MUID Gas R&D Results

Introduction

This note summarizes the results of the ORNL gas R&D effort to date.

All tests were performed with a cosmic muon stand in the "close" geometry shown in this figure.

Figure 1. Cosmic ray test stand used in these studies.

The "far" geometry is more realistic since we expect muons to be nearly normally incident, but we believe the "close" geometry to be good enough and the much higher count rates (due to solid angle) make it much more convenient. The amplitudes are somewhat lower in the "far" geometry as shown in the following figure.

Figure 2. Signal amplitudes for "close" and "far" geometries.

This note does not summarize the electronics R&D effort since it is not yet mature. But, some facts are known which are important to understanding the direction of the gas studies.

Many of the gas tests were performed with a readout scheme having coaxial cables and no in-panel preamplification. This is contrary to the baseline advocated at the Costa Mesa meeting we used. The reason for this is that we discovered that the primary source of noise (at least in our lab) was the tube itself, due to the large tube capacitance (roughly one nF for a ganged pair of tubes). Therefore preamplification did not help much. We still advocate an in-panel preamplifier in order to guard against pickup on the long (> 15 m) signal cable runs before the FEMs.

Figure 3. Iarocci tube readout scheme used for these tests.

We discovered that a constant fraction discriminator buys us about 15 nsec in the width of the drift time distribution. This can be worth several percentage points in per-gap trigger efficiency. This benefit of the CFD is due to the fact that the shaped signals had a very slow rise time (roughly 15 nsec), and our discriminator level was not very far below the smallest signals. Thus the smallest signals need almost their full rise time to cross the threshold, increasing the width of the distribution by that value. The CFD eliminates this signal "walk" by always triggering at a particluar point on the signal (for example, the peak). So far we have compared performance with and without the CFD only for Ar/isoButane and with a particular shaping filter.

.Figure 4. Drift time distribution with leading edge and constant fraction discriminators.

An Exhaustive Search

A prioritized criteria list for gas selection follows:

• The gas must allow operation with high efficiency over a relatively long HV plateau. Long here is considered to be 10% of the nominal voltage to guard against performance degradation with changes in atmospheric pressure.

• The gas must be fast enough that high efficiency is achieved within a narrow time window (roughly 80 nsec). MUID is in the trigger which means that any particle whose signal in the LST arrives at the FEM more than 106 nsec after the earliest possible signal will be effectively lost. We lose some these 106 nsec to variations in propagation time along the LST anode wire, different flight times, and different cable lengths, thus the 80 nsec figure of merit.

• The gas must be relatively inexpensive. The MUID has a volume of roughly 50 cubic meters. Past experience has shown that the LSTs should be flushed at a rate of at least once per day. Until we show that a lower flow rate is possible we have to go with this flow rate as a baseline. Even for bottles of "cheap" gas (Argon, etc) a year's gas budget is >$25K. Initial cost estimates for recycling at this flow rate appear to be prohibitive (>$150K).

• Preferably the gas would not be flammable (or even explosive), although we have been approved to run the standard LST mixture (25/75 Ar/isoButane).

• The gas should allow successful operation in proportional mode. The desire to run in proportional mode is primarily motivated by longevity concerns associated with the difficulty in servicing the panels after installation into PHENIX. Streamer mode is harder on the tubes. Also, streamer mode requires higher operating voltages and thus significantly more expensive HV supplies.

Many of the gases failed because the efficiency plateau was very short. This is because of the large inherent tube noise (see above) which required us to go to relatively high voltage to get the signals above the noise. We had to get near breakdown for many gases before we achieved any efficiency.

• Ar/CO2 mixtures - In general these mixtures were too slow and had unacceptably short efficiency plateaus. Bubbling the mixture through alcohol did not have a significant affect in extending the plateau to higher voltage.

• P10 - We didn't even get to 100% efficiency with this gas before breakdown started. Alcohol didn't help.

• Ar/isoButane - The "proportional" mixtures (high in Argon) also suffered from a short effiency plateau before breakdown. The "streamer" mixtures (high in isoButane) had the expected long effiency plateau, but not until the voltage was high enough that they were operating in streamer mode. All of these mixtures were too slow.

• CF4/isoButane - This mixture worked beautifully. It was much faster than any other gas we tested and had a very long efficiency plateau. The only problem is that it is prohibitively expensive if we need to flush once per day, even at a concentration of 10%.

• CH4 - The plateau was very short.

• isoButane - Pure isoButane worked very nicely. We were worried about the possibility of this being an explosive (not just flammable) mixture and were warned about isoButane polymerization. But, this pointed us on the right track - we needed to find a gas to dilute the isoButane without detrimentally affecting it.

• isoButane/N2 - We thought that N2 might be a good diluter gas, but the chambers were audibly sparking before we achieved any efficiency.

• isoButane/CO2 - The combination of these gases, in a wide variety of ratios, satisfied all requirements. This is a proportional mode gas, and both components are relatively inexpensive. We may want to run with a flammable (but not extraordinary) concentration of isoButane. Results for efficiency, single's rates and timing are summarized in two figures below. The first figure shows the efficiency and single's plateaus, as functions of the HV setting, for the different ratios. The plateaus are quite wide and are within reach of the HV supplies already ordered. The second figure shows the effiency versus gatewidth for the same mixtures. All show saturation at very high efficiency by 80 nsec.

  

  Figure 5. Efficiency and single's plateaus as functions of high voltage for different ratios of isoButane and CO2.

  

  Figure 6. Inefficiency as a function of LVL1 trigger gate width for different ratios of isoButane and CO2.

  Note that the timing plots were made with the signals going through a constant-fraction discriminator. If the CFD buys us 15 nsec for this gas mixture, shaping parameters, etc. then it buys roughly five percent in efficiency. We are about to start optimizing the CFD and shaping parameters for this gas mixture.

  The ratio of the two gases in our mixture does not seem to be critical for performance. Although somewhere between 90% CO2 and pure CO2 the behavior must deteriorate since pure CO2 has a negligible efficiency plateau and is relatively slow. So far, the only difference we see between different ratios is the gas gain. The figure below shows the most probable value of the signal amplitude for different gas ratios. We may want to optimize the gas gain (a 50/50 mixture) or minimize safety hazard and expense (10/90 isoButane/CO2?). This decision has not been made.

  Figure 7. Most probable signal amplitude for different ratios of isoButane and CO2.

Conclusions

We have found a gas mixture (isoButane/CO2) which satisfies all of our requirements. There is a very wide range of acceptable ratios, and ratio of the two gases has not been decided. But, we believe the mixture will contain between 25 and 60% isoButane. Our plan is to optimize electronics with a 25/75 isoButane/CO2 mixture and return to the question of the exact ratio later.