Design considerations
Overview of ideal design: While our time-based personal samplers might be the only ones of their type, the design principle itself is not novel. It comprises a slot inlet and a moving collection surface, and if clockwork were exchanged for battery motor, it could have been built anytime in the last 100 years. The Burkard / Hirst sampler - a large impaction device designed for sampling outdoor aerosols over 7 day periods - has been around since 1952, functions on the same principles.
A more advanced and complex design would allow collection with very small resolutions of time, or with accurate cut points between the events. For better resolution, the sampling surface could be in the form of a spiral track, as on a record or CD, or around a drum like a thread on a screw. If solely mechanical, either the orifice or surface could move to accommodate this.
If the purpose permits a series of single samples, but each of different and controllable intervals, the sampling surface could advance intermittently, as directed by the person sampling associated with pre-designated activities, and all this monitored by a time/location marker. This would allow a much smaller sampling surface, sufficient perhaps to collect 20 or 30 samples with a couple of mm between them.
I would be interested in other options.
The mechanism of collection is based on impaction, which limits the particle size of collection, typically >2-4 micron diameter. If instead, a time-based sampler was restricted to serial sampling of separate events within one device (second option above), then samples could be drawn through a filter, such as electret, which would collect particles of all sizes, probably with >90% efficiency (depending on several factors). The surface area of electret required for this purpose would probably be of the order of 1 sq cm, (providing it was thick enough, something like H&V Technostat 150+, or a nanofibre matrix) but this depends on the electret and sampling flow rate which affects the pressure drop, the type of pump, the battery life, noise and other factors.
Another feature which would be useful from the perspective of clinical disease, would be to size-separate the collected sample, as small and large particles may have different clinical effects. In particular into factions larger than 10 microns (largely retained in the upper airways), and smaller than ~3 microns, (inhalable into small airways) are of particular interest. While time-based impaction samplers that separate particles on the basis of size do exist, they are the size of a small filing cabinet (link); designing to do this with a personal sampler is much more challenging.
Who is the user/what are their needs?
I thought I learnt years ago that designing something for which there is a limited commercial market for, is an exercise fraught with risks, ('you must understand the 'pain in the market'?' was how it was expressed) ... I sometimes wonder how well I have really learnt this.
The most likely user scenario involves ordinary people (children, adults) going about their normal 24/7 lives. Thus they require something that is simple to manage, not too large or conspicuous and is non-threatening to them or others who see it in use. These factors affect choice about where it is worn on the body, the size of the unit, how the air pump is connected, any noise it makes, what the device is analogous to if seen.
If it shares an aesthetic like an iPhone, or an iWatch, then it is probably dismissed at first glance, whereas if it looks like a criminal monitoring device or a security issue, then that is a problem.
An alternative design approach to having a discretely worn and visible device would be to make the devices (sampler plus pump) able to be worn completely out of sight. For example, both were kept in a small backpack or shoulder bag or worn under the clothes and attached to a silicone tube (low particle losses) which had its inlet on the lapel/shoulder, in the breathing zone.
What resolution of time is required/how large is the device: There probably is no single answer. The longer the pathlength of collection, (other things being equal) the greater the resolution of events. The two models, 1 and 2 probably lie at the boundaries in terms of size and what is required depends on the need and intended purpose.
The width of the model 2 slot outlet is ~0.7 mm, with more deposition in the centre and tapering off over perhaps 0.5 mm either side, (electret probably has a wider deposition site than on adhesive film). So, depending on the level of certainty, events corresponding to about 1-1.5 mm of pathlength can be resolved. The model 2 sampling wheel has a circumference of 180 mm, which is 0.25 mm movement per minute using a 12 hour clock motor. On this basis, the accuracy of resolving events is around 6 minutes. This also presents an opportunity for confounding when cutting up the tape; the accuracy of events is in seconds.
