Application and Performance Test of a Small Aerosol Sensor for the Measurement of Aerosolized DNA Strands

There is an immediate need for aerosol measuring sensors to monitor the size and concentration of DNA strand aerosols in PCR systems. We evaluate the performance of a previously developed small aerosol sensor for its potential use as a component in measuring DNA strand aerosols in PCR system chambers. A detailed derivation of the working principle is presented along with the principles used to determine the dimensions of the stages and the operational parameters. After characterizing the aerosolized DNA strands, experiments were conducted to identify their relationship with measured currents. The experimental results indicate that for aerosolized Escherichia coli DNA strands, the sensor is capable of measuring concentrations from 10 cm to 10 cm (from 10 cm to 10 cm for particles smaller than 102 nm) and sizes from 100 bp to 1000 bp. There was a slight difference between the results of the sensor and its theoretical model. The sensor exhibited good sensitivity to different concentrations and can detect every 150 bp of strands, indicating its effectiveness in monitoring ultrafine DNA strand aerosols for PCR systems and other applications.


INTRODUCTION
Nowadays, polymerase chain reaction (PCR) based methods, in particular quantitative PCR are predominantly used as a molecular biological tool to replicate desoxyribonucleic acid (DNA), and can create copies of specific fragments of DNA by cycling through three temperature steps (Postollec et al., 2011).Many systems based on quantitative PCR amplification of specific sequence have been developed (Chartier et al., 2003;Le et al., 2010).There are usually more than one reaction chambers in order to achieve high-throughput PCR amplification and to conduct many sequential PCR experiments simultaneously.During a PCR reaction, handling and processing sample fluid, and evaporation of samples may result in their invasion into the surrounding gas and subsequently forming DNA aerosol.It is possible for this aerosol to cause contamination and cross-talk of the PCR, and poses a risk of carry-over from experiments to experiments such that enlarges the possibility of failed amplification (Zhang et al., 2006).
Therefore, there is an immediate need of aerosol measuring devices to monitor the size and concentration of DNA strand aerosol in real time and to provide a warrant for decontamination and disinfection in chambers of PCR systems.
There are two general real-time ways to measure aerosol categorized according to their measurement principles.One is optical technique which detect aerosol by way of light, but most of the instruments based on this technique are not able to detect particles smaller than the wavelength of light that been used, namely, they lose sensitivity when they come to small particles (usually smaller than 100 nm (Kulkarni et al., 2011)) and consequently cannot be used in the measurement of DNA strands.Condensation particle counters (CPC) are capable of detecting particles down to about 1 nm by condensation of supersaturated vapour onto particles and the following detection by light scattering (Winklmayr et al., 1991).
Another method is based on electrical technique.Differential mobility analysis, combined with nanoelectrospray and online condensation particle counting, has been successfully applied in determining the mobility diameter of single-stranded and double-stranded DNA ranging from 6 kDa to 900 kDa (Mouradian et al., 1997;Guha et al., 2012).For this analytical technique, the nanoelectrospray converts a dilute analyte solution into a series of droplets, which are then neutralized and converted into solvent-free nanoparticles.These nanoparticles are sized according to their gas-phase electrophoretic mobility in a differential mobility analyzer (DMA).As the voltage of the DMA is ramped, nanoparticles of increasing size sequentially exit the DMA and enter a CPC where they are counted.Coupling of nanoelectrospray and macro-ion mobility spectrometry with inductively coupled plasma mass spectrometer to determine the size of large biomolecules (proteins and DNA) was also demonstrated by Carazzone et al. (2008).Although these techniques have been used with satisfactory accuracy and precision (Laschober et al., 2007), the involved instruments are large in size, expensive and complicated to use, which make them cannot be integrated into PCR systems.Some examples of portable particle monitoring apparatuses combining differential mobility analysis and condensation particle counting technique were developed, such as WPS™ (Liu et al., 2010), NanoScan (Tritscher et al., 2013) and Kanomax PAMS 3300 (Hsiao et al., 2016), whereas their sizes are still too large to be used in the limited space of PCR system chambers.There are also some aerosol sensors smaller in size purely based on diffusion charging and electrical detection, including NanoMonitor (Marra et al., 2010), DiSCMini (Fierz et al., 2011) and NanoCheck (Kulkarni et al., 2011).In these sensors, aerosol is mixed with unipolar ions and then they attach to the aerosol particles by diffusion.Next, charges on the particles are measured using Faraday electrometer after removal of excess ions.These sensors can determine a number-weighted mean diameter and the number concentration of the particles in a diameter range from tens of nanometer to hundreds of nanometer.However, they have not been evaluated for the measurement of DNA strands.
In this study, tests of measuring aerosolized Escherichia coli DNA strands were carried out with a small aerosol sensor (SAS) to investigate its feasibility of monitoring the size and concentration of DNA strand aerosol in PCR systems.In our previous work (Yang and Zhu, 2014;Zhang et al., 2016), a SAS was developed based on particle diffusion charging and inducted-current measurement, and its performance was tested for ultrafine NaCl aerosol particles.Fig. 1 schematically depicts the chip of the SAS.It consists of three stages (the charging stage, the precipitation stage and the sensing stage) made of pairs of printed circuit boards (PCBs).The two PCBs are bonded by a nonconductive film adhesive with fixed thickness h and there are copper layers connecting to ground to separate each stage and prevent the cross-talk among them.
However, dimensions of each stage and operation parameters of the SAS must be redetermined experimentally because of their different charging properties between DNA strands and NaCl particles.In this paper, a detailed derivation of the working principle was presented.Then principles of determining the dimensions of stages and the operation parameters were given for the purpose of meet the need of measuring aerosolized DNA strands.After characterization of the DNA particles, experiments were conducted to verify the applicability of the SAS in DNA

