Is the arsenic present as arsenopyrite or as an oxide? It seems to me it would be easier to depress the arsenopyrite if that is the arsenic bearing mineral. If however if the arsenic has value, you might be able to float it with starvation amount of a short chain xanthate collector without losing much Sb.

Also:

Is the As mineralization locked or liberated at target grind size?
Is the As present within the Sb mineralization?

Need more detail to discuss this question.

My speculation is that the arsenic is present as arsenopyrite and oxide because after tests the recovery of antimony and arsenic were 75% and 36% respectively. And the composition of concentrate for Sb and As is 52% and 2.63% As indicating that the arsenic and antimony are upgraded (ore composition 10% Sb and 0.78% As). The reagent setup of this test was as follows: activator lead nitrate at 250 g/t, Frother/secondary collector-Aero float at 60 g/t, collector SIBX at 60 g/t, depressant Dextrin at 300 g/t and arsenic depressant Cyanide at 80 g/t.

There is no sufficient information about mineralogy studies however the target grind is 70% minus 75 microns, the reason for 70% is that the ore is talcose in nature and when grounded further metallurgical response will be hampered. The fact that normal antimony reagent regime upgrade arsenic it might be possible that As is present within the Sb mineralization.

How do I overcome these challenges?

I was Plant Manager at CMM for 6 years, left Jan 2010. I don't know if the mineralogy changes but it was always suphide orebody. At times we processed older stockpiles / old stopes causing the Sb to weathered which jeopardize the Sb recovery.

I don't know if you changed the process but we achieved 75% Sb recovery with in spec As in final con by the following:
Optimized process, by installing Jameson pre roughers in 2008/9. It was to extract the predominant stibnite (fast floater) and then float the bertharite / goodmandite (slow floaters) at the middling / scavs. But if you pull to hard on the pre rougher/roughers you recover As, but then we Installed 2nd re-cleaner (Jameson) to sweeten final con grade / reduce As.
Spray only clean water (NO PROCESS WATER) at the cleaner / re-cleaners just underneath the froth surface to clean the bubbles from As. This clean water spray was the biggest improvement from various other trials to control As.
Ensure to keep pulling hard to avoid the sumps (mainly no1 & 2) to overflow. Once sumps heavy overflow and feed back into the circuit you battle to maintain the correct regent control and you just poison the circuit. Once circuit was unbalanced it took as days to stabilize it.

I would suggest to find out mineralogy of As and Sb at first. At least general sulfide mineralogy characterization should be done to make any further suggestions. There are hundreds of minerals that can incorporate As besides arsenopyrite. Suggesting presence of the mineral in the sample and make a flow sheet based on a suggestion may not be very effective. Both Sb and As can be present in many minerals together and their separation may not be possible, especially if you have oxidized material. It seems tempting to look for the way in the darkness of ignorance of mineralogy of the matter they process.

It appears that this As is locked with Sb I am not quite sure how because I ran further test to float arsenic using Xanthate (collector) and copper (activator) and Aeroflot(frother), the results suggest this i.e. grade of Sb on As concentrate is 20.6% average while arsenic is 9.8% , on the Sb concentrate however the Sb grade is 45% average and As is in the region of 1.68% marketable Sb concentrate must have As content of not more than 0.50% depending on the customer. If I may ask which arsenic mineral is prone to be locked with stibnite?

You might find this discussion interesting:

http://is.gd/Zd6Cuw

If you can access a copy of Guilbert, J.M. and Park, C.F., Jr., The Geology of Ore Deposits you will find some good details:

http://is.gd/pXT149

Deposits Related to Sub-aerial Volcanism Epithermal Silver-Gold Deposits / Chinese Antimony Deposits / Bulk Low-Grade Silver-Gold Deposits / Carlin-Type Gold Deposits / Other Deposits Related to Sub-aerial Volcanism

I can help you sort out the mineralogy of the ore if you want to send me a sample. As suggests a little mineralogy can really help to sort out some of these issues. Often the sulphosalt phases can change slightly as you move through a deposit. Knowing the composition of the Feed material before you start to process the ore may be far more valuable than additional assays or flot work. You may have intergrowths of various As-Sb minerals that may be causing the As to be capture in your concentrate.