In model 1, the functional diameter of the disk is around 22 mm, (inside the 25 mm overall diameter) and the slot 0.5 mm (as we used after modifying). The circumference of the disk moves at ~0.1 mm per minute, suggesting exposure events of 10 minutes apart can be resolved. Of course if transitory high exposure occurs close to where a cut-point is made, this could present a misclassification problem.
Thus the optimum size presents the design dilemma as to the optimum diameter. Is it better to have a small diameter sampling disk so the device is small and discrete, or a larger disk which provides more time/sample resolution.
While total times of 1 hour (mount wheel on minute hand) or 12 hours (on hour hand) or a week (on 7 day clock) work ok with a single rotation of a clock, the device is not optimum if for example a total sampling time of 4 hours is contemplated. In this case something like a controlled stepper motor, as on a CD player, would be optimal.
Face (disk) vs Edge (strip) collection: Another design issue is: should the sample be collected on the face of the disk, (as in model 1) or around the side / edge, (as in model 2). Again, pros and cons. Originally, with model 1 it was intended to use a press-blot immunostaining system and use of 25 mm disks of MCE as the protein binding membrane. These were going to be processed as a single unit. In model 2 it was thought that processing a single large disk was not so feasible and the analysis of strips was easier.
The use of a disk allowed processing of the 12 (or 24) hours as a single sample which would later be analysed in terms of the site/time of deposition. Our attempt is shown in the supplement in the 2016 PLoS paper. The paper can be download here, however some websites do not provide access to the supplement (the publisher's site does) and the supplement is provided here.
As discussed in the supplement, we ended up cutting up the sample (both electret and adhesive) into segments associated with individual time/activities and quantifying allergen on each by amplified ELISA. This was because we could not accurately quantify the chemiluminescence, and the ELISA was more sensitive. Ultimately the chemiluminescence (or some single-sample processing system) would be our preference, but that awaits further development.
Slot width / pressure drop. Essentially the narrower the slot, the narrower the width of the band of deposition, and the higher the back-pressure; that is the demand on the pump. Some multislot, multistage air samplers do have very narrow slots, such as the Marple Personal Sampler, and can collect down to ~0.5 microns, (see here), although most impaction samplers collect down to 2-5 microns, see review by Estelle Levetin here.
The formula for back pressure and slot width can be found here (download it from paragraph 21). I am grateful to Bill Lindsley for this information. Applying this formula to our sampler, a width slot of 0.7 mm (x 5 mm) has a pressure drop of around 0.6 inches of water and cut point of 2.6 microns, whereas a wider slot of 1 mm has a pressure drop of 0.25 inches of water and a cut point of 3.7 microns. These numbers are relevant to the later discussion about a combined sampler-pump.
A few words about the nature of allergen exposure: Some of the biological exposures that people measure, involve (or presume) relatively uniformly sized particles which are relatively uniformly and consistently distributed in the air. This is definitely not the case for most allergen sources. Our experience has mainly been with mite allergens, but this also applies to other allergen sources we have worked with including cat, cockroach, wheat and also applies to pollens (which occur both as large whole pollen grains and small fragmented particles (Google 'thunderstorm asthma') and fungi, (which both range in size 3-70 microns, depending on species, but also occur as the condia/spores and as smaller hyphal fragments).
With mite, the main sources of allergen are the feces, but mites also deposit allergen on the material they are feeding on, the allergen leaches from feces to other dust particles at high humidity, (we have shown but not published both of these) and the mite bodies, or cast exoskeletons also fragment. The feces also aggregate with other dust components. The 'size' of unfragmented airborne mite allergen during personal exposure has been demonstrated using the Halogen assay by us, see here and here. It's predominantly associated with large particles, at least during the day, although we have a recent letter in JACI-IP suggesting this may be different at night, see here. Earlier than that, in a different lab and using different methods (cascade impactor), I identified feces as the main source of disturbed allergen, and showed that the size distribution of airborne particles declines rapidly following their initial disturbance, see here. The methods are imperfect and may not fully reflect 'reality': its possible that the cascade impactor fragments aggregates, and so biases towards smaller particles, and its also possible that Halogen assays are less good at detecting smaller particles that carry very small amounts of allergen, and so bias towards observation of larger particles.