Working Principle of the SAS
For the sensor chip shown in Fig. 1, a controlled aerosol flow ϕ is driven through the three stages successively by means of a pump situated at the outlet of the sensing stage.After entry the aerosol particles are electrically charged inside the charging stage via diffusion charging (Adachi et al., 1985), which is enabled by a corona discharge formed between a tungsten needle-tip electrode mounted on one of the PCBs and a planar electrode on the opposite PCB.A sufficiently high positive voltage V c generated by a highvoltage module is imposed on the needle-tip electrode to locally ionize the carrier gas.The produced positive ions drift from the discharge tip to the opposite electrode and a positive ion cloud is formed in the gap between the two electrodes.The aerosol particles become charged when they pass through the ion cloud region via diffusion charging.
After charging the particles enter into the precipitation stage and then the sensing stage, both of which are composed of two parallel planar electrodes plated on PCBs.The constant voltage V s imposed on the sensing electrodes produces an electric field E s =V s /h that is capable of precipitating substantially all the electrically-charged particles in the aerosol flow passing between the plates.One of the sensing plates is connected via an electrometer to a reference potential.Currents measured by the electrometer represent the precipitated amount of charges carried by the particles per unit time in the sensing stage.Since we need to obtain two characteristic parameters including particle number concentration N and average size d p,av of particles, two different inducted-current values are required during one measurement cycle.Thence, a biased block-shaped voltage pulse varying between V p = V 1 and V p = V 2 is applied to the electrodes of the precipitation stage.At the low level (V p = V 1 ), nearly all the charged particles traverse the precipitation stage unhindered but all the residual ions are removed because of their mobility disparities, while there are also a portion of charged particles captured inside the precipitation stage at the high level (V p = V 2 ) due to the slightly higher electric field E p = V 2 /h.Accordingly, two different current values are acquired in the sensing stage (I s = I 1 at When the low-level voltage is imposed on the precipitation stage, the number of particles in size d p in the aerosol flowing through the sensor during time period ∆t is n d p = φN(d p )∆t, where N(d p ) is the size-distribution function of the aerosol, φ is the aerosol flow through the sensor.The quantity of charge carried by the particles in size d p is p i r p p 0 ( , , ) ( ) , where q is the amount of elementary charges attracted on the particles, e = 1.6 × 10 -19 C, f q the fraction of particles in size d p that are charged with q elementary charges.It is the function of d p , q, and the product of ion density N i and the residence time t r .For an arbitrary distribution of aerosol, taking the example for that shown in Fig. 2, the total charge quantity on all the particles is i r p p 1 0 lim ( , , ) ( ) And the inducted current in the sensing stage will be (2) Similarly, when the high-level voltage is imposed on the precipitation stage, the current is where ξ q (d p , E p ) is the precipitated proportion of particles in size d p charged with q elementary charges under the effect of electrical field E p in the precipitation stage.At normal temperature and pressure, the Reynolds number of the flow in the sensor is Re = 65000 vh ≤ 130 when h ≤ 400 µm and v ≤ 5 m s -1 , which indicates the flow is laminar.
The slip flow regime specified by the Knudsen number Kn = 0.1328 × 10 s -6 /d p lying in the range from 0.885 to 13.2 for the aerosol particles in size ranging from 10 nm to 300 nm occurs in the aerosol system.Then ξ q (d p , E p ) can be calculated according to the balance between drag force and electrostatic force for parallel plate structure of the precipitation stage as on the non-mixing assumption, where L p is the length of the electrode plates of the precipitation stage, η is the viscosity of carrier gas, v is the average air speed in the stage, C c is the slip correction factor, which is nearly inversely proportional to d p for spherical particles (Allen and Raabe, 1985), namely, The average number of elementary charges on a particle with diameter d p can be expressed as As regards the direct charging stage of our sensor, the discharge current is about a few tenths of milliamperes by the action of an imposed voltage of nearly 1 kV.The order of magnitude of the estimated product N i t r is 10 13 m -3 s.Under such conditions, Q(d p ) is roughly proportional to d p for the range of 10 nm < d p < 300 nm (Fuchs, 1963;Pui et al., 1988;Kulkarni et al., 2011), namely, Q(d p ) = C 2 d p and C 2 is a constant.Such that Eq. ( 2) can be simplified as where p,av 1 lim ( ) According to Eq. ( 5), we can derive where f q < 1 and q is usually less than 2 in the diameter range of 10-300 nm (Pui et al., 1988).Thus the second term at the right side of Eq. ( 7) can be ignored and then we yield 2 2 2 2 q i r p p 2 p 0 ( , , ) ( ) Based on Eqs. ( 4) and ( 8) and the definition N = p 1 lim ( ) (3) can be simplified as Combining Eq. ( 9) with Eq. ( 6), the following expressions of number-averaged size and number concentration for a definite sensor can be obtained.They can be calibrated experimentally in specific applications.For instance, measurement of monodispersed particles with known size and concentration could establish the relationships between the size and concentration with the corresponding I 1 and I 2 .Accordingly, the values of coefficients S N and S dp could be obtained.

