Sieves are excellent tools for separating particulate matter based on size. It is how flour is separated from chaff in milling (wire mesh) and how Mary Berry sifts flour when baking a cake, it is how water is separated from cooked pasta (colander) and it is part of the process of preparing sand for building. It therefore seems likely that at least once a day something is filtered on our behalf, and that we therefore probably all filter something or other at least once each week for ourselves.
Cellular membranes and the membranes of the compartments (organelles) within cells are excellent filters. They need to be; waste must be lost and nutrients from the surrounding medium must be taken up. Additionally, the correct water balance (osmotic pressure) must be maintained, something that requires the cell to be able to move water in either direction. This is a lot of work for a something that is only two molecules thick.
Lipid membranes are therefore sophisticated filters that put the preparation of cooking ingredients distinctly in the shade. However, with this range of jobs comes a degree of vulnerability. The permeability of membranes is combatted with careful control mechanisms: it is not useful to the cell to take on more water than is necessary, or fail to take up nutrients such as metal ions, that it needs. In the case of metal ions, the control mechanism is particularly well developed, but also elegantly simple.
Metal ions typically have one or two positive charges. Group I elements like sodium and potassium have a single charge, Group II elements like calcium and magnesium have two positive charges. Such positive charges are electrically attracted to negatively charged species – something at the basis of the observation that opposites attract.
But what would happen if we had a permeable membrane between a medium containing negatively charged ions and one containing positively charged ions? There would be a strong, electrical desire for ions to move across that membrane. This attraction between ions of dissimilar charges across a membrane is known as an electrical gradient. There is a high concentration of positive charges on one side that is attracted to a high concentration of negative charges on the other side. But what if this movement of ions is not useful to the cell? What if it needs to retain a high concentration of positive ions within the cell and prevent the entry of negatively charged ions from the medium? Normally, this function is performed by membrane proteins called ion channels that have evolved to transport ions through a pore. This process requires energy, but it provides very fine control of the movement of ions. These channels are useful as they mean that the cell has control over the movement of these ions.
Some times however, a bulk movement of ions is required – for example in the movement of a neuronal impulse. In this case, the fine control of some types of ion channel is not quite as necessary. For this purpose, a mechanism that does not even require a protein has been exploited by cells. This mechanism occurs through voltage gated lipid channels. As the name suggests, the movement of ions through such channels is controlled by voltage. If the potential difference (measured in volts, V) is sufficiently high, the gate (pore) will open and the ions are lost to the far side of the membrane. Recent work by Blicher and Heimburg  has shown that this is comparable to that of protein channels.
In light of the efficacy of these lipid channels, you may wonder why protein ion channels ever evolved. It has been tentatively suggested (Blicher and Heimburgh ), that differing pH and temperature influence the size and behaviour of lipid pores, rendering them somewhat inconsistent. Protein channels are more dependable and provide a greater consistenty, allowing the cell to retain control of ion movement irrespective of changing conditions. However, this has yet to be tested formally.
 A. Blicher, T. Heimburg, Voltage-Gated Lipid Ion Channels. PLoS ONE, 2013, 8, e65707. doi:10.1371/journal.pone.0065707