Periodic Reporting for period 4 - CHROMIUM (CHROMIUM)
Reporting period: 2021-04-01 to 2022-09-30
The recent discovery of the Higgs particle has left us in no doubt that the Standard Model (SM) of particle physics is correct in very large part. This theory has managed to describe almost everything we see involving three of the four known forces of nature: the Strong, Weak and Electromagnetic forces.
However, another slightly less recent discovery, that of neutrino mass, is in tension with the SM's successes. In order for the Weak force to be properly described in the theory,
neutrinos were assigned exactly zero mass: however it has been proved that while their masses are very small, they are definitely not zero. Over the last few decades, particle physicists mounted a number of experiments looking for, and finally measuring, a quantity known as Charge Parity (CP) violation in sub-atomic particles called quarks. This manifests itself as a difference in interaction rates between matter and antimatter particles and the reason for looking for this difference was an attempt to explain what happened to the anti-matter which must have existed in the early universe. The simple explanation was that as the Universe expanded, transitions between matter and anti-matter were stopped at some energy density and over time, the small excess of matter that had been produced led to the total annihilation of all the antimatter, leaving just a small amount of matter which is what we see around us in the Universe. Unfortunately, the CP violation observed in the quark sector was much too small .
However, the discovery of neutrino mass gives rise to the possibility of a mechanism which can explain the matter that we see.
The overarching goal of this proposal is to measure, or constrain further, the neutrino CP violating angle \dcp,
by measuring the size of the difference between the quantum mechanical flavor oscillations of neutrinos and anti-neutrinos, taking advantage of a
unique, and manifestly overlooked opportunity provided by the NuMI muon neutrino beam, produced at the Fermi National Accelerator Laboratory.
Neutrinos only interact via the Weak Interaction, and this produces an exceptionally acute problem for precision measurements. As the needed precision
increases, so too does the detector mass, until a statistically significant improvement over existing measurements requires either decades of data taking
or a detector that up to now would be beyond acceptable cost. There is a new effort in the USA to develop a \$600M experiment to search for this effect along with many
other potentially naturally occurring processes such as proton decay and supernova, in a planned future neutrino beam. This could provide results at the earliest in 2030.
Our approach is to use the existing neutrino beam and to build a vastly cheaper detector with the sole purpose of measuring the beam neutrinos.
The novel detector concept is named CHIPS, for Cherenkov detectors In PitS which will be located in the Wentworth 2W flooded taconite pit near Aurora, Minnesota. This pit intersects the NuMI neutrino beam. The concept pushes on costs of the detector by using the natural body of water to support the detector volume, avoiding a very strong and costly mechanical structure. It uses the water overburden to shield from cosmic rays, making use of the time window that the beam is delivered and the knowledge of the neutrinos direction, to avoid having to be positioned under a very large overburden of rock. The first year will deploy a 20~kt volume vessel. Neutrinos from the NuMI beam interact in the large volume of water and produce charged particles which in turn produce Cherenkov radiation in the water volume. This is predominantly blue light which can be detected by sensitive light detectors called Photo-Multiplier Tubes (PMTs). Funds from Templeton Foundation could enable the full instrumentation of the vessel within 2 years, enabling physics measurements to start very quickly. The NuMI beam will deliver neutrinos for the next 7-8 years and so time is of the essence to get started. The data collection would continue beyond the time frame of this project, potentially for as long as the NuMI beam is operational in light of the challenges faced by the weakly interacting nature of the neutrinos. Our estimate is that if hints from other, much less precise experiments are correct, dCP could be known within 5 years.
The new technical concept is that of a submerged radome, providing a light tight, and water tight barrier within which are mounted a very large number of PMTs. The radome is filled with purified water which will be continuously circulated and cleaned making the water very transparent to the Cherenkov light being produced. The PMTs are traditional, and very well understood, low technical risk, photon detectors. They have been deployed in water Cherenkov detectors before but the technical challenge here is to combine the signals from so many PMTs under the water, and use local computer power also under the water to identify the neutrino events coming from the NuMI beam which arrive at a known time and within a very short time window. The electronic signals have to be transported to shore, up to 300m away, and ideally the entire detector readout can be transmitted on one or two long cables. This again provides very large cost savings compared to traditional detectors.
