And especially three points :
in which chromosomes is it located (especially for the human case) ?
how do we know about it ?
does the proportion and composition vary a lot from one eukaryot to another ?
We know about nuclear DNA having a mitochondrial origin mainly in two ways: (1) a sequence in the nucleus is found to closely match a sequence found in mitochondria, or (2) mitochondrial proteins are found to be encoded by the nuclear genome but not by the mitochondrial genome, and those proteins seem likely to have been necessary for sustaining life of the mitochondrial free-living ancestor.
The integration of copies of mitochondrial DNA into the nuclear genome is an ongoing process. Pieces of DNA can float around and occasionally be spliced into a chromosome. Most such integrated pieces are entirely nonfunctional (the functional version remaining in the mitochondria) and can pretty much be considered junk despite being replicated when the chromosome is replicated. Useless gene-like pieces are often called nuclear mitochondrial pseudogenes or NUMTs (see, e.g., pseudogene.net). NUMTs are of technical importance because they interfere with genome studies: the sequences of the NUMTs can easily be mistaken for true mitochondrial DNA leading to false conclusions about mitochondrial genomes.
A more general term is endosymbiotic gene transfer, which can be used to describe nuclear DNA which was derived from DNA originally residing in mitochondria but now is no longer found there. Some of this is nonfunctional, but much is functional, coding for proteins which after translation are sent to the mitochondria. In fact, the overwhelming majority of mitochondrial proteins are coded by the nucleus, many of which probably had a mitochondrial origin.
Eukaryotes do vary in how much nuclear DNA is of mitochondrial origin. A majority of functional genes are probably ancient and so tend not to vary a lot between organisms. On the other hand, nonfunctional NUMTs are often relatively recently established and so often vary widely between organisms.
Particular nuclear chromosomes are not associated with mitochondrial origins; the integration of free floating DNA pieces is mostly a fairly random process.
For more detail, you might want to see Timmis et al's review, "Endosymbiotic Gene Transfer".
Interesting question. I researched this a bit now and the phenomenon is termed "numt" for "nuclear mitochondrial DNA". This term descrives the transfer of cytoplasmic mitochondrial DNA sequences into the separate nuclear genome of a eukaryotic organism. It seems that most of these sequences are inactive. This list at pseudogene.net has a large number of sequences in a lot of different organisms, for humans insertion between 0,2 and 14,2 kb are listed. Most of these sequences are detected in the deep sequencing projects underway these days and by comparative sequence analysis. An example is this paper, where they found mtDNA on chromosome 1 of homo sapiens, which was still part of the mitochondria in homo neanderthalensis. For more information you can go to the list of papers I list below, this should be a good starting point:
The papers should all be accessible, if not, let me know.