Determination of the Dimensions and Operation Parameters of the Sensor Chip
Dimensions of each stage and operation parameters of the sensor can be determined according to electrostatic principle and some experimental results.It is also a tradeoff between normal operation of corona discharge and measurement of weak current signals.This is because it is better to reduce the flow rate to gain a higher charging efficiency and a less disturbance to the discharge.However, the electrometer prefers larger flow volume in order to relieve the requirement of higher measurement accuracy.
The three stages are integrated into a whole minimized sensing ship and the flow channels inside are identical in width and height for ease of manufacturing.The height of the channel h is determined by the compromise between the corona inception voltage and the leakage current in the sensing stage and the width w is educed from the experimental results of ion drift scope of corona discharge.The velocity of the aerosol flow v is the maximum one that will not disturb the discharge severely.
The low-level voltage V 1 applied on the precipitation stage and the length of its electrode plate L p should satisfy the relationship V 1 L p = h 2 v/eB air for the sake of removing all the residual ions, where B air is the mobility of air molecular.The value of V 1 was verified experimentally by gradually increasing of it when only pure air was driven through the sensor until the current observed in the sensing stage was the same as that when there was not any air flow.However, there must not be too much loss of the smallest particles been measured due to this electric field.The loss ratio caused by the electric field can be estimated as η = neV 2 B d p L p /(h 2 v), where B d p is the particle mobility.According to this estimation, the loss ratio of particles with diameter of 20 nm is less than 0.6% in an actual design if the diffusion loss was neglected.Theoretically, the loss ratio of particles larger than 20 nm should be much smaller than this estimated value.The selection of high-level voltage V 2 should be ensured that the smallest particles must not be completely removed and meanwhile, the largest ones should be partly precipitated.Moreover, there must be a sufficient difference between V 1 and V 2 to guarantee a proper discrimination of I 1 with I 2 by the electrometer.
The length of the sensing electrodes L s in the sensing stage relates to the applied voltage V s by L s ≥ h 2 v/(eV s B m ) so as to collect all the charged particles.B m is the mobility of the largest particles to be measured.And V s must not be too high to avoid gas ionization between the electrodes.For one type of design, V s was set to 5 V when L s = 40 mm.
The completed sensor is shown in Fig. 3, in which only the sensor chip is assembled inside the instrument while other components could be connected with the chip through pipelines and wires from the inside out in practical applications.The size of the whole device is about 12 × 12 × 5 cm and that of its main components are given in Table 1.

Characterization of DNA Aerosol Particles
In our study, Escherichia coli DNA particles were  generated by aerosolization of DNA solutions suspended with monodisperse DNA strands, which were made from separation and purification of DNA markers (DL 1000 and DL 5000, Takara Bio Inc.) using HPLC method.To examine the size of the aerosolized DNA strands, they were collected on filter papers and then tested by SEM.Fig. 4 shows one of the SEM images of the DNA particles collected by air filter, the effective size of which is characterized as their strand length.It is consistent with the relationship d p ≈ 0.34•N t (Robert, 2011), where N t is the number of base pairs (bp) composing the strands.We can also find from SEM experiments that the strands were less likely to curl themselves when N t < 1500 bp.Thence the effective size of the aerosolized DNA strands could be obtained by N t using above transform formula.
The aerosol generator (ATM-220, TOPAS) used to aerosolize the DNA solutions with different concentration of suspended DNA strands was calibrated employing a CPC (3776, TSI) to obtain the number concentration of that dispersed in aerosol.During the experiments, inlet pressure and concentration of DNA solution were adjusted to obtain DNA aerosol with specific number concentration.