However, another slightly less recent discovery, that of neutrino mass, is in tension with the SM's successes. In order for the Weak force to be properly described in the theory,
neutrinos were assigned exactly zero mass: however it has been proved that while their masses are very small, they are definitely not zero. Over the last few decades, particle physicists mounted a number of experiments looking for, and finally measuring, a quantity known as Charge Parity (CP) violation in sub-atomic particles called quarks. This manifests itself as a difference in interaction rates between matter and antimatter particles and the reason for looking for this difference was an attempt to explain what happened to the anti-matter which must have existed in the early universe. The simple explanation was that as the Universe expanded, transitions between matter and anti-matter were stopped at some energy density and over time, the small excess of matter that had been produced led to the total annihilation of all the antimatter, leaving just a small amount of matter which is what we see around us in the Universe. Unfortunately, the CP violation observed in the quark sector was much too small .
However, the discovery of neutrino mass gives rise to the possibility of a mechanism which can explain the matter that we see.
The overarching goal of this proposal is to measure, or constrain further, the neutrino CP violating angle \dcp,
by measuring the size of the difference between the quantum mechanical flavor oscillations of neutrinos and anti-neutrinos, taking advantage of a
unique, and manifestly overlooked opportunity provided by the NuMI muon neutrino beam, produced at the Fermi National Accelerator Laboratory.
Neutrinos only interact via the Weak Interaction, and this produces an exceptionally acute problem for precision measurements. As the needed precision
increases, so too does the detector mass, until a statistically significant improvement over existing measurements requires either decades of data taking
or a detector that up to now would be beyond acceptable cost. There is a new effort in the USA to develop a \$600M experiment to search for this effect along with many
other potentially naturally occurring processes such as proton decay and supernova, in a planned future neutrino beam. This could provide results at the earliest in 2030.
Our approach is to use the existing neutrino beam and to build a vastly cheaper detector with the sole purpose of measuring the beam neutrinos.
The novel detector concept is named CHIPS, for Cherenkov detectors In PitS which will be located in the Wentworth 2W flooded taconite pit near Aurora, Minnesota. This pit intersects the NuMI neutrino beam. The concept pushes on costs of the detector by using the natural body of water to support the detector volume, avoiding a very strong and costly mechanical structure. It uses the water overburden to shield from cosmic rays, making use of the time window that the beam is delivered and the knowledge of the neutrinos direction, to avoid having to be positioned under a very large overburden of rock. The first year will deploy a 20~kt volume vessel. Neutrinos from the NuMI beam interact in the large volume of water and produce charged particles which in turn produce Cherenkov radiation in the water volume. This is predominantly blue light which can be detected by sensitive light detectors called Photo-Multiplier Tubes (PMTs). Funds from Templeton Foundation could enable the full instrumentation of the vessel within 2 years, enabling physics measurements to start very quickly. The NuMI beam will deliver neutrinos for the next 7-8 years and so time is of the essence to get started. The data collection would continue beyond the time frame of this project, potentially for as long as the NuMI beam is operational in light of the challenges faced by the weakly interacting nature of the neutrinos. Our estimate is that if hints from other, much less precise experiments are correct, dCP could be known within 5 years.
The new technical concept is that of a submerged radome, providing a light tight, and water tight barrier within which are mounted a very large number of PMTs. The radome is filled with purified water which will be continuously circulated and cleaned making the water very transparent to the Cherenkov light being produced. The PMTs are traditional, and very well understood, low technical risk, photon detectors. They have been deployed in water Cherenkov detectors before but the technical challenge here is to combine the signals from so many PMTs under the water, and use local computer power also under the water to identify the neutrino events coming from the NuMI beam which arrive at a known time and within a very short time window. The electronic signals have to be transported to shore, up to 300m away, and ideally the entire detector readout can be transmitted on one or two long cables. This again provides very large cost savings compared to traditional detectors.
The global project has accelerated over the past year in order to try to make up the lost time from the grant start. The detector can only be deployed seasonally and we did not succeed with the 2018 deployment before season end October 1st. There are several work strands where there has been progress, and several progressional developments outside of the ERC-funded part of the detector which are important.
\section{ Design and Simulation : Work Strand 1}
Final design has been verified by detailed simulations. This has included overall layout and full simulation of the light cones (to increase effective photocathode area). Deep learning techniques have been applied to our reconstruction which have resulted in significant gains in efficiency, as shown in the Figures called vtxX_energy.png vxY_energy.png vtxZ_energy.png and energy_energy.png.
{PMT Plane Design - work strand 2}.
