Learning is all about asking questions.
Deep learning is about asking thought-provoking questions that have an unknown or indefinite answer.
An essential question is one that guides a learner to combine existing information with personal knowledge and experience to come to a conclusion. An essential question seeks an answer that requires the collection and analysis of known information, and the synthesis of thoughts based on that analysis to reach a conclusion.
Therefore, an essential question usually involves researching the topic, speaking with experts, and doing experimentation or other original research.
It SHOULD involve drawing conclusions, and being able to explain why the conclusion one has reached makes sense based on the best information available.
Often, learning starts with questions that are "closed ended" or have definite answers; e.g., "Who were the first U.S. Senators elected to Congress from the state of New Jersey?"
Essential questions require background knowledge, but then go further, encouraging exploration of the topic and mental growth by the learner. They are "open-ended"; e.g., "Should there be term limits for Senators and Representatives, as there are for the President?" Answering this question requires a knowledge of what term limits are, the functions of senators and representative, how those functions compare to the presidency. But it also requires personal perspective, backed by reasoning based on factual information.
Essential questions might address a gap in our knowledge. If no one has successfully constructed a thirty foot tall layer cake, one could pose the question, "Is it possible to overcome the structural limitations of cake to construct a thirty foot tall layer cake?" Again, a lot of background knowledge is needed to address this question; to answer it, one might need to conduct experiments. And the final conclusion might be qualified - perhaps it's possible to construct such a cake, but in order to do so, the definition of what a "cake" is might need to be stretched beyond edibility.