'Personal exposure', that is what I think of as what you breath in, can also be very local; this depends on the sources and mixing. If you are vacuuming the floor, it can be presumed that the airborne particles are fairly well mixed into the air due to its movement and turbulence, if however you are moving around, at home or in bed, you are probably within a small bubble of particles, created from disturbed reservoirs, which itself moves with the thermal air current that occur around the body. When standing there is a plume of such air from the feet, up the body, moving at around 0.5-1 M / second. This allergen-cloud is generated by movement of items like clothing or bedclothing and is probably quite local, or at least dilutes with increasing distance from the body. Precisely sampling of what is inhaled therefore means collecting what is within what is referred to as the 'breathing zone' a not-well-defined concept, generally thought of as around 20-30 cm from the nose. This might mean that sampling 20 cm and 50 cams from the nose, could make a difference in perceived exposure, (~2? fold), particularly if the site was not within the local aerosol plume.
The shape and downward facing direction of the nose also means that it has its own sampling characteristics relative to the particles in its proximity. The same issue apply to the orifice used to sample particles.
A further factor affecting any measurement of exposure, is what has been referred to by Gert Doekes as, 'the quantum nature' of some allergens. Natural exposure is sometimes composed of a relatively few, dispersed, large particles individually carrying a relatively large amount of allergen plus many more smaller particles generally carrying much less. If you are only sampling small volumes, relative to the distribution of the large particles, then each sample may collect small and relatively variable numbers of them and thus any single sample contains a lot of uncertainty about the average number in a more general exposure. This is mainly a problem with low sampling flow rates and/or low exposure.
My memory of our observations is that when you get to an exposure of above ~30 allergen-carrying particles in a halogen assay, the results are consistent with measurement of allergen by ELISA assay; measuring very low levels of exposure involves large errors as to whether you encountered one, two, three or none of the large particles within the volume of air sampled. Of course, in life, you do what you can.
A more advanced and complex design would allow collection with very small resolutions of time, or with accurate cut points between the events. For better resolution, the sampling surface could be in the form of a spiral track, as on a record or CD, or around a drum like a thread on a screw. If solely mechanical, either the orifice or surface could move to accommodate this.
If the purpose permits a series of single samples, but each of different and controllable intervals, the sampling surface could advance intermittently, as directed by the person sampling associated with pre-designated activities, and all this monitored by a time/location marker. This would allow a much smaller sampling surface, sufficient perhaps to collect 20 or 30 samples with a couple of mm between them.
I would be interested in other options.
The mechanism of collection is based on impaction, which limits the particle size of collection, typically >2-4 micron diameter. If instead, a time-based sampler was restricted to serial sampling of separate events within one device (second option above), then samples could be drawn through a filter, such as electret, which would collect particles of all sizes, probably with >90% efficiency (depending on several factors). The surface area of electret required for this purpose would probably be of the order of 1 sq cm, (providing it was thick enough, something like H&V Technostat 150+, or a nanofibre matrix) but this depends on the electret and sampling flow rate which affects the pressure drop, the type of pump, the battery life, noise and other factors.
Another feature which would be useful from the perspective of clinical disease, would be to size-separate the collected sample, as small and large particles may have different clinical effects. In particular into factions larger than 10 microns (largely retained in the upper airways), and smaller than ~3 microns, (inhalable into small airways) are of particular interest. While time-based impaction samplers that separate particles on the basis of size do exist, they are the size of a small filing cabinet (link); designing to do this with a personal sampler is much more challenging.
Who is the user/what are their needs?
I thought I learnt years ago that designing something for which there is a limited commercial market for, is an exercise fraught with risks, ('you must understand the 'pain in the market'?' was how it was expressed) ... I sometimes wonder how well I have really learnt this.
