8.83 mm in all cases.
ABA+B
Thickness h / mass W2.18 mm / 0.574 g4.46 mm / 1.174 g6.64 mm / 1.747 g
Balance reading Wrepul0.839 g1.797 g2.788 g
(8-13) L43.27 mm42.13 mm41.04 mm
(8-14) μ2.40 × 10−8 Wb·m5.07 × 10−8 Wb·m7.62 × 10−8 Wb·m
(8-16) Id0.1799 Wb/m20.1855 Wb/m20.1874 Wb/m2
Ni (linear approximation)0.8200.6310.451
(8-17) Bdi0.0325 Wb/m20.0685 Wb/m20.1030 Wb/m2
Hdi (the point is (−Hdi, Bdi))117.4 kA/m93.2 kA/m67.2 kA/m
From the least-squares lineBr ≈ 0.198 Wb/m2, Hc ≈ 141 kA/m

The values of this analysis and the internal values of the model used to generate the readings are different things. The model determines the apparent remanent magnetization from the remanent magnetization of the isotropic sintered body, Ir = 0.189 Wb/m2, the rough guide for the I–H coercivity, 246 kA/m, and the demagnetizing factors given by the curve approximation as a function of the dimension ratio (A 0.625, B 0.440, A+B 0.337). Eqs. (8-13)–(8-16) are the same relation as the model’s equation for the repulsive force, so the Id calculated back from the readings agrees with the model’s Id. The difference comes from the following two points. At a dimension ratio of 0.25, the linear approximation of the demagnetizing factor gives 0.82, larger than the curve (0.62), which changes the positions of Bdi and Hdi. Also, the estimated Hc is the coercivity of the B–H curve, a different quantity from the coercivity of the I–H curve. Since it is the field at which B = μ0H + I becomes 0, its magnitude does not exceed Ir/μ0 ≈ 150 kA/m. As a result, the estimated Br (0.198 Wb/m2) comes out about 5 % larger than the model’s Ir, and Hc (141 kA/m) comes out smaller than the model’s 246 kA/m. Both are differences due to the approximations in the procedure of “correcting for the demagnetizing field and estimating with a straight line,” so the simulator displays the two sets of values separately.

10. How the XRD-Like Plot Is Made

The “XRD-like” plot that can be selected in the graph of ④ has calculated values only for the positions of the lines. For the hexagonal system, 1/d2 = 4(h2 + hk + k2)/3a2 + l2/c2; for the orthorhombic and cubic systems, the interplanar spacing is obtained from the respective formulas, and it is converted to 2θ with Bragg’s law λ = 2d sin θ (Cu Kα1, λ = 1.5406 Å). The lattice constants are BaFe12O19: a = 5.89 Å, c = 23.18 Å (Ref. 9); α-Fe2O3: a = 5.034 Å, c = 13.75 Å; goethite: a = 4.608, b = 9.956, c = 3.021 Å (Ref. 10); and NaCl: a = 5.640 Å. For BaCO3, only the two strong lines (d = 3.72 and 3.65 Å) are drawn, and two-line ferrihydrite is drawn as two broad humps at d ≈ 2.5 and 1.5 Å (Ref. 10). The lines of BaFe2O4 are omitted.

The height of a line is the mass fraction of the phase multiplied by the relative intensity of a representative line (schematic; the structure factor is not calculated), and differences in scattering strength between phases are ignored. The lines are broadened according to the crystallite size G with the Scherrer equation β = 0.9λ/(G cos θ) (Ref. 11). The plot is meant for seeing that the humps of the precursor are low and broad and that the BaFe12O19 lines appear and sharpen on calcination; it cannot be used for comparison with measured patterns. For the principle of XRD, see the X-ray diffraction simulator.

11. Simplifications and Range of Applicability

12. Items to Confirm (with Prof. Mochizuki)

The reagent masses and water volumes, the liquid volume after removing the supernatant (about 125 mL), the filter paper (5B), the drying (105 °C, about 20 h), the calcination (about 3 °C/min, 925 °C for 2 h, lidded, furnace cooling), and the conditions for pressing, sintering, and magnetization were checked against the lab manual and are no longer provisional values. The VEL values of the solubility products and of the heat of dissolution and specific heat were also the same as the given conditions of the question in lab manual §8.1. The atmosphere of the heat treatment, the dimensions of the balance, and the range that may be made public were settled by Prof. Mochizuki’s reply (2026-09-25) (items 1, 4, and 5 below). The items that remain to be confirmed are as follows.

