ReadMe file for PALMS (Particle Analysis by Laser Mass Spectrometry) during ACE-1 at Cape Grim March 3, 1997 D. M. Murphy, D. S. Thomson, A. M. Middlebrook, and M. E. Schein FILES: The instrument measured either the positive or negative ion spectrum of each particle after it is hit with a pulse from a 193 nm excimer laser. For this reason, there are two sets of files: MSCGNEG and MSCGPOS. There are two files of titles for each column (TITLES.CGxxx) in ASCII format, one title per line, with units or a short description after a comma. Some of these are reproduced at the end of this file. There are two files containing the number of columns and the number of particles for each date (DataInfo.CGNEG and DataInfo.CGPOS). There are two files of data for each date. They are in ASCII format, 3E13.5,then E11.3 for the remainder of the columns. Each line contains the data from one particle. EXPERIMENT DESCRIPTION Important note about times: The times of each particle have been shifted by 270 s to account for the estimated transit time in the long inlet from the 55 m level of the tower. This correction is good to about +- 270 s. This experiment description is taken from a draft of a paper intended for the JGR ACE-1 special issue. Look there for updates.... We measured the mass spectra of about 20,000 individual particles at Cape Grim. The PALMS instrument was described by Murphy and Thomson (1995; 1997). Two significant changes have been made since those early measurements. First, the excimer laser used for ionization was operated at 193 nm at Cape Grim rather than 248 nm as in the earlier measurements. This change was based on laboratory results showing that pure sulfate particles are difficult to ionize at 248 nm (Thomson et al., in press). Most particles tested ionize more uniformly at 193 nm than at 248 nm, with the notable exception of extremely pure sulfuric acid particles. These remain difficult to ionize at 193 nm (Thomson et al., in press). A combination of several changes improved the limit to the smallest particles analyzed from 0.3 to 0.16 µm diameter since the Idaho Hill measurements reported by Murphy and Thomson (1997). The He-Ne laser for particle detection was replaced with a diode pumped doubled YAG laser (Amoco) with 80mW output at 532 nm. The shorter wavelength is scattered more efficiently by small particles. The avalanche photodiode detector was replaced with a small PMT (Hamamatsu). Finally, an anamorphic prism was used to produce a focal spot for the YAG laser that was smaller along the axis of the particle beam than across it. This increased the intensity of the light and sharpened the time response of the trigger system without missing a larger fraction of the aerosol beam, as would occur if the focal spot were simply made smaller in both dimensions. In order to simplify data analysis, the time of flight mass spectrometer was changed from a pulse focus design to a reflectron design (Mamyrin et al., 1973). The custom designed ion reflector has two stages, is 17 cm deep, and has a 14° included angle between the incident and reflected beams. The total drift tube length in the mass spectrometer was also increased to 135 cm. Our previous pulse focus design performed well but the analysis of the mass spectra proved difficult to automate because of the complicated relationship between ion arrival time and ion mass in the pulsed design. In contrast, the ion arrival time in a reflectron is simply proportional to the square root of the mass to charge ratio. Small magnets were added next to the flight tube between the ion reflector and the detector to eliminate false peaks caused by ions hitting grids or metal surfaces in the ion reflector and releasing secondary electrons. In negative ion mode, some of these secondary electrons reached the microchannel plate detector. For our particular mass spectrometer design, these peaks were about 2% of the intensity of their parent peaks and appeared at about 60% of the parent peaksą masses. A field of a few gauss was sufficient to bend the electrons into the flight tube wall without significantly disturbing the much heavier ions. The PALMS instrument was located in a trailer at the base of the sampling and communications tower at Cape Grim. Air was obtained from a common inlet which ran from the 55 m level to near the base of the tower. The base of the tower is 90 m above sea level. Air flow over the cliff face results in an estimated 110 m effective height above sea level [Baines and Murray, 1994]. This inlet was a 10 cm diameter stainless steel tube with at least a 29 liter per minute (lpm) flow rate, with more when filter samples were being taken by other groups. A 2.5 cm diameter copper tube (except for a short PVC section for electrical isolation) with a 22 lpm flow rate led to just inside the trailer. Finally, the approximately 1 lpm flow into the PALMS instrument was sampled through about 2 m of ~8 mm diameter copper tubing and about 10 cm of conductive silicone tubing. At times, a differential mobility analyzer (DMA) was inserted into the ambient flowstream to reject large particles. When used, the DMA was connected to the inlet 2 pieces of Tygon tubing, each about 2 m long. Particle sizes and compositions reported here should be regarded as dry. In an effort to avoid heating the particles in the inlet lines, the copper line outside the trailer was insulated and wrapped in white tape. The lines inside the trailer were also insulated, except for the last section to the PALMS instrument. We monitored the relative humidity in the inlet at the entrance to the