Solid motors burn a fuel-and-oxidiser mix cast inside the casing; liquid engines pump separate propellants into a chamber; cryogenic engines burn super-cold liquid hydrogen and liquid oxygen.
Every rocket stage burns a fuel with an oxidiser. What changes is the form the propellant takes. Indian rockets use all three kinds, often on the same vehicle.
In a solid motor, the fuel and oxidiser are mixed into a rubbery solid and packed into the casing, with a hole down the middle. When it is lit, the surface of that hole burns. Designers shape the hole to control how the thrust changes during the burn.
ISRO's solid motors use a propellant called HTPB. Examples: PSLV's S139 first stage (139 tonnes of propellant), LVM3's S200 boosters (about 205 tonnes each) and all three main stages of SSLV. In Indian rockets they do the heavy lifting at lift-off: PSLV's S139 gives up to 4,800 kN.
A liquid engine keeps its fuel and oxidiser in separate tanks and pumps them into a combustion chamber, where they mix and burn. NASA describes getting this mixing and burning right, without blowing out the flame, as very complex.
India's main liquid engine is the Vikas. It powers PSLV's second stage, GSLV's second stage and its four L40 strap-ons, and LVM3's L110 core, which has two of them. GSLV's second stage burns UH25 with nitrogen tetroxide (N2O4). PSLV's fourth stage and SSLV's trimming module burn MMH with an oxidiser called MON.
A cryogenic-engine burns liquid hydrogen with liquid oxygen. Both must be kept extremely cold to stay liquid. This makes the engines efficient but hard to build. GSLV's upper stage uses the CE-7.5 engine; LVM3's C25 stage uses the CE-20, with 28 tonnes of propellant and 200 kN of thrust.
The difficulty is real. On GSLV-F10 in August 2021, pressure in the liquid hydrogen tank did not build up properly. The fuel booster turbopump then delivered too little hydrogen, and the mission was lost. ISRO's review traced it to a likely leak in a vent and relief valve.
Follow GSLV's record before and after India's own cryogenic upper stage started flying successfully. See GSLV's flights
Why put the cryogenic stage at the top? The rocket equation answers it: Δv = Veq × ln(m_full / m_empty).
Take LVM3's C25 stage with a 4,000 kg GTO payload (fairing and adapters ignored):
From ISRO's figures (200 kN, 28 t in about 646 s), the average Veq is about 4,620 m/s, so Δv ≈ 4,620 × 1.475 ≈ 6,800 m/s.
Now imagine the same stage with a less efficient engine, Isp = 300 s (Veq ≈ 2,940 m/s): Δv ≈ 2,940 × 1.475 ≈ 4,340 m/s. That is about 2,500 m/s less from the same mass.
Efficiency matters most at the top because the upper stage must carry the payload the whole way. At lift-off, raw thrust matters more: each S200 burns about 205 t in about 134 s, an average of roughly 1,530 kg/s.
The GSLV-F10 failure shows the trade-off. The cryogenic stage deviated at 297.3 s after lift-off, and the onboard computer aborted the mission at 307 s. ISRO's committee recommended an active pressurisation system for the hydrogen tank, a stronger vent and relief valve, and automatic checks of more cryogenic parameters before lift-off.
The third stage, the Cryogenic Upper Stage, with the CE-7.5 engine. It burns liquid hydrogen with liquid oxygen.
The upper stage carries the payload all the way to orbit, so extra speed per kilogram of propellant there adds directly to payload. At lift-off, high thrust matters more than efficiency.
Liquid hydrogen tank pressure did not build up normally, so the fuel booster turbopump supplied too little hydrogen to the engine. The most likely cause was a leak in the vent and relief valve.