Phsopher and others are absolutely correct. The atoms of space - elements of the "spin network" - that loop quantum gravity assumes to be the elementary building blocks of space don't look like ordinary atoms we know, but they share their basic properties such as having a characteristic size and distance from each other (assumed to be the Planck length). That's the typical length of the edges of the "spin network", for example.
This simple property that the spin network shares with the "ordinary atoms" is enough to see that conventional cosmology is impossible in the framework. The spin network is a material, analogous to solids, and it would have to create new atoms (or links of the "spin network") for the Universe to expand.
There are other ways to see that all models assuming that the empty space is made out of anything similar to "atoms" contradict basic, well-established, and somewhat "obvious" features of physics. The most obvious one is that the atoms pick a preferred reference frame, much like the luminiferous aether did in the 19th century. That's excluded by special relativity and all evidence supporting special relativity. It's enough for the atoms to be able to carry a nonzero entropy in empty space and relativity is doomed because the entropy density is the time-component of a 4-vector. If it is nonzero, Lorentz invariance must be (heavily) broken, by an amount proportional to the (gigantic) Planckian entropy density. The "material" of the spin network would also slow down moving objects hugely: anything made out of anything resembling normal atoms is closer to a crystal than a vacuum.
Every hint that theories with structure filling empty space able to carry nonzero entropy (information about the detailed arrangement of the atoms) can be viable 103 years after special relativity are balderdash. LQG and dozens of other naive models of physics have been dead for more than a century but some very slow people haven't yet noticed.