Experiment Setup
Using different DNA markers (from DL 1000-100 to 1000-1000 corresponding to effective size of DNA particles from 34 nm to 340 nm) and strand solutions with different number concentration, DNA aerosol samples with particle number concentration of 10 2 -10 5 cm -3 were generated by aerosolization.The experimental setup shown in Fig. 5 allows aerosolization of DNA strands from their liquid suspensions and the DNA aerosol detection by the sensor, in which all the circuits and the micropump are integrated with the sensor given in Fig. 3.The oscilloscope is used to monitor the stability of corona discharge.Three repeated measurements were conducted under each condition for evaluating statistics.

RESULTS AND DISCUSSION
Fig. 6 shows the relationship between the number concentration N of DNA strands and the measured I 1 -I 2 in the cases of different strand sizes.I 1 -I 2 increases linearly with the number concentration, which is similar to that predicted by Eq. ( 11).However, there are intercepts with the horizontal abscissa.Linear fits were conducted and the relationship could be expressed as where the units are cm -3 (for N) and pA (for I 1 and I 2 ), respectively.St(N) in Fig. 6(a) is the standard deviation of (I 1 -I 2 ) values of particles with different sizes for certain number concentration N. The St(N) value is relatively larger for higher concentrations, but all of them are smaller than 0.05.Therefore, the number concentration is supposed to  be independent of size in this research.In most cases, the sensor could measure the number concentration of DNA strands ranging from 10 2 cm -3 to 10 5 cm -3 , but it failed when the size was less than 102 nm.In our experiments, the lowest measurable concentration of the sensor was 10 3 cm -3 for particles in the size range of 34-102 nm.It is because the smaller the particles are, the less possibility for them to be charged, such that the induced current is too weak to be sensed by the electrometer (which has a resolution of 50 fA) in the sensing stage.
As for the particle size detection, there are also linear relationships between I 1 /(I 1 -I 2 ) and the size of DNA strands in the range of 34-340 nm.However, the slopes and offsets of these relationships vary with number concentration, as is evident in Fig. 7.They can be summarized as where K(N) and b(N) represent slope and offset as a function of N, respectively, which can be expressed as Fitting of the results in the cases of different number concentration is shown in Fig. 8.The difference between the experimental results and the theoretical model is supposedly attributed to the slip correction factor C c , which was considered to be reversely proportional to d p .It is true for sphere particles, but may not be suitable for DNA strands since the charging level depends on shape of particles (Oh et al., 2004).Another reason may be that not all residual ions were removed during the low level voltage applied on the precipitation stage.This level cannot be set too high to bring about too much loss of charged particles to be measured.
Based on the above mentioned analysis, the aerosol sensor can figure out the number concentration of aerosolized DNA strands from the relationship between N and I 1 -I 2 given by Eq. ( 14).And then derives the size from the relationships between d p and I 1 /(I 1 -I 2 ) described by Eqs ( 15)-( 17).Comparison between the measured results and the reference data is shown in Figs. 9 and 10.Note that the sensor exhibits a good detectability to different number concentration in the range of 10 2 -10 5 cm -3 (10 3 -10 5 cm -3 for particles in size less than 102 nm), but the measurement error becomes larger for the lower concentration of 10 2 cm -3 , that is with a relative error of 33.3%.As for the size, the sensor is capable of detecting every 50 nm (corresponding to 150 bp) of the strands with a relative error of 64%-6.9%.One source of error may come from the random distribution of ions on the chains before they enter into the sensor.

CONCLUSIONS
A previously developed small aerosol sensor based on diffusion charging and electrical detection was used to detect DNA strand aerosols in real time.The sensor is capable of measuring concentrations ranging from 10 2 -10 5 cm -3 (10 3 -10 5 cm -3 for particles smaller than 102 nm) and sizes ranging from 34-340 nm (corresponding to 100-1000 bp) for aerosolized Escherichia coli DNA strands.The sensor accurately detects different concentrations and sizes of DNA strands, demonstrating its ability to monitor ultrafine DNA strand aerosols.Our future research aims to improve the performance of the sensor, including measurement accuracy.

Fig. 2 .
Fig. 2. One type of size distribution of aerosol.

Fig. 3 .
Fig. 3.The photo of the aerosol sensor.

Fig. 6 .Fig. 7 .
Fig. 6.Number concentration N of DNA strands as a function of I 1 -I 2 for different strand sizes.(a) figure with logarithmic scales; (b) figure with linear scales.

Table 1 .
Sizes of the main components of the sensor (units: cm).