Glad you got it sorted!
As for the Sample rate, the higher the better in terms of sound quality but at the expense of more CPU.
The way that samples work is that if you imagine sound in the real world it is a continuous stream of values, and so sample rate means that you capture a selected amount of those values at regular intervals, so the more you use the more accurate to real sound it will be. Now the reason that sample rate exists is because it would require an infinitive number of samples per second to accurately represent reality (although we as humans probably have our own sample rate limit on what we hear anyway, just like our vision works around 60 FPS I think).
Because in an instrument such as Spark or any other VST instruments, real time processing means that for every single sample every filter, envelope, every real-time parameter etc... must be calculated. (They are actually calculated in blocks dictated by your ASIO buffer size and you can save some CPU sometimes by calculating once per block but that is probably complicating the issue for this discussion
So bearing in mind all of the above realtime calculations needed to make Spark process all of it's 16 tracks in real time and produce it's final sound output, using 48k will need 48,000 calculations per second, and 96k will require 96,000 per second. So 96k will use significantly more CPU power to process but will sound more natural. Bit Rate is more important in recorded WAVs in my opinion (although others may disagree) because this increases the accuracy of each sample by using a higher resolution data type to store the value with very little CPU overhead.
Regardless of bit rate and sample rate, the most important thing for Computer Recording is the end result, you are generally going to end up needing a 16-bit 44k WAV because this is still the standard due to CD's using this format. If your mastered file ends up being 24-bit 96k how you convert this to 16bit 44.1k is very
important. You must use some form of high quality dithering to achieve it, otherwise you you will lose lots of high frequency data, without dithering the information just gets shaved off resulting in a dull sound. Dithering will recalculate every sample to make it sound as close as possible to the original. If you don't have any software to do this, an excellent free tool is R8brain from Voxengo which is superb for this job.
It's good practice to work at a high bitrate within a DAW (I generally work at 32-bit in Cubase as it internally works at 32-bit anyway), mix out at 32-bit and do any further processing/mastering in Sound Forge etc.. at this bit rate, keep this file and then produce the formats you require from that one using dithering, ie; make your CD version. In terms of sample rate, I usually work at 48k which I find adequate, but if I am recording Acoustic Instruments I may go up to 88 or 96 where the extra quality can be noticeable. In a nutshell, do all you creation and processing at the highest your CPU can afford, and then make your final distribution versions from that master using dithering.
I probably haven't really answered your question... If your CPU is dealing with the sample rate at 96k you might as well
Hopefully the above will give you some understanding of what this stuff means.