The original design outlined in the ERC proposal called for 31 PMTs per plane, the plane was constructed of perspex sheets as shown in Figure~\ref{fig:nikhefplane}(left). During the summer of 2015 it was concluded that this was not an optimal design due to difficulty with maintaining water tightness across the large number of flat joints. A different approach was developed on the fly during that prototype deployment, shown in Figure \ref{fig:nikhefplane} (right) which has been carried through to the final design. The design and specifics of the detector planes are complete. The design of 4 types of planes is complete and construction of the planes has been completed. Factories were set up in Minnesota and in Wisconsin. 3500/5500 PMTs and bases for the front and end-cap sections have been delivered and potted and all other subsystems for detector planes have been delivered. 10 planes have been fully tested with readout during the winter months in a dark room at the site. Data shows dark rates of the PMTs is acceptable reaching about 100Hz/tube at 4C. All the endcap planes have been constructed, cabled and water tested. A small-format White Rabbit (WR) board has been developed at Nikhef for insertion into the central water tight container shown in the figure PK-switch.png.
We originally foresaw veto PMTs in the top and side "veto-volumes" to aid with cosmic ray rejection. However, it was realised that not only did that require at least an extra 2m in diameter for the whole structure along with the associated costs but it would increase the complexity of the water circulation system as well as the construction itself. The main use of veto counters will be on the top-cap of the detector to veto cosmic rays entering the top of the detector.
{\bf Veto and Rear PMTs - work strand 3}. A design based on the front plane design has been developed using borrowed PMTs from our French colleagues. The electronics for these PMTs has been developed from scratch using innovative design using components that are commercially available and present in many mobile phones. The rear wall instrumentation was originally foreseen to be 100 large PMTS, instrumented with the PARISROC readout system for which there were funds for hardware and technical effort at UCL. The PMTs were to be donated by UCL. However, we were lent 500 3" Hamamatsu R6091's from our colleagues from the NEMO-3 collaboration and together with
electronics developed at UW we have developed a novel idea of a distributed readout system for particle physics based on commercially available electronic components. There are three distinct aspects of this hardware development: a small-format Cockroft-Walton (CW) positive HV generator base (based on a design from the COUPP experiment at FNAL) to drive the PMTs; a "microdaq" board which contains a STM32F446 microprocessor and sits on the PMT, delivering self-triggered time-over-threshold information to the single-board computer as well as control signals to the CW board; and the WR signal fan-out board to deliver the absolute time to every \microdaq. The CW is controlled via PWM signals which adjust the resonant frequency to change the overall magnitude of the HV. The advantage of the positive HV base is that the PMTs can be submerged in water without loss of functionality which is often observed with a negative HV. Miniature negative HV CW bases are much easier to build, without the need for a large transformer which typically must be specially fabricated. We have innovated on this by using two commercially available inductors on either side of the base board as seen in the figure pmtbasemicrodaq.png
The WR system is fast becoming the new standard for 1\,ns absolute timing across Ethernet. Software to be developed on the BB single-board computer will collect up the signals from up to 32 PMTs via LVDS links and deliver buffered data to the network via Ethernet. The ensemble will provide a very inexpensive DAQ+HV system for similar applications at the target level of \$50/PMT.
{\bf Deployment, Commissioning and Data Analysis - work strand 4}. UW covered the cost of the extensive re-engineering of the steel structure since the original proposal. This has resulted in a reduction in the cost of more than a factor of 2. We have been working together with a company in China who have made significant contributions to the custom design and who will provide the structure at \$171,000. There are two components to these savings. One was the idea to suspend one end-cap structure from the other, using Dyneema strings to attach the endcaps to each other and as the mounting points for the PMT planes, instead of a rigid wall structure. This is not only a cheaper solution but gives us the opportunity to increase the size of the detector without a large cost increase due to the increased structure cost. The other component was the investment in the simplification of the structure for construction.Steel frame designed and purchased in China, liner, floating dock, and water system have been delivered. Ground work has been carried out for the deployment, such as electricity supply and ground preparation.
We had a review of the structure in May 2017 and most of the suggestions have been taken on board. There were no show stoppers at that review. We have worked with an external engineering company (Barr Engineering) who have reviewed the structure for safety and are satisfied the structure adheres to all necessary regulations. The floating dock will be constructed from units already delivered.
The deployment approach is shown in Figure~\ref{fig:structure}. The endcaps are shown in a simulated picture floating in the water in chipinthewater.png before the walls are attached in the left picture. The floating dock supporting winches from which the detector is lowered into the water is shown in chipssuspended.png. The status of the structure after season 2018 is shown in figure structureslide.png.
{\bf DAQ subsystem}
The system arcitechture diagram is shown in Figure \ref{fig:system}(left) and the prototyped board labelled "Badgerboard" in the left figure is shown at right. The DANOUT fans-out decoded PPS and 10\,MHz clock to the planes, and funnels incoming PMT data to the Ethernet switch, then to the surface via the WR-LEN and fibre-optic cable.