The most likely user scenario involves ordinary people (children, adults) going about their normal 24/7 lives. Thus they require something that is simple to manage, not too large or conspicuous and is non-threatening to them or others who see it in use. These factors affect choice about where it is worn on the body, the size of the unit, how the air pump is connected, any noise it makes, what the device is analogous to if seen.
If it shares an aesthetic like an iPhone, or an iWatch, then it is probably dismissed at first glance, whereas if it looks like a criminal monitoring device or a security issue, then that is a problem.
An alternative design approach to having a discretely worn and visible device would be to make the devices (sampler plus pump) able to be worn completely out of sight. For example, both were kept in a small backpack or shoulder bag or worn under the clothes and attached to a silicone tube (low particle losses) which had its inlet on the lapel/shoulder, in the breathing zone.
What resolution of time is required/how large is the device: There probably is no single answer. The longer the pathlength of collection, (other things being equal) the greater the resolution of events. The two models, 1 and 2 probably lie at the boundaries in terms of size and what is required depends on the need and intended purpose.
The width of the model 2 slot outlet is ~0.7 mm, with more deposition in the centre and tapering off over perhaps 0.5 mm either side, (electret probably has a wider deposition site than on adhesive film). So, depending on the level of certainty, events corresponding to about 1-1.5 mm of pathlength can be resolved. The model 2 sampling wheel has a circumference of 180 mm, which is 0.25 mm movement per minute using a 12 hour clock motor. On this basis, the accuracy of resolving events is around 6 minutes. This also presents an opportunity for confounding when cutting up the tape; the accuracy of events is in seconds.
In model 1, the functional diameter of the disk is around 22 mm, (inside the 25 mm overall diameter) and the slot 0.5 mm (as we used after modifying). The circumference of the disk moves at ~0.1 mm per minute, suggesting exposure events of 10 minutes apart can be resolved. Of course if transitory high exposure occurs close to where a cut-point is made, this could present a misclassification problem.
Thus the optimum size presents the design dilemma as to the optimum diameter. Is it better to have a small diameter sampling disk so the device is small and discrete, or a larger disk which provides more time/sample resolution.
While total times of 1 hour (mount wheel on minute hand) or 12 hours (on hour hand) or a week (on 7 day clock) work ok with a single rotation of a clock, the device is not optimum if for example a total sampling time of 4 hours is contemplated. In this case something like a controlled stepper motor, as on a CD player, would be optimal.
Face (disk) vs Edge (strip) collection: Another design issue is: should the sample be collected on the face of the disk, (as in model 1) or around the side / edge, (as in model 2). Again, pros and cons. Originally, with model 1 it was intended to use a press-blot immunostaining system and use of 25 mm disks of MCE as the protein binding membrane. These were going to be processed as a single unit. In model 2 it was thought that processing a single large disk was not so feasible and the analysis of strips was easier.
The use of a disk allowed processing of the 12 (or 24) hours as a single sample which would later be analysed in terms of the site/time of deposition. Our attempt is shown in the supplement in the 2016 PLoS paper. The paper can be download here, however some websites do not provide access to the supplement (the publisher's site does) and the supplement is provided here.
As discussed in the supplement, we ended up cutting up the sample (both electret and adhesive) into segments associated with individual time/activities and quantifying allergen on each by amplified ELISA. This was because we could not accurately quantify the chemiluminescence, and the ELISA was more sensitive. Ultimately the chemiluminescence (or some single-sample processing system) would be our preference, but that awaits further development.
Slot width / pressure drop. Essentially the narrower the slot, the narrower the width of the band of deposition, and the higher the back-pressure; that is the demand on the pump. Some multislot, multistage air samplers do have very narrow slots, such as the Marple Personal Sampler, and can collect down to ~0.5 microns, (see here), although most impaction samplers collect down to 2-5 microns, see review by Estelle Levetin here.