trailer and near optical particle counters that shared a sample line with the PALMS inlet until inside the trailer. Despite the insulation, the air warmed enough that relative humidities were typically in the 30to 40% range, as opposed to typically 70 to 90% outside. We also measured the temperature in the sample lines. The measured temperatures and relative humidities, compared to ambient data from by the Cape Grim station, were consistent with warming as the only cause of the drop in relative humidity. When particles were sampled through the DMA, a diffusion drier in the DMA further reduced the sample line humidity. REFERENCES: Murphy, D. M., and D. S. Thomson, Laser ionization mass spectroscopy of single aerosol particles, Aer. Sci. Technol., 22, 237-249, 1995. Murphy, D. M., and D. S. Thomson, Chemical composition of single aerosol particles at Idaho Hill: Negative ion measurements, J. Geophys. Res., 1997. Murphy, D. M., and D. S. Thomson, Chemical composition of single aerosol particles at Idaho Hill: Positive ion measurements, J. Geophys. Res., 1997. DATA REDUCTION: Each particle produces three digitized waveforms and a number of other sampled data channels. The waveform of the light scattered by the particle as it goes through the YAG laser beam is converted to a height and width. The data from the grid portion of the detector is used only to check the data from the microchannel plate. Those data consist of about 16000 8 bit samples of detector output spaced 5 ns apart. They are converted to detector current, which involves the calibration of the logarithmic amplifier prior to the digitizer. Then peaks are identified, with a complicated algorithm to keep from identifying noise as peaks and blips on the sides of peaks as separate peaks. Then the mass scale is calculated for each particle: what masses corres- pond to the time of each peak. This was generally automated but took a lot of hand checking and correction. More than one peak of the same mass was allowed. The automated routines did about 80% of the spectra without inter- vention and about another 15% were kicked back for a manual check but were OK. About 5% of the spectra were reduced manually. The worst problems in peak identification in the Cape Grim data were the doublets from the cluster ions like (NaCl)Cl- or (NaCl)Na+. These can undergo unimolecular decay in the mass spectrometer and end up producing doublets separated by a fraction of a mass. Such doublets play havoc with the automated mass scale assignment. Occasionally, the biggest peaks had concave tops to also produce doublets. (I believe that this is due more to ion space charge effects than to detector saturation.) A subjective confidence was assigned to each spectrum: 8 if the automated analysis only, otherwise 1-10 with 10 best. Low confidence spectra (between 2 and 5% of all spectra) were not used in later analysis. A very few spectra also were rejected if the unidentified peaks were a large fraction of the total ion current. A few small unidentified peaks (i. e. no definitive mass number) were allowed. All peaks were integrated with an algorithm that took into account shoulders of other peaks and baseline shifts. The algorithm is good to about 5%. All peaks are expressed as a fraction of the total ion current because lab work shows that this is much more stable than absolute peak heights. The last step in data reduction is to assign chemical meaning to the peaks; that is subject to further revision and is not included in these files. COLUMN HEADINGS: - Note that 86400*Date+Time (in double precision) gives Macintosh system time. This system time is used by IGOR and some other data handling software. - Note that the joulemeter data are not always reliable. The autoscaling on the joulemeter didn't work properly so the data are often off-scale. - The inlet line temperature and humidity were taken just outside the trailer. Humidity was with a Vaisala sensor. DATE, Local time (days from 1904) TIME, Local time (seconds) INDEX, just the number of the spectrum on that day SCATTER, Scatter pulse height SCATWIDTH, scatter pulse width SIZE, -1 polydisperse else DMA diameter DetSignal, Total detector signal (Coulombs) NumPKS, Number of peaks in spectrum CONF, Subjective confidence (10 best; 8 auto) NDFilter, Neutral density filter on excimer JMeter, joulemeter excimer pulse energy ScaleA, Mass scale t=a+bSQRT(m) ScaleB, Mass scale t=a+bSQRT(m) INLETTE, Inlet line temperature INLETRH, Inlet relative humidity ROOMT Trailer temperature MSHigh, Total of all peaks > 319 (fraction of ion current) MSUnlist, Total of all unlisted peaks (fraction of ion current) MSNoID, Total of all unidentified peaks (fraction of ion current) MS1, is peak at mass 1 atomic unit (fraction of ion current) MS6 MS7 MS11 MS11.5; this is the only non-integer mass retained (from Na++, positive only) MS12 MS13 MS14 ... MS319 There are diffent sets of "derived" variables for the positive and negative spectra. These are intended to ease computations. For example, DCL in the negative spectra includes both mass 35 and 37, makes sure that the isotope ratio is reasonable, and applies corrections based on mass 36 and 38 to mass 37 if it looks like there are organics in the spectrum. DCLCLU totals up all of the various cluster ion peaks that contain chlorine. A few derived variables, like DSN and DMO, were calculated using linear least squares on the isotopic pattern. The code used to derive the derived variables is very complex and not fully documented here. Contact Dr. Murphy if you need details of how a variable was calculated.