Try increasing the flotation temperature "unless otherwise", there is a relationship between pyrite recovery and Flo T.

As suggested, the mineralogy is important in selecting the appropriate flowsheet.

Just as an aside, the gold can be either

Free,
Locked in the arsenopyrite or
Present as aurostibnite or a combination or all three.

What is the reason for wanting to recover the arsenic - environmental or perceived gold content? You also mentioned talc - a naturally floating mineral. How much is present? This is a tricky separation since both stibnite and arsenopyrite require activation to float successfully. So after activation, both minerals will tend to float. Presumably the significantly greater abundance of stibnite would consume most of the activator but no doubt there would be enough remaining to float any unoxidisedarsenopyrite.

Arsenopyrite oxidises more readily than stibnite and depressing the arsenopyrite (controlled aeration or hydrogen peroxide conditioning), as suggested by Richard, from a bulk Sb-As concentrate may be an option (the tailings would be As rich). If Sb-As composites are present then regrind may be required. Interesting challenge but some mineralogical studies on the feed as well as the concentrate and tailings products would prove invaluable and guide your decision making.

For complicated problems we have an innovative technology to handle even finer particles up to 45 microns with very good 100% bubble washer. We have used for our burning problem of reducing ISM in Zinc concentrate form 8% to 3%. Further testing with RGBM of column tail gave further good results.

YOUR PROBLEM:

Initially try regrinding concentrate and analyze under microscope to find liberation studies. In INDIA we have many labs that can do this job at cheaper prices. If you are successful then use COLUMN FLOTATION 10 to 12 meter height. In Jameson cell there is crowding of froth. In INDIA NMDC lab and NML labs have column pilot model cell.

I will have to try hydrogen peroxide conditioning before flotation, and reduce lead nitrate. What I have noticed is that when too much lead nitrate is added other sulphides are activated including arsenic bearing ones. I do not have a quantity for talc as our lab does not have facilities to asses this, but we depress it with dextrin at 300g/t. gold on this ore is mostly gravity recoverable, after gravity I was left with 1.25g/t on the tail sample. The flow sheet I use here is mostly Wemco cells for roughing, rough concentrate is cleaned and re-cleaned by the Jameson cells.

If you have panning plate you can test. Or you can test on a lab table. As sp. gr difference between gold and other minerals is wide. Take advantage of this property. Now you get Arseno pyrite and Sb. Now you try HGMS at 20000 gauss where Arsenopyrite will separate as Mag con. as they are feebly magnetic also called as para magnetic. All our suggestions are only guide lines may work may not. You are the real judge of your mineral to know what treatment is required.

I was 6 years at CM as Process Manager. The only good way we separated the As was spraying clean (no process) water on the bubbles at the cleaners / re cleaners. Ensure the regrind mill grind is around 75% p80. If float sumps 1 & 2 overflow then you have problems reintroduced the pulp back into the circuit.

I agree 100% with you. We too got good results with 100% washing each bubble with clean water. Technology developed by us. Column flotation: It produces very small size bubbles which when washed gives goods and clean concentrate. Jameson cell bubbles sizes?. Good washing system to wash each bubble is developed by us and was successful in reducing silica in zinc concentrate. This system is patented. You need to purchase the system.

I grew up in Phalaborwa and was involved in a process improvement project at CM during 2001/2 and we developed various models for the process such as recovery/particle size, retention time, conditioning time, bubble surface flux, froth depth etc. The head grade then was 1.5% Sb while the recoveries were 85% at 58% Sb conc. grade. You should be pushing 90% recovery with a 10%Sb head grade. How can I help?

10% Sb content in excavating Sb-ore? It is for manual sorting! I was fighting with 1.5% in my floatation plant - if we have higher Sb content we had to diluted inflow material into flotation plant (for example with road drivels) Arsenic? After flotation purification you will gain monomimeral antimony concentrate -what is standard procedure. What is copper (Cu) in your raw Sb-ore and in your Sb-concentrate (if you produce any)? What is your metallurgy limit for Cu-content? And finally, what is "the mother" mineral for copper?