  1. Atmosphere for calcination and sintering (resolved: Prof. Mochizuki’s reply, 2026-09-25): The student laboratory uses air. In research, N2 is sometimes flowed through a tube furnace, or heat treatment is done in a glove box filled with Ar. This simulator treats only air (§2).
  2. How washing ends: The practical session says “until the pH of the supernatant is about 7,” but in this model (a closed system) the pH stops in the 9s, and even when equilibrated with CO2 in the air it is about 8 by calculation (§6.2). After about how many washes is pH 7 actually judged to be reached? Also the standing time (not specified in the lab manual; this simulator uses 10 min, provisional).
  3. Examples of practical-session results: Typical values for the XRD of the dried and calcined powders, the mass loss on calcination, the relative density of the sintered body, and the apparent remanent magnetization Wrepul. If available, we would like to match the schematic coefficients to them.
  4. Dimensions of the balance (resolved: Prof. Mochizuki’s reply, 2026-09-25): Thickness of the reference magnet href = 1.4020 × 10−2 m; distance from the lower face of the reference magnet to the upper surface of the sample stage L′ = 3.7349 × 10−2 m (values written on the balance). The readings in ⑤ and the numbers in §9.2 and §9.3 were calculated with these values.
  5. Range that may be made public (resolved: Prof. Mochizuki’s reply, 2026-09-25): The lab manual is that of the former Tokyo Institute of Technology and is not distributed in classes at Tokyo University of Science, so all calculated values may be shown. Accordingly, we put the theoretical yield 3.4747 g, BaO:Fe2O3 = 1:5.487, the liquid temperature of 54.6 °C when the NaOH is dissolved all at once, the 4.501 g of NaOH needed for Fe(OH)3, the theoretical mass loss of 24.89 %, the X-ray density of 5.302 g/cm3 (a reference row in §9.2 of the text and in the readout of ⑤; the true density 5.374 of the lab manual is used for the relative density), the hydraulic gauge reading of about 12.0 MPa, and the analysis of §9.3 (L, μ, md, Id, Ni, Bdi, Hdi, and the estimated Br and Hc) in the text and on the screen. The wording of VEL’s feedback comments and questions has not been copied.
  6. Values of the solubility products: Which should be used on the screen, the given conditions of the lab manual (Fe(OH)3 4 × 10−36, BaCO3 8.1 × 10−9) or the values from the CRC Handbook that this simulator uses?
  7. Treatment of Ba when Na2CO3 is scarce (the provisional model of §4.4): May it be shown in class?
  8. Critical single-domain size of 0.3 µm (provisional), and whether to show the rough guide for the coercivity on the screen.
  9. Starting Fe/Ba and the phases of the calcined powder: Under the default conditions of this model, 2.8 % of BaFe2O4 remains at a starting ratio of 10.97 (§8.4). Is BaFe2O4 seen in the XRD of the calcined powder in the practical session?
  10. How the precursor is named: We wrote the Fe side as “a hydrous iron oxide (often written Fe(OH)3).” The example in the lab manual’s report is Fe(OH)3, and in the literature it is two-line ferrihydrite.

Appendix A. What the Simulator (coprecip-simulator-en.html) Shows

What this appendix explains What each tab of the companion simulator coprecip-simulator-en.html draws, what each operation means, and by which approximation the numbers shown on the screen are calculated. The derivations of the equations are in §3–§10 of the main text, so here we collect how to read the screen and the settings that appear only in the simulator.

A.1 The page as a whole — five steps and how the sample is handed on

The five tabs are arranged in the order of the steps of the practical session. Every tab has a diagram of the apparatus at the upper left, an enlarged schematic of the inside of the material at the upper right, and a graph below them, and below these a “short comment” (what the current operation changed) and a table of readout values, with the controls on the right (below, on a smartphone). The apparatus diagram, internal diagram, graph, and numbers are all drawn from a single state (sample) within the page, so they never disagree.

