Dear Michael,
I would like to refer to your Luminous Landscape article on digital sensor cleaning.
First, one thing has to be considered when using compressed CO2 or other compressed gas for sensor cleaning: during spraying the compressed gas cools down due to expansion. Anyone can make this experience with cleaning sprays. One can even produce dry ice (temperature_ roughly -110 deg, i.e. -78 deg. C) simply by expanding CO2 from bigger, high-pressure bottles. Chip cleaning by spraying too long may cool down the chip cover material severely and, by generating stress therein, may cause damage. This is the hour of the good old hand squeezed blower bulb.
Second, besides a possible contribution of electrical charge, the higher sensitiy of your D1 to dust particles may be related to two parameters: To the distance of the dust-covered chip-protecting surface to the light-sensitive surface, as well as to the angle of the light cones contributing to individual pixel illumination on the sensor. These two parameters contribute to the "sharpness" of the dust signal, very like the type of illumination and the film holders' glass did in the old days of darkroom enlargers: diffuse, indirect illumination vs. direct illumination (condenser or even spot illumination), the latter producing sharper images, but generating sharp, nasty dust signals on conventional prints.
As a consequence, dust will be more pronounced with a lens producing narrower light cones by design, and if one closes the field stop. This enables the dust particles to absorb or scatter a higher fraction of the light on its way to the target pixels. Likewise, thinner protection covers on the chip's sensitive surface will enhance (sharpen) signals caused by dust. Do you have informations on the chips regarding cover thickness?
One could imagine to take a real unsharp image under identical imaging conditions (at least at the same f stop) if the dust cannot be removed. A neutral diffusor in front of the lens may help to produce such a "background" image. This image, additionally corrected to represent only the dust signals, could be digitally subtracted later on from the orginal image. Such "mottle" removal is common practice in video enhanced contrast microscopy, although probably easier to achieve there (monochrome cameras, constant background etc.). it may be worth a test with digital cameras, if not already done.
Best regards,
Guenter