The formula for back pressure and slot width can be found here (download it from paragraph 21). I am grateful to Bill Lindsley for this information. Applying this formula to our sampler, a width slot of 0.7 mm (x 5 mm) has a pressure drop of around 0.6 inches of water and cut point of 2.6 microns, whereas a wider slot of 1 mm has a pressure drop of 0.25 inches of water and a cut point of 3.7 microns. These numbers are relevant to the later discussion about a combined sampler-pump.
A few words about the nature of allergen exposure: Some of the biological exposures that people measure, involve (or presume) relatively uniformly sized particles which are relatively uniformly and consistently distributed in the air. This is definitely not the case for most allergen sources. Our experience has mainly been with mite allergens, but this also applies to other allergen sources we have worked with including cat, cockroach, wheat and also applies to pollens (which occur both as large whole pollen grains and small fragmented particles (Google 'thunderstorm asthma') and fungi, (which both range in size 3-70 microns, depending on species, but also occur as the condia/spores and as smaller hyphal fragments).
With mite, the main sources of allergen are the feces, but mites also deposit allergen on the material they are feeding on, the allergen leaches from feces to other dust particles at high humidity, (we have shown but not published both of these) and the mite bodies, or cast exoskeletons also fragment. The feces also aggregate with other dust components. The 'size' of unfragmented airborne mite allergen during personal exposure has been demonstrated using the Halogen assay by us, see here and here. It's predominantly associated with large particles, at least during the day, although we have a recent letter in JACI-IP suggesting this may be different at night, see here. Earlier than that, in a different lab and using different methods (cascade impactor), I identified feces as the main source of disturbed allergen, and showed that the size distribution of airborne particles declines rapidly following their initial disturbance, see here. The methods are imperfect and may not fully reflect 'reality': its possible that the cascade impactor fragments aggregates, and so biases towards smaller particles, and its also possible that Halogen assays are less good at detecting smaller particles that carry very small amounts of allergen, and so bias towards observation of larger particles.
'Personal exposure', that is what I think of as what you breath in, can also be very local; this depends on the sources and mixing. If you are vacuuming the floor, it can be presumed that the airborne particles are fairly well mixed into the air due to its movement and turbulence, if however you are moving around, at home or in bed, you are probably within a small bubble of particles, created from disturbed reservoirs, which itself moves with the thermal air current that occur around the body. When standing there is a plume of such air from the feet, up the body, moving at around 0.5-1 M / second. This allergen-cloud is generated by movement of items like clothing or bedclothing and is probably quite local, or at least dilutes with increasing distance from the body. Precisely sampling of what is inhaled therefore means collecting what is within what is referred to as the 'breathing zone' a not-well-defined concept, generally thought of as around 20-30 cm from the nose. This might mean that sampling 20 cm and 50 cams from the nose, could make a difference in perceived exposure, (~2? fold), particularly if the site was not within the local aerosol plume.
The shape and downward facing direction of the nose also means that it has its own sampling characteristics relative to the particles in its proximity. The same issue apply to the orifice used to sample particles.
A further factor affecting any measurement of exposure, is what has been referred to by Gert Doekes as, 'the quantum nature' of some allergens. Natural exposure is sometimes composed of a relatively few, dispersed, large particles individually carrying a relatively large amount of allergen plus many more smaller particles generally carrying much less. If you are only sampling small volumes, relative to the distribution of the large particles, then each sample may collect small and relatively variable numbers of them and thus any single sample contains a lot of uncertainty about the average number in a more general exposure. This is mainly a problem with low sampling flow rates and/or low exposure.
My memory of our observations is that when you get to an exposure of above ~30 allergen-carrying particles in a halogen assay, the results are consistent with measurement of allergen by ELISA assay; measuring very low levels of exposure involves large errors as to whether you encountered one, two, three or none of the large particles within the volume of air sampled. Of course, in life, you do what you can.
Proudly powered by Weebly