Each step takes over the sample from the previous step as it is. If you change the starting amounts in ①, then ② and the later steps are made again. When you open a later tab without having done the previous step, the sample obtained by automatically running the previous step to the end under the current conditions is used, and this is shown in the yellow band above each tab (the automatic washing in ③ is 13 washes). However, if you open ④ while ③ is partway through (in the middle of washing, or up to filtration), the state of ③ is left unchanged and ④ is calculated for a powder assumed to be a filtered and dried copy of it, so you can return to ③ and continue from where you left off. If you open ⑤ without having run ④ to the end, the powder obtained by calculating the calcination to the end under the current conditions of ④ is used. The row above the tabs, “① Solution → ② Precipitate (precursor) → ③ Dried powder (precursor) → ④ Calcined powder → ⑤ Sintered body (magnet),” shows how far you have progressed.

Adding #solution, #precip, #wash, #calcine, or #form to the end of the address opens that tab. ?embed=1 gives a display for embedding with the heading and introduction hidden (the tabs remain); writing something like ?only=calcine displays only that tab. Animations stop when you switch tabs or when the screen is hidden.

Table A1. Correspondence between the tabs and the main text
TabWhat to look at on the screenMain text
① Weigh and dissolveOrder of weighing and dissolving, ions in the liquid, pH, heat of dissolution, comparison of the starting ratio with the stoichiometry and the base budget§2, §3, A.2
② Dropwise addition and precipitationPouring direction and the change of pH, precipitated fractions of Fe and Ba, supersaturation, grains of the precursor§4, §5, A.3
③ Wash, filter, and drypH and electrical conductivity of the supernatant, peptization and outflow, salts remaining in the filter cake, contents of the dried powder§6, §7, A.4
④ CalcinationTemperature program, phase changes, mass change, XRD-like plot, rough guides to the magnetism§8–§10, A.5
⑤ Press, sinter, and magnetizeOrder of granulation, pressing, sintering, polishing, magnetization, and measurement on the balance; grains, pores, and direction of magnetization; bulk density; balance reading Wrepul and increase Wrepul − W; analysis as in the lab manual ((8-13)–(8-18))§9.2, §9.3, A.6

A.2 ① Weigh and dissolve

“Weigh and dissolve” draws the weighing on the balance and the dissolving in the order NaOH → Na2CO3 (solution A) and FeCl3·6H2O → BaCl2·2H2O (solution B) (the time of each frame is shortened). As in video B1, the containers are a plastic beaker for NaOH, weighing paper for Na2CO3, a large balance tray for the iron chloride, and a small balance tray for the barium chloride. The default reagent masses and water volumes are the values of lab manual §8.1. The thermometer shows the estimate of the heat of dissolution (§3), and if “How the NaOH is added” is set to “All at once,” “Splash hazard” is displayed. The number of ions in the internal diagram is proportional to the amount of substance: each Fe3+, Ba2+, CO32−, and H+ drawn stands for 1 mmol, and the more numerous Na+, Cl−, and OH− are thinned out so that each one drawn stands for 8 mmol. The area of the box is proportional to the liquid volume. The graph places the “BaFe12O19 that can be made from Fe (nFe/12)” beside the “amount that can be made from Ba (nBa)” and shows that the smaller one determines the theoretical yield, and whether the base added exceeds the amount needed to precipitate the Fe (3nFe). The sliders are the starting Fe/Ba (8–13; it sets the amount of BaCl2·2H2O), the amount of FeCl3·6H2O (the concentration of solution B), and the amounts of NaOH and Na2CO3. Inputs that are out of range or not numbers are clamped to the end of the range before the calculation.

