The number of stars in the observable universe has been discussed in the question How do we estimate $10^{23}$ stars in the observable universe? My answer to this does little more than repeat the conventional wisdom, but Pulsar's answer presents a very interesting alternative approach that ends up at about the same number. The number of stars in the observable universe turns out to be around $10^{23}$, but this is a pretty loose approximation and it could easily be an order of magnitude higher or lower.
The number of atoms in silicon dioxide can be rather more precisely determined since the molecular weight of silicon dioxide is $60$, so $60$ grammes contains $3 \times 6.023 \times 10^{23}$ atoms, which is about $1.8 \times 10^{24}$ atoms.
So $10^{23}$ atoms is about 3.3 grammes of sand. But bear in mind that order of magnitude uncertainty in the number of stars.