A total of 16 planes will receive the WR signals from one WR end node. The WR end node will connect up to the GPS driven master clock on the shore via fiber-optic cable.
We are attempting to use the NuMI spill signal as a trigger. This has not been tried for the remote detectors, as both MINOS and \nova have utilised a very wide gate and have aligned the beam spill time to the activity in the detector after the fact. There is an accelerator signal (TClock A5) which is issued at the start of the accelerator cycle, approximately 1.5 sec from the start of the neutrino beam spill. Using the \nova Time Distribution Unit (TDU) and a GPS receiver at FNAL we will encode the time of the A5 and write that to linux box which will run the spill server software. The spill server and the remote DAQ machine will communicate the encoded time via a dedicated ssh tunnel. Hopefully the encoded time will arrive at the remote DAQ machine within the 1.5 sec. We will be testing this in the next month.
\section{Schedule}
The original timescale called for the first endcap to be deployed in summer of 2017 followed by the rest of the detector in the summer of 2018. The initial 2017 window was excluded as the funding decision from ERC was past the 2016 cycle. Due to delivery issues the 2018 window closed without completion of deployment. The project is ambitious and carries substantial risk of failure to deploy owing to its inherent complexity, but we are on track to complete the deployment in 2019.
\section{ Design and Simulation : Work Strand 1}
Final design has been verified by detailed simulations. This has included overall layout and full simulation of the light cones (to increase effective photocathode area). Deep learning techniques have been applied to our reconstruction which have resulted in significant gains in efficiency, as shown in the Figures called vtxX_energy.png vxY_energy.png vtxZ_energy.png and energy_energy.png.
{PMT Plane Design - work strand 2}.
The original design outlined in the ERC proposal called for 31 PMTs per plane, the plane was constructed of perspex sheets as shown in Figure~\ref{fig:nikhefplane}(left). During the summer of 2015 it was concluded that this was not an optimal design due to difficulty with maintaining water tightness across the large number of flat joints. A different approach was developed on the fly during that prototype deployment, shown in Figure \ref{fig:nikhefplane} (right) which has been carried through to the final design. The design and specifics of the detector planes are complete. The design of 4 types of planes is complete and construction of the planes has been completed. Factories were set up in Minnesota and in Wisconsin. 3500/5500 PMTs and bases for the front and end-cap sections have been delivered and potted and all other subsystems for detector planes have been delivered. 10 planes have been fully tested with readout during the winter months in a dark room at the site. Data shows dark rates of the PMTs is acceptable reaching about 100Hz/tube at 4C. All the endcap planes have been constructed, cabled and water tested. A small-format White Rabbit (WR) board has been developed at Nikhef for insertion into the central water tight container shown in the figure PK-switch.png.
We originally foresaw veto PMTs in the top and side "veto-volumes" to aid with cosmic ray rejection. However, it was realised that not only did that require at least an extra 2m in diameter for the whole structure along with the associated costs but it would increase the complexity of the water circulation system as well as the construction itself. The main use of veto counters will be on the top-cap of the detector to veto cosmic rays entering the top of the detector.
{\bf Veto and Rear PMTs - work strand 3}. A design based on the front plane design has been developed using borrowed PMTs from our French colleagues. The electronics for these PMTs has been developed from scratch using innovative design using components that are commercially available and present in many mobile phones. The rear wall instrumentation was originally foreseen to be 100 large PMTS, instrumented with the PARISROC readout system for which there were funds for hardware and technical effort at UCL. The PMTs were to be donated by UCL. However, we were lent 500 3" Hamamatsu R6091's from our colleagues from the NEMO-3 collaboration and together with
electronics developed at UW we have developed a novel idea of a distributed readout system for particle physics based on commercially available electronic components. There are three distinct aspects of this hardware development: a small-format Cockroft-Walton (CW) positive HV generator base (based on a design from the COUPP experiment at FNAL) to drive the PMTs; a "microdaq" board which contains a STM32F446 microprocessor and sits on the PMT, delivering self-triggered time-over-threshold information to the single-board computer as well as control signals to the CW board; and the WR signal fan-out board to deliver the absolute time to every \microdaq. The CW is controlled via PWM signals which adjust the resonant frequency to change the overall magnitude of the HV. The advantage of the positive HV base is that the PMTs can be submerged in water without loss of functionality which is often observed with a negative HV. Miniature negative HV CW bases are much easier to build, without the need for a large transformer which typically must be specially fabricated. We have innovated on this by using two commercially available inductors on either side of the base board as seen in the figure pmtbasemicrodaq.png
The WR system is fast becoming the new standard for 1\,ns absolute timing across Ethernet. Software to be developed on the BB single-board computer will collect up the signals from up to 32 PMTs via LVDS links and deliver buffered data to the network via Ethernet. The ensemble will provide a very inexpensive DAQ+HV system for similar applications at the target level of \$50/PMT.