A.3 ② Dropwise addition and precipitation

“Start dropwise addition” solves the equilibria of §4 for each mixture obtained by adding the liquid being dropped in a little at a time, and draws the pH, the precipitated fractions, and the supersaturation (pouring finishes in 7 s on the screen; the time in the real experiment is the volume poured ÷ the rate). As in the practical session, the apparatus diagram pours the liquid in portions along a glass rod (the practical session says “in several portions” with no number specified, so the 5 portions on the screen are schematic) and stirs with the glass rod in between (no stir bar, stirrer, or pH meter is used; the pH is displayed as a calculated value). After pouring is finished, the beaker is covered with a watch glass and left to stand. The stirring-strength slider is a scale for comparing conditions (rpm equivalent). The upper panel of the graph shows the pH and the lower panel shows the precipitated fractions of Fe and Ba; the thin dotted lines are the planned curves up to the end of pouring. The horizontal axis is the fraction poured (end of pouring = 100 %), so the curves can be compared even when the volume of the liquid poured changes with the direction. If you press “Keep last result” and then change the conditions, the previous curves are overlaid as gray dashed lines (dotted for Ba). In the internal diagram, the brown grains are hydrous iron oxide (1 grain = 1 mmol of Fe), □ is BaCO3 (1 square = 1 mmol), the small blue-green dots are incorporated Ba (provisional; 1 dot = 0.5 mmol), and the needles are goethite. The fraction of □ lying next to the brown grains is matched to the uniformity (§5). If you set the pouring direction to “Pour A into B (reverse),” you can see the pH stay acidic up to about 30 % (during which Fe precipitates), then rise abruptly, at which point Ba precipitates, and also the carbonate that entered during the acidic stage escaping as CO2 (estimate). Use “Keep last result” to overlay it on the default direction and compare. Examples of conditions to compare: the direction (B→A and A→B); 2 g of NaOH in ① (insufficient base: Fe stays in solution and Ba does not precipitate); 0 g of Na2CO3 in ① (Ba does not become BaCO3); and aging at 60 °C for 120 min (goethite).

A.4 ③ Wash, filter, and dry

“Wash once” performs one cycle of standing → removing the supernatant with a Komagome pipette → filling to 300 mL with warm water (1.5 s on the screen). “Until the supernatant pH stops falling” stops when the drop in pH per wash becomes less than 0.03 or after 40 washes. In the apparatus diagram, the beaker is covered with a watch glass during standing, the supernatant is removed with a 5 mL Komagome pipette into a waste bottle, and warm water brought to about 60 °C in a 500 mL beaker on a hot plate is poured in with a 20 mL Komagome pipette. When the slurry is acidic (insufficient base), the button changes to “Until the supernatant pH approaches 7” and stops when the pH reaches 6.8 or higher or after 40 washes. The pH axis is normally 6–15, and is extended downward if there is an acidic supernatant. The short comment gives a warning when, because the standing time is short, more than 2 % of grains that have not settled completely are removed in one wash (or when the cumulative loss exceeds 10 % of the precipitate of ②). Pressing “Keep last result” lays the pH and electrical-conductivity curves underneath as thick gray lines and lists the number of washes, the pH, the cumulative loss of precipitate, and the Fe/Ba of the solid in the readout (since the pH and conductivity are determined almost entirely by how much the liquid is diluted, the curves tend to overlap, and differences in water temperature or standing time show up in the cumulative loss). If you switch tabs in the middle of washing, the remaining washes are calculated without animation and reflected. The default values of 125 mL of liquid after removal, 60 °C, 300 mL, filter paper 5B, and 105 °C for 20 h are the values of lab manual §8.1, and the standing time of 10 min is provisional. “Filter with suction” draws the sequence of wetting the filter paper → checking the suction → pouring → rinsing the beaker → suction until cracks appear → closing the stopcock and switching off → transferring to a watch glass together with the filter paper. After drying, the diagram shows the filter cake broken into dark-brown small pieces and the filter paper stained orange, and “Grind in a mortar and transfer to the crucible (video B2)” draws the weighing and grinding of video B2 (the portion set aside for XRD is not included in the calculation). The graph shows the pH of the supernatant (left axis) and the electrical conductivity (right axis, logarithmic), with red dots for the washes in which peptization occurred. The internal diagram shows the supernatant (Na+, Cl−, OH−, and so on decrease with each wash) and the layer of precipitate; on peptization, the grains come apart and float in the supernatant. Amounts of less than one dot are not drawn, so once they become small, read them from the numbers. After filtration it becomes the filter cake (the Na+ and Cl− remaining in the mother liquor are drawn enlarged, 1 dot = 0.1 mmol), and after drying it becomes the dried powder (NaCl crystals are drawn, 1 crystal = 0.1 mmol). Examples of conditions to compare: water temperatures of 60 °C and 20 °C (at 20 °C, peptization starts around the 7th wash and the precipitate is washed away); filtering after 2 washes (NaCl and NaOH remain, the grains become larger in ④, and NaFeO2 forms); filter paper Type 2 (the filtrate becomes cloudy); and drying at 60 °C for 5 h (still damp).

A.5 ④ Calcination

“Start heating” plays back, in 10 s on the screen, the result of advancing the calculation of §8 along the temperature program (heating → holding) in steps of 120 s or less. The color of the window of the electric furnace is a rough guide to the temperature, and H2O↑ and CO2↑ indicate that the gas is being released at that time. The internal diagram is a cross-section of a grain; with the phase fractions as areas, it is painted from the outside in the order BaFe12O19 (dark gray; the hexagonal plates are crystals), BaFe2O4 (yellow), α-Fe2O3 (red), and hydrous iron oxide (brown), and BaCO3 is drawn as □ next to the grain (farther away if the mixing is poor). When NaCl is molten, the edge of the grain turns blue. The graph can be switched among “Phases and temperature,” “Mass change (TG-like),” and “XRD-like (positions calculated)” (§10), and “Keep last result” overlays the previous result as dashed lines (phase fractions in the same color as each phase; TG and XRD in gray). While there is still almost no BaFe12O19 (less than 1 %), the crystallite size is displayed as “—.” For a sample with little Na2CO3, the Ba incorporated into the Fe precipitate (the provisional model) and the BaO formed when it dehydrates are shown in the phase table and graph as “incorporated Ba (provisional) and BaO” (the phases add up to 100 %). Under “Sample to calcine,” besides the dried powder made in ③, you can choose comparison samples that differ from the default procedure in just one respect (Fe/Ba 13, Fe/Ba 8, 2 washes, reverse direction, no Na2CO3). Examples of conditions to compare: holding temperatures of 700, 800, 925, and 1100 °C (unreacted → intermediate phase → nearly complete → coercivity drops through grain growth); Fe/Ba of 13 and 8 (which of α-Fe2O3 and BaFe2O4 remains); the reverse direction (slower even at the same temperature); and 2 washes (at 925 °C, about 11 % NaFeO2 and 83 % BaFe12O19, and the saturation magnetization falls).

A.6 ⑤ Press, sinter, and magnetize

Each press of “Next step” advances one step at a time, in this order: ① grind the calcined powder in a mortar and put 3 g in a 30 mL beaker; ② add 3 drops of 5 % PVA aqueous solution and granulate; ③ uniaxially press 2 pieces (φ 10 mm, 98.1 MPa, 1 min); ④ sinter (3 °C/min → 1100 °C for 2 h → furnace cooling); ⑤ polish (#400 → #800) and measure W, D, and h; ⑥ magnetize (DC 10 A, 1 min; operated by the staff); ⑦ measure the apparent mass from repulsion on the balance (A, B, A+B) (the conditions are from lab manual §8.3–8.5). “Advance to the end” goes all the way to the end at once, and “Reset” returns to the beginning. The sample is the final state of the calcined powder of ④, and if you change the sample or temperature in ④, ⑤ starts over from the beginning.

The internal diagram draws, in order, the ground grains → grains bound together by PVA → the pressed green body (with many gaps) → the sintered body in which the grains have grown and joined (the white dots of pores are placed at triple points of grain boundaries, and their area fraction is matched to 1 − relative density), and in the sintered body, arrows along the easy axis of each grain show the direction of magnetization. The colors indicate phases (BaFe12O19 dark gray, α-Fe2O3 red, BaFe2O4 yellow, NaFeO2 green), and the proportions are matched to the calculated phases after sintering. The graph shows the relative densities of the green body and the sintered body (left) and the rough guide for the increase in the displayed value, Wrepul − W (right; displayed to 0.001 g). When the measurement is finished, the readout lists “Analysis as in the lab manual” (L, μ, md, Id, the linearly approximated Ni, Bdi, and Hdi obtained from the balance readings with Eqs. (8-13)–(8-18), and Br and Hc estimated by least squares through the 3 points; §9.3) and, separately, “Internal values of the model” (the curve-approximated N, Ir, Id at the intersection, and the rough guide for the I–H coercivity). The hydraulic gauge reading in the pressing diagram is 98.1 MPa × (1.000/2.86)2 ≈ 12.0 MPa (die diameter 1.000 cm, ram diameter 2.86 cm; lab manual §8.3). If you press “Keep last result” and then change the conditions, the previous result is overlaid as gray dashed frames. The sliders set the amounts of powder used for green bodies A and B (the thickness changes), and when a green body becomes 6 mm or thicker, a “Too thick” message asking you to press again is shown and you cannot proceed to sintering. Examples of conditions to compare: the thicknesses of A and B (Wrepul − W is larger the thicker the sample, and grows faster than the ratio of the volumes); setting the sample of ④ to “vs. 2 washes (salt remains)” or “vs. no Na2CO3 (provisional)” (less BaFe12O19, so Wrepul is smaller); and running ④ at 700 °C (the reaction proceeds during sintering). The model and the provisional values are in §9.2.