{\bf Deployment, Commissioning and Data Analysis - work strand 4}. UW covered the cost of the extensive re-engineering of the steel structure since the original proposal. This has resulted in a reduction in the cost of more than a factor of 2. We have been working together with a company in China who have made significant contributions to the custom design and who will provide the structure at \$171,000. There are two components to these savings. One was the idea to suspend one end-cap structure from the other, using Dyneema strings to attach the endcaps to each other and as the mounting points for the PMT planes, instead of a rigid wall structure. This is not only a cheaper solution but gives us the opportunity to increase the size of the detector without a large cost increase due to the increased structure cost. The other component was the investment in the simplification of the structure for construction.Steel frame designed and purchased in China, liner, floating dock, and water system have been delivered. Ground work has been carried out for the deployment, such as electricity supply and ground preparation.
We had a review of the structure in May 2017 and most of the suggestions have been taken on board. There were no show stoppers at that review. We have worked with an external engineering company (Barr Engineering) who have reviewed the structure for safety and are satisfied the structure adheres to all necessary regulations. The floating dock will be constructed from units already delivered.
The deployment approach is shown in Figure~\ref{fig:structure}. The endcaps are shown in a simulated picture floating in the water in chipinthewater.png before the walls are attached in the left picture. The floating dock supporting winches from which the detector is lowered into the water is shown in chipssuspended.png. The status of the structure after season 2018 is shown in figure structureslide.png.
{\bf DAQ subsystem}
The system arcitechture diagram is shown in Figure \ref{fig:system}(left) and the prototyped board labelled "Badgerboard" in the left figure is shown at right. The DANOUT fans-out decoded PPS and 10\,MHz clock to the planes, and funnels incoming PMT data to the Ethernet switch, then to the surface via the WR-LEN and fibre-optic cable.
A total of 16 planes will receive the WR signals from one WR end node. The WR end node will connect up to the GPS driven master clock on the shore via fiber-optic cable.
We are attempting to use the NuMI spill signal as a trigger. This has not been tried for the remote detectors, as both MINOS and \nova have utilised a very wide gate and have aligned the beam spill time to the activity in the detector after the fact. There is an accelerator signal (TClock A5) which is issued at the start of the accelerator cycle, approximately 1.5 sec from the start of the neutrino beam spill. Using the \nova Time Distribution Unit (TDU) and a GPS receiver at FNAL we will encode the time of the A5 and write that to linux box which will run the spill server software. The spill server and the remote DAQ machine will communicate the encoded time via a dedicated ssh tunnel. Hopefully the encoded time will arrive at the remote DAQ machine within the 1.5 sec. We will be testing this in the next month.
\section{Schedule}
The original timescale called for the first endcap to be deployed in summer of 2017 followed by the rest of the detector in the summer of 2018. The initial 2017 window was excluded as the funding decision from ERC was past the 2016 cycle. Due to delivery issues the 2018 window closed without completion of deployment. The project is ambitious and carries substantial risk of failure to deploy owing to its inherent complexity, but we are on track to complete the deployment in 2019.
The development of the electronics for the loaned PMTs has produced an unforseen innovation.
The idea is for a distributed system that can be used for large arrays of PMTs based on commercially available electronic components. It can be seen as part of a movement to distribute intelligence towards the detector, a natural way forward as electronic components gets cheaper and smarter. This approach of the microprocessor on the PMT has been pioneered at UW, with a more sophisticated microDAQ prototype (V2) deployed at the South Pole this winter (2017/18) for IceTop tests.
The MicroDAQ relies on commercially tested components and could potentially be sold as a commercial system with the timing fan-out and CW base. The CW base is not necessary, but can easily be modified for different PMTs if needed as it works together and is controlled by the MicroDAQ.
The idea is for a distributed system that can be used for large arrays of PMTs based on commercially available electronic components. It can be seen as part of a movement to distribute intelligence towards the detector, a natural way forward as electronic components gets cheaper and smarter. This approach of the microprocessor on the PMT has been pioneered at UW, with a more sophisticated microDAQ prototype (V2) deployed at the South Pole this winter (2017/18) for IceTop tests.
The MicroDAQ relies on commercially tested components and could potentially be sold as a commercial system with the timing fan-out and CW base. The CW base is not necessary, but can easily be modified for different PMTs if needed as it works together and is controlled by the MicroDAQ.