A.7 The calculation model and notes

Table A2. Constants used and the kind of source
QuantityValueKind and source
Standard atomic weightsH 1.008, C 12.011, O 15.999, Na 22.990, Cl 35.45, Fe 55.845, Ba 137.327Literature values (Ref. 12)
Kw, Ka1, Ka2, hydrolysis of Fe3+10−14, 4.45 × 10−7, 4.69 × 10−11, 6.3 × 10−3General chemical constants (25 °C)
KspTable 2Literature values (Ref. 13)
Heat of dissolution and specific heat of NaOH44.51 kJ/mol, 4.179 J/(g·K)Given conditions of the question in lab manual §8.1 (VEL uses the same values)
Limiting molar conductivitiesNa+ 50.1, Cl− 76.3, OH− 198.0, H+ 349.8 S·cm2/mol, etc.General values (from VEL)
Conditions for starting amounts, water volumes, washing, filtration, drying, and calcinationTable 1Lab manual §8.1–8.2 and videos B1 and B2 (the standing time and pouring rate are provisional; the atmosphere is air: Prof. Mochizuki’s instructions (2026-09-25))
Uniformity, aggregate size, washing, filtration, drying, bound water§5–§7VEL estimates (schematic)
Incorporation of BaMaximum 0.5; 5 % per washProvisional (added in this simulator)
Na+ adsorbed on the surface0 → maximum at pH 8–10 (only above the isoelectric point ≈ 8); 22–40 % released per washEstimate (the amount released is from VEL; the pH condition was added in this simulator)
Rate constants of calcination, crystallite growth, NaCl§8.2VEL estimates (the completion temperature is the upper end of the literature range; §8.3)
Unwashed Na → NaFeO2Na2O above 400 °C; time constant 10 min at 700 °CEstimate (added in this simulator, following the Na-based impurities reported by Zahn 2024)
Magnetism72 A·m2/kg, 5500 Oe, Dsd 0.3 µm, Ku 3.3 × 105 J/m3Literature values, VEL estimates, provisional (§9)
XRD§10Line positions are calculated; heights are schematic
Conditions for pressing, sintering, and magnetization3 g + 3 drops of PVA; φ 10 mm, 98.1 MPa, 1 min; 1100 °C, 2 h; 10 A, 752 000 A/m, 1 min; true density 5.374 g/cm3; reference magnet 1 × 10−6 Wb·m; ram diameter 2.86 cmLab manual §8.3–8.5
Dimensions of the balance href, L′14.020 mm, 37.349 mmProf. Mochizuki’s instructions (2026-09-25; values written on the balance)
Analysis as in the lab manualEqs. (8-13)–(8-18), N = 1 − 0.73 h/D, least squares through 3 pointsLab manual §8.5 (§9.3)
Densities of the green body and sintered body, polishing, demagnetization curve55 %, 80 %, 0.10 mm, §9.2Provisional (schematic)

VEL (the virtual laboratory) is a separate teaching tool that carries out the same experiment on a three-dimensional screen, and is in preparation for release. This simulator takes over only VEL’s models of aqueous chemistry and heat treatment (functions and constants), and moves them to a lightweight screen for class use. The wording of VEL’s assignments and feedback comments and its internal materials are not used. Unlike VEL, this simulator directly shows the phase fractions and the Fe/Ba of the solid.

The numbers on this page are results calculated with the same functions as the simulator. The rate constants of calcination, the coefficients of washing and filtration, and the magnetic coefficients are estimates chosen for teaching, and are not predictions for any specific experiment. The standing time, pouring rate, density of the sintered body, and so on are provisional values (§12). The atmosphere of the heat treatment (air) and the dimensions of the balance are from Prof. Mochizuki’s instructions (2026-09-25).

Companion simulator: Barium Ferrite — From a Coprecipitated Precursor to a Magnetic Material

References

  1. X. Liu, J. Wang, J. Ding, M. S. Chen, Z. X. Shen, “The effects of mechanical activation in synthesizing ultrafine barium ferrite powders from co-precipitated precursors”, J. Mater. Chem. 10, 1745 (2000). doi:10.1039/a909568c — Checked against the abstract only.
  2. W. Roos, “Formation of chemically coprecipitated barium ferrite”, J. Am. Ceram. Soc. 63, 601 (1980). — Checked against the summary only.
  3. D. Zahn, M. Diegel, A. Valitova, J. Dellith, S. Dutz, “Influence of synthesis parameters on the properties of barium hexaferrite nanoparticles prepared by co-precipitation”, Nanomaterials 14, 992 (2024). doi:10.3390/nano14120992
  4. S. R. Janasi, M. Emura, F. J. G. Landgraf, D. Rodrigues, “The effects of synthesis variables on the magnetic properties of coprecipitated barium ferrite powders”, J. Magn. Magn. Mater. 238, 168 (2002). — Checked against the abstract only.
  5. K. Haneda, C. Miyakawa, H. Kojima, “Preparation of high-coercivity BaFe12O19”, J. Am. Ceram. Soc. 57, 354 (1974). — Checked through citation in a secondary source.
  6. J. Y. Kwak et al., J. Korean Chem. Soc. 56 (2012). S. L. Hu, J. Liu, H. Y. Yu, Z. W. Liu, J. Magn. Magn. Mater. 473, 79 (2019). P. Shepherd, K. K. Mallick, R. J. Green, J. Magn. Magn. Mater. 311, 683 (2007). — All checked against the abstracts only. Examples in which the temperatures of the phase changes differ greatly depending on the precursor.
  7. S. H. Mahmood, I. Bsoul, “Tuning the magnetic properties of M-type hexaferrites”, arXiv:1707.07243. — On the Fe:Ba starting ratio and the coexisting phases, and the phase changes of sol–gel precursors (citing Zhong et al. 1997).
  8. A. M. Ginstling, B. I. Brounshtein, “Concerning the diffusion kinetics of reactions in spherical particles”, J. Appl. Chem. USSR 23, 1327 (1950).
  9. R. C. Pullar, “Hexagonal ferrites: A review of the synthesis, properties and applications of hexaferrite ceramics”, Prog. Mater. Sci. 57, 1191 (2012). doi:10.1016/j.pmatsci.2012.04.001 — A review of representative values of lattice constants and magnetic properties (the full text was not checked; the values were cross-checked against literature that cites the review and the Wikipedia article “Barium ferrite”).
  10. R. M. Cornell, U. Schwertmann, The Iron Oxides, 2nd ed., Wiley-VCH (2003). — Lattice constants of iron oxides and goethite, and the diffraction humps of two-line ferrihydrite.
  11. P. Scherrer, “Bestimmung der Größe und der inneren Struktur von Kolloidteilchen mittels Röntgenstrahlen”, Nachr. Ges. Wiss. Göttingen, 98 (1918).
  12. T. Prohaska et al., “Standard atomic weights of the elements 2021 (IUPAC Technical Report)”, Pure Appl. Chem. 94, 573 (2022). doi:10.1515/pac-2019-0603
  13. J. R. Rumble (ed.), CRC Handbook of Chemistry and Physics, table “Solubility product constants.”
  14. VEL (the virtual laboratory, Mochizuki Research Group, in preparation for release), src/sim/coprecip.js, thermal_ferrite.js, reaction.js, and magnetics.js (as of 2026-09-24). — The source from which the models of aqueous chemistry and heat treatment and the estimates were ported.
  15. Lab manual for Materials Science Experiments (Inorganic Materials) of the former Tokyo Institute of Technology, 2025 edition (§7, §8.1–8.5) (in Japanese). Not distributed in classes at Tokyo University of Science. Practical-session videos (A1–A3, B1–B4). — Used to confirm the procedures and conditions (reagent masses, water volumes, washing, filtration and drying, calcination, pressing, sintering, and magnetization, and the method of measuring the apparent remanent magnetization and the equations for analyzing the results). The text has not been reproduced.
  16. Prof. Mochizuki’s instructions (2026-09-25). — The atmosphere of the heat treatment (air in the student laboratory), the values of the thickness href of the reference magnet of the balance and the distance L′, and